- Email: [email protected]
Oligomerization of opioid receptors
Methods 27 (2002) 358–365 www.academicpress.com
Oligomerization of opioid receptors Ivone Gomes, Julija Filipovska, Bryen A. Jordan,1 and Lakshmi A. Devi* Department of Pharmacology, New York University School of Medicine, MSB 408, 550 First Avenue, New York, NY 10016, USA Accepted 4 June 2002
Abstract Opioid receptors belong to the family of G-protein-coupled receptors characterized by their seven transmembrane domains. The activation of these receptors by agonists such as morphine and endogenous opioid peptides leads to the activation of inhibitory Gproteins followed by a decrease in the levels of intracellular cAMP. Opioid receptor activation is also associated with the opening of Kþ channels and the inhibition of Ca2þ channels. A number of investigations, prior to the development of opioid receptor cDNAs, suggested that opioid receptor types interacted with each other. Early pharmacological studies provided evidence for the probable interaction between opioid receptors. More recent studies using receptor selective antagonists, antisense oligonucleotides, or animals lacking opioid receptors further suggested that interactions between opioid receptor types could modulate their activity. We examined opioid receptor interactions using biochemical, biophysical, and pharmacological techniques. We used differential epitope tagging and selective immunoisolation of receptor complexes to demonstrate homotypic and heterotypic interactions between opioid receptor types. We also used the proximity-based bioluminescence resonance energy transfer assay to explore opioid receptor–receptor interactions in living cells. In this article we describe the biochemical and biophysical methods involved in the detection of receptor dimers. We also address some of the concerns and suggest precautions to be taken in studies examining receptor–receptor interactions. Ó 2002 Elsevier Science (USA). All rights reserved. Keywords: Immunoprecipitation; Western blotting; Crosslinking; Bioluminescence resonance energy transfer
1. Biochemical techniques to study opioid receptor oligomerization To demonstrate receptor–receptor interactions using biochemical methods different epitope-tagged receptors (Flag- and myc-tagged versions of the same receptor) are expressed in heterologous cells. After membrane solubilization myc-tagged receptors are immunoprecipitated using anti-myc polyclonal antibodies; the immunoprecipitates are subjected to separation by SDS–polyacrylamide gel electrophoresis (SDS–PAGE). The associated receptors in the complex are visualized by Western blot with anti-Flag monoclonal antibodies. A signal is detected in the blots only if there is an association between the myc- and Flag-tagged receptors.
*
Corresponding author. Fax: +212-263-7133. E-mail address: [email protected] (L.A. Devi). 1 Present address: Department of Biochemistry, NYU School of Medicine, 550 First Avenue, New York, NY 10016, USA.
1.1. Experimental protocols 1.1.1. Transient transfection and crosslinking We have typically used opioid receptors that have been tagged at the N termini with Flag, myc, or HA epitope tags. Human embryonic kidney 293 (HEK-293) or COS cells are used for the transient expression of these receptors since these cells are easy to grow, easy to transfect, express heterologous proteins at fairly high levels, and do not normally express opioid receptors. To transiently express these receptors we routinely use calcium phosphate-mediated transfection. In this method a calcium phosphate–DNA precipitate is formed by slowly mixing Hepes-buffered saline with a solution containing calcium chloride and the DNA to be transfected. This precipitate adheres to the surface of the cells to be transfected and ‘‘glycerol shock’’ increases the amount of DNA absorbed by the cells. HEK-293 or COS cells (American Type Culture Collection, Manassas, VA) are grown in the growth medium [Dulbecco’s modified Eagle’s medium (DMEM,
