Research article

The orexin OX1 receptor exists predominantly as a homodimer in the basal state: potential regulation of receptor organization by both agonist and antagonist ligands

Tian-Rui Xu, Richard J. Ward, John D. Pediani, Graeme Milligan


It is unclear what proportion of a G-protein-coupled receptor is present in cells as dimers or oligomers. Saturation bioluminescence resonance energy transfer studies demonstrated the orexin OX1 receptor to be present in such complexes. Forms of this receptor containing a minimal epitope tag, with the C-terminus linked to yellow fluorescent protein or modified at the N-terminus to incorporate a SNAP tag, migrated in SDS/PAGE gels as monomers, indicating a lack of covalent interactions. Solubilization with dodecylmaltoside, followed by Blue native-PAGE, indicated that the receptor constructs migrated predominantly as anticipated for dimeric species with evidence for further, higher-order, complexes, and this was true over a wide range of expression levels. Addition of SDS prior to separation by Blue native-PAGE resulted in much of the previously dimeric, and all of the higher-order, complexes being dissociated and now migrating at the size predicted for monomeric species. Expression of forms of the OX1 receptor capable of generating enzyme complementation confirmed that solubilization itself did not result in interaction artefacts. Addition of the endogenous agonist orexin A enhanced the proportion of higher-order OX1 receptor complexes, whereas selective OX1 antagonists increased the proportion the OX1 receptor migrating in Blue native-PAGE as a monomer. The antagonist effects were produced in a concentration-dependent manner, consistent with the affinity of the ligands for the receptor. Homogeneous time-resolved fluorescence resonance energy transfer studies using Tag-Lite™ reagents on cells expressing the SNAP-tagged OX1 receptor identified cell-surface OX1 homomers. Predominantly at low receptor expression levels, orexin A increased such fluorescence resonance energy transfer signals, also consistent with ligand-induced reorganization of the homomeric complex.

  • Blue native-PAGE (BN-PAGE)
  • homodimer
  • orexin OX1 receptor
  • receptor organization


In recent years the concept that GPCRs (G-protein-coupled receptors) exist in cells exclusively as independent, non-interacting, monomeric species [1] has largely been overtaken by a view that they can exist as dimeric, or even oligomeric, complexes [2,3]. However, apart from members of the class C, metabotropic, glutamate-like receptors that exist as obligate homomers or heteromers [4] and where disulfide bonds between the large extracellular Venus flytrap-like domains act to partially stabilize the multi-protein complex, there is little evidence to indicate that receptor dimers/oligomers are linked covalently. This implies that for the large family of class A, rhodopsin-related, GPCRs, the extent of dimeric or oligomeric assembly must be related to the interaction affinities between individual GPCR protomers. This may potentially vary significantly between individual GPCRs, particularly if, as has been suggested in a number of studies [3], different transmembrane elements contribute to interactions between individual GPCRs. Despite this, relatively little is known about the dynamics and regulation of GPCR oligomer formation and/or stability, and even less is known about the proportion of individual GPCRs that is present as dimers/oligomers. However, there is some direct evidence to suggest differences in affinity of such interactions between even closely related GPCRs. For example, using FRAP (fluorescence recovery after photobleaching) techniques it was demonstrated that although the β2-adrenoceptor appears to form a stable oligomeric complex, parallel studies on the β1-adrenoceptor suggested that it forms a more transient complex [5]. Equally, by employing total internal reflectance fluorescence microscopy Hern et al. [6] have shown that dimers of the muscarinic M1 receptor appear to associate and dissociate on a time scale of seconds.

The orexin receptors OX1 and OX2 are members of the rhodopsin-like family of GPCRs [7]. Ligand pairing of these receptors with the peptides orexin A and orexin B, both derived from the precursor prepro-orexin [7], instigated a wide range of studies on their biological function that have centred on the regulation of sleep/wakefulness and in the control of feeding and appetite [8,9]. On the basis of such studies, agonism of orexin receptors has been suggested as a means to treat narcolepsy, whereas orexin receptor antagonists have been promoted as a potential treatment of sleep disorders such as insomnia. Indeed, the combined OX1 and OX2 antagonist almorexant has undergone late-stage clinical trials in this area [10]. In previous studies we have shown that when expressed from an inducible promoter in Flp-In™ T-REx™ HEK (human embryonic kidney)-293 cells, modified forms of the human OX1 receptor migrate on SDS/PAGE gels as apparently single polypeptides [11]. This is in contrast with a number of other class A GPCRs that show complex migration patterns, often associated with differential N-glycosylation. In the present study, we use the human OX1 receptor to explore the propensity of this GPCR to form homodimers/oligomers and, via BN-PAGE (Blue native-PAGE) [12,13], demonstrate that at a range of expression levels the OX1 receptor exists predominantly as a homodimer in the basal state. This is the first study to define the proportion of a GPCR present in cells as a homodimer. We also use combinations of cell-surface htrFRET [homogeneous time-resolved FRET (fluorescence resonance energy transfer)] [14,15] and BN-PAGE to explore whether ligand binding to the OX1 receptor alters receptor quaternary organization and an enzyme complementation strategy [16] based on a split Renilla luciferase variant [17] to ensure that membrane receptor solubilization by DDM (n-dodecyl-β-D-maltoside) prior to BN-PAGE does not inherently produce OX1 receptor interaction artefacts.



[3H]SB-674042 [1-(5-(2-fluorophenyl)-2-methyl-thiazol-4-yl)1-((S)-2-(5-phenyl-(1,3,4)oxadiazol-2-ylmethyl)-pyrrolidin-1yl)-methanone] was from GE Healthcare. Orexin A was a gift from GlaxoSmithKline or was supplied by Bachem UK. Rluc8 (Renilla luciferase 8) [17] cDNA was a gift from Dr S.S. Gambhir, Stanford University School of Medicine, Stanford, CA, U.S.A. SB-334867 [N-(2-methyl-6-benzoxazolyl)-N′-1,5-naphthyridin-4-yl urea] and SB-408124 {N-(6,8-difluoro-2-methyl-4-quinolinyl)-N′-[4-(dimethylamino)phenyl]urea} were from Tocris Biosciences. Polyclonal rabbit anti-[EGF (epidermal growth factor) receptor] antibody and rabbit anti-HA (haemagglutinin) antibody were from Santa Cruz Biotechnology. Polyclonal rabbit VSV-G (where VSV is vesicular stomatitis virus) antiserum was generated in-house. Flp-In™ T-REx™ HEK-293 cells and Lipofectamine™ 2000 transfection reagent were from Invitrogen. Complete™ protease inhibitor cocktail tablets were from Roche Diagnostics. Monoclonal mouse anti-VSV-G antibody and all other materials were supplied by Sigma–Aldrich or Fisher Scientific.

