Biochemical Journal

Research article

Cholesterol-induced conformational changes in the oxytocin receptor

Sabine Muth , Anja Fries , Gerald Gimpl

Abstract

Recent studies suggest that cholesterol binding is widespread among GPCRs (G-protein-coupled receptors). In the present study, we analysed putative cholesterol-induced changes in the OTR [OT (oxytocin) receptor], a prototype of cholesterol-interacting GPCRs. For this purpose, we have created recombinant OTRs that are able to bind two small-sized fluorescence-labelled ligands simultaneously. An OTR antagonist was chosen as one of the ligands. To create a second ligand-binding site, a small-sized α-BTB (bungarotoxin binding) site was inserted at the N-terminus or within the third extracellular loop of the OTR. All receptor constructs were functionally active and bound both ligands with high affinity in the nanomolar range. Measurements of the quenching behaviour, fluorescence anisotropy and energy transfer of both receptor-bound ligands were performed to monitor receptor states at various cholesterol concentrations. The quenching studies suggested no major changes in the molecular environment of the fluorophores in response to cholesterol. The fluorescence anisotropy data indicated that cholesterol affects the dynamics or orientation of the antagonist. The energy transfer efficiency between both ligands clearly increased with increasing cholesterol. Overall, cholesterol induced both a changed orientation and a decreased distance of the receptor-bound ligands, suggesting a more compact receptor state in association with cholesterol.

  • α-bungarotoxin
  • bungarotoxin-binding site
  • cholesterol
  • human embryonic kidney (HEK) cell
  • oxytocin receptor

INTRODUCTION

GPCRs (G-protein-coupled receptors) comprise the largest and most diverse receptor superfamily involved in signal transduction across membranes [1]. Owing to their integration into the membrane bilayer with seven transmembrane helices, GPCRs are in close contact with membrane lipids. Among those lipids, cholesterol is one of the most abundant in the plasma membrane of eukaryotic cells. Cholesterol is a multifunctional lipid. It regulates the fluidity and the phase behaviour of the membrane bilayer, serves as a precursor of steroid hormones and some vitamins, is essentially involved in the formation of lateral membrane microdomains (e.g. lipid rafts and caveolae) [2,3], and plays a crucial role in the function and organization of membrane proteins, e.g. receptors and ion channels [46].

Direct GPCR–cholesterol interactions have been suggested for the OT (oxytocin), galanin and serotonin1A receptors [79]. To fulfil their function in signal transduction, GPCRs require conformational flexibility. There is some evidence that cholesterol is able to stabilize the conformation of receptors, e.g. the β2-adrenergic receptor [10,11] and the OTR (OT receptor) [12]. Structural data have supported the stabilizing role of cholesterol for the conformation of GPCRs [10,13]. Based on homology, a so-called ‘cholesterol consensus motif’ has been defined that is found in more than 40 class A GPCRs including the cholesterol-dependent OTR [10]. Thus cholesterol binding is rather widespread among the GPCRs, and the OTR can be regarded as a prototype of these cholesterol-interacting receptors.

In the present study, we asked whether cholesterol-dependent changes of receptor conformations are detectable. For this purpose, the fluorescence properties of specifically bound ligands were exploited to report on the molecular behaviour and environment of the OTR. Specifically, the goal was to find an appropriate combination of two ligands that are able to bind simultaneously to one receptor molecule. Under certain conditions, changes of the relative distance or orientation of the bound ligands [e.g. measurable by FRET (fluorescence resonance energy transfer)] will reflect structural changes of the receptor. The two ligands required for this study should be as small as possible and possess non-overlapping binding sites at the receptor molecule. An OTR antagonist was chosen as one of the receptor ligands. Other known high-affinity ligands of the OTR are expected to overlap (at least partially) with the antagonist receptor-binding site and are therefore excluded as a potential further receptor binding partner. In order to create a second ligand-specific binding site at the OTR, we introduced a small-sized epitope comprising 13 amino acids (WRYYESSLEPYPD) at different positions of the receptor. Henceforth referred to as α-BTB [αBT (α-bungarotoxin) binding] peptide, this oligopeptide is able to bind to the snake-venom α-BT with high affinity in the nanomolar range [14,15]. It has been demonstrated that receptors tagged with this BTB site can be visualized by addition of fluorescence-labelled α-BT [1618]. The α-BT–BTB reporter system has several advantages. First, α-BT has a molecular mass of only ~8 kDa, and thus it is considerably smaller than other reporter proteins, such as antibodies or any of the fluorescent proteins. Secondly, α-BT is membrane impermeant and, when applied to living cells, allows a selective labelling of functional BTB-tagged receptors at the plasma membrane excluding signals from the cell interior. Moreover, receptor tracking and pulse–chase studies are possible.

We provide evidence that the usage of this reporter system in combination with an appropriate high-affinity antagonist allows monitoring of conformational changes of the OTR.

EXPERIMENTAL

Reagents

Reagents were provided from the following supplier: Alexa Fluor® 568 and Alexa Fluor® 647 carboxylic acid, succinimidyl ester, α-BT-Alexa Fluor® 647, α-BT–tetramethylrhodamine, Lipofectamine™ 2000, Invitrogen; primers and sequencing oligonucleotides, Genterprise; Mowiol, Calbiochem; restriction enzymes, and Phusion DNA polymerase, Fermentas and NEB; Petri dishes, Sarstedt; penicillin, streptomycin, BSA, Roth; fetal bovine serum (Gold), L-glutamine, DMEM (Dulbecco's modified Eagle medium), minimal essential medium, trypsin and hygromycin B, PAA; plasmid purification kit NucleoSpin, Macherey & Nagel; gel extraction and PCR purification kits, Qiagen; F792 antagonist [designated as OTAN (OT antagonist)], Ferring Pharmaceutics; [Tyrosyl-2,6-3H]OT (NET-858, 48.5 and 32 Ci/mmol), NEN Du Pont de Nemours. DChol (6-dansyl-cholestanol) was synthesized by our group [19]. All other substances, unless stated otherwise, were purchased from Sigma–Aldrich.

Synthesis of OTAN-Alexa Fluor® 568 (OTAN-A568) and OTAN-Alexa Fluor® 647 (OTAN-A647)

Alexa Fluor® 568 and Alexa Fluor® 647 carboxylic acid, succinimidyl ester (A568 and A647 respectively) (each 1 mg; ~1.3 μmol) were first dissolved in 60 μl of dimethylformamide (anhydrous). The OTAN (Ferring compound 792, designated here as OTAN) (0.8 mg; 0.9 μmol) was dissolved in 80 μl of dimethylformamide (anhydrous) and was mixed with solutions A568 and A647 respectively. Triethylamine (1 μl) and 300 μl of methanol were added to this mixture. The pH value of the mixture was monitored at >8. The reaction was allowed to proceed for approx. 20 h under light protection. The process of the reaction was controlled by HPLC analysis of small aliquots at different time points. The reaction was stopped by the addition of 4 ml of solution A [0.1% TFA (trifluoroacetic acid) in H2O]. An aliquot of this reaction mixture was injected on to an HPLC reversed-phase column (Vydac C8: 5 μm). Elution was performed with the following gradient of solution A (H2O, 0.1% TFA) and solution B (90% acetonitrile, 10% H2O, 0.1% TFA) at a flow rate of 1 ml/min: from 10% B to 30% B for 10 min and from 30% B to 90% B for 40 min. Peak fractions containing the reaction product were collected and analysed. The calculated molecular masses of both OTAN-A568 and OTAN-A647 were confirmed by MS (Ciphergen Biosystems).