1046-2023/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved. PII: S 1 0 4 6 - 2 0 2 3 ( 0 2 ) 0 0 0 9 4 - 4
I. Gomes et al. / Methods 27 (2002) 358–365
Gibco-BRL, Gaithersburg, MD) containing 10% fetal bovine serum (FBS; Gibco-BRL) and 1% penicillin– streptomycin (Gibco-BRL)]. Cells from a confluent plate are split in a 1:5 ratio into 10-cm dishes (BectonDickinson, Falcon) the day before transfection. Care must be taken to ensure that the cells are well separated from each other. On the day of the transfection medium is replaced with fresh growth medium preferably 2 h prior to transfection. The calcium phosphate–DNA precipitate is generated as follows: 500 ll of 2 Hepesbuffered saline (HeBS: 16.4 g NaCl, 11.9 g Hepes acid, 0.21 g Na2 HPO4 ), the pH of which has been adjusted to exactly 7.05 with 5 M NaOH, is filter sterilized and placed in a 15-ml sterile centrifuge tube. In another tube 50 ll of filter sterilized 2.5 M calcium chloride is mixed with 10–50 lg of DNA (ethanol precipitated to sterilize) and sterile distilled water is added to a final volume of 500 ll. The DNA/CaCl2 solution is added dropwise to the HeBS solution with continuous stirring and mixed for another 5 s. The precipitate is allowed to form for 20 min at room temperature; it should have a fine sandy appearance when viewed under the microscope. The precipitate is distributed evenly (dropwise) over the cells and gently mixed with the medium. The cells are incubated for 4 h or overnight at 37 °C and the medium is removed. To ‘‘glycerol shock’’ 2 ml of 10% sterile glycerol (in culture medium) is added to the cells. Following a short incubation (3 min) at room temperature 5 ml of PBS is added and mixed, and the solution is removed. The cells are washed twice with 5 ml of PBS and replaced with complete medium. The cells are collected for analysis after 48–72 h. For crosslinking analysis, cells are chilled to 4 °C and incubated with crosslinkers in 2% dimethyl sulfoxide (DMSO) in phosphate-buffered saline (PBS) at 4 °C; 5 mM dithiobis(succinimidyl propionate) (DSP, Pierce, Rockford, IL) for 30–60 min, 5 mM bis[b-(4-azidosalicylaminido)ethyl]disulfide (BASED, Pierce) for 2–15 min, or 5 mM N-5-azido-2-nitrobenzoyloxysuccinimide (ANB-NOS, Pierce) for 1–3 min. The reaction is terminated by incubation with 50 mM Tris–Cl, pH 7.4, for 15 min. The cells are washed with ice-cold PBS prior to solubilization of membranes (described below). When examining the effect of agonists on the level of dimers, cells are incubated with either different doses of ligands for 10 or 30 min or with the same dose of the ligand for different periods in growth medium at 37 °C. At the end of the incubation period, cells are washed with cold PBS and chilled to 4 °C prior to crosslinking and/or solubilization. 1.1.2. Determining the level of opioid receptor expression To determine the level of expression of opioid receptors in the transiently transfected cells we use radioligand binding assays. A universal opioid antagonist, [3 H]diprenorphine, is used to determine the total level of
359
opioid receptors and selective radiolabeled ligands (e.g., DAMGO for l, Deltorphin II or DPDPE for d and U69593 for j receptors) are used to determine the expression levels of specific types of receptors. Typically, 5 105 transfected cells are suspended in 100 ll of 50 mM Tris–Cl pH 7.4. Then 100 ll of [3 H]diprenorphine in 50 mM Tris–Cl, pH 7.4, is added at different concentrations (0.1–10 nM) to determine Bmax (maximal amount of specific binding). Nonspecific binding is determined in the presence of unlabeled diprenorphine (1 lM final concentration). Cells are incubated with ligands for 1 h at 37 °C. The bound radiolabeled ligands are collected on Whatman (Clifton, NJ) GF-B filters, and filters are washed three times with ice-cold 50 mM Tris–Cl, pH 7.4. The bound radioactivity is determined following an overnight incubation of filters in scintillation fluid. The Bmax and IC50 values are determined from displacement curves using GraphPad Prism 2.0. The protein concentration is estimated using bovine serum albumin and BCA assay reagent (Pierce). 