Purification and Functional Reconstitution of Monomeric μ-Opioid Receptors

Despite extensive characterization of the μ-opioid receptor (MOR), the biochemical properties of the isolated receptor remain unclear. In light of recent reports, we proposed that the monomeric form of MOR can activate G proteins and be subject to allosteric regulation. A μ-opioid receptor fused to yellow fluorescent protein (YMOR) was constructed and expressed in insect cells. YMOR binds ligands with high affinity, displays agonist-stimulated [35S]guanosine 5′-(γ-thio)triphosphate binding to Gαi, and is allosterically regulated by coupled Gi protein heterotrimer both in insect cell membranes and as purified protein reconstituted into a phospholipid bilayer in the form of high density lipoprotein particles. Single-particle imaging of fluorescently labeled receptor indicates that the reconstituted YMOR is monomeric. Moreover, single-molecule imaging of a Cy3-labeled agonist, [Lys7, Cys8]dermorphin, illustrates a novel method for studying G protein-coupled receptor-ligand binding and suggests that one molecule of agonist binds per monomeric YMOR. Together these data support the notion that oligomerization of the μ-opioid receptor is not required for agonist and antagonist binding and that the monomeric receptor is the minimal functional unit in regard to G protein activation and strong allosteric regulation of agonist binding by G proteins.

Opioid receptors are members of the G protein-coupled receptor (GPCR) 2 superfamily and are clinical mainstays for inducing analgesia. Three isoforms of opioid receptors, , ␦, and , have been cloned and are known to couple to G i/o proteins to regulate adenylyl cyclase and K ϩ /Ca ϩ ion channels (1)(2)(3). An ever growing amount of data suggests that many GPCRs oligomerize (4,5), and several studies have suggested that -opioid receptors (MORs) and ␦-opioid receptors heterodimerize to form unique ligand binding and G protein-activating units (6 -10). Although intriguing, these studies utilize cellular overexpression systems where it is difficult to know the exact nature of protein complexes formed between the receptors.
To study the function of isolated GPCRs, our laboratory and others have utilized a novel phospholipid bilayer reconstitution method (11)(12)(13)(14)(15)(16). In this approach purified GPCRs are reconstituted into the phospholipid bilayer of a high density lipoprotein (HDL) particle. The reconstituted HDL (rHDL) particles are monodispersed, uniform in size, and preferentially incorporate a GPCR monomer (14,15). Previous work in our lab has shown that rhodopsin, a class A GPCR previously proposed to function as a dimer (17)(18)(19), is fully capable of activating its G protein when reconstituted as a monomer in the rHDL lipid bilayer (15). Moreover, we have demonstrated that agonist binding to a monomeric ␤ 2 -adrenergic receptor, another class A GPCR, can be allosterically regulated by G proteins (14). This led us to determine whether a monomer of MOR, a class A GPCR that endogenously binds peptide ligands, is the minimal functional unit required to activate coupled G proteins. We additionally investigated whether agonist binding to monomeric MOR is allosterically regulated by inhibitory G protein heterotrimer.
To study the function of monomeric MOR we have purified a modified version of the receptor to near homogeneity. A yellow fluorescent protein was fused to the N terminus of MOR, and this construct (YMOR) was expressed in insect cells for purification. After reconstitution of purified YMOR into rHDL particles, single-molecule imaging of Cy3-labeled and Cy5-labeled YMOR determined that the rHDL particles contained one receptor. This monomeric YMOR sample binds ligands with affinities nearly equivalent to those observed in plasma membrane preparations. Monomeric YMOR efficiently stimulates GTP␥S binding to G i2 heterotrimeric G protein. G i2 allosteric regulation of agonist binding to rHDL⅐YMOR was also observed. Single-particle imaging of binding of [Lys 7 , Cys 8 ]dermorphin-Cy3, a fluorophore-labeled agonist, to rHDL⅐YMOR supports the notion that the rHDL particles contain a single YMOR. Taken together, these results suggest that a monomeric MOR is the minimal functional unit for ligand binding and G protein activation and illustrate a novel method for imaging ligand binding to opioid receptors.