OTR constructs

BTB tags were introduced at different positions of the OTR. The plasmid pfmOTRh encoding the human OTR including small epitope tags for FLAG (f), Myc (m) at the N-terminal and poly-His (h) at the C-terminal part of the receptor was used as a starting point for the cloning of the different constructs described below. The cloning of the similar plasmid pmOTRf has been described previously [7]. For the first construct, the plasmid pfmOTRh was cut with HindIII and BamHI. To generate the double-strand BTB-adapter at the N-terminus of the OTR, the following primers were annealed: 5′-AGCTTGGTACCATGGGATGGAGATACTACGAGAGCTCCCTGGAGCCCTACCCTGACTTG-3′ (forward) and 5′-GATCCAAGTCAGGGTAGGGCTCCAGGGAGCTCTCGTAGTATCTCCATCCCATGGTACCA-3′ (reverse). This adapter consists of a BTB site (peptide WRYYESSLEPYPD) (Figure 1A) including a START codon (bold), a KpnI site (underlined) necessary for the next cloning reaction (see below) and cohesive ends compatible for ligation with BamHI and HindIII. The primers were annealed in a thermocycler and were ligated to the vector pfmOTRh digested with BamHI and HindIII. The resulting expression plasmid was designated pOTR-BTBe1. In a second construct, the EGFP (enhanced green fluorescent protein) tag was introduced at the C-terminus of the OTR construct OTR-BTBe1. For this purpose, pOTR-BTBe1 (~6.6 kb) was digested with KpnI that cuts the plasmid at two sites. The first site is within the BTB adapter before the start codon, the second site is shortly before the stop codon of the OTR. The insert (~1.2 kb) encoding the OTR-BTB was purified. The vector pEGFP-N3 (Clontech) was cut with KpnI, dephosphorylated to prevent self-ligation and was ligated with the insert OTR-BTB. The plasmid that contained OTRGFP-BTB in the correct orientation was designated pOTRGFP-BTBe1. In the third construct, the BTB tag was inserted into the third extracellular loop (e4, see Figure 1B). PCRs were employed to introduce two restriction sites (EcoRI and EcoRV) into the cDNA sequence encoding this loop. The PCRs were performed with Phusion Polymerase using the following primers: P1, 5′-AAGCGTGTACGGTGGGAGGTC-3′; P2, 5′-GGGATATCCGCCCCCGGAATTCCCGGCATCCCAGACGCTCCACATCTG-3′; P3, 5′-GGGAATTCCGGGGGCGGATATCCCAACGCGCCCAAGGAAGCCTCGGCC-3′; and P4, 5′-CAGAATAGAATGACACCTACTC-3′. The primers P2 and P3 possess at their 5′-end cleavage sites for EcoRI and EcoRV, and at their 3′-end nucleotides complementary to the cDNA encoding the OTR. Additionally, P2 and P3 were complementary to each other at their 5′ ends. Using the plasmid pfmOTRh as template, the first two PCRs were performed with the primers P1 and P2 (PCR1) and P3 and P4 (PCR2), yielding products of ~1.2 and ~0.6 kb respectively. The PCR products were purified and used in stoichiometric amounts together with the primers P1 and P4 for a third PCR. In the first round of this reaction, the complementary products of PCR1 and PCR2 hybridize and undergo a fill-in reaction at their 3′ ends. Thereafter, the PCR starts and amplifies a ~1.7 kb product that encodes the OTR with the inserted EcoRI and EcoRV sites. This linear 1.7 kb DNA sequence was purified, and digested with NheI and KpnI. The ~0.6 kb NheI–KpnI fragment was isolated and cloned into the vector pfmOTRf cut by NheI and KpnI. The resulting plasmid now contained the coding sequence for the OTR including the EcoRI and EcoRV sites at positions appropriate for the insertion of the BTB adapter oligonucleotides. The plasmid was cleaved with EcoRI and EcoRV and the BTB adapter was inserted. The following primers were hybridized to obtain the BTB adapter with compatible ends for ligation with EcoRI and EcoRV: 5′-AATTCCTGGAGATACTACGAGAGCTCCCTCGAGCCCTACCCTGACG-3′ (forward) and 5′-CGTCAGGGTAGGGCTCGAGGGAGCTCTTCGTAGTATCTCCAGG-3′ (reverse). Following successful ligation the resulting clone was designated pfmOTRBTBe4. The identities of the various constructs were confirmed by DNA sequencing.

Figure 1 OTR constructs used in the present study

(A) The BTB peptide (dotted circle) comprises only 13 amino acids (sequence as indicated), forms a hairpin-like structure, and binds with high affinity to α-BT [structure from NCBI (National Center for Biotechnology Information], MMDB ID:17181, PDB (Protein Data Bank) ID: 1HAJ) [15]. (B) The human OTR and its putative binding domains for OT, peptide antagonists (dotted squares), G-proteins (dashed squares). Black circles, residues conservative for the whole GPCR family; grey circles, residues conservative within the OT/vasopressin receptor family; residues marked with ‘Y’, putative N-glycosylation sites; e1–e4, extracellular domains; i1–i3, intracellular loops; dashed line, transmembrane helices 1 and 2 and upper parts of transmembrane helix 7 are previously defined binding sites for OTR antagonists.

Cell culture, transfection and expression

All cells were cultured in monolayers in complete DMEM supplemented with 100 μg/ml of penicillin, 100 μg/ml of streptomycin and 10% (v/v) fetal bovine serum. HEK-293 cells (human embryonic kidney-293 cells) were transfected with the different constructs using Lipofectamine™ 2000. To obtain stably expressing cells, the transfected cells were incubated for 24 h at 37 °C, split and selected with 1 mg/ml G418. Several neomycin-resistant cell clones were propagated and screened for [3H]OT binding activity.

Membrane preparation

Cells were centrifuged at 100 g for 10 min. The cell pellet was washed twice with PBS and resuspended in homogenization buffer (5 ml per 100 ml of cells) containing 20 mM Hepes, pH 7.4, 5 mM EDTA and a protease inhibitor cocktail composed of bacitracin, soyabean trypsin inhibitor, leupeptin and PMSF. The suspension was homogenized using a Polytron PT-10. This was followed by homogenization using a Dounce glass homogenizer (10 strokes). Subsequently, the homogenate was centrifuged at 40000 g for 30 min, and the pellet was washed once with homogenization buffer. For some experiments, these ‘crude membranes’ were further purified by sucrose-density-gradient centrifugation as described previously [20]. Briefly, the membranes (5–10 mg of protein) was resuspended in homogenization buffer and was layered on top of a stepwise gradient consisting of 3.5 ml of 60% (w/v) sucrose and 4 ml of 35% (w/v) sucrose prepared in homogenization buffer. After centrifugation at 32000 rev./min for 90 min (SW-41 rotor) the membranes at the upper 0–35% sucrose interface were collected and diluted with binding buffer containing 20 mM Hepes, pH 7.4, 5 mM MgCl2, trypsin inhibitor and bacitracin. The membranes were centrifuged for 1 h at 50000 rev./min (Ti60 rotor), were resuspended in binding buffer and were shock-frozen in liquid nitrogen and stored at −70 °C.

The Bradford method was used to quantify the protein content of the samples, with BSA as a standard.

Receptor-binding assays

To measure the ligand-binding activity, membranes (100 μg) were incubated with [3H]OT (5–10 nM) in a total volume of 100 μl of binding buffer for 30 min at 30 °C. Association experiments were performed by incubating the membranes with 3 nM [3H]OT at 30 °C for various times until equilibrium binding was achieved (30 min). Then, the dissociation of the ligand–receptor complex was initiated by the addition of a 1000-fold excess of unlabelled OT. The binding reaction was stopped by the addition of ice-cold binding buffer (10 mM Tris, pH 7.4 and 5 mM MgCl2) and bound ligand was separated from free ligand by rapid filtration over Whatman GF/F filters presoaked with 0.3% poly(ethylenimine). Filters were incubated with 10 ml of scintillation cocktail (Filter-Count, PerkinElmer) and were counted. Non-specific binding was determined in the presence of a 700–1000-fold excess of unlabelled OT.

In saturation experiments with fluorescent ligands (α-BT-A568 or OTAN-A568), membranes (50–100 μg) were incubated for 30 min at 30 °C with increasing ligand concentrations in binding buffer (100 μl). Unbound ligand was removed by centrifugation (5 min, 21000 g, 4 °C), followed by two washing steps with 1 ml each of ice-cold binding buffer. Finally, the membranes were resuspended in 1 ml of methanol, in which the fluorescence was measured spectrofluorimetrically (Quantamaster, PTI) using λex and λem of 578 and 600 nm respectively.

All assays were carried out in triplicate. Data analysis of the binding studies was performed using non-linear curve-fitting algorithms (GraphPad). Graphical output was performed by Sigmaplot version 8.0 (Systat Software).

Radioligand-binding assays on whole cells stably transfected with OTRs were carried out as described above. Data were analysed using LIGAND and IC50 values were converted into pKi values (−log of the inhibition constant Ki) using the Cheng–Prusoff equation: Ki=IC50/(1+([L]/KD), where IC50 is the concentration of competing ligand that displaces 50% of the specific binding of the radioligand, [L] is the concentration of the radioligand and KD is the dissociation constant of the radioligand–receptor interaction.