1.1.3. Lysis of cells and solubilization of membranes A number of different buffer combinations can be used to solubilize proteins from membranes. There are two important considerations in the choice of solubilization buffer: one is the efficient release of the protein and the other is retention of its physical and functional properties. The conditions used for solubilization should be as harsh as possible to ensure the quantitative release of the protein of interest while avoiding solubilization of background proteins and as gentle as possible so as to retain most of its functional properties including recognition of the receptor by the antibody. Variables that can drastically affect the solubilization of proteins are salt concentration, type of detergent, presence of bivalent cations, and pH. In addition, many extraction procedures can cause the release of proteases into the solubilization buffer. Since protease digestion can become a problem during some of the functional analyses care should be taken to minimize its effects. This is done by maintaining the samples at 4 °C as much as possible and supplementing the solubilization buffer with protease inhibitors. We routinely use a protease inhibitor cocktail (Sigma Chemical, St. Louis, MO) in our solubilization and immunoprecipitation buffers. In addition, 100 mM iodoacetamide, a capping agent that S-carboxymethylates cysteines, is used in all stages of sample preparation to prevent the artifactual associations by free sulfhydryl groups that could be exposed during solubilization. For the isolation of receptor–receptor interacting complexes we have used different types of solubilization buffers: (i) Buffer G (50 mM Tris–Cl, pH 7.4, containing 300 mM NaCl, 1% Triton X-100, 10% glycerol, 1.5 mM MgCl2 , and 1 mM CaCl2 ); (ii) RIPA buffer (50 mM Tris–Cl, pH 8, containing 150 mM NaCl, 1% Nonidet
360
I. Gomes et al. / Methods 27 (2002) 358–365
P-40 (NP-40), 0.5% deoxycholate, 0.1% sodium dodecyl sulfate (SDS), and 1 mM CaCl2 ); (iii) NP-40 solubilization buffer (10 mM Tris–Cl, pH 8, containing 1% NP-40, 150 mM NaCl, 1 mM EDTA, 10% glycerol, and 1 mM CaCl2 ); (iv) CHAPS buffer (0.5% in 50 mM Tris–Cl, pH 7.4; (v) dodecyl maltoside solubilization buffer (0.5% dodecyl b-maltoside in 50 mM Tris–Cl, pH 7). To solubilize receptors from whole cells the following protocol is used: Approximately 48–72 h after transfection cells are washed twice with 5 ml of PBS at room temperature. The dish is placed on ice and 1.5 ml of prechilled solubilization buffer containing protease inhibitors (15 ll), 1 mM phenylmethylsulfonyl fluoride (PMSF), and 10 mM iodoacetamide is added to each plate. The cells are collected using a rubber policeman into an Epperdorf tube. Following incubation for 60 min at 4 °C in a rocking shaker the extract is subjected to centrifugation at 15,000g for 20 min at 4 °C and the supernatant is transferred to a fresh Eppendorf tube. The protein concentration is estimated as described above. 1.1.4. Immunoprecipitation During immunoprecipitation, tagged receptors are isolated from the mixture of proteins in the detergentsolubilized cell lysate/membranes by means of a specific antibody directed against the tag on one of the receptors. The antibody in the immunocomplex is then allowed to bind to protein A–Sepharose beads and the unbound proteins are removed by washing the beads; this leaves the antibody–receptor complex bound to the beads. The immunoprecipitated material bound to the protein A beads is then subjected to further analysis by size fractionation using SDS–polyacrylamide gel electrophoresis (SDS–PAGE) and visualization of the second receptor in the complex using antibody directed against its epitope tag. Polyclonal antibodies are commonly used for immunoprecipitation because they generally bind to multiple sites on the antigen and therefore have a greater avidity for