DNA constructs

The plasmids for expression of the constructs VSV-G–OX1–eYFP (enhanced yellow fluorescent protein) and VSV-G–SNAP–OX1 were generated as described previously in [11] and [18] respectively. pcDNA3.1-VSV-G-OX1-Rluc8 was generated by replacing the eYFP of pcDNA3.1-VSV-G-OX1-eYFP with Rluc8. The primers were 5′-CGATCGATGCGGCCGCCATGGCTTCCAAGGTGTACGACCC-3′ (forward primer with NotI site) and 5′-TCGTCTCGAGTTACTGCTCGTTCTTCAGCACGCGCT-3′ (reverse primer with XhoI site). pcDNA3.1-VSV-OX1-Rluc8 N-terminal split (amino acids 1–229) and pcDNA3.1-VSV-G-OX1Rluc8 C-terminal split (amino acids 230–311) were generated using the same strategy. The primers for the Rluc8 N-terminal split were 5′-CGATCGATGCGGCCGCCATGGCTTCCAAGGTGTACGACCC3′ (forward primer with NotI site) and 5′-TCGTCTCGAGTTAGCCTCCCTTAACGAGAGGGATCTC-3′ (reverse primer with XhoI site). The primers for the Rluc8 C-terminal split were 5′-CGATCGATGCGGCCGCCAAGCCCGACGTCGTCCAGATTGT-3′ (forward primer with NotI site) and 5′-TCGTCTCGAGTTACTGCTCGTTCTTCAGCACGCGCT-3′ (reverse primer with XhoI site). pcDNA3-FLAG-α1b-adrenoceptor-eYFP was constructed as described previously [19], and pcDNA3-FLAG-α1b-adrenoceptor-Rluc8 was generated by replacing the eYFP of pcDNA3-FLAG-α1b-adrenoceptor-eYFP with Rluc8. The primers were 5′-CGATGGTACCATGGCTTCCAAGGTGTACGACCC-3′ (forward primer with KpnI site) and 5′-TCGTGCGGCCGCTTACTGCTCGTTCTTCAGCACGCGCT-3′ (reverse primer with NotI site). To generate a minimally tagged form of the OX1 receptor, a stop codon was added to the end of the OX1 sequence of the pcDNA5/FRT/TO VSV-G-OX1-eYFP construct. The primers used were 5′-GTCACCACAGTGCTGCCCTGAGCCGCCGTGAGCAAGGG-3′ (forward) and 5′-CCCTTGCTCACGGCGGCTCAGGGCAGCACTGTGGTGAC-3′ (reverse).

Generation and maintenance of stable Flp-In™ T-REx™ HEK-293 cells

To generate Flp-In™ T-REx™ HEK-293 cells able to inducibly express the VSV-G–OX1, VSV-G–OX1–eYFP [11] or VSV-G–SNAP–OX1 [18] constructs, cells were co-transfected with the plasmid pOG44 and the desired cDNA in pcDNA5/FRT/TO (Invitrogen) at a ratio of 9:1 using Lipofectamine™. After 48 h, the medium was supplemented with 200 μg·ml−1 hygromycin to select for stably transfected cells. Pools of cells were established and tested for inducible expression by the addition of 1 μg·ml−1 doxycycline for 48 h followed by screening for VSV-G or SNAP-tag protein expression by Western blotting. A further cell line was established using the VSV-G–OX1–eYFP-inducible Flp-In™ T-REx™ HEK-293 cells as a base. These cells were transfected with pcDNA3.1-HA-OX1-eYFP [20], and clones resistant to the presence of 1 mg·ml −1 G418 were screened initially for the presence of eYFP in the absence of doxycycline (as VSV-G–OX1–eYFP is not expressed in this situation) and subsequently to detect anti-HA antibody immunoreactivity.

Co-immunoprecipitation studies

The cells described above were untreated or VSV-G–OX1–eYFP expression was induced by treatment with doxycycline for 24 h. Cells were harvested and treated with lysis buffer (150 mM NaCl, 0.01 mM Na3PO4, pH 7.4, 2 mM EDTA, 0.5% DDM and 5% glycerol plus Complete™ protease inhibitor cocktail tablets) on a rotating wheel for 30 min at 4 °C. Samples were then centrifuged for 30 min at 100000 rev./min at 4 °C in a Beckman TLA 100.2 rotor, and the supernatant was transferred to a fresh tube and incubated with anti-VSV-G antibody–agarose beads (Sigma) for 2 h at 4 °C on a rotating wheel. Samples were subsequently washed four times with lysis buffer. The bound receptors were eluted with 0.2 mg·ml−1 VSV-G peptide (Sigma) in lysis buffer and the eluates were separated by BN-PAGE. Immunoblotting with anti-HA and anti-VSV-G antibodies was then performed.

HEK-293T cell culture and transfection

HEK-293T cells [HEK-293 cells expressing the large T-antigen of SV40 (simian virus 40)] were maintained in DMEM (Dulbecco's modified Eagle's medium) supplemented with 0.292 g·l−1 L-glutamine, 1% antibiotic mixture and 10% (v/v) newborn calf serum at 37 °C in a 5% CO2 humidified atmosphere. The cells were transfected using Lipofectamine™ 2000 reagent as described in the protocols from the supplier.

Cell membrane preparation

Pellets of cells were frozen at −80 °C for a minimum of 1 h, thawed and resuspended in ice-cold TE buffer (10 mM Tris/HCl, pH 7.4, and 0.1 mM EDTA) supplemented with Complete™ protease inhibitor cocktail. Cells were homogenized on ice by 40 strokes in a glass–Teflon homogenizer followed by centrifugation at 1000 g for 5 min at 4 °C to remove unbroken cells and nuclei. The supernatant fraction was removed and passed through a 25-gauge needle ten times before being transferred to ultracentrifuge tubes and subjected to centrifugation at 50000 rev./min for 30 min at 4 °C in a TLA 100.2 rotor. The resulting pellets were resuspended in ice-cold TE buffer. Protein concentration was assessed and membranes were stored at −80 °C until required.

[3H]SB-674042-binding assays

Saturation binding curves were established by the addition of 5 μg of membrane protein to assay buffer (25 mM Hepes, pH 7.4, 500 mM NaCl and 2.5 mM MgCl2), supplemented with 0.3% BSA containing up to 12 nM [3H]SB-674042. Non-specific binding was determined in the presence of 30 μM SB-408124. Reaction mixtures were incubated for 90 min at 25 °C, and bound ligand was separated from free ligand by vacuum filtration through GF/C filters (Brandel). The filters were washed twice with ice-cold 1× PBS (120 mM NaCl, 25 mM KCl, 10 mM Na2HPO4 and 3 mM KH2PO4, pH 7.4) and bound ligand was estimated by liquid-scintillation spectrometry.

[3H]SB-674042-binding assays and DDM solubilization

Flp-In™ T-REx™ HEK-293 cells harbouring VSV-G–OX1–eYFP were untreated or induced with doxycycline for 24 h. The cells were harvested and lysed as described for cell membrane preparations and the supernatants collected after the low-speed spin. A total of 250 μg of protein was added to assay buffer containing up to 5 nM [3H]SB-674042. Non-specific binding was determined in the presence of 3 μM SB-408124. Reaction mixtures were incubated for 90 min at 25 °C and then divided into two aliquots, which were spun at 14000 rev./min for 15 min at 4 °C in an Eppendorf F45-30-11 rotor, and the supernatants were removed. The first pellets were washed with 400 μl of assay buffer and then resuspended in 200 μl of assay buffer for bound ligand to be determined by liquid-scintillation spectrometry. The second pellets were resuspended in 100 μl of assay buffer supplemented with 0.5% DDM and 5% glycerol before being incubated at 4 °C on a rotating wheel for 30 min. These samples were centrifuged at 100000 rev./min for 60 min at 4 °C in a TLA 100.2 rotor, and 50 μl of the supernatant was then applied to a Micro Bio-Spin P-6 column (Bio-Rad Laboratories) to separate bound from free [3H]SB-674042. Protein-bound [3H]SB-674042 eluted from the column was determined by liquid-scintillation spectrometry and values were corrected to allow for only half of the sample being applied to the Micro Bio-Spin column.