Alteration of the cholesterol content in the membranes

Extraction of cholesterol from the plasma membranes was carried out as described previously [7]. Briefly, membranes (5 mg/ml) were incubated with MβCD (methyl-β-cyclodextrin; final concentration 30 mM) for various times at 30 °C. Thereafter, the membranes were washed twice and were resuspended in assay buffer (20 mM Hepes, pH 7.4 and 5 mM MgCl2). According to this protocol 70–85% of the initial cholesterol (35 μg/mg of protein) could be removed from the membranes. To enrich the plasma membranes with cholesterol, we employed cholesterol–MβCD inclusion complexes [7]. Cholesterol (final concentration 3 mM) was added to an aqueous solution of MβCD (40 mg/ml). The mixture was overlaid with nitrogen, and was continuously vortex mixed under light protection for 24 h at 30 °C in a thermomixer. The solution was filtered prior to use. The cholesterol-depleted membranes were incubated with cholesterol–MβCD (final concentration of 0.3 mM cholesterol in complex) for various times to restore their cholesterol content.

Lipid extraction and determination of cholesterol

Lipids, including cholesterol, were extracted using the method of Bligh and Dyer, slightly modified as described previously [21]. Briefly, 200 μl of membranes (10 μg–1 mg of protein) and 0.75 ml of chloroform/methanol (1: 2, v/v) were vigorously mixed for 10 min at 30 °C in a thermomixer and centrifuged for 10 min at 21000 g. The supernatant was mixed with 250 μl of chloroform and 250 μl of water and was centrifuged for 30 min at 21000 g. The bottom lipid phase was evaporated under a N2 atmosphere and was dissolved in propan-2-ol.

Cholesterol was assayed spectrophotometrically using a diagnostic kit (R-Biopharm AG) performed in a microscale dimension. The protein concentration was determined by the Bradford assay (Roti-Quant and Roti-Nanoquant, Roth) using BSA as a standard.

Calcium measurements

For measurements of the cytosolic Ca2+ concentrations, cells grown on Petri dishes to approx. 80% of confluency were loaded with fura-2 AM (acetoxymethyl ester; 1.5 μM) for 30 min at 37 °C. Thereafter, the cells were scraped from the Petri dishes and were resuspended in HBS buffer (20 mM Hepes, pH 7.4, 150 mM NaCl, 5 mM KCl, 1 mM MgCl2, 1 mM CaCl2 and 1 mM glucose). Aliquots of the suspension (6×105 cells) were added to pre-warmed HBS and transferred into a cuvette that was placed into a thermostated (37 °C) holder. The cell suspension was continuously mixed using a magnetic stirrer. OT was applied to the cells and the changes of the [Ca2+]i were monitored spectrofluorimetrically (Quantamaster, PTI). The λem was set at 510 nm and dual-λex were performed at 340 and 380 nm respectively (approximately 2 ratios/s). The [Ca2+]i was calculated by using the ratio 340/380 nm. The Rmax value was obtained after Triton X-100 was added to the samples. Thereafter, minimum fluorescence (Rmin) was obtained by chelating calcium with 5 mM EGTA and increasing the pH above 8.3. For the calculations of [Ca2+]i, a KD of 225 nM between Ca2+ and fura-2 was assumed.

Fluorescence quenching experiments

Collisional quenching of free and bound probes, OTAN-A568 and α-BT-A647, was analysed fluorimetrically using the hydrophilic quenchers, KI (potassium iodide) and CoCl2 (cobalt chloride) as well as the lipid-soluble quencher TEMPO (2,2,6,6-tetramethylpiperidine-N-oxyl). Measurements were performed in assay buffer with 50 nM of ligand. To monitor the quenching of membrane-bound probes, the ligands were added to 50 μg of plasma membranes prepared from the indicated cells and were bound to equilibrium (30 min, 30 °C). Excess ligand was washed off by centrifugation. Fluorescence was measured in a thermostated (20 °C) cuvette after sequential additions of the quencher (from stock solutions of 1 M for CoCl2 and KI) under continuous stirring. For KI, the stock was prepared in 10 mM NaS2O3 to prevent air-induced oxidation. The effects of dilution and ionic strength were calibrated by sequential additions of 1 M KCl to control samples. The quencher TEMPO was added from a 500 mM stock solution in ethanol. The hydrophobic quenching of TEMPO was verified using membranes in which the fluorescent cholesterol analogue DChol has been incorporated. For this purpose, 50 μg of plasma membranes were incubated in a volume of 1 ml with 3 μl of 3 mM DChol–MβCD at 4 °C for 20 min. Thereafter, membranes were washed twice with assay buffer and were used for experiments. DChol–MβCD was prepared as described in [19].

λex and λem were set at the corresponding maxima, i.e. 578 and 600 nm for OTAN-A568, 647 and 670 nm for α-BT-A647, and 360 and 510 nm for DChol. The results were background corrected and were plotted according to the Stern–Volmer equation, F0/F=1+KSV [Q], where F0/F is the ratio of fluorescence intensity in the absence and presence of the quencher (Q). The Stern–Volmer constants KSV were determined from the slope of F0/F as a function of the quencher concentration.

Fluorescence anisotropy and FRET measurements

Plasma membranes from OTR-expressing cells (50 μg) with normal or altered cholesterol content were incubated with OTAN-A568 or α-BT-A647 (50 nM each) for 30 min at 30 °C. After washing-off unbound ligands by centrifugation (for 20 min at 32000 g at 4 °C) the membranes were resuspended in assay buffer, and the anisotropy of the corresponding fluorophore was measured. For that purpose, the membranes were transferred to a thermostated cuvette in a spectrofluorimeter (Quantamaster, PTI) and were continuously stirred at 20 °C. Samples with OTAN-A568 or α-BT-A647 were λex at 578 or 650 nm respectively. The λem was measured at 600 nm for OTAN-A568 and 670 nm for α-BT-A647. Slit-widths of 5 nm were used at the excitation and emission site. The steady-state fluorescence, r, was determined according to r=(IVVIVHG)/(IVV+IVHG), where IVV and IVH are the fluorescence intensities observed with the excitation polarizer in the vertical position, and the analysing emission polarizer in both the vertical (IVV) and the horizontal (IVH) configurations. The factor G was used to correct for the unequal transmission of differently polarized light.

For most FRET measurements, the receptor-bound ligands OTAN-A568 and α-BT-A647 were used as donor and acceptor fluorophores respectively. For a series of control experiments, the receptor-bound antagonists OTAN-A568 and OTAN-A647 were used as donor–acceptor pairs. The ligands were added either separately or together to plasma membranes prepared from OTR-BTBe1 and OTR-BTBe4 cells. After washing-off unbound ligands by centrifugation (for 20 min at 32000 g at 4 °C), the membranes were placed into a thermostated (20 °C) cuvette under continuous stirring. The samples were λex at 578 nm and the λem fluorescence was scanned from 585 to 700 nm. Energy transfer, E, was calculated from the measured donor fluorescence intensity at 578 nm in the presence and absence of the acceptor according to E=1−(FDA/FD), where FD and FDA are the donor intensities of samples containing only donor-labelled ligand (OTAN-A568) and samples with both donor- and acceptor-labelled ligands (OTAN-A568/α-BT-A647 or OTAN-A568/OTAN-A647) respectively.

Fluorescence microscopy

For live-cell microscopy, cells were grown in small Petri dishes into which coverslips (diameter of 18 mm) were placed. Prior to addition of the cells, coverslips and dishes were coated with poly-D-lysine. The coverslips were placed into a homemade chamber system where temperature and humidity are controlled. Unless stated otherwise, the cells were perfused with imaging buffer (10 mM Hepes, pH 7.4, 145 mM NaCl, 4.5 mM KCl, 1.2 mM MgCl2, 1.2 mM CaCl2 and 10 mM glucose) containing the indicated concentrations of ligands/reagents. Microscopical images were acquired either on an epifluorescence system (Olympus Cell-R, Hamamatsu ORCA-ER) or on a confocal laser-scanning microscope (Leica LSC-SP1, equipped with argon 488 nm, diode-pumped solid-state 561 nm and HeNe 633 nm laser) using the objectives ×63, 1.3 NA (numerical aperture) oil (Leica LSC-SP1) and ×100, 1.4 NA oil (Olympus system) respectively. Appropriate filter settings were used. If required, cells were fixed with 3.7% (w/v) paraformaldehyde for 20 min at 23 °C. Then, the cells were washed twice with PBS and were embedded with mowiol.