the antigen. They have another advantage in that the interaction is stronger, leading to very stable antigen–antibody–protein A complexes, which could be subjected to relatively harsh washing procedures. However, the disadvantage to using these antibodies is a possible higher background due to binding to nonspecific proteins. This could be prevented by preclearing the cell lysates with normal serum (rabbit serum) and/or protein A beads. A detailed protocol for the immunoprecipitation of epitope-tagged opioid receptors is as follows: Approximately 150 lg of the solubilized protein is placed in an Eppendorf tube, about 1 lg (5 ll) of c-myc polyclonal antibody (200 lg=ml, Santa Cruz Biotechnology) and 12 ll of protease inhibitor cocktail (Sigma Chemical) are added, and the volume is adjusted to 1.2 ml with cold solubilization buffer. The mixture is incubated overnight
on a rocker at 4 °C. Protein A beads (Sigma) are equilibrated in the solubilization buffer and 100 ll of a 50% slurry (protein A in buffer) is added to each tube. Following incubation at 4 °C on a rocker for 2 h the mixture is subjected to centrifugation at 14,000 rpm for 1 min at 4 °C. The beads are washed three times with 500 ll of solubilization buffer containing protease inhibitors (supernatant removed from the immunoprecipitate with the help of an insulin syringe). Sample buffer (2X) with or without 50 mM dithiothreitol (DTT) is added and incubated at 60 °C for 15 min before subjecting the sample to SDS–PAGE and Western blotting (see below). 1.1.5. Cell surface labeling and immunoprecipitation HEK cells transiently expressing epitope-tagged receptors are washed with ice-cold PBS and incubated with 5 lg=ml anti-Flag antibody for 2 h at 4 °C. Cells are washed and lysed with the solubilization buffer (Buffer G) containing protease inhibitor cocktail for 1 h at 4 °C. The lysate is incubated with 15 ll of protein A–Sepharose 4B beads for 16 h at 4 °C. The immunoprecipitates (containing cell surface receptors) are collected, washed twice with Buffer G, and eluted with 30 ll of nonreducing sample buffer. Ten microliters of the eluate are subjected to SDS–PAGE/Western blotting analysis with anti-myc antibody as described below. 1.1.6. Western blot analysis This technique, when used in combination with immunoprecipitation, is a powerful tool that can be used for the detection of the second receptor in the complex and to study the specific interactions between proteins. Proteins separated by gel electrophoresis are transferred to nitrocellulose membranes and the membrane is incubated with a known protein or protein mixture to eliminate nonspecific binding of proteins to the membrane. The specific receptor is then localized using a primary antibody against the unique epitope tag on the second receptor followed by a secondary antibody conjugated to horseradish peroxidase. The sensitivity of the Western blotting procedure depends on the type of detection method used and the amount of antigen. Also, it is important that the antibodies can recognize denatured epitopes since sample preparation and electrophoresis conditions are likely to denature the epitope. One of the major advantages in using monoclonal antibodies for immunoblotting lies in the specificity since they bind to only one epitope and thus they provide an elegant tool for the identification of a particular region of the antigen. The major disadvantage is the possibility of cross-reactivity with other closely related proteins since the epitope usually comprises only four or five amino acids. The detailed protocol used for the visualization of receptor–receptor complexes is as described below: Approximately 70 ll of 2 sample buffer (120 mM
I. Gomes et al. / Methods 27 (2002) 358–365