Epifluorescence imaging of SNAP-tag proteins in live cells

Cells able to express VSV-G–SNAP–OX1 were grown on coverslips that had been cleaned with ethanol and treated with 0.1 mg·ml−1 poly-D-lysine. SNAP-tag-specific dye substrates were diluted in complete DMEM from a 1 mM stock solution to give a labelling solution of 5 μM with respect to the SNAP dye substrate. The cell medium was replaced with labelling solution and incubated at 37 °C in 5% CO2 for 30 min. Cells were washed three times with complete DMEM and once with Hepes physiological saline solution (130 mM NaCl, 5 mM KCl, 1 mM CaCl2, 1 mM MgCl2, 20 mM Hepes, pH 7.4, and 10 mM D-glucose). Coverslips were then transferred to a microscope chamber where they were imaged using an inverted Nikon TE2000-E microscope equipped with a ×40 (numerical aperture 1.3) oil-immersion Pan Fluor lens and a cooled digital photometrics Cool Snap-HQ charge-coupled device camera (Roper Scientific).

HtrFRET studies

Cells induced to express VSV-G–SNAP–OX1 were seeded at 100000 cells per well in solid black 96-well plates (Greiner BioOne) that had been treated with 0.1 mg·ml−1 poly-D-lysine. The growth medium was replaced with 100 μl of a mixture containing the fixed optimal concentrations of donor (Tag-lite SNAP-Lumi4-Tb) and acceptor (Tag-lite SNAP-Red) in 1× labelling medium (all from Cisbio Bioassays). Plates were incubated for 1 h at 37 °C in a 5% CO2 humidified atmosphere and were subsequently washed four times in labelling medium. Plates were either read directly after this or were subjected to ligand treatment first. For this, appropriate drug concentrations were added to the plates after the last labelling medium wash; they were then incubated at 37 °C for the required time and analysed using a PHERAstar FS HTRF®-compatible microplate reader (BMG Labtech). The emission signal from the Tag-lite SNAP-Lumi4-Tb cryptate (620 nm) and the FRET signal resulting from the acceptor Tag-lite SNAP-Red (665 nm) were recorded. The specific htrFRET signal was calculated by subtracting the 665 nm signal obtained from un-induced cells (i.e. not expressing the receptor), but which had been labelled, from the 665 nm signal of cells expressing the receptor. This corrected 665 nm value was then used to calculate the 665/620 FRET ratio.


Flp-In™ T-REx™ 293 cells induced with doxycycline to express VSV-G–OX1–eYFP or VSV-G–SNAP–OX1 were harvested in PBS and lysed in lysis buffer on a rotating wheel for 30 min at 4 °C. Samples were then centrifuged for up to 60 min at 100000 rev./min at 4 °C using a TLA 100.2 rotor, and the supernatants were collected. A total of 18 μg of solubilized supernatant plus 4 μl of G250 additive (Invitrogen) was loaded on to each lane of NativePAGE™ Novex® 3–12% Bis-Tris Gels (Invitrogen). In certain experiments, various concentrations of SDS were added prior to this stage. After migration at room temperature (24 °C), proteins were transferred (90 min at 25 V) on to a PVDF membrane that had been pre-wetted for 30 s in methanol. The membrane was then fixed in 8% acetic acid for 15 min and immunoblotted with an anti-VSV-G antibody or VSV-G antiserum. For SDS/PAGE, the same set of solubilized supernatants was mixed with 2× SDS loading buffer and, after heating to 65 °C for 15 min, samples were resolved on 4–12% NuPAGE Bis-Tris gels (Invitrogen) and subsequently immunoblotted to detect proteins of interest.

BRET (bioluminescence resonance energy transfer) assays

HEK-293T cells were co-transfected with receptor–Rluc8 and receptor–eYFP constructs in six-well plates. At 24 h post-transfection, cells were trypsinized and transferred to poly-D-lysine-coated white-walled 96-well plates (for BRET and luminescence reading) and black-walled 96-well plates (for eYFP reading). At 48 h post-transfection, cells in black-walled 96-well plates were washed with HBSS (Hanks balanced salt solution; Invitrogen) and the expression of eYFP was measured on a PHERAstar FS microplate reader (BMG Labtech). Cells in white-walled 96-well plates were washed and incubated in HBSS at 37 °C for 30 min, then coelenterazine-h was added (5 μM final concentration) and incubated for another 10 min in the dark at 37 °C, and BRET was measured on a PHERAstar FS reader. For BRET saturation assays, cells were co-transfected with a fixed amount of the Rluc8-tagged receptor construct and increasing amounts of the eYFP-tagged receptor construct. Net BRET signal was plotted as a function of eYFP (energy acceptor) divided by Rluc8 (energy donor) fusion expression (eYFP/Rluc8). The curves were fit using a non-linear regression equation assuming a single binding site with Prism v. 5.0 software. The hetero-titration curve of OX1–Rluc8/α1b-adrenoceptor–eYFP was achieved by transfecting a fixed amount of OX1–Rluc8 (0.5 μg/well) and increasing amounts of α1b-adrenoceptor–eYFP. The hetero-titration curve of α1b-adrenoceptor–Rluc8/OX1–eYFP was achieved in a similar manner.

Enzyme complementation assay of solubilized receptors

HEK-293T cells were transfected with OX1–Rluc8 N-terminal split or OX1 receptor–Rluc8 C-terminal split respectively, or co-transfected with both forms. Transfected cells were solubilized in lysis buffer. Samples were then centrifuged for 15 min at 15000 g at 4 °C and supernatants were collected. Supernatants [100 μg (45 μl)] from mixed single transfections or co-transfections were added into 45 μl of HBSS in a white-walled 96-well plate. Rluc8 activity was measured by adding 10 μl of 50 μM coelenterazine-h and reading on a PHERAStar FS instrument.