RESULTS

Expression and activity of the recombinant OTRs

We characterized the ligand-binding behaviour and the physiological responsiveness of the various OTR constructs expressed in HEK-293 cells. Three receptor constructs have been created. In the first construct, the BTB tag (Figure 1A) has been incorporated at the N-terminus (e1) (see Figure 1B). In the second construct, EGFP has been fused at the C-terminus of the OTR that additionally bears the BTB tag at the N-terminus. In the third construct, BTB was placed into the middle of the third extracellular loop e4 (Figure 1B). The corresponding constructs were designated OTR-BTBe1, OTRGFP-BTBe1 and OTR-BTBe4 respectively. Cells expressing EGFP at the C-terminus of the OTR (OTRGFP-BTBe1) were created to compare the trafficking of the OTR using the BT–BTB system against the GFP reporter system. HEK-293 cells expressing the same OTR–EGFP fusion protein as found in the construct OTRGFP-BTBe1, but without the BTB tag, have been characterized [21]. The [3H]OT-binding affinity of the later receptor construct was in the nanomolar range similar to that found for the untagged OTR [21]. The data of ligand-binding parameters are summarized in Table 1. In saturation experiments with [3H]OT, high-affinity radioligand binding was observed for all OTR constructs. OTRs with both EGFP and BTB tags exhibit a slightly reduced affinity in [3H]OT binding. The expression level did not differ among the membranes from the three receptor constructs OTR-BTBe1, OTRGFP-BTBe1 and OTR-BTBe4. Moreover, all three OTR constructs were functionally active. In response to OT, they initiated transient increases of the intracellular calcium concentration. Supplementary Figure S1 (available at http://www.BiochemJ.org/bj/437/bj4370541add.htm) shows the percentage amplitudes of the corresponding calcium signals as dependent on the OT concentration. The EC50 values of these responses correlate well with the corresponding KD data of the three constructs (Table 1).

View this table:
Table 1 Ligand binding and activity of OTR constructs

Binding data were obtained from membranes prepared from HEK-293 cells that have been stably transfected with the indicated constructs. All experiments were performed in triplicate and results shown are means±S.E.M.

Characteristics of the fluorescent ligands used in the present study

Next, we analysed the properties of the fluorescent ligands used in the present study. For this purpose, the following ligands were analysed with respect to their potency as competitors of the [3H]OT binding to membranes from HEK-293 cells expressing the OTR: the agonist OT, the partial agonist AVP (Arg-vasopressin), the antagonist OTAN, and the fluorescent antagonists OTAN-A568 and OTAN-A647. All ligands were able to displace the radioligand from the OTR in a dose-dependent fashion (Figure 2A). The IC50 values are given in the legend to Figure 2. According to the equation of Cheng–Prusoff, the following corresponding Ki values were calculated: 4.5 nM for OT, 36.8 nM for AVP, 2.0 nM for OTAN, 22.3 nM for OTAN-A568 and 21.5 nM for OTAN-A647. Thus, the fluorescent ligands OTAN-A658 and OTAN-A647 were shown to displace [3H]OT binding with an approximately 10-fold less potency as compared with the unlabelled antagonist (for the structure, see Figure 2B) [22]. Thus the attachment of the Alexa Fluor® fluorophores maintained a reasonably high binding affinity to the OTR. Further assays (e.g. the blocking of the OT-induced calcium response) demonstrated that the antagonist retained its antagonistic behaviour after its labelling with fluorophore (results not shown).

Figure 2 OTR binding of the fluorescent antagonists OTAN-A568 and OTAN-A647 in comparison with related unlabelled receptor ligands

(A) Displacement of [3H]OT binding to membranes from HEK-293 OTR cells by AVP (filled circles), OTAN-A568 (filled triangles), OTAN-A647 (filled squares), OT (open circles) and OTAN (open squares). Radioligand displacement was assessed by incubating membranes from HEK-293 OTR cells (100 μg) and 10 nM [3H]OT with the various peptides at the indicated concentrations. Values of specific binding of [3H]OT were expressed as a percentage of the maximum radioligand binding which was obtained in the absence of a competitor. Data of a representative experiment are shown. Curve fitting was performed using the logistic function y=100/[1+(x/IC50)b]. The fitted parameters were as follows: for OT, IC50=31 nM, b=1.62; for AVP, IC50=254 nM, b=1.37; for OTAN, IC50=14 nM, b=1.57; for OTAN-A568, IC50=154 nM, b=1.79; for OTAN-A647, IC50=150 nM, b=1.38. (B) Structure of OTAN conjugated to the fluorophore group R (=A568 or A647). (C) Normalized excitation and emission fluorescence spectra of OTAN-A568 (solid line) and OTAN-A647 (broken line) (each 50 nM) in 20 mM Hepes, pH 7.4.

The spectral properties of both OTAN-A568 and OTAN-A647 in different solvents were analysed (Table 2). Only slight changes of the corresponding excitation and emission maxima were measured in polar compared with non-polar solvents. The relative emission intensities of OTAN-A568 varied from 53 to 162%, with the highest values detected in methanol. The environmental sensitivity of the fluorescence of OTAN-A647 was slightly lower than that of OTAN-A568. Overall, the fluorescence was relatively robust against changes of pH and high salt. Normalized excitation and emission spectra of both OTAN-A568 and OTAN-A647 dissolved in buffer are shown in Figure 2(C).

View this table:
Table 2 Spectral properties of the fluorescent antagonist OTAN-A568 and OTAN-A647

Measurements were performed with 100 nM each of OTAN-A568 or OTAN-A647. Results are means±S.D. (n=3).

Binding of α-BT-A568 to BTB-tagged OTRs

The binding behaviour of α-BT-A568 to the different BTB-containing OTR constructs was explored. The kinetics of α-BT-A568 binding to cells expressing the receptor OTR-BTBe1 is illustrated in Figure 3(A). For the other constructs OTR-BTBe4 and OTRGFP-BTBe1, the association and dissociation curves of α-BT-A568 occurred with similar kinetics (results not shown). At 16 °C, equilibrium of α-BT-A568 binding was achieved at ~30 min (Figure 3A, filled circles). Receptor-bound α-BT-A568 remained stable for at least 90 min (Figure 3A, open circles). Dissociation of cell-surface bound α-BT-A568 was initiated when excess unlabelled α-BT was administered to these cells. Within ~20 min approximately 70% of the α-BT-A568 dissociated from the OTR (Figure 3A, open squares). Best fitting of the kinetic binding data was obtained by one-phase exponential equations (see the legend to Figure 3).

Figure 3 Binding behaviour of α-BT-A568 to BTB-tagged OTRs

(A) Kinetics of α-BT-A568 binding to HEK-293 cells transfected with the OTR construct OTR-BTBe1. To determine association data, cells (~105) were incubated for 0–40 min with 2 μg/ml α-BT-A568 in Hepes-buffered culture medium at 16 °C. Excess ligand was washed off by centrifugation and cell-bound fluorescence was measured by spectrofluorimetry (filled circles). After equilibrium binding had been achieved, the cells were further incubated in the absence of fluorescent α-BT (open circles). Dissociation of cell-surface-bound α-BT-A568 was negligible for at least 90 min. Rapid dissociation of α-BT-A568 was initiated by administration of excess unlabelled α-BT (open squares). The association data were fitted according to the equation y=a(1−ebx). For OTR-BTBe1 cells, the calculated parameters for the association data (filled circles) were a=102, b=0.184. Fitting of the dissociation data was performed according to the equation y=y0+aebx and yielded the parameters y0=22.7, a=2654.8, b=0.0886. (B) Saturation binding of membranes from HEK-293 cells expressing OTR-BTBe1, OTR-BTBe4 or OTRGFPe1 receptors. The membranes (100 μg) were incubated to equilibrium (40 min, 22 °C) with increasing concentrations of α-BT-A568. The data were fitted according to the equation y=Bmaxx/(KD+x). (C) OT-induced internalization of α-BT-A568 bound to the cell surface of HEK-293 cells expressing OTR-BTBe1 receptors. Cells cultivated on a coverslip in small Petri dishes were transferred to an imaging chamber on a confocal microscope (constant temperature 30 °C) and were incubated for 30 min with α-BT-A568 (2 μg/ml). After a short washing step, OT (1 μM) was added. In time-lapse experiments, the sequestration of receptor-bound α-BT-A568 from the cell surface was recorded for the indicated ROIs, and was analysed by Leica confocal software. The kinetic data were plotted and fitted according to the equation y=a+bekx, with τ=ln2/k. The average value of τ calculated for three typical ROIs was 6.2 min. Inset, cells with the indicated ROIs (scale bar, 10 μm).