Tris–Cl, pH 6.8, containing 4% SDS, 20% glycerol and 0.002% bromphenol blue with or without 50 mM DTT) is added to the immunoprecipitates (described in the previous section). Urea (8 M) could be added to the sample buffer and/or to the gel to disrupt oligomeric arrays and sharpen the specific bands. The samples are incubated at 60 °C for 15 min (it is important not to subject the complexes to higher temperatures, especially boiling, since this may lead to nonspecific aggregation of proteins). The samples are briefly cooled and about 10–15 ll is subjected to SDS–PAGE on 8% gels at 90 V until the dye front reaches the bottom of the gel. Urea gels are preferred when greater separation of protein is sought. The separated proteins are transferred to Protran nitrocellulose membranes (Schleicher & Schuell, Keene, NH) in a transfer buffer (composed of 25 mM Tris, 192 mM glycine, 20% methanol) for 16 h at 30 V. The membranes are briefly rinsed in TBS (50 mM Tris– Cl, pH 7.4, containing 150 mM NaCl), stained for 1 min with Ponceau S (Sigma) to visualize the bands, and washed with 20 ml TTBS (TBS containing 0.05% Tween 20) to remove Ponceau S stain. Following overnight incubation at 4 °C with 25 ml of 5% nonfat dried milk in TTBS to block nonspecific binding the membranes are rinsed in TTBS and incubated with 25 ml anti-Flag monoclonal antibody [1:500; 8 lg=ml in TTBS containing 5% protease-free bovine serum albumin and 0.01% sodium azide, (Sigma)] for 2 h at room temperature in a shaker. The membranes are washed six times (5 min each) with 20–25 ml TTBS and incubated with 25 ml anti-mouse IgG conjugated to horseradish peroxidase (1:5000 dilution in 5% nonfat dried milk in TTBS, Vector Laboratories, Burlingame, CA) for 1–2 h at room temperature in a shaker. Following incubation, the membranes are subjected to six washes (5 min each) with 20–25 ml TTBS. Approximately 8 ml of SuperSignal Chemiluminiscent Substrate (Pierce) is added. Following an incubation of 5 min, the membranes are exposed to X-ray films (Eastman Kodak, Rochester NY) for different periods to obtain multiple exposures that differ in the intensity of signal.
361
receptor interaction does not depend on a high level of receptor expression [1,2]. Another concern with the immunoprecipitation studies is the possibility of artifactual receptor aggregation during solubilization/immunoprecitipation procedures due to the inherent hydrophobic nature of GPCRs; this is addressed with a number of controls. One of the important controls is to mix cells individually expressing one or the other tagged receptor. These mixed cells are then subjected to the same solubilization and immunoprecipitation conditions as the cells expressing both epitope-tagged receptors. Under these conditions if the dimers are observed only in cells coexpressing both receptors (and not in the mixed cells) this would imply that the association between proteins is not the result of artifactual aggregation. Finally, a variety of agents can be used to study the biochemical properties of the receptor oligomers. (a) Detergents that differ in their physical properties have been used to help address a concern with detergent-induced association [2]. We have previously used a variety of combinations of different detergents to study the hydrophobic nature of opioid receptor oligomers [3]. (b) Capping agents that carboxymethylate free sulfhydryls (on cysteines) that are exposed during solubilization of membranes have been used to examine nonselective associations [4]. We have used iodoacetamide in solubilization and immunoprecipitation buffers, especially following treatment of cells with reducing agents, to address the involvement of disulfide bonds in the associations between receptors [2,5,6]. (c) Crosslinking agents have been used to examine detergent sensitive associations; these allow for stringent conditions of washing during the isolation of receptor complexes. A variety of crosslinking agents that differ in their functional properties have been used to investigate the nature of associations of a number of cell surface proteins including GPCRs [1,7]. Therefore, when appropriate controls are used immunoprecipitation can provide valuable information on receptor–receptor associations.