EXPERIMENTAL PROCEDURES
Materials-G protein baculoviruses encoding rat G␣ i2 , His 6 -G␤ 1 , and G␥ 2 were provided by Dr. Alfred G. Gilman (University of Texas Southwestern, Dallas, TX). DNA encoding human -opioid receptor (NM 000914.2) was provided by Dr. John R. Traynor (University of Michigan, Ann Arbor, MI). Expired serum was generously provided by Dr. Bert La Du (University of Michigan, Ann Arbor, MI). Spodoptera frugiperda (Sf9) and Trichoplusia ni (HighFive TM ) cells, pFastBac TM baculovirus expression vectors and Sf900 TM serum-free medium were from Invitrogen. InsectExpress TM medium was purchased from Lonza (Allendale, NJ). N-Dodecyl-␤-D-maltoside was from Dojindo (Rockville, MD). All of the lipids were from Avanti Polar Lipids (Alabaster, AL). [ 3 H]diprenorphine (DPN, 54.9 Ci/mmol) and [ 35 S]GTP␥S (1250 Ci/mmol) were obtained from PerkinElmer Life Sciences. EZ-Link TM NHS-Biotin reagent was from Pierce. Ovomucoid trypsin inhibitor was purchased from United States Biological (Swampscott, MA). GF/B and BA85 filters, Cy3 and Cy5 NHSester mono-reactive dyes, Cy3-maleimide dye, and Source 15Q and Superdex 200 chromatography resins were from GE Healthcare. Talon TM resin was from Clontech. Bio-Beads TM SM-2 absorbant resin was from Bio-Rad. Chromatography columns were run using a BioLogic Duo-Flow protein purification system from Bio-Rad. Amicon Ultra centrifugation filters were from Millipore (Billerica, MA). Amino acids for ligand synthesis were obtained from Advanced ChemTech (Louisville, KY) or Sigma-Aldrich. All other chemicals and ligands were from either Sigma-Aldrich or Fisher Scientific.
YFP--Opioid Receptor Fusion Protein Expression and Purification-Baculoviruses were created using transfer vectors (pFastBac TM ) that encoded a fusion protein of an N-terminal cleavable hemagglutinin signal sequence (MKTIIALSYIF-CLVF), a FLAG epitope (DYKDDDD), a decahistidine tag, the monomeric and enhanced yellow fluorescent protein (Clontech), and the human MOR. High titer viruses (10 7 -10 8 plaque-forming units/ml) were used to infect Sf9 or High-Five TM suspension cultures at a multiplicity of infection of 0.25 to 1. FLAG-His 10 -mEYFP-MOR (YMOR) was expressed for 48 -52 h in the presence of 1 M naltrexone (NTX). The cells were resuspended in Buffer A (50 mM Tris⅐HCl, pH 8.0, 50 mM NaCl, 100 nM NTX, and protease inhibitors (3.2 g/ml leupeptin, 3.2 g/ml ovomucoid trypsin inhibitor, 17.5 g/ml phenylmethanesulfonyl fluoride, 16 g/ml tosyl-L-lysine-chloromethyl ketone (TLCK), 16 g/ml tosyl-L-phenylalanine chloromethyl ketone (TPCK)) and lysed by nitrogen cavitation. Supernatants from a 500 ϫ g spin were then subjected to 125,000 ϫ g spin for 35 min, and pellets containing membrane fractions were resuspended in Buffer A and either stored at Ϫ80°C for later binding assays or further processed to purify receptor.
All of the purification steps were performed sequentially and at 4°C or on ice. Membrane preparations were diluted to 5 mg/ml in Buffer A plus 1% DDM (w/v) and 0.01% cholesteryl hemisuccinate (CHS) (w/v) and gently stirred for 1 h to solubilize YMOR out of the membrane. Detergent extracted YMOR was enriched via Talon TM metal affinity chromatography resin (Clontech) in Buffer A plus 0.1% DDM and 0.01% CHS and eluted with Buffer A plus 0.1% DDM, 0.01% CHS, and 150 mM imidazole. Fractions containing YMOR (based on Coomassie staining after SDS-PAGE separation) were pooled, diluted 5-fold in Buffer B (20 mM Hepes, pH 8.0, 5 mM MgCl 2 , 0.1% DDM, 0.01% CHS, 100 nM NTX, and phenylmethanesulfonyl fluoride-TLCK-TPCK protease inhibitors), and eluted from a 1-ml Source 15Q strong anion exchange column with a 50 -300 mM NaCl linear gradient. Peak fractions were identified by radioligand binding assays using [ 3 H]DPN (2-4 nM). Peak fractions were then pooled and concentrated on Amicon Ultra centrifugation filters (10-kDa molecular mass cut-off). As a final purification step this YMOR sample was resolved based on size using a Superdex 200 gel filtration column (GE Healthcare) in Buffer C (20 mM Hepes, pH 8.0, 100 mM NaCl, 0.1% DDM, 0.01% CHS, 100 nM NTX, and protease inhibitors). Coomassie staining after SDS-PAGE was used to identify fractions containing YMOR, which were pooled and concentrated to ϳ1-2 M. Glycerol was added to a final concentration of 10% (v/v), and the samples were flash-frozen in liquid nitrogen and then stored at Ϫ80°C until further use.