Plasma membranes prepared from HEK-293 cells expressing the BTB-tagged OTRs were used for saturation experiments with α-BT-A568 (Figure 3B). All binding data were best fitted by one-site models. Accordingly, the following Bmax and KD values were calculated: 0.46 pmol/mg and 10.9 nM for OTR-BTBe1, 0.35 pmol/mg and 8.0 nM for OTR-BTBe4, 0.55 pmol/mg and 9.9 nM for OTRGFP-BTBe1. Thus all receptor constructs showed similar high affinity for the ligand α-BT-A568 (KD~10 nM).

Finally, we analysed whether receptor-bound α-BT-A568 becomes internalized in response to OT. For this purpose, living HEK-293 cells expressing OTR-BTBe1 were imaged at constant temperature (30 °C) on a confocal microscope. After an incubation period of 30 min with α-BT-A568 (2 μg/ml) and a short washing step, OT (1 μM) was added. The amount of receptor-bound α-BT-A568 at the cell surface decreased within minutes. To quantify this process, receptor sequestration from the cell surface was recorded for several ROIs (regions of interest). The kinetics of agonist-induced receptor internalization is illustrated in Figure 3(C) for three different ROIs from the indicated cells (Figure 3C, inset). Each data point represents the α-BT-A568 fluorescence of the indicated ROI in a time-lapse experiment. The data were fitted according to exponential equations and yielded an average half-time of ~6.2 min for receptor sequestration.

Binding of fluorescent ligands to BTB-tagged OTRs

We have created two fluorescent ligands of high specificity for BTB-tagged OTRs: a fluorescent antagonist such as OTAN-A568 (or OTAN-A647) and a fluorescence-labelled α-BT as α-BT-A568. Their binding to HEK-293 cells expressing the OTR-BTBe1 was further studied by imaging experiments. When OTAN-A568 or OTAN-A647 (each 100 nM) was applied to these cells, a distinct fluorescence labelling of the plasma membrane appeared (Figure 4A, left-hand panel for OTAN-A568 and Figure 4D, right-hand panel for OTAN-A647). In control experiments with untransfected HEK-293 cells, the plasma membrane remains unstained (Figure 4A, middle panel, compare with the corresponding transmission image in the right-hand panel).

Figure 4 Binding of fluorescent ligands to HEK-293 cells expressing BTB-tagged OTRs

(A) The antagonist OTAN-A568 (100 nM) was applied to HEK-293 cells expressing OTR-BTBe1 (left-hand panel). OTAN-A568 did not label untransfected HEK-293 cells (middle panel). The right-hand panel shows a transmission image of the cells present in the middle panel. (B) Alexa Fluor® 568 labelled α-BT (α-BT-A568) was added for 10 min at 30 °C to HEK-293 cells expressing OTR-BTBe1. Then, the cells were stimulated or not (left-hand panel) with 1 μM OT for 3 min (middle panel) or for 20 min (right-hand panel) at 30 °C. (C) To induce receptor internalization, cells expressing OTRGFP-BTBe1 were treated with OT for 15 min at 30 °C. During this early time of agonist-induced endocytosis, internalized receptors labelled with EGFP (green, left panel) and α-BT-A568 (red, middle panel) co-localize (merge, right-hand panel). (D) After prolonged incubation with OT (60 min), co-localization of EGFP-labelled receptors (green) and α-BT-A568 (red)-containing structures decreases. Middle panel, magnified inset of the marked region of the left-hand panel. Right-hand panel, labelling of HEK-293 cells expressing OTR-BTBe1 with antagonist OTAN-A647 (scale bar in magnified inset, 2 μm; other scale bars, 10 μm).

In Figure 3(C), we have already shown that receptor-bound α-BT-A568 undergoes OT-induced sequestration from the plasma membrane within minutes. In cells expressing OTR-BTBe1, stimulation by OT leads to a substantial decrease of surface-bound α-BT-A568 (Figure 4B, left-hand panel) and translocation of fluorescence from the plasma membrane into vesicles (Figure 4B, middle and right-hand panels). Does the receptor-bound ligand α-BT-A568 dissociate from the receptor somewhere after the internalization process? To address this issue, we used HEK-293 cells expressing the OTR with two different tags, the BTB-tag and a covalently attached EGFP protein at the receptors' C-terminus, i.e. the construct OTRGFP-BTBe1. The timing of receptor-associated α-BT-A568 and EGFP fluorescence in response to OT were compared. Up to a period of ~15 min of agonist-induced endocytosis, internalized receptors labelled with EGFP (Figure 4C, left-hand panel) and α-BT-A568 (Figure 4C, middle panel) revealed nearly perfect co-localization (Figure 4C, merge). However, at prolonged incubation times (>30 min) with OT, co-localization of EGFP-labelled receptors and α-BT-A568-containing structures decreases (Figure 4D). The right-hand panel of Figure 4(D) shows the magnified inset of a region where separation of both fluorescence signals is particularly pronounced. Thus fluorescent α-BT remains associated with the receptor during the early phases of endocytosis. Later on (after 30–60 min of agonist stimulation) both molecules separate from each other and translocate into different populations of endosomes. The further fate of α-BT-A568 (e.g. its possible degradation) has not been explored in the present study.

Cholesterol dependence of ligand binding to BTB-tagged OTRs

Agonist binding of the OTR strongly depends on the presence of cholesterol. We compared the binding of the ligands OT (i.e. radiolabelled agonist [3H]OT), OTAN-A568 (fluorescent antagonist) and α-BT-A568 (fluorescent α-BT) to the OTR-BTBe1 with respect to their cholesterol dependence. For that purpose, plasma membranes from HEK-293 cells expressing OTR-BTBe1 were first incubated to equilibrium (30 min, 30 °C) with [3H]OT (100 nM), OTAN-A568 (1 μM) or α-BT-A568 (2 μg/ml). Then, these membranes were (or were not) further subjected to cholesterol depletion by treatment with MβCD (10 mM). This treatment decreased the cholesterol content from 201±5 μg/mg of protein to 62±4 μg/mg of protein (i.e. ~70% reduction). In cholesterol-depleted membranes, receptor binding of [3H]OT and OTAN-A568 was dramatically decreased (Figure 5A). When cholesterol was restored to normal (i.e. ~200 μg/mg of protein) by incubating the cholesterol-depleted membranes in the presence of the cholesterol donor DChol–MβCD, binding of both [3H]OT and OTAN-A568 was fully recovered. In contrast, the binding of α-BT-A568 to the receptor remained unaffected, irrespective of the membrane cholesterol content (Figure 5A). Thus, in contrast with agonists and antagonists, α-BT binds the BTB-tagged OTR OTR-BTBe1 independent of the presence of cholesterol in the membrane. The same ligand-binding behaviour was obtained when, instead of OTRBTBe1, the construct OTR-BTBe4 was used (results not shown).

Figure 5 Cholesterol dependence of ligand binding to BTB-tagged OTRs

(A) Membranes (100 μg) from HEK-293 cells expressing OTR-BTBe1 were incubated with the indicated ligands [3H]OT (100 nM), OTAN-A568 (1 μM), α-BT-A568 (2 μg/ml) for 30 min at 30 °C. Then, the membranes were treated in the absence or presence of the cholesterol acceptor MβCD (final concentration 10 mM) for 30 min at 30 °C. To restore the cholesterol content of cholesterol-depleted membranes, MβCD-treated membranes were incubated with DChol–MβCD (0.3 mM) for 20 min at 30 °C. Excess ligands and MβCD/DChol–MβCD were washed-off by centrifugation and the receptor-bound ligands (radioactivity or fluorescence) were counted. The cyclodextrin treatments changed the cholesterol content from 201±5 μg/mg of protein (untreated) to 62±4 μg/mg of protein (i.e. ~70% reduction), and back to normal (200±10 μg/mg of protein). (B) Membranes (50 μg) from HEK-293 cells expressing OTR-BTBe4 were incubated with OTAN-A568 (1 μM) for 30 min at 30 °C. Successive depletion of cholesterol was achieved by increasing exposure times of the membranes to the cholesterol acceptor MβCD (10 mM at 30 °C) yielding the indicated cholesterol contents. For cholesterol enrichment, membranes were incubated for different times with DChol–MβCD (0.3 mM at 30 °C) to increase their cholesterol content to the indicated values. Excess ligands and MβCD/DChol–MβCD were washed-off and the receptor-bound ligand was measured. All data are means±S.D. (n=3).