Flp-In™ T-REx™ HEK-293 cells harbouring a VSV-G–human OX1 receptor–eYFP (VSV-G–OX1–eYFP) construct at the Flp-In™ locus [11] were induced to express this polypeptide by treatment with doxycycline for either 22 or 36 h. Imaging the distribution of the eYFP tag in such cells indicated that a high proportion of the induced receptor was present at the cell surface (Supplementary Figure S1A at Lysates of these samples were resolved by SDS/PAGE and immunoblotted with an anti-VSV-G antibody. As anticipated from previous studies [11], although no anti-VSV-G immunoreactivity was detected without pre-treatment with doxycycline, an apparently single polypeptide with a molecular mass of ∼90 kDa was present following doxycyline treatment (Supplementary Figure S1B). Although this is consistent with detection of a single, monomeric, form of this OX1 receptor construct with limited micro-heterogeneity of N-linked glycosylation or other related modifications, such studies cannot interrogate the potential quaternary organization of the receptor in the cell membrane. To address this, VSV-G–OX1–eYFP was co-transfected into HEK-293T cells with a form of the VSV-G–OX1 receptor C-terminally tagged with the Renilla luciferase variant Rluc8 [18] (VSV-G–OX1–Rluc8). BRET was measured after addition of coelenterazine-h (Figure 1). Co-transfection of VSV-G–OX1–Rluc8 and VSV-G–OX1–eYFP in various ratios, followed by addition of coelenterazine-h, resulted in BRET signals that increased over a range of low energy-acceptor to energy-donor ratios, but subsequently reached a maximal level and saturated at higher energy-acceptor to energy-donor ratios (Figure 1A). Such studies are consistent with at least a proportion of the OX1 receptors forming dimeric or oligomeric complexes [21]. Substantially lower BRET signals were produced at the same eYFP (energy acceptor) to Rluc8 (energy donor) ratio when VSV-G–OX1–Rluc8 was co-transfected with α1b-adrenoceptor–eYFP or when the energy acceptor/energy donor configuration was reversed by co-transfection of VSV-G–OX1–eYFP and α1b-adrenoceptor–Rluc8 (Figure 1B).

Figure 1 BRET assays indicate that VSV-G–OX1–eYFP forms dimers/oligomers

VSV-G–OX1–eYFP and VSV-G–OX1–Rluc8 were transiently co-expressed in various ratios in HEK-293T cells. (A) Following addition of coelenterazine-h, BRET was measured. (B) In equivalent studies, VSV-G–OX1–eYFP was co-expressed with α1b-adrenoceptor–Rluc8 or VSV-G–OX1–Rluc8 was co-expressed with α1b-adrenoceptor–eYFP. BRET was recorded and signals at eYFP/Rluc8=0.1 are displayed. Results shown are means±S.E.M. and are representative of three independent experiments.

To attempt to explore the proportion of OX1 receptors that exist in such complexes we initially assessed the fraction of OX1 receptors that could be solubilized by treatment with the detergent DDM. Lysates from both un-induced cells and those induced to express VSV-G–OX1–eYFP were treated with the OX1 receptor antagonist [3H]SB-674042 (5 nM) in the absence and presence of a second, unlabelled, OX1 receptor antagonist, SB-408124 (3 μM), to define non-specific binding. Although no specific binding of [3H]SB-674042 was detected without treatment of the cells with doxycycline (Figure 2), 3.7±0.08 pmol·(mg of protein)−1 of specific [3H]SB-674042 binding was present after induction (Figure 2). Such lysates were subjected to treatment with 0.5% DDM. Following sustained centrifugation (100000 rev./min for 1 h at 4 °C using a TLA 100.2 rotor), the non-solubilized material was resuspended and the presence of specific [3H]SB-674042-binding sites was assessed. Recovery in this fraction was 24.7% of the specific [3H]SB-674042 binding of the lysate, indicating that approximately 75% of the OX1 receptor construct was solubilized or unfolded by this treatment. Specific binding of [3H]SB-674042 was also assessed in the solubilized fraction and recovered after passage through a Micro Bio-Spin P-6 column. 75.8±1.0% of the specific binding sites potentially solubilized by treatment with DDM (and therefore 57.1% of the binding sites in the initial lysate) were recovered by this process (Figure 2), indicating the stability of VSV-G–OX1–eYFP in 0.5% DDM. Subsequently, cell lysates from un-induced Flp-In™ T-REx™ HEK-293 cells and those induced to express VSV-G–OX1–eYFP for either 22 or 36 h were treated with 0.5% DDM in the same way and, following sustained centrifugation, the material in the supernatant was subsequently resolved by BN-PAGE after the addition of G250. In such gels the vast majority of the induced anti-VSV-G immunoreactivity migrated with mobility corresponding to reference proteins of some 180 kDa, potentially consistent with a dimeric form of VSVG–OX1–eYFP (Figure 3A). Indeed, sustained exposure of such immunoblots was required to observe a small proportion of an apparent 90 kDa species (Figure 3A) and, at this level of exposure, a series of apparent higher-order species could also be detected (Figure 3A). There are potential issues with the use of BN-PAGE to assess the molecular mass and organization of solubilized transmembrane proteins [22], but addition of SDS to such samples is anticipated to result in their resolution and migration as monomers if the protein is not simply aggregated irreversibly or the migration reflects a monomeric protein still encased in the detergent used for solubilization [22]. Various concentrations of SDS were added to samples prior to resolution by BN-PAGE. At concentrations of SDS below 0.1%, no clear differences were noted in mobility of the VSV-G reactive complex. However, at 0.1% SDS, there was the appearance of a small degree of immunoreactivity consistent with a monomer and this was greatly increased at 1% SDS (Figure 3B). Importantly, migration in BN-PAGE of the EGF receptor (175 kDa) expressed endogenously by the cells was at the size anticipated for a monomer and was unaffected by the addition of 1% SDS to samples prior to BN-PAGE (Figure 3C). Addition of EGF is known to generate a non-covalently associated dimeric complex of the receptor [23,24], and after addition of 100 ng·ml−1 EGF for 5 min prior to solubilization with 0.5% DDM, a substantial proportion of the EGF receptor now migrated in BN-PAGE with the size anticipated for a dimeric complex (Figure 3C) and this was reversed by addition of 1% SDS (Figure 3C). To further explore the possibility that the DDM solubilization resulted in artificial aggregation of the OX1 receptor, populations of HEK-293T cells were transfected individually with forms of the VSV-G–OX1 receptor C-terminally tagged with either amino acids 1–229 (Rluc8N) or amino acids 230–311 (Rluc8C) of Rluc8. These fragments are known to be able to complement and generate luciferase activity if they are linked to polypeptides that interact and hence bring the Renilla luciferase fragments into close proximity [16,17]. Following initial solubilization with 0.5% DDM, the supernatant samples were mixed. However, this process did not result in Rluc8 complementation and luciferase activity (Figure 4). In contrast, solubilization of lysates of HEK-293T cells transfected to co-express VSV-G–OX1–Rluc8N and VSV-G–OX1–Rluc8C produced a high level of luciferase activity (Figure 4), indicating the presence of a dimeric or oligomeric population of the OX1 receptor in the DDM-solubilized fraction derived from the co-transfected cells that was not dissociated by detergent extraction.

Figure 2 DDM solubilizes a high proportion of VSV-G–OX1–eYFP

Flp-In™ T-REx™ HEK-293 cells harbouring VSV-G–OX1–eYFP were treated with (dox) or without (no dox) doxycycline for 24 h. Cell lysates were incubated with 5 nM [3H]SB-674042 with or without 3 μM SB-408124 to define total and non-specific binding. Samples were then treated as described in the Experimental section. A specific binding of [3H]SB-674042 of 100% corresponds to 3.7±0.08 pmol·(mg of protein)−1. Results shown are means±S.E.M.; four other experiments produced similar results.