1.2. Concerns and controls
2. Biophysical techniques to monitor opioid receptor oligomerization in living cells
A concern with the biochemical studies is the level of expression of receptors in transiently transfected cells that can reach ‘‘nonphysiological’’ levels. We have addressed this issue by generating stable cell lines that express the interacting receptors at 1:1 ratio and at various levels, i.e., low (100,000 receptors/cell), medium (500,000 receptors/cell), and high (2,000,000 receptors/cell); the receptor numbers are determined by binding assays. In all three types of cell lines we observed coimmunoprecipitation of the receptors, albeit to varying extents, supporting the notion that the receptor–
As discussed above, biochemical studies have been used to demonstrate that opioid receptors can physically associate with each other to form dimers/oligomers [3,8]. Furthermore, pharmacological analysis has suggested that these complexes may play an important role in ligand binding, signaling, and/or internalization of opioid receptors on ligand binding [3,8]. Although crucial in establishing the concept of dimerization for opioid receptors and G-protein-coupled receptors in general, biochemical studies do not allow examination of receptor–receptor interactions or their modulation by
362
I. Gomes et al. / Methods 27 (2002) 358–365
ligands in live cells. Therefore we have recently employed bioluminescence resonance energy transfer (BRET), a proximity-based biophysical method, to examine receptor–receptor associations under physiological conditions in living cells. This assay is based on a naturally occurring phenomenon in which the energy generated by a luminescent donor, Renilla luciferase (Rluc) interacting with its substrate coelenterazine, is transferred to a fluorescence acceptor, a mutant form of green fluorescent protein (GFP). The energy transfer can occur only if the donor and the acceptor are in close ), thus allowing examination proximity (less than 100 A of close interactions between the donor- and acceptortagged opioid receptors (Fig. 1). 2.1. Experimental protocols 2.1.1. Preparation of tagged receptors We chose to tag opioid receptors with Rluc as the donor and mutant forms of GFP (yellow fluorescent protein, YFP, or enhanced green fluorescent protein,
EGFP) as the acceptor. Coelenterazine h (Molecular Probes) is used as an Rluc substrate. The emission maximum (470 nm) of coelenterazine h oxidized by Rluc is sufficient to excite YFP at one of its two excitation peaks (470 and 514 nm). The emission maximum of YFP at 530 nm allows good separation of light emissions from the donor and the acceptor. This increases the signal-to-noise ratio at the acceptor emission wavelength and thus increases the ability to detect BRET signal, i.e., receptor–receptor interactions with lower levels of protein expression. EGFP that has an excitation maximum at 488 nm can also be used as an acceptor. However, this acceptor molecule emits maximally at 509 nm, which partially overlaps with the emission spectrum of oxidized coelenterazine h. Thus, only very strong or close interactions can be detected due to the low signal-tonoise ratio at this wavelength (compare Figs. 2A and B). The donor Rluc or acceptor YFP is genetically fused to the C termini of opioid receptors to allow for membrane expression of the fusion proteins. A set of vectors for Rluc fusion (pRluc-N1, -N2, and -N3) can be ob-
Fig. 1. Principle of bioluminescence resonance energy transfer assay. Upper panels: Schematic representation of seven transmembrane opioid receptors fused to Renilla luciferase (Rluc) and yellow fluorescent protein (YFP). When they are far apart the light from the luminescent donor Rluc cannot excite the fluorescent acceptor YFP (A). If two differentially tagged receptors interact and are brought close together the acceptor is excited and emits light at 530 nm (B). Lower panels: Typical spectra obtained in the absence (A) and in the presence (B) of receptor–receptor interactions with a peak at 470 nm (A) and peaks at 470 and 530 nm (B).
I. Gomes et al. / Methods 27 (2002) 358–365
363
Fig. 2. Intensity and resolution of the BRET assay depend on the type of acceptor. Magenta curve: Light emission spectrum with cells expressing d opioid receptors fused to Renilla luciferase (dluc) and yellow fluorescent protein (dYFP) (A) or enhanced green fluorescent protein (dEGFP) (B). The dark blue curve represents the spectrum obtained with cells expressing dLuc alone.
tained from Packard Bioscience and the YFP fusion vector (pEYFP-N1) from Clontech. To minimize the possibility that Rluc or YFP will interfere with the normal functioning of the opioid receptors the entire sequence of the opioid receptor’s C terminus is preserved. Only the stop codon is mutated and fused inframe with the donor or acceptor molecule. A small intervening sequence between the C terminus of the opioid receptor and Rluc or YFP of 6–10 amino acids is introduced to allow for relative flexibility of the donor and acceptor. To maximize the interactions between opioid receptor molecules tagged with donor and acceptor tags and consequently the BRET signal, we chose HEK-293 cells that do not express endogenous opioid receptors and have high levels of trimeric G-proteins. This allows proper coupling of ectopically expressed opioid receptors to the cell’s signaling machinery. A transfection method that maximizes the fraction of cells cotransfected with the tagged opioid receptors should be used. In our case that is the calcium phosphate transfection. The level of expression of the tagged opioid receptors influences the ability to detect a BRET signal and also the specificity of interaction. This is determined by binding of radiolabeled ligands (described previously). A balance between high levels of expression required for BRET signal detection and relatively low levels of expression, as seen under physiological conditions, that allow specific interactions should be sought. This is achieved by using different amounts and ratios of Rluc and YFP-tagged receptors. Specificity of the BRET signal could be confirmed by competition with specific (opioid receptors) and nonspecific (other seven-transmembrane proteins) untagged competitors whose level of expression can be estimated by ligand binding assays.