In Vitro HDL Reconstitution-HDL particles were reconstituted according to previously reported protocols (14). Briefly, 21 mM sodium cholate, 7 mM lipids (1-palmitoyl-2-oleoyl-snglycero-3-phosphocholine (POPC) and 1-palmitoyl-2-oleoylsn-glycero-3-[phospho-rac-(1-glycerol)] (POPG) at a molar ratio of 3:2), purified YMOR (DDM-solubilized), and 100 M purified apoA-1 were solubilized in 20 mM Hepes, pH 8.0, 100 mM NaCl, 1 mM EDTA, 50 mM sodium cholate. The final concentration of YMOR varied from 0.2 to 0.4 M, but YMOR always comprised 20% of the total reconstitution volume. In some reconstitutions the lipid component was modified such that porcine polar brain lipid extract (Avanti Polar Lipids) was used in addition to POPC and POPG for a final concentration of 7 mM lipids at a molar ratio of 1.07:1.5:1 brain lipid:POPC: POPG. Following incubation on ice (1.5-2 h), the samples were added to Bio-Beads TM (Bio-Rad, 0.05 mg/ml of reconstitution volume) to remove detergent and to form HDL particles. Particles containing YMOR were purified via M1 anti-FLAG immunoaffinity chromatography resin (Sigma) and eluted with 1 mM EDTA plus 200 g/ml FLAG peptide. To assess the efficiency of HDL reconstitution, the total protein concentration of rHDL⅐YMOR was compared with the concentration of active reconstituted YMOR. FLAG affinity column-purified rHDL⅐YMOR was resolved from BSA and FLAG peptide on a Superdex 200 gel filtration column, and the peak fraction was analyzed for protein content with Amido Black staining (20), whereas [ 3 H]DPN saturation assays were used to measure active YMOR content. YMOR reconstituted into HDL was stored on ice until further use.
G Protein Addition to rHDL⅐YMOR Particles-G i2 heterotrimer (G␣ i2 -His 6 G␤ 1 -G␥ 2 ) was expressed in Sf9 cells and purified as previously described (21). The concentration of active G protein was determined by filter binding of 20 M [ 35 S]GTP␥S (isotopically diluted) in 30 mM NaHepes, pH 8.0, 100 mM NaCl, 50 mM MgCl 2 , 1 mM EDTA, 0.05% C 12 E 10 , and 1 mM dithiothreitol after 1 h of incubation at room temperature. Purified G protein heterotrimer was added to preformed, FLAG-purified rHDL⅐YMOR particles at a molar ratio of 1:10 receptor to G protein. G i2 was purified in 0.7% CHAPS, and the final volume of G protein added to rHDL⅐YMOR was such that the final CHAPS concentration was well below its critical micelle concentration. Reconstituted receptor and G protein samples were then incubated in Bio-Beads TM for 30 -45 min at 4°C to remove residual detergent. were measured for radioactivity on a liquid scintillation counter, and the data were fit with one-site saturation, one-site competition, or two-site competition binding models using Prism 5.0 (GraphPad, San Diego, CA).
[ 35 S]GTP␥S Binding Assay-100-l volume reactions were prepared containing 1 g of total membrane protein from YMOR expressing HighFive TM cells or ϳ50 -60 fmol of YMOR incorporated into rHDL particles in 30 mM Tris⅐HCl, pH 7.4, 100 mM NaCl, 5 mM MgCl 2 , 0.1 mM dithiothreitol, 10 or 1 M GDP (membranes or HDL particles), and 0.1% BSA. Membrane assays and rHDL assays were incubated with 10 nM isotopically diluted [ 35 S]GTP␥S (12.5 Ci/mmol). YMOR samples were incubated with increasing concentrations of agonists (1 pM to 1 mM) for 1 h at room temperature, then rapidly filtered through GF/B (membrane samples) or BA85 filters (HDL samples), and washed three times with 2 ml of ice-cold 30 mM Tris⅐HCl, pH 7.4, 100 mM NaCl, 5 mM MgCl 2 . The samples were measured for radioactivity on a liquid scintillation counter, and the data were fit with a log dose-response model using Prism 5.0.