The observed cholesterol dependence of OTAN-A568 binding to the OTR was analysed in more detail. For this purpose, purified plasma membranes from HEK-293 cells expressing OTR-BTBe4 were prepared and their cholesterol concentration was varied by using the cyclodextrin method. Successive depletion of cholesterol was achieved by increasing exposure times of the plasma membranes to the cholesterol acceptor MβCD (10 mM at 30 °C) yielding a reduction of cholesterol from 201±6 μg/mg of protein (untreated=100%) down to 59±7 μg/mg of protein. To obtain cholesterol-enriched membranes, plasma membranes were incubated for different times with the cholesterol donor DChol–MβCD (0.3 mM at 30 °C). This increased their cholesterol content up to 300±7 μg/mg of protein (~50% above untreated membranes). The binding of OTAN-A568 to membranes with various cholesterol contents is shown in Figure 5(B). A marked decrease in OTAN-A568 binding was observed when membrane cholesterol was reduced below ~130 μg/mg of protein (i.e. ~35% decrease of cholesterol). We also performed saturation analysis with increasing OTAN-A568 concentrations to both cholesterol-depleted (~35% below normal) and cholesterol-enriched (~50% above normal) membranes from HEK-293 cells expressing OTR-BTBe4, yielding the following KD and Bmax values: 18±4 nM and 0.6±0.1 pmol/mg of protein for cholesterol-enriched, 27±6 nM and 0.5±0.1 pmol/mg of protein for the cholesterol-depleted membranes (see Supplementary Figure S2 at http://www.BiochemJ.org/bj/437/bj4370541add.htm). When membranes were further cholesterol depleted, the affinity to OTAN-A568 was drastically diminished (KD>100 nM) similar to that previously described for the ligand OT [21].

Probing the accessibility of bound ligands by collisional quenching and dependence on cholesterol

The water-soluble quenchers KI and CoCl2 were used to probe the accessibility of free compared with bound fluorescently labelled ligands OTAN-A568 and α-BT-A647. For both quenchers, linear Stern–Volmer plots were obtained which indicated collisional quenching. This is demonstrated in Supplementary Figure S3(A) (available at http://www.BiochemJ.org/bj/437/bj4370541add.htm) for OTAN-A568 bound to plasma membranes prepared from OTR-BTBe4 cells. The Stern–Volmer constant KSV represent the slopes of F0/F as a function of the quencher concentration and provide information about the degree of accessibility of the fluorophore to the quencher. The KSV values are displayed in Table 3. For both fluorescent ligands, quenching by the cationic cobalt was more pronounced than by the anionic iodide. Notably, the quenching constants of the free and receptor-bound ligands OTAN-A568 and α-BT-A647 were more or less the same. Surprisingly, non-specifically bound OTAN-A568 appears to be slightly more exposed to the iodide quencher than the free or receptor-bound antagonist.

View this table:
Table 3 Fluorescence quenching constants for the ligands OTAN-A568, OTAN-A647 and αBT-A647

Plasma membranes were prepared from cells expressing the OTR-BTBe1 receptor and were incubated with increasing concentrations of the indicated quencher. The Stern–Volmer constants (KSV) were determined as described in the legend of Supplementary Figure S3 (at http://www.BiochemJ.org/bj/437/bj4370541add.htm). Values represent the means±S.E.M. of at least four experiments (n.d., not determined).

Additionally, we performed experiments with the lipid-soluble quencher TEMPO. To verify whether TEMPO is able to quench molecules within the hydrophobic environment, the fluorescent cholesterol analogue DChol previously synthesized by our group [19] was incorporated into the plasma membrane using DChol-loaded MβCD and its quenching by TEMPO was measured. As shown in Supplementary Figure S3(B) (open circles), DChol was efficiently quenched by TEMPO (KSV ~ 80 M−1). In contrast, quenching of the fluorescence of receptor-bound OTAN-A568 by TEMPO was negligible (Supplementary Figure S3B, filled circles).

Next, we asked whether changes of the cholesterol content in the plasma membrane from OTR-BTBe4-expressing cells affected the solvent accessibility of the receptor-bound antagonist. Lowering the cholesterol level beneath a certain threshold can dramatically decrease the binding of antagonist OTAN-A568 as shown in Figure 5(B). Thus we decreased the membrane cholesterol only slightly, remaining above the level that was necessary to support the high-affinity binding of the antagonist as shown above. This was achieved by a substantial shortened exposure time of the membranes to the MβCD (10 mM for 5 min at 30 °C). Under these conditions, ~20% of the initial cholesterol was extracted (=‘−Chol’, i.e. cholesterol reduction from 201±6 μg/mg of protein to 155±5 μg/mg of protein) and >95% of OTAN-A568 remained bound as compared with untreated membranes (=‘Con’). To obtain cholesterol-enriched membranes, plasma membranes were incubated with DChol–MβCD (0.3 mM for 5 min at 30 °C) to increase their cholesterol content by ~20% (=‘+ Chol’, i.e. cholesterol increase to 247±8 μg/mg of protein) above untreated membranes. OTAN-A568 was bound to cholesterol-depleted (‘−Chol’), cholesterol-enriched (‘+Chol’) and untreated (‘Con’) membranes and quenching was performed with KI, CoCl2 or TEMPO (Supplementary Figure S3C, white, grey and black bars). In all cases, the corresponding quenching constants (KSV) did not differ significantly among membranes with different cholesterol contents. Similar KSV values were obtained in quenching experiments with membranes from OTRBTBe1 cells (results not shown). Thus the solvent accessibility of the bound antagonist is not affected by the cholesterol level of the membrane.

Rotational freedom of bound ligands and dependence on cholesterol

To assess the degree of rotational freedom of the receptor-bound ligands, fluorescence anisotropy measurements were performed. Plasma membranes from OTR-BTBe4-expressing cells with normal (‘Con’), decreased (‘−Chol’) or increased (‘+Chol’) cholesterol amounts were prepared as described above. Figure 6 shows the fluorescence anisotropy values determined for the receptor bound antagonist OTAN-A568 (open bars) and α-BT-A647 (closed bars) respectively, in membranes with different cholesterol contents. The corresponding cholesterol concentrations are given in the legend of Figure 6. The fluorescence anisotropy (r) of receptor-bound OTAN-A568 was significantly increased in cholesterol-enriched membranes (r=0.246±0.015) compared with control (r=0.21±0.010) or cholesterol-depleted membranes (r=0.20±0.017). All of these values are markedly higher than those measured for unbound OTAN-A568 (r=0.065±0.011) or for OTAN-A568 non-specifically bound to BSA (r=0.124±0.013). For α-BT-A647, the receptor-bound anisotropy values were only slightly higher than that measured for the free ligand (r=0.166±0.013). The higher anisotropy value for unbound α-BT-A647 compared with free OTAN-A568 is expected in view of the fact that the rate of rotational diffusion is inversely correlated with the size of the molecule; α-BT-A647 is several fold larger (~8 kDa) than OTAN-A568 (~1.6 kDa). In membranes with different cholesterol contents, the fluorescence anisotropy data of receptor-bound α-BT-A647 were almost identical.

Figure 6 Anisotropy of fluorescent ligands bound to the BTB-tagged OTR and dependence on cholesterol

Plasma membranes from OTR-BTBe4-expressing cells were either left untreated (‘Con’), were depleted of cholesterol by MβCD (‘−Chol’) or were cholesterol-enriched using DChol–MβCD (‘+Chol’) as described in the legend of Supplementary Figure S3 (at http://www.BiochemJ.org/bj/437/bj4370541add.htm). The cholesterol concentrations for plasma membranes ‘Con’, ‘−Chol’ and ‘+Chol’ were 201±6, 155±5 and 247±8 μg/mg of protein respectively. The membranes (50 μg) were incubated with OTAN-A568 (50 nM) (open bars) or α-BT-A647 (50 nM) (filled bars) (50 μg) for 30 min at 30 °C. Excess ligand was washed-off by centrifugation and the fluorescence anisotropy was measured. The data are means±S.D. (n=3). *Significant difference compared with control (t test, P<0.005).