2.1.2. Receptor expression and the BRET assay The protocol for transient transfection of HEK cells is essentially as described in Section 1.1. The amounts and ratio of plasmid vectors expressing Rluc and YFP tagged opioid receptors are optimized in each individual case; this differs for different combinations of opioid receptors. The ratio becomes very important when examining the formation of heterodimers between opioid receptors that are also known to homodimerize. We typically use 0:25 lg of each recombinant plasmid DNA and incubate cells with the DNA–calcium phosphate precipitate for 16–18 h. The cells are collected 36–72 h posttransfection and used for analysis. The level of receptor expression is determined using the radioligand binding assay as described earlier (Section 1.1). For BRET assay, about 36 h after transfection cells are detached from plates by incubation with PBS containing 1 mM EDTA for 2 min, collected by centrifugation, washed with PBS, and resuspended to 1–2 106 cells=ml with PBS containing 1 mM EDTA. Cells (2 ml) are placed in a cuvette, coelenterazine is added to 5 lM final concentration, and the suspension is quickly mixed with a pipette. Light emission is immediately monitored from 420 to 590 nm at 5-nm intervals for 0.5 s using a FluoroMax-2 spectrometer (Jobin Yvon-Spex Instruments S.A.) with closed excitation slit. A peak at about 470 nm generated on the addition of coelenterazine is detected in all samples expressing luciferase-tagged opioid receptors (Fig. 2). A second peak around 530 nm is detected in samples where Rluc-tagged receptors are coexpressed with YFP-tagged receptors (Fig. 2A) and at 509 nm with EGFP-tagged receptors (Fig. 2B). Alternatively, instead of monitoring light emission for the entire spectrum, the peak emissions at 470 and 530 nm (if YFP is used as an acceptor) or at
364
I. Gomes et al. / Methods 27 (2002) 358–365
509 nm (if EGFP is used as an acceptor) are monitored. In this case a spectrophotometer (BRET counter) with appropriate filters can be used to measure light emissions. In cases where light emission is monitored from 470 to 530 nm, the BRET signal can be represented as the difference between the area under the emission spectrum of the cotransfected cells (Fig. 2) and that of the luciferase-tagged opioid receptor transfected alone (Fig. 2). In other cases where light emission is monitored only at 470 and 530 nm the BRET signal is represented as BRET ratio. This is calculated as the ratio of the emission at 530 nm to the emission at 470 nm obtained when Rluc-tagged receptors and YFP-tagged receptors are coexpressed, standardized to the ratio when Rluc-tagged receptor is expressed alone [4]. 2.2. Concerns and controls The BRET assay can give very valuable information about the proximity and likelihood of interactions of different proteins in the context of live cells. Therefore, this assay excludes the possible effect of different detergents and membrane preparations on protein–protein interactions. However, certain concerns should be addressed because it relies on transient transfections and the ectopic expression of proteins from strong viral promoters. One important issue that is always being considered when examining protein–protein interactions in transiently transfected cells is the level of expression of the proteins of interest. Very often the level of protein expression that generates the best signal-to-noise ratio for the assay is not necessarily its ‘‘physiological level’’ of expression. To optimize the signal-to-noise (BRET to luciferase) ratio as well as the specificity of interactions a
variety of transfection protocols and expression conditions that allow for optimal expression of fusion proteins should be considered. The level of expression should be quantiated by independent methods (ligand binding assay). To maximize BRET signal, i.e., donor– acceptor associations as compared with donor–donor and acceptor–acceptor combinations, we have varied the levels of Rluc to GFP/YFP fusion proteins. As demonstrated in Fig. 3A in the case of d–d interactions, increasing the level of the acceptor dEGFP gives a stronger BRET signal (Fig. 3A). The increase was significant when a 1:3 instead of a 1:1 donor:acceptor ratio was tested (compare red and magenta curves in Fig. 3A). This could be due to the increased