Single-molecule Imaging of Reconstituted Cy3-and Cy5-YMOR-Purified YMOR (ϳ200 pmol) was incubated with NHS-ester Cy3 or Cy5 mono-reactive dye (GE Healthcare) in 20 mM Hepes, pH 8.0, 100 mM NaCl, 5 mM MgCl 2 , 6 mM EDTA, 0.1% DDM, 0.01% CHS, 100 nM NTX for 30 min at 25°C and then 1 h at 4°C. Conjugation reactions were quenched by the addition of Tris⅐HCl, pH 7.7 buffer (10 mM final). Cy-labeled YMOR was then separated from free dye using a 12-cm Sephadex G-50 column. The final dye to protein molar ratio was 3.8:1 for Cy3-YMOR and 3.7:1 for Cy5-YMOR, determined by absorbance at 280 nm for total protein, 550 nm for Cy3 dye, and 650 nm for Cy5 dye as measured on a NanoDrop ND-1000 spectrophotometer. A separate aliquot of YMOR was co-labeled with both Cy3 and Cy5 for final molar ratios of 2:1 Cy3: YMOR and 1.8:1 Cy5:YMOR. HDL reconstitutions of Cy3-YMOR alone, Cy5-YMOR alone, a mixture of Cy3-and Cy5-YMOR, and Cy3-Cy5-YMOR were then performed. Reconstitutions were performed as previously described, using apoA-1 that had been biotinylated at a 3:1 molar ratio using EZ-Link NHS-Biotin according to the manufacturer's protocol (Pierce). Reconstituted samples were diluted 5000-fold in 25 mM Tris⅐HCl, pH 7.7, and injected into a microfluidic channel on a quartz slide that was previously coated with biotinylated polyethylene glycol and treated with 0.2 mg/ml streptavidin to generate a surface density of ϳ0.05 molecules/m 2 . An oxygen scavenging system of 10 mM Trolox (Sigma-Aldrich), 100 mM protocatechuic acid, and 1 M protocatechuate-3,4-dioxygenase was included in the sample dilution (22). Following a 10-min incubation to allow binding of the biotin-HDL⅐Cy-YMOR complex to the streptavidin-coated slide, the channel was washed with ice-cold 25 mM Tris⅐HCl, pH 7.7, containing the oxygen scavenging system. An Olympus IX71 inverted microscope configured for prism-based total internal reflection fluorescence (TIRF) and coupled to an intensified CCD camera was used to image Cy3 and Cy5 fluorophores (CrystaLaser, 532 nm; Coherent CUBE laser, 638 nm; Chroma band pass filters HQ580/60 and HQ710/130 nm). Fluorophore intensity time traces were collected for 30 -100 s at 10 frames/s. Time traces were analyzed for photobleaching with in-house software (MatLab 7.0).

Synthesis of [Lys 7 , Cys 8 ]Dermorphin and Labeling with Cy3
Dye-[Lys 7 , Cys 8 ]dermorphin (Tyr-D-Ala-Phe-Gly-Tyr-Pro-Lys-Cys-NH 2 ) was synthesized on Rink resin (solid support) using an Applied Biosystems 431A peptide synthesizer and standard Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry. The samples were characterized on a Waters reverse phase HPLC using a Vydac C18 (Protein and Peptide) 10-micron column. The samples were run on a linear gradient of 0 -45% acetonitrile in an aqueous phase containing 0.1% trifluoroacetic acid at 35°C and monitored at 254 and 230 nm (supplemental Fig. S3). Peptides were similarly purified on a Waters semipreparative reverse phase HPLC using a Vydac C18 10-micron column at room temperature. The purified peptide was labeled with Cy3-maleimide (GE Healthcare) according to the manufacturer's instructions using a ratio of 1.5:1 peptide to fluorophore and repurified by HPLC as before. The labeled peptide was further purified via semi-preparative HPLC using a 5-micron Vydac C18 column as described above. The potency and efficacy of [Lys 7 , Cys 8 ]dermorphin-Cy3 at MOR were con-firmed in radiolabeled [ 3

Expression of a Functional
Allosteric regulation of agonist binding to opioid receptors by G proteins has been well established in plasma membrane preparations of brain homogenates and overexpression systems (26 -30) and was also observed for YMOR expressed in insect cells. In the absence of G i2 , DAMGO competed [ 3 H]DPN (0.5 nM) binding in a concentration-dependent manner with a K i of ϳ580 nM (Fig. 1D). In membranes expressing both YMOR and Gi 2 , DAMGO exhibited a biphasic mode of inhibition of [ 3 H]DPN binding with a K i hi of ϳ2.9 nM and a K i lo of ϳ1.5 M (fraction of K i hi ϳ0.53). The addition of 10 M GTP␥S to the YMOR ϩ Gi 2 membranes eliminated the high affinity DAMGO-binding site (K i ϭ ϳ300 nM), illustrating that YMOR is allosterically regulated by G proteins. Therefore, YMOR expressed in HighFive TM cells proved fully functional in regards to G protein coupling.