FRET between receptor-bound ligands and dependence on cholesterol

Cholesterol-induced conformational changes of the OTR might lead to changes in the distance and/or orientation of two ligands simultaneously bound to one receptor molecule. To address this issue, we used FRET measurements with different combinations of appropriate FRET pairs. As a prerequisite for FRET, the emission spectrum of the donor has to overlap with the excitation spectrum of the acceptor fluorophore. The dyes Alexa Fluor® 568 and Alexa Fluor® 647 attached to the ligands OTAN or α-BT form such a FRET pair. Plasma membranes prepared from cells expressing untagged (OTR) and BTB-tagged OTRs (OTR-BTBe1 and OTR-BTBe4) were incubated with different combinations of fluorescent ligands, OTAN-A568/α-BT-A647 or α-BT-A568/α-BT-A647. After washing-off unbound ligands, FRET signals were analysed spectrofluorimetrically. The results are shown in Table 4. Negligible ligand binding and no FRET signals were observed with control membranes from cells expressing untagged OTRs. This confirms the specificity of both ligands and excludes any contribution of energy transfer due to non-specific ligand binding. When the ligand combination OTAN-A568 and α-BT-A647 was applied to membranes prepared from cells expressing the BTB-tagged OTR constructs OTR-BTBe1 and OTR-BTBe4, FRET between these ligands was clearly observed. Significantly larger FRET values were obtained for the construct OTR-BTBe4 compared with OTR-BTBe1 (Table 4). A typical FRET curve is displayed in Figure 7(A) (emission scans from 585 to 700 nm with excitation at the donor excitation maximum of 578 nm). Energy transfer occurs from the FRET-donor OTAN-A568 (D, donor-only curve) to the FRET-acceptor α-BT-A647 (A, acceptor-only) yielding a decreased fluorescence intensity of the donor emission (peak at ~600 nm) concomitant with increased sensitized emission of the acceptor fluorescence (peak at ~670 nm) (DA, donor–acceptor curve).

View this table:
Table 4 FRET between fluorescent ligands bound to the OTR

Plasma membranes were prepared from cells expressing untagged (OTR, negative control) and BTB-tagged OTRs (OTR-BTBe1 and OTR-BTBe4). The indicated combination of ligands was added to the membranes and FRET measurements were performed spectrofluorimetrically as described in the Experimental section. Results are means±S.D. (n=4).

Figure 7 FRET between the ligands OTAN-A568 and α-BT-A647 bound to the OTR-BTBe4 and dependence on cholesterol

(A) Plasma membranes prepared from OTR-BTBe4-expressing cells were incubated with OTAN-A568 (donor, D) and α-BT-A647 (acceptor, A) either separately or together. After washing-off any unbound ligands by centrifugation, the membranes were resuspended in assay buffer, and were transferred to a thermostated (20 °C) cuvette under continuous stirring. The samples were excited at 578 nm and the emission fluorescence was scanned from 585 to 700 nm. The three curves are emission scans of donor-only (D), acceptor-only (A) and donor–acceptor (DA) membranes. The background-subtracted and normalized fluorescence intensities are shown. Emission maxima of donor and acceptor probe are indicated. (B) Plasma membranes from OTR-BTBe4-expressing cells were treated or not with MβCD/DChol–MβCD to change their cholesterol content as indicated using the same experimental conditions described in Figure 5(B). OTAN-A568 and α-BT-A647 were both added to the membranes as donor and acceptor ligands respectively. After washing-off excess ligands and MβCD/DChol–MβCD, FRET measurements were performed spectrofluorimetrically as described in the Experimental section. The percentages indicate the cholesterol changes compared with untreated membranes. All data are means±S.D. (n=3).

Finally, we used OTR-BTBe4 to analyse the influence of cholesterol on FRET between receptor-bound OTAN-A568 and α-BT-A647. For this purpose, the plasma membranes from OTR-BTBe4-expressing cells were adjusted to six different cholesterol concentrations within the range from 130 to 300 μg/mg of protein by treatments with MβCD/DChol–MβCD. For better comparison, we selected the same cholesterol concentrations (within the range mentioned above) used to analyse the cholesterol dependence of ligand binding to BTB-tagged OTRs (see section above and Figure 5B). FRET efficiency between OTAN-A568 and α-BT-A647 clearly increased in parallel with the cholesterol content in the membranes from OTR-BTBe4-expressing cells as shown in Figure 7(B). The corresponding FRET efficiencies and membrane cholesterol levels were positively correlated with a correlation coefficient close to 1 (r2=0.9641).

OTRs that are localized in close proximity within the plasma membrane, e.g. forming dimers or clusters, could also contribute to and disturb intramolecular FRET signals. To prove this possibility, we co-incubated BTB-tagged OTRs with the ligand combination α-BT-A568 and α-BT-A647, as well as with the antagonist combination OTAN-A568 and OTAN-A647. However, no FRET signals were detectable in all of these samples (Table 4). This suggests that intermolecular receptor interactions, if present at all, occur rather infrequently, and should not contribute significantly to the observed FRET signals.

DISCUSSION

A direct interaction of the OTR with cholesterol is supported by the following features: (i) agonist binding and receptor stability are highly dependent on cholesterol and exhibit a strong requirement on the specific structure of cholesterol [7,12]; (ii) the OTR is markedly more stabilized in cholesterol-rich microdomains than in cholesterol-poor domains [21]; (iii) the receptor displays a specific dependence on cholesterol even in its solubilized form [12,21]; (iv) two putative cholesterol-binding domains are found within the OTR, termed the ‘cholesterol consensus motif’ [10] and ‘cholesterol recognition amino acids consensus’ sequence [23,24].

To explore putative conformational changes of the OTR and dependence on cholesterol, we exploited the fluorescence properties of specifically bound ligands at various cholesterol levels. For this purpose, we have created recombinant OTRs that are able to bind two small-sized ligands simultaneously. To achieve this goal, a short (13-amino acid) α-BTB site was inserted at the N-terminus (construct OTR-BTBe1) or within the third extracellular loop connecting transmembrane helices 6 and 7 (construct OTR-BTBe4). Cells expressing EGFP at the C-terminus of the OTR (OTRGFP-BTBe1) were produced to compare the trafficking of the OTR tagged with BT–BTB against EGFP. All receptor constructs were functionally active and bound both ligands, a specific receptor antagonist (OTAN) and α-BT with high-affinity. The simultaneous receptor binding of both α-BT and antagonist is supported by several observations: (i) approximately equal amounts of binding sites have been calculated in saturation studies with both ligands. (ii) Even if a saturating concentration of one ligand (e.g. OTAN) has occupied the OTRs, similar amounts of the second ligand (e.g. labelled α-BT) were still able to bind to the preoccupied receptors. Any displacement of the prebound first ligand by application of the second ligand has not been observed. (iii) OT that shares a partly common binding site with OTAN induces internalization of all α-BT-bound receptors, with a nearly perfect co-localization during the early phases (up to 30 min) of endocytosis. Thereafter, co-localization of ligand and receptor diminished gradually, indicating different timings for both molecules. The majority of OTRs are known to recycle back to the plasma membrane, whereas the ligand probably becomes degraded in lysosomes [25].

Agonist and, to a slightly lesser extent, antagonist binding of the OTR strongly depended on the presence of cholesterol, in agreement with our previous studies [7,12,26]. In contrast, fluorescent α-BT (α-BT-A568) bound to all BTB-tagged OTR constructs independent of the cholesterol status in the membrane. This might be expected for construct OTR-BTBe1 in which α-BT binds at the N-terminus presumably distant from the membrane bilayer. The fact that this is also observed for construct OTR-BTBe4 suggests that cholesterol-induced changes in the receptor conformation are not accompanied by significant changes in the position of the small extracellular loop 3.