Figure 3 BN-PAGE shows that VSV-G–OX1–eYFP migrates consistent with being predominantly a dimer

BN-PAGE was used to resolve proteins extracted from Flp-In™ T-REx™ HEK-293 cells able to express VSV-G–OX1–eYFP on demand with 0.5% DDM. (A) Samples were transferred on to a PVDF membrane and immunoblotted to detect the VSV-G tag, with two different exposures of the same samples shown. The migration of protein molecular mass markers is shown in the left-hand lane. (B) Prior to separation by BN-PAGE, samples from cells induced to express VSV-G–OX1–eYFP for 24 h had various concentrations of SDS added. Equivalent anti-VSV-G immunoreactivity is shown. (C) Equivalent BN-PAGE experiments were performed on cells treated with or without EGF and with and without 1% SDS before separation. The migration of the EGF receptor (EGFR) was then detected with a polyclonal EGF receptor antiserum following transfer of samples on to a PVDF membrane. Each element of the Figure was reproduced in at least three independent experiments. DOX, doxycycline; kD, kDa.

Figure 4 Luciferase enzyme complementation assays demonstrate that DDM solubilization does not generate artificial OX1 receptor dimers

HEK-293T cells were transiently transfected to express either VSV-G–OX1–Rluc8N or VSV-G–OX1–Rluc8C (left-hand bar) or were transfected to co-express this pairing (right-hand bar). Following solubilization using 0.5% DDM and sample preparation, Renilla luciferase activity was measured. Results shown are means±S.E.M. of six independent experiments.

The inducible nature of expression from the T-REx™ locus of Flp-In™ T-REx™ HEK-293 cells [11,15,25] allowed different amounts of VSV-G–OX1–eYFP to be expressed by varying the time of treatment with doxycycline (Figure 5). Ligand-binding studies employing the OX1 receptor antagonist [3H]SB-674042 allowed measurement of the total number of OX1 receptor-binding sites present at each time point (Figure 5A). This indicated that over a 48 h period of induction of VSV-G–OX1–eYFP, levels of the construct in cell lysates increased from being undetectable up to some 6 pmol·(mg of protein)−1. BN-PAGE was again used to resolve samples from cells induced for various times that were solubilized using 0.5% DDM and showed that across the full range of expression levels achieved the vast majority of the receptor construct migrated with an apparent mass consistent with a dimer (Figures 5B and 5C). As earlier, addition of 1% SDS to the samples prior to resolution resulted in a much greater proportion of the immunodetected protein migrating to a position consistent with a monomer (Figure 5B). Interestingly, although addition of orexin A (1 μM or 0.1 μM) prior to treatment with DDM had limited effects on the mobility of VSV-G–OX1–eYFP in BN-PAGE (Figure 6A), addition of either of the OX1 receptor antagonists SB-334867 and SB-408124 (1 μM) resulted in a significant proportion of VSV-G–OX1–eYFP migrating to the position of a monomer in BN-PAGE (Figure 6A). This effect of SB-334867 was produced in a concentration-dependent manner with pEC50=7.9±0.17 (Figures 6B and 6C), close to the reported binding affinity of this ligand for the OX1 receptor in preparations of membranes expressing this receptor [18].

Figure 5 VSV-G–OX1–eYFP is predominantly a dimer at a range of expression levels

Flp-In™ T-REx™ HEK-293 cells harbouring VSV-G–OX1–eYFP at the Flp-In™ locus were either not induced (0 h) or induced to express this polypeptide by treatment with doxycycline for various times. (A) Cell lysate samples were used to measure VSV-G–OX1–eYFP expression based on the specific binding of [3H]SB-674042. (B) Following 0.5% DDM solubilization, samples were resolved by BN-PAGE. In both cases, parental Flp-In™ T-REx™ HEK-293 (Flp-In) cells provided an extra ‘no expression’ control. In (B) samples of cells induced with doxycycline for 24 h were treated with buffer or 1% SDS before electrophoresis on a Blue-native gel. Quantification, based on densitometry, of the apparent monomer and dimer levels are shown in (C). Results shown are means±S.E.M.; each element of the Figure was derived from three independent experiments. kD, kDa.

Figure 6 OX1 antagonists, but not orexin A, modify mobility of VSV-G–OX1–eYFP on BN-PAGE

Samples were prepared for BN-PAGE from un-induced (-) and 24 h-doxycycline (DOX)-treated (+) Flp-In™ T-REx™ HEK-293 cells harbouring VSV-G–OX1–eYFP. Prior to solubilization with 0.5% DDM, cells were treated with vehicle or with orexin A (OxA; 1 μM or 0.1 μM), or with the OX1 antagonists SB-334867 or SB-408124 (1 μM) for 40 min. Samples were subsequently resolved by BN-PAGE and immunoblotted with an anti-VSV-G antibody. Experiments akin to those of (A) were performed (B) and quantified (C) after the addition of various concentrations of SB-334867. A further set of experiments produced similar results. kD, kDa.

The addition of large tags to transmembrane proteins may alter the avidity of the interaction and/or produce interaction artefacts. In order to exclude the possibility that the eYFP fused to the OX1 C-terminus was influencing substantially the monomeric/multimeric state of the OX1 receptor, we modified the construct by introducing a stop codon at the end of the OX1 sequence. The VSV-G–OX1 expression construct so produced was also used to generate a Flp-In™ T-REx™ HEK-293 cell line to allow inducible and controlled expression. Following induction, these cells were also subjected to ligand treatments and analysis by BN-PAGE (Figure 7A). Apart from enhanced mobility consistent with lack of the eYFP tag (Figure 7A) VSV-G–OX1 behaved in a similar manner to VSV-G–OX1–eYFP. Orexin A had limited effects, whereas SB-334867 again resulted in a higher proportion of the OX1 receptor migrating to a position consistent with that of a monomeric species. Treatment of these cells with various concentrations of SB-334867 also resulted in concentration-dependent alteration in the migration pattern of VSV-G–OX1 to favour the monomer, although the ligand appeared to be somewhat less potent (pEC50=7.33±0.28) at this construct (Figures 7B and 7C).

Figure 7 VSV-G–OX1 behaves similarly to VSV-G–OX1–eYFP on BN-PAGE

(A) Samples were prepared for BN-PAGE from un-induced (-) and 24 h-doxycycline (DOX)-induced (+) Flp-In™ T-REx™ HEK-293 cells harbouring VSV-G–OX1. Prior to solubilization with 0.5% DDM, cells were treated with vehicle or with orexin A (OxA; 1 μM), or with the OX1 antagonist SB-334867 (1 μM) for 40 min. A VSV-G–OX1–eYFP sample is also shown for comparison. Samples were subsequently resolved by BN-PAGE and immunoblotted with an anti-VSV-G antibody. As in Figure 6, equivalent experiments to those in (A) were performed using various concentrations of SB-334867 (B) and were quantified (C). Four separate experiments were performed with similar results. KD/kD, kDa.