YMOR Purification-The capacity of a variety of detergents (zwitterionic, polar, and nonionic detergents such as Triton X-100, digitonin, Nonidet P-40, CHAPS, C 12 E 10 , and n-octyl-␤-glucoside) to solubilize YMOR from insect cell membranes was assessed by anti-FLAG Western blot analysis. Extraction of YMOR with DDM proved to be most efficient (data not shown). The addition of CHS (0.01% w/v) improved the stability of the solubilized receptor as measured by [ 3 H]DPN binding, consistent with the stabilizing effects of cholesterol moieties observed for solubilized ␤ 2 AR (31) (supplemental Fig. S1). The expression levels of YMOR and the efficiency of DDM solubilization, assessed by [ 3 H]DPN binding and anti-FLAG Western blots, were enhanced in the presence of naltrexone. Therefore this antagonist was present (100 nM) during all subsequent purification steps.
DDM-extracted YMOR was purified through a series of chromatographic steps including metal chelate (Talon TM ), anion exchange (Source 15Q), and size exclusion (Superdex 200) chromatography columns. Peak fractions from each column were determined by Coomassie staining, anti-FLAG antibody Western blotting, and [ 3 H]DPN binding. Fractions displaying the highest [ 3 H]DPN binding capacity and appropriate molecular weight were pooled and subjected to the next chromatographic step. Superdex 200 peak fractions were pooled, concentrated, flash-frozen in liquid N 2 , and then stored at Ϫ80°C until later use. A representative silver-stained SDS-PAGE separation of each enrichment step is shown in Fig. 2A. Yields of YMOR were typically ϳ50 g of Ͼ95% pure receptor/ liter of insect culture. YMOR, deemed pure based on silver staining, bound [ 3 H]DPN at ϳ20% of the predicted binding sites calculated from the molecular mass. This loss of activity is likely due to the detrimental freeze/thaw process.
Reconstitution of YMOR into High Density Lipoprotein Particles-The reconstitution of YMOR into high density lipoprotein (rHDL) particles was performed with a 500-fold molar excess of apoA-1 to YMOR (250-fold excess of rHDL) to favor reconstitution of a single YMOR molecule/rHDL particle. Resolution of rHDL particles containing YMOR (rHDL⅐ YMOR) with size exclusion chromatography (SEC) suggested an apparent Stokes diameter of 10.5 nm (Fig. 2B).

FIGURE 2. Purification of YMOR from HighFive TM insect cells and reconstitution into HDL particles.
A, YMOR was extracted from membranes in the presence of 100 nM naltrexone with 1% n-dodecyl-␤-D-maltoside and 0.01% cholesteryl hemisuccinate and enriched on a Talon TM metal affinity column. The Talon TM pool containing YMOR (predicted 73-kDa molecular mass) was applied to a Source 15Q anion exchange column followed by a size exclusion gel filtration column (Superdex 200). Samples of the purification steps were resolved by SDS-PAGE and silver-stained. YMOR was enriched to ϳ95% purity (GF peak). B, purified YMOR was reconstituted into HDL particles (see "Experimental Procedures") and resolved by size exclusion chromatography (Superdex 200). Fractions were analyzed for total protein content (UV absorbance), active YMOR ([ 3 H]DPN binding), and YFP fluorescence. UV absorbance showed a major peak corresponding to rHDL particles (stokes diameter of ϳ10.5 nm). The elution volume of active YMOR corresponded with the rising slope of the rHDL peak. YFP fluorescence eluted as two peaks, corresponding to the active YMOR and a larger YMOR aggregate that was not incorporated into rHDL particles. This aggregated YMOR was not active based on [ 3 H]DPN binding. The data were normalized such that the maximal value for each parameter was set to 100%. tional YMOR in particles that eluted slightly earlier than the main UV absorbance peak (Stokes diameter of ϳ10.3 nm), indicating the slightly larger size of rHDL⅐YMOR compared with rHDL. The YFP fluorescence of the SEC fractions indicated an additional peak that elutes near in the Superdex 200 void volume, suggestive of aggregated YMOR. Because these fractions did not bind [ 3 H]DPN, we surmised that this receptor population was inactive. This aggregated and inactive receptor is consistent with the previous observation that ϳ80% of the purified YMOR is inactive post freeze/thaw and prior to reconstitution into HDL.
The  (Fig. 2C). This significant disruption of high affinity ligand binding for the soluble receptor is not surprising, because detergents are known to decrease the ligand affinity of opioid receptors (32) and disrupt GPCRs in general (33). Replacement of detergents with phospholipids reverses the deleterious effects of the detergent and restores YMOR conformation to one that binds [ 3 H]DPN with native, membrane-bound affinity.