The solvent accessibility of the fluorophores attached to both ligands was assessed by the influence of various types of quenchers. The accessibility of the water-soluble quenchers KI and CoCl2 did not change upon association with the receptor. This was expected for α-BT-A647 as the position of the fluorophore attached to α-BT should not change dramatically. However, iodide quenching of both Alexa Fluor®-labelled ligands was relatively low. For example, in the cases of the secretin and cholecystokinin receptors, iodide quenching of Alexa Fluor® 488-coupled ligands yielded several fold higher KSV values [27,28]. The higher quenching by Co2+ compared with the anionic iodide might be explained by the electrostatic attraction of cationic Co2+ to the negatively charged Alexa Fluor® fluorophore. Notably, the solvent accessibility of the bound antagonist was not affected by the cholesterol level of the membrane. Additionally we used TEMPO, a small nitroxide-carrying quencher that is able to examine the possible location of a fluorescent ligand in a hydrophobic microenvironment [29,30]. While TEMPO efficiently quenched a fluorescent cholesterol analogue, a substantial quenching of the fluorescent receptor-bound ligands OTAN-A568 and α-BT-A647 was not observed, irrespective of the cholesterol content of the membrane. Thus none of the fluorophores attached to the ligands appear to be located in a hydrophobic pocket or very close to the membrane. Interestingly, an antagonist structurally very similar to OTAN binds in a central receptor pocket with close contacts to the first two transmembrane segments [31]. The results of the quenching experiments suggest that the fluorophore of OTAN-A568 located at the acyclic side chain of the peptide protrudes out of these transmembrane segments. Notably, the polarity around the fluorophores of the receptor-bound ligands was completely unaffected by cholesterol.

Our fluorescence anisotropy data showed that cholesterol had no influence on the rotational motion of α-BT-A647, whereas an increased cholesterol content restricted the rotational motion of receptor-bound OTAN-A568. Thus cholesterol acting as a stabilizer of the OTR [12] might decrease the rotational freedom of the antagonist placed within the binding pocket.

To specify putative cholesterol-induced receptor changes, FRET measurements were performed with different combinations of ligands containing appropriate FRET-pair fluorophores. Using the ligand combination OTAN-A568 and α-BT-A647, FRET between these ligands was clearly observed for both OTR-BTBe1 and OTR-BTBe4 receptors. However, the construct OTR-BTBe4 revealed considerably larger FRET values compared with OTR-BTBe1. We provide two lines of evidence that the FRET signals measured here are caused by intramolecular FRET, i.e. described changes of receptor conformation are not due to intermolecular FRET (e.g. the presence of receptor dimers/oligomers). First, no FRET signals were detectable for the constructs OTR-BTBe1 or OTR-BTBe4 when the FRET pair α-BT-A568/α-BT-A647 was used. Secondly, we have employed antagonist OTAN-A647 that could act as a potential FRET acceptor for OTAN-A568 in the case when receptor dimerization occurs. Again, we observed no energy transfer in experiments using the donor/acceptor pair OTAN-568/OTAN-A647 bound to the receptor OTR-BTBe4. Nevertheless, it is important to note that our results do not rule out that OTR dimers are present in cells expressing either of these constructs. Actually, OTR dimers/oligomers have been described in HEK-293T cells [HEK-293 cells expressing the large T-antigen of SV40 (simian virus 40)], COS cells and mammary glands of lactating rats using sensitive detection methods such as bioluminescence resonance energy transfer or lanthanide-based time-resolved fluorescence [3234]. Perhaps, the higher sensitivities of these latter methods are more appropriate for the detection of intermolecular interactions.

The observed cholesterol-induced increase in FRET between OTAN-A568 and α-BT-A647 could be caused either by changes in the orientation or in the distance of the fluorophores. The fluorescence-quenching data suggested no major changes in the molecular environment of the fluorophores in response to modulation of the cholesterol content. This would support the conclusion that the cholesterol-induced increase in FRET is primarily caused by decreasing the distance between the fluorophores. This interpretation of the data is illustrated in Figure 8. Accordingly, the antagonist-binding site is closer to the BTB site in extracellular loop e4 than to the BTB site in loop e1. Additional cholesterol decreases the distance between antagonist-binding site and position e4. Whether the distance between antagonist-binding site and the e1 site is also diminished by cholesterol, could not be proved because the corresponding FRET value was very low and close to the detection limit, possibly due to a high structural flexibility of the N-terminus of the receptor. However, the Alexa Fluor® 568 dye revealed a low sensitivity to environmental changes and the fluorescence anisotropy of OTAN-A568 was found to be affected by cholesterol. Thus cholesterol-induced changes in the dynamics of OTAN-A568 and possibly in its orientation might contribute to the observed changes in the FRET signal. The dipole orientation factor (κ2) cannot be directly measured and is the major uncertainty in the calculation of the Förster radius (R0). Provided that both FRET probes exhibit free isotropic motion, κ2 is assumed to have a value of 2/3. Under this assumption, an R0 value of 82 Å (1 Å=0.1 nm) was calculated for the dye pair Alexa Fluor® 568 and Alexa Fluor® 647 (from the ‘Molecular Probes Handbook’). If all the changes in the FRET signal would come from the change in the distance between the fluorescently labelled ligands, one can calculate an ~10 Å decrease in distance between OTAN-A568 and α-BT-A647 in response to cholesterol levels ranging from ~130 μg/mg of protein (i.e. ~35% below normal) to ~270 μg/mg of protein (i.e. ~30% above normal). Alternatively, cholesterol-induced alterations in the orientation of OTAN-A568 could lead to changes in κ2 and, thus an altered R0 value. Most probably, cholesterol causes a combination of both changes in orientation and distance of the receptor-bound ligands.

Figure 8 Model of the OTR with the binding sites for antagonist (A) and α-BT (e1 and e4)

The model (top view of the OTR from the extracellular side) was constructed by homology modelling using ‘EasyPred3D’ with rhodopsin as a template [40]. The UCSF (University of California, San Francisco) Chimera programme (http://www.cgl.ucsf.edu/chimera) was employed for graphical display. TM, transmembrane domains; E, extracellular loops; e1 and e4, positions at E1 and E4 where BTB sites have been included.

Cholesterol is known to stabilize some GPCRs such as rhodopsin [35,36], the β2-adrenergic receptor [10,11], the A2a adenosine receptor [37] and the OTR [12]. As cholesterol combines small size, flexibility (aliphatic side chain) and rigidity (tetracyclic ring system) in one molecule, it is ideally suitable to fill shallow grooves and cavities at the receptor surface, e.g. by occupying so-called non-annular lipid-binding sites [38]. Cholesterol binding may increase the intramolecular occluded surface area of the receptor, leading to enhanced protein stability [10]. Molecular dynamics simulations also provided evidence that surface-bound or embedded cholesterol is essential for the stabilization and functioning of diverse membrane proteins [36,39]. Overall, cholesterol binding appears to be mostly associated with reduced conformational flexibility and tighter packing of the protein structure. For the OTR, the results of the present study fully agree with the conclusion that cholesterol supports a more compact structure. The approach described in the prsent study might also be applicable to other membrane proteins aimed to analyse corresponding conformational changes of the protein in their native membrane environment.

AUTHOR CONTRIBUTION

Sabine Muth and Anja Fries generated recombinant OTR constructs, characterized the function of these receptors and performed fluorescence microscopy. Gerald Gimpl synthesized fluorescent antagonists, conducted the biophysical measurements, performed fluorescence microscopy, designed the study and wrote the manuscript.

FUNDING

G. G. was supported by a grant from Bundesministerium für Bildung und Forschung (BMBF) [grant number CZE01-027].

Acknowledgments

We thank Dr Per Melin (Ferring Pharmaceutics, Sweden) for providing the antagonist F792, designated here as OTAN. We thank Christa Wolpert for excellent technical assistance.

Abbreviations: AVP, Arg-vasopressin; α-BT, α-bungarotoxin; BTB, bungarotoxin binding; CoCl2, cobalt chloride; DChol, 6-dansyl-cholestanol; DMEM, Dulbecco's modified Eagle's medium; EGFP, enhanced green fluorescent protein; FRET, fluorescence resonance energy transfer; GPCR, G-protein-coupled receptor; HEK-293 cell, human embryonic kidney cell; MβCD, methyl-β-cyclodextrin; NA, numerical aperture; OT, oxytocin; OTAN, OT antagonist; OTR, OT receptor; KI, potassium iodide; ROI, region of interest; TEMPO, 2,2,6,6-tetramethylpiperidine-N-oxyl; TFA, trifluoroacetic acid

References

View Abstract