It was at least conceptually possible that the apparent 180 kDa form of VSV-G–OX1–eYFP on BN-PAGE corresponded to the interaction of a monomeric VSV-G–OX1–eYFP receptor plus a second, undefined, protein with similar molecular mass. To assess this we established a further Flp-In™ T-REx™ HEK-293 cell line in which VSV-G–OX1–eYFP was located at the inducible locus, whereas HA-OX1-eYFP was expressed constitutively (Figure 8). In the absence of doxycycline treatment only HA–OX1–eYFP was detected, whereas following induction with doxycycline co-expression of the two forms of the OX1 receptor was confirmed by immunoblotting of BN-PAGE-resolved cell lysates with anti-HA or anti-VSV-G antibodies (Figure 8). Following extraction with 0.5% DDM and clearance by centrifugation of lysates from doxycycline-treated cells, immunoprecipitation was performed with anti-VSV-G antibody–agarose beads. Elution of the beads with the VSV-G peptide was followed by BN-PAGE and immunoblotting of the eluates. Anti-HA immunoreactivity was detected both at an apparent mass of 180 kDa and as higher-order species, but even with sustained exposure of blots, no protein consistent with a HA–OX1–eYFP monomer could be detected (Figure 8A). However, following addition of 1% SDS prior to resolution by BN-PAGE, the majority of the anti-HA immunoreactivity now migrated with an apparent mass of 90 kDa (Figure 8A). The presence of HA–OX1–eYFP migrating at 180 kDa following immunoprecipitation with anti-VSV-G antibody from DDM extracts of these cells is consistent with the 180 kDa anti-HA-immunoreactive species being an HA–OX1–eYFP/VSV-G–OX1–eYFP dimer, and this is largely disassembled by treatment with 1% SDS (Figure 8A). In equivalent studies, the eluates of the anti-VSV-G immunoprecipitation were also resolved by BN-PAGE and immunoblotted. VSV-G–OX1–eYFP was detected as both 180 kDa and 90 kDa species (Figure 8B), and addition of 1% SDS also greatly increased the relative proportion of the 90 kDa species (Figure 8B).

Figure 8 Co-immunoprecipitation studies show that the 180 kDa species is an OX1–eYFP dimer

Flp-In™ T-REx™ HEK-293 cells in which VSV-G–OX1–eYFP was located at the inducible locus while HA–OX1–eYFP was expressed constitutively were untreated or treated with doxycycline (DOX). Lysates of these cells were prepared. Following extraction of such lysates with 0.5% DDM and clearance by centrifugation, co-immunoprecipitation (CO-IP) was performed with anti-VSV-G antibody–agarose beads. These were subsequently eluted with VSV-G peptide. Samples of the lysates and eluates of the immunoprecipitates were treated with or without 1% SDS, resolved by BN-PAGE and immunoblotted to detect either the HA (A) or VSV-G (B) epitope tags. The longer exposure time in (A) is included to indicate that no monomeric HA–OX1–eYFP could be detected without treatment of the sample with SDS. Similar results were produced in three further experiments. kD, kDa.

To explore potential OX1 receptor dimerization in a further, distinct, cell system and with a different form of the receptor, we also generated a version of the OX1 receptor that was modified at the N-terminus by addition of an N-terminal leader sequence, derived from the metabotropic glutamate receptor 5, linked in-frame to the VSV-G epitope tag sequence and the 20 kDa SNAP-tag protein, which is based on mammalian O6-alkylguanine-DNA-alkyltransferase [26], to generate VSV-G–SNAP–OX1. Such SNAP-tagged constructs have recently been shown to be suitable to allow detection of cell-surface protein–protein interactions via htrFRET [14,15,26]. Flp-In™ T-REx™ HEK-293 cells harbouring VSV-G–SNAP–OX1 were also produced and induced with doxycycline. VSV-G–SNAP–OX1 is predicted to have a slightly lower molecular mass than VSV-G–OX1–eYFP and, indeed, this was noted when lysates from cells induced to express VSV-G–SNAP–OX1 were resolved by SDS/PAGE alongside those expressing VSV-G–OX1–eYFP (Figure 9A) and, following solubilization with 0.5% DDM, BN-PAGE (Figure 9B). Again, however, a high proportion of VSV-G–SNAP–OX1 also migrated to a position consistent with that of a dimeric species (Figure 9B). However, when compared with VSV-G–OX1—eYFP, there was evidence for a greater proportion of VSV-G–SNAP–OX1 migrating as apparent higher-order species (Figures 9B and 9C). As for VSV-G–OX1–eYFP, the mobility of a significant fraction of VSV-G–SNAP–OX1 was increased to that anticipated for a monomer following addition of 1% SDS prior to BN-PAGE, and the potentially higher-order complexes were completely removed by this treatment (Figure 9C). However, the effect of 1% SDS on the proportion of the 90 kDa species detected was not as marked as for VSV-G–OX1–eYFP. Whether with relatively limited expression of VSV-G–SNAP–OX1 induced with a low concentration of doxycycline, or a higher level of the receptor construct produced by a maximally effective concentration of doxycycline, addition of orexin A resulted in the appearance of much more pronounced levels of an apparently higher-order complex of VSV-G–SNAP–OX1 (Figure 10) and although not possible to detect at low-level expression, addition of either SB-334867 or SB-408124 to cells expressing higher levels of VSV-G–SNAP–OX1 resulted in a greater proportion of the receptor construct migrating in BN-PAGE to a position consistent with that of a monomer (Figure 10).

Figure 9 VSV-G–OX1–eYFP and VSV-G–SNAP–OX1 receptor behave similarly on BN-PAGE and SDS/PAGE

Lysates from Flp-In™ T-REx™ HEK-293 cells induced to express VSV-G–OX1–eYFP or VSV-G–SNAP–OX1 were resolved by (A) SDS/PAGE or (B) BN-PAGE and immunoblotted to detect the VSV-G tag. Expression controls were provided by un-induced (- DOX) Flp-In™ T-REx™ HEK-293 cells harbouring VSV-G–OX1–eYFP. (C) The effects of varying the concentration of SDS on the mobility of VSV-G–SNAP–OX1 in BN-PAGE was assessed. Experiments were repeated on more than three other sample sets. kD, kDa.

Figure 10 Regulation of VSV-G–SNAP–OX1 receptor migration in BN-PAGE by receptor ligands

Experiments akin to those of Figure 7 were performed following treatment of cells induced to express either low (1 ng·ml−1 doxycycline) or higher (100 ng·ml−1 doxycycline) VSV-G–SNAP–OX1 with either orexin A (OxA), SB-334867 or SB-408124 for 40 min. A second independent experiment produced similar results. dox, doxycycline; kD, kDa.

Following induction with doxycycline (10 ng·ml−1), cell-surface VSV-G–SNAP–OX1 was identified with the cell-impermeant dye SNAP 549 (Supplementary Figure S2A at Co-addition of a single concentration (10 nM) of SNAP-Lumi4-Tb as potential htrFRET energy donor and various concentrations of SNAP-Red as a potential htrFRET energy acceptor resulted in a bell-shaped htrFRET signal measured at 665 nm (Supplementary Figure S2B), consistent with the detection of cell-surface VSV-G–SNAP–OX1 dimers/oligomers. When using 10 nM SNAP-Lumi4-Tb, the optimal concentration of SNAP-Red was approximately 160 nM (Supplementary Figure S2B). At higher concentrations of SNAP-Red, the FRET signal decreased (Supplementary Figure S2B). This reflects that in this situation an increasing fraction of cell-surface VSV-G–SNAP–OX1 dimers will probably have bound a molecule of SNAP-Red to each protomer, limiting FRET (Supplementary Figure S2B). Optimal FRET is expected to be obtained when the maximal proportion of dimers bind a molecule of SNAP-Lumi4-Tb to one protomer and a molecule of SNAP-Red to the other protomer (Supplementary Figure S2B). Indeed, increasing concentrations of SNAP-Red outcompeted and prevented binding of SNAP-Lumi4-Tb as shown by the decline in SNAP-Lumi4-Tb signal at 620 nm that simply reports binding of this reagent (Supplementary Figure S2B). In the absence of induction of VSV-G–SNAP–OX1, no htrFRET signal was produced (Supplementary Figure S2B and Figure 11A), but such signals were generated following induction of VSV-G–SNAP–OX1 (Figures 11B–11D) with the extent of htrFRET signal dependent upon the extent of induction of the receptor (Figures 11B–11D). Although the basis and significance of this is uncertain, the basal 665/620 nm htrFRET signal declined slightly over a 60 min period (Figure 11). This was little affected by treatment with a concentration (1 μM) of the OX1 antagonists SB-334867 and SB-408124 expected from their measured Ki values to occupy the bulk of the receptor population (Figure 11). However, at lower receptor expression levels, the endogenous agonist orexin A produced a substantial increase in 665/620 nm htrFRET signal that was noted at the shortest time period monitored and maintained throughout a 60 min period (Figure 11). This is at least consistent with the agonist promoting the formation of higher-order receptor complexes as observed in the BN-PAGE studies (Figure 10) or altering the quaternary structure of the dimeric/oligomeric complexes.