YMOR Is Monomeric When Incorporated into rHDL-To determine the number of YMOR molecules present in each rHDL particle, we assessed the degree of receptor co-localization between Cy3-and Cy5-labeled YMOR using single-particle imaging. Purified YMOR was labeled with Cy3-or Cy5reactive fluorescent dyes and reconstituted into HDL using biotinylated ⌬(1-43)-His 6 -apoA-1. This biotin-rHDL⅐Cy-YMOR complex was then incubated on a streptavidin-coated microfluidic slide and excited with 532-and 638-nm lasers. Fluorescence emission was detected using prism-based SM-TIRF. Reconstituted Cy3-YMOR and Cy5-YMOR were visualized as mono-disperse fluorescent foci (Fig. 3, A and B). When Cy3-YMOR and Cy5-YMOR were mixed prior to reconstitution (Cy3-YMORϩCy5-YMOR), a low level of co-localization of the two fluorophores (3.4%) was observed (Fig. 3, C and  E). A false-positive co-localization signal of 2.8 and 2.2% was observed for the Cy3-YMOR and Cy5-YMOR samples, respectively (Fig. 3E). As a positive control, co-localization for YMOR labeled with both Cy3 and Cy5 was also measured (Fig. 3D). Considering that our method does not allow for the differentiation between YMOR labeled with multiple Cy probes of the same fluorophore versus co-localization of multiple YMORs labeled with the same Cy probe, we may underestimate colocalization by as much as a factor of 2. Thus we estimate an upper limit for co-localization, i.e. incorporation of two or more YMORs into a single rHDL particle, of ϳ1.8% ((3.4% ϫ 2) Ϫ (2.8% ϩ 2.2%)). These data suggest that HDL reconstitution resulted in a sample containing mostly monomeric YMOR. Therefore the ligand binding and G protein coupling characteristics of the HDL-reconstituted YMOR presented here are indicative of the properties of a monomeric receptor.
To assess the amount of active versus inactive YMOR incorporated into rHDL, we compared the total protein concentration of purified rHDL⅐YMOR particles to the maximal concen-tration of YMOR based on [ 3 H]DPN saturation binding assays. Reconstituted YMOR was purified using FLAG affinity resin and resolved by SEC into a sample containing only apoA-1 and YMOR (based on SDS-PAGE and Coomassie staining). Amido Black staining of these peak fractions indicated a protein concentration of ϳ5.2 g/ml. Considering that the protein sample consists of two apoA-1 molecules (molecular mass, ϳ25,500 Da) and one YMOR (molecular mass, ϳ73,500 Da), the rHDL⅐YMOR sample has a total molecular mass of 124,500 Da and therefore a molar concentration of ϳ42 nM. Saturation binding assays on the rHDL⅐YMOR SEC peak indicated a molar concentration of ϳ31 nM for YMOR which bound ligand. Given the inherent limitations of protein detection assays at the low concentration of our samples, these data suggest that at least 75% of the YMOR in rHDL is active based on [ 3 H]DPN binding.
Monomeric YMOR Binds Antagonists with High Affinity-Reconstituted and monomeric YMOR binds antagonists with affinities similar to those observed in the plasma membrane. The antagonists naloxone and naltrexone and the -specific peptide CTAP (D-Phe-Cys-Tyr-D-Trp-Arg-Thr-Pen-Thr-NH 2 ) compete [ 3 H]DPN binding in similar fashion when in HighFive TM cell membrane preparations or in reconstituted Purified YMOR was labeled with Cy3 or Cy5 fluorescent dyes and reconstituted separately or together into biotin-labeled rHDL. The particles were imaged using single-molecule total internal reflection fluorescence microscopy on quartz slides coated with streptavidin. Representative overlay images of reconstituted rHDL⅐Cy3-YMOR (A), rHDL⅐Cy5-YMOR (B), rHDL⅐(Cy3-YMOR ϩ Cy5-YMOR) (C), and rHDL⅐Cy3-Cy5-YMOR (D) are shown. Quantification of Cy3 and Cy5 co-localization (E) showed that when both Cy3-YMOR and Cy5-YMOR were mixed together prior to reconstitution (mixture, 4c), only ϳ3.4% of rHDL particles contained two labeled receptors, compared with a false-positive co-localization signal of 2.8 and 2.2% observed for the rHDL⅐Cy3-YMOR and rHDL⅐Cy5-YMOR samples. YMOR co-labeled with both Cy3 and Cy5 was also imaged as a positive control for co-localization. Approximately 24% of rHDL⅐Cy3-Cy5-YMOR particles (co-labeled, 4d) exhibited colocalization, indicating that not all YMOR received a secondary fluorescent dye under the labeling conditions. SEPTEMBER 25, 2009 • VOLUME 284 • NUMBER 39 HDL particles (Fig. 4A). Therefore reconstitution into HDL allows YMOR to adopt a conformation that appears identical to receptor in plasma membranes in terms of high affinity antagonist binding.

Functional G Protein Coupling by -Opioid Receptor Monomer
Monomeric YMOR Functionally Couples Inhibitory G Proteins-We next analyzed the capacity of monomeric YMOR to functionally couple to heterotrimeric G protein. Functional measurements included allosteric regulation of agonist binding (Fig. 4B) and agonist-mediated stimulation of [ 35 S]GTP␥S binding (Fig. 4C). Anti-FLAG affinity column purification of rHDL⅐YMOR was performed to remove "empty" rHDL particles, and purified G i2 heterotrimer was added at a 10:1 G i2 : rHDL⅐YMOR molar ratio. The concentration of purified G protein was determined by [ 35 S]GTP␥S binding assays, and receptor concentration was measured with [ 3 H]DPN saturation assays of rHDL⅐YMOR samples. G i2 addition results in high affinity agonist binding (Fig. 4B). Morphine competed [ 3 H]DPN binding with a K i hi of ϳ1.7 nM and a K i lo of ϳ320 nM, whereas DAMGO inhibited with a K i hi of ϳ7.6 nM and a K i lo of ϳ1.8 M. As in membranes the addition of 10 M GTP␥S to the monomeric rHDL⅐YMORϩG i2 uncouples the G protein and results in a single-site, low affinity agonist competition curve. Similarly, when rHDL⅐YMOR was not coupled to G i2 , DAMGO competed [ 3 H]DPN binding with micromolar affinity (supplemental Fig. S2).
Although G i2 was added to rHDL⅐YMOR at a 10:1 molar ratio in these reconstitutions, the observed high and low agonist affinities illustrate that not all YMOR was coupled to G i2 . In fact, the agonist competition assays indicate that ϳ54% of the receptor was coupled to G i2 . These data are consistent with previous studies on the ␤ 2 AR where the addition of purified G protein heterotrimer to rHDL particles in the absence of detergents resulted in 90 -95% G protein loss, largely because of aggregation (14). Indeed, SEC analysis of rHDL⅐YMOR before and after G i2 addition confirmed that a large amount of the heterotrimer aggregates during its addition (data not shown). As such, the two populations of ϳ54% coupled and ϳ46% uncoupled YMOR correspond well with the expected final G protein to YMOR ratio of 0.5-1:1.

DISCUSSION
Intense research over the past decade has focused on the existence and functional consequences of GPCR oligomerization. Recently, our laboratory and others have taken advantage of a unique phospholipid bilayer platform in the form of rHDL particles (13)(14)(15)(16).
Using this system we previously demonstrated the functional reconstitution of two prototypical GPCRs, rhodopsin and ␤ 2 AR (14,15). Having illustrated that these receptors are monomeric following reconstitution, we observed complete functional coupling to their respective G proteins. For the ␤ 2 AR, agonist binding to monomeric receptor is allosterically regulated by the stimulatory heterotrimeric G protein G␣ s ␤␥. Therefore GPCR dimerization is not required for functional G protein coupling in the prototypical class A GPCRs rhodopsin and ␤ 2 AR.