Figure 11 At low expression levels cell-surface VSV-G–SNAP–OX1 dimer/oligomer htrFRET signals are enhanced by orexin A

htrFRET signals assessed as the 665/620 nm ratio were measured over time in Flp-In™ T-REx™ HEK-293 cells not induced (A) and induced to express VSV-G–SNAP–OX1 by treatment for 24 h with differing concentrations of doxycycline (dox) (BD). In cells expressing the construct, htrFRET signals were monitored over a 60 min period. These were essentially unaltered by the presence of the OX1 antagonists SB-334867 and SB-408124 (1 μM), but at low expression levels (B) were increased and subsequently maintained in the presence of orexin A (OxA; 1 μM). Results shown are means±S.E.M. and are representative sets taken from three independent experiments.


Although a large number of studies have reported results consistent with the idea that GPCRs can exists as dimers and/or higher-order oligomers [2,3] a number of issues remain unresolved. Among the most vexing of these include the proportion of any GPCR that is a dimer or oligomer at steady state and whether this might be regulated by receptor ligands. Although a substantial literature exists on the effects and lack of effects of ligands on the quaternary organization of GPCRs [27], much of it is difficult to analyse with clarity either because there is contradictory evidence or because a single approach was used to inform the conclusions.

A group of resonance energy transfer-based techniques have become a popular means to assess GPCR quaternary structure [14,28]. In the present study, each of the saturation-BRET experiments employing a pair of C-terminally eYFP- and Renilla luciferase-tagged forms of the OX1 receptor and htrFRET studies employing an N-terminally SNAP-tagged form of this receptor generated data consistent with the presence of at least a proportion of this receptor as a quaternary complex over the range of expression levels that could be controlled either by altering the amount of cDNA used in transient cellular transfections or through the use of an inducible expression system in cells stably harbouring forms of the OX1 receptor. However, despite these observations, it remains challenging to analyse and interpret such studies in terms of the proportion of the receptor within such complexes [29,30]. Furthermore, effects of ligands in such studies are also challenging to interpret. For example, although orexin A increased the htrFRET signal in cells expressing VSV-G–SNAP–OX1 this was only observed at relatively low receptor construct expression, in which situation the FRET signal consistent with the presence of OX1 dimers was also relatively low, and this was not observed following addition of OX1 receptor antagonists. This is similar to effects reported recently by Alvarez-Curto et al. [15] for the muscarinic M3 receptor using an equivalent approach. This may reflect that the monomer–dimer equilibrium is such that at low receptor expression levels a limited number of dimers are present, that this is increased with higher-level receptor expression that would favour such interactions and that agonists promote or stabilize the quaternary complex. However, there are other possible interpretations, including that such alterations in signal may reflect movement of the receptor towards clathrin-coated pits in preparation for internalization, for example, and hence a possible increase in potential ‘bystander’ resonance energy transfer effects [21,31] rather than a specific effect on the organization of receptor complexes. It is noteworthy that in recent times a series of studies have suggested both that GPCR–GPCR interactions may be dynamic [5,6,32] and that the extent of dimerization may vary significantly between even closely related receptors [5]. This may be defined by the affinity of different GPCR protomers for one another and, if so, implies that at equal expression densities the proportion of monomers/dimers/oligomers may be quite different for different receptors. Indeed, as ligand-induced movements of helices that have been implicated as GPCR-dimer interfaces have been detected [33], then ligands might, at least in certain cases, modify these interactions. It is important, however, to consider evidence to the contrary. Although FRAP microscopy studies have suggested that the β1-adrenoceptor may form a lower-affinity and more transient quaternary complex than the β2-adrenoceptor [5], saturation-BRET studies [31], in which similar so-called ‘BRET50’ values are observed for β1–β1-adrenoceptor interactions as for β2–β2-adrenoceptor interactions, have been interpreted as being consistent with these receptors having similar interaction affinities [31].

It is also noteworthy that in the absence of well-characterized antibodies for many GPCRs, experiments that explore organizational structure are highly reliant on molecularly modified forms of the receptors and, as in the case of a number of the experiments used in the present study, the tags used are themselves substantial proteins. These may generate artefacts by promoting or inhibiting receptor interactions and/or ligand-induced reorganization. Because of some of these issues, we decided additionally to address the proportion of OX1 receptor constructs present as quaternary complexes in a range of other ways. Importantly, for the BN-PAGE studies we demonstrated initially that much of the OX1 receptor could be solubilized from cell lysates using the detergent DDM in a form that retained binding of the antagonist [3H]SB-674042. Furthermore, in parallel studies employing enzyme complementation we also demonstrated that treatment with DDM did not result in non-specific aggregation of the OX1 receptor or, indeed, in the disassembly of pre-formed dimers. To do so we employed split-luciferase technology [16,34] in which two inherently non-functional fragments of Renilla luciferase that are able to recombine to generate a functional enzyme when brought into close proximity were linked to the C-terminal tail of forms of the OX1 receptor. Following co-transfection of these forms, solubilization with DDM resulted in soluble luciferase activity, whereas following separate transfections of each construct, solubilization followed by mixing did not reconstitute enzyme activity. These studies are akin to the ‘mixing’ controls used in many co-immunoprecipitation approaches to detect authentic GPCR dimers [35,36]. These are a key set of experiments because, although purified β2-adrenoceptors appear able to form dimers, and indeed potential tetramers, spontaneously when reconstituted into lipid bilayers [37], extraction from baculovirus-infected insect cells with 1% DDM is sufficient to promote monomerization of the receptor [38]. Indeed the capacity to solubilize receptors has been integral to efforts to obtain atomic-level structural information [39,40] and this has often resulted in monomerization of the receptor. However, this need not intrinsically be the case. For example, atomic-level structures of the CXCR4 chemokine receptor reveal a consistent homodimer configuration over a number of individual crystal forms [41] indicating that, at least in this specific case, combinations of protein engineering and the presence of a CXCR4 ligand was sufficient to preserve dimer contacts throughout solubilization with a combination of 0.5% DDM and 0.1% cholesteryl hemisuccinate [41]. It is also worthy of note, however, that the most notable effects of the agonist orexin A were observed when using the SNAP-tagged construct where a 20 kDa protein with enzymatic activity is placed on the extracellular face, and although this does not alter the binding affinity of the antagonist [3H]SB-674042 or of orexin A, it might alter the orientation of the receptor in response to binding the agonist.