In contrast, the accumulation of considerable biochemical and biophysical evidence suggests that the -, ␦-, and -opioid receptors may function in a fashion that is dependent on their oligomerization. Analysis of receptor overexpression in co-transfected cells suggested the existence of ␦-␦ homodimers (34) and demonstrated unique pharmacology resulting from -␦ (6, 7, 10) and ␦heterodimers (35). Bioluminescence resonance energy transfer studies have shown that homo-and heterodimers can be formed by all three opioid receptor isoforms (8). These findings have promoted the notion that opioid receptor oligomerization is important for function (36,37). However, with co-transfection systems the final profiles of and ␦ monomers, homodimers, and heterodimers are unknown. In light of these reports, we rationalized that it was vital to determine whether the monomeric form of the MOR behaves in a pharmacologically similar manner as the native, membrane-bound form. The HDL reconstitution system was used to investigate the function of monomeric MOR. Reconstitution of MOR required the purification of active receptor in large enough quantities for subsequent biochemical manipulation and analysis. Previous reports of MOR purification from endogenous or recombinant sources have yielded either low quantities (38 -42) or poor agonist binding affinities (43)(44)(45). Several aspects were key to the expression and purification of YFP-MOR (YMOR): a cleavable hemagglutinin signal sequence at the N terminus of the receptor (46), the presence of naltrexone during expression, and the inclusion of cholesteryl hemisuccinate and naltrexone during the entire purification process. These modifications contrib- FIGURE 5. Single-molecule imaging of Cy3-labeled agonist binding to rHDL⅐YMOR ؉ G i2 confirms that the majority of YMOR is monomeric when reconstituted into HDL. Binding of [Lys 7 , Cys 8 ]dermorphin-Cy3, a fluorescently labeled MOR-specific agonist, to rHDL⅐YMORϩG i2 was observed with prism-based total internal reflection fluorescence microscopy. YMOR was reconstituted into HDL particles using biotinylated apoA-1, followed by G i2 heterotrimer addition at a 30:1 G protein to receptor molar ratio. Reconstituted HDL (A) or rHDL⅐YMORϩG i2 (B) were then incubated with a saturating concentration (500 nM) of [Lys 7 , Cys 8 ]dermorphin-Cy3, adhered to a streptavidin-coated quartz slide, and washed with ice-cold 25 mM Tris⅐HCl, pH 7.7 buffer. Bound [Lys 7 , Cys 8 ]dermorphin-Cy3 was continuously excited at 532 nm to observe photobleaching of the fluorophore. C, representative fluorescence intensity traces for a one-and two-step photobleach event are shown. The arrows indicate photobleach events. D and E, quantification of photobleaching showed that ϳ95% of the bound [Lys 7 , Cys 8 ]dermorphin-Cy3 bleached in a single step, suggesting that 95% of rHDL particles contained monomeric YMOR. [Lys 7 , Cys 8 ]dermorphin-Cy3 binding was reversible, as shown the addition of 5 M NTX. A minimum of four slide regions and 200 fluorescent spots were counted for each sample.
uted to the stabilization of YMOR, leading to an increase in yields during the chromatography process as well as increasing the specific activity of the detergent solubilized receptor. Reconstitution into HDL facilitated the isolation of a monomeric form of YMOR that couples to and activates heterotrimeric G proteins. Furthermore, we illustrate that the inhibitory G protein G␣ i ␤␥ can allosterically regulate agonist binding to monomeric MOR. These results suggest that G protein regulation of monomeric receptors is likely a phenomenon common to all G proteins and class A GPCRs. The capacity of MOR to couple to various isoforms of G i/o heterotrimers in a differential manner are currently under investigation in our laboratory.
Taken together our results demonstrate the functionality of monomeric MOR, illustrating that the opioid GPCR does not require dimerization to bind ligands and signal to G proteins. However, these data do not refute the existence of opioid receptor oligomerization in cells. It remains possible that homo-and heterodimerization create unique ligand binding entities that may be subject to differential regulation, desensitization, and internalization. One of our future goals is to isolate various oligomeric states of opioid receptors and other GPCRs in rHDL particles and compare their activities toward ligands and signaling partners directly. A major technical obstacle is the formation of "anti-parallel" receptor dimers, where the N termini for the reconstituted GPCRs in the oligomer are on opposite sides of the phospholipid bilayer. Indeed, reconstitution of two GPCRs in a single rHDL particle has been previously demonstrated for rhodopsin (13,16), but a detailed analysis by nanogold labeling and single-particle electron microscopy of the reconstituted "dimers" reveals that a significant fraction were anti-parallel (16). It is plausible, and even likely, that an antiparallel GPCR dimer will result in suboptimal or disrupted G protein coupling. The development of approaches to ensure parallel GPCR dimer incorporation into rHDL particles to confidently study the functional relevance of opioid receptor dimerization is currently a major priority in the laboratory.
HDL reconstitution of YMOR also provided a platform for analysis of ligand binding using single-molecule microscopy. In this study we examined the reversible binding of a -opioid receptor specific agonist [Lys 7 , Cys 8 ]dermorphin (47) with SM-TIRF. To the best of our knowledge these data represent the first reported observation of a peptide agonist binding to an isolated GPCR in a lipid bilayer using single-particle imaging. We are currently refining our methods to resolve the kinetics of ligand binding to YMOR, as well as visualizing ligand binding to opioid receptors in intact cells. Peptide receptors such as the opioid family are particularly amenable to fluorescence spectroscopy because the relatively large ligand size can tolerate the incorporation of fluorophores without drastically impairing binding affinities. This single-molecule visual approach of studying ligand binding to opioid receptors, utilizing labeled agonists and antagonists with unique binding affinities toward receptor homo-and heterodimers, may potentially address opioid receptor oligomerization in physiologically relevant tissue preparations.