BN-PAGE has recently been employed to study regulation of the dimeric status of the muscarinic M1 receptor and the effects of the MT7 toxin on this [42]. These studies indicated that the toxin either bound to and stabilized or favoured production of a dimeric form [42]. There are a range of potential issues with the use of BN-PAGE to study the quaternary structure of transmembrane proteins [22], not least in the case of GPCRs, because although this varies significantly between individual family members [43,44], detergent solubilization often results in denaturation and aggregation [43,44]. Initially, we therefore optimized the BN-PAGE procedure by performing preliminary studies with the single-transmembrane-pass EGF receptor. Following DDM solubilization, we demonstrated that this receptor migrated in BN-PAGE to a position anticipated for the monomer and that a substantial proportion of the receptor migrated to the position anticipated of a dimer following addition of EGF prior to solubilization with DDM. Importantly, this did not reflect non-specific aggregation or some irreversible modification because addition of 1% SDS to the sample prior to resolution by BN-PAGE restored the mobility of the EGF receptor to that anticipated for the monomer. Following DDM solubilization, VSV-G–OX1–eYFP migrated on BN-PAGE predominantly to the position predicted for a dimer. Importantly, after treatment with 1% SDS, a substantial proportion of VSV-G–OX1–eYFP now migrated to a position consistent with that of a monomer, suggesting that at least the bulk of the receptor was not aggregated and/or denatured, consistent with the ligand-binding studies. Although the extent of this effect varied somewhat between experiments, a number of key points should be noted. First, this occurred across a substantial range of expression levels as controlled by the extent of induction of the receptor constructs. Secondly, this effect of SDS was more pronounced when studying VSV-G–OX1–eYFP than the VSV-G–SNAP–OX1 construct. We do not have a ready explanation for this, but the SNAP tag is a recently developed means of introducing covalently linked labels into proteins [45,46] and rather little is known about its specific properties and any undesirable effects it might have in such studies. Thirdly, it was at least conceptually possible that much of the OX1 receptor constructs migrating to a position consistent with an OX1 receptor dimer on BN-PAGE might reflect a monomer of the OX1 receptor in association with an unidentified GPCR-interacting protein. To address this we generated cells expressing HA–OX1–eYFP constitutively but also able to express VSV-G–OX1–eYFP on demand. Following induction of VSV-G–OX1–eYFP, anti-VSV-G immunoprecipitation, elution and subsequent BN-PAGE, immunoblotting demonstrated the anti-HA immunoreactivity to migrate to the position of the predicted OX1–eYFP dimer. This must contain VSV-G–OX1–eYFP as well as HA–OX1—eYFP, and pre-addition of 1% SDS resulted in a large proportion of HA–OX1–eYFP now migrating to the position expected for this monomer. This provides clear support that the protein detected with lower mobility was a HA–OX1–eYFP+VSV-G–OX1–eYFP dimer.

Not all of the OX1 receptor construct that migrated in BN-PAGE to the position consistent with a receptor dimer was disaggregated by treatment with 1% SDS prior to resolution. It is certainly possible that some of this represents aggregated protein as it is well established that many GPCRs are not highly stable after detergent extraction [43,44]. However, it should also be noted that even in SDS/PAGE a substantial fraction of many GPCRs migrate with an apparent mass that is consistent with them behaving as an SDS-resistant dimer [47,48] and, indeed, in many studies this characteristic has been used to support other evidence of GPCR quaternary structure.

Interestingly, although the addition of orexin A to samples expressing VSV-G–OX1–eYFP had little noticeable effect on the mobility of the construct in BN-PAGE, the agonist produced a marked alteration in pattern when using VSV-G–SNAP–OX1. Now, a substantial proportion of the receptor migrated at a size consistent with production of a higher-order oligomer. Moreover, this was reversed entirely by addition of 1% SDS prior to resolution by BN-PAGE. This is at least consistent with agonist-mediated production of a large, reversible, quaternary complex. However, as this was not replicated when studying VSV-G–OX1–eYFP, it may indicate that the SNAP tag is at least partly responsible for the observations. As SNAP-tagged receptors have not been studied in this manner previously, it will be important for such possible effects to be monitored in future studies, not least because this emerging technology is becoming a popular approach [14,30]. As well as the effects of orexin A, addition of OX1 receptor antagonists increased the proportion of OX1 receptor constructs migrating in BN-PAGE to a position consistent with a receptor monomer, and this was the case for both VSV-G–SNAP–OX1 and VSV-G–OX1–eYFP. This may indeed indicate that binding of antagonists reduces the affinity of OX1 protomers for each other and modulates the equilibrium between receptor mono-mers and dimers. A second possibility, however, as receptor ligands are known to stabilize GPCR structure [49], is that the antagonists are limiting receptor denaturation. It was clear that this effect of OX1 receptor antagonists was concentration-dependent and that the EC50 for these effects was similar to the binding affinity of the ligands. Recent studies have suggested that rather than promoting monomerization, the M1-muscarinic-receptor-selective antagonist pirenzipine can promote dimerization [50], but there is no inherent reason to assume that the effects of ligands on such complexes will be consistent across family members and meaningful insight will only emerge with the publication of further studies. Although we did perform a series of densitometry studies, these were designed to provide information on the concentration of antagonist ligands that produced half-maximal effects in altering the apparent dimer–monomer equilibrium rather than being designed to provide absolute values for this ratio. Densitometry is susceptible to film-saturation artefacts and limited reproducibility between experiments. As such, we have avoided attempting to overinterpret these studies in quantitative terms.

The present study provides novel insight into the proportion of orexin OX1 receptors that exist as dimers and hint at a general means to assess this for GPCRs that are either inherently sufficiently stable to withstand denaturation upon solubilization or have been engineered to enhance such stability [43,44].


Tian-Rui Xu and Richard Ward generated reagents, performed most of the experiments and analysed the data. John Pediani performed the microscopy. Graeme Milligan devised the programme of work, contributed to experimental design and wrote the paper.


This work was supported by the Medical Research Council [grant number G0900050].


We thank Sarah Cumming (College of Medical, Veterinary and Life Sciences, University of Glasgow) for assistance with a series of preliminary experiments.

Abbreviations: BN-PAGE, Blue native-PAGE; BRET, bioluminescence resonance energy transfer; DDM, n-dodecyl-β-D-maltoside; DMEM, Dulbecco's modified Eagle's medium; EGF, epidermal growth factor; eYFP, enhanced yellow fluorescent protein; FRAP, fluorescence recovery after photobleaching; FRET, fluorescence resonance energy transfer; GPCR, G-protein-coupled receptor; HA, haemagglutinin; HBSS, Hanks balanced salt solution; HEK, human embryonic kidney; HEK-293T, cells, HEK-293 cells expressing the large T-antigen of SV40 (simian virus 40); htrFRET, homogeneous time-resolved FRET; Rluc8, Renilla luciferase 8; VSV, vesicular stomatitis virus


View Abstract