probability of donor– acceptor association when increasing the concentration of one of the components. Further increase in the acceptor molecules does not significantly increase the BRET signal (compare magenta and green curves in Fig. 3A). It is possible that increasing acceptor expression leads to increased acceptor–acceptor associations (that do not contribute to BRET signal). It is also possible that increasing the protein expression leads to nonspecific interactions driven by mass action. To examine if the total level of protein expressed affected the BRET signal we varied the level of receptor expression while keeping the donor:acceptor ratios constant (Fig. 3B). Our results show that under these conditions the BRET signal does not vary even with a five fold increase in the level of expression; the level of receptor expression varied from 200 to 1000 fmol/mg of protein (determined by radioligand binding) in these studies. In Fig. 3B a 1:3 donor:acceptor ratio is presented; however, similar results were obtained with a 1:1 ratio (data not shown). Another important consideration especially when examining receptor heterodimerization is the differences
Fig. 3. Determination of optimal conditions for specific BRET signal. Homotypic association of d (A) and l (B) opioid receptors monitored by BRET in live cells. (A) The level of BRET signal with varying ratios of donor and acceptor fusion proteins: red (1:1), magenta (1:3), and green (1:4). (B) The level of BRET signal while varying the total amount lluc þ lEGFP DNA while keeping the ratio constant (1:3): magenta (4 lg), and green (20 lg). The dark blue curve represents the spectrum obtained with cells expressing lluc alone. Numbers correspond to micrograms of each DNA used for transfection.
I. Gomes et al. / Methods 27 (2002) 358–365
365
3. Summary
Fig. 4. Interaction (heterodimerization) of d and j opioid receptors. A BRET signal (peak at 509 nm) is detected when dluc is used as a donor and jEGFP as an acceptor (magenta curve) as well as when jluc is the donor and dEGFP the acceptor (green curve). The dark blue curve represents the spectrum obtained with cells expressing dluc alone. Similar results were obtained with jluc (not shown).
in the relative efficiency of expression of each of the fusion proteins. We have addressed this by using both combinations (jluc-dEGFP as well as dluc-jEGFP) of receptors. As shown in Fig. 4, association of j and d opioid receptors can be detected with either jRluc as donor and d-EGFP as acceptor or dRluc as donor and jEGFP as acceptor. These data further support our biochemical and pharmacological data that explored j–d receptor interactions [2]. To determine the specificity of the interactions between different opioid receptors fused to Rluc and YFP, a competition assay with untagged opioid receptors and other unrelated proteins is used [9]. The use of unrelated proteins that do not interact with opioid receptors under the same transfection conditions ideally serves as a negative control. Since such an experiment would involve simultaneous transfection of three different expression vectors the relative levels of expression of the proteins are difficult to control. The simplest controls universally used are the Rluc or YFP vectors (without the recombinant protein) in combination with Rluc- or YFP-tagged opioid receptors respectively.
We have used classic biochemical methods and modern biophysical methods to examine interaction of opioid receptors [10]. By coimmunoprecipitation studies we have shown that opioid receptors form homotypic as well as heterotypic associations in the absence of the agonist treatment. Results from BRET assays have extended these observations and demonstrated that the receptor–receptor association can be seen in live cells. This is very important since biochemical studies require solubilization of membranes with different detergents; this, in addition to disrupting the cellular environment of the receptors, may also affect their association of other regulatory proteins. BRET assays monitor interactions within the dynamic environment of the cell membrane, allowing us to assess modulation of these interactions by different ligands as a crucial aspect of the functioning of opioid receptors.
Acknowledgments This work was supported in part by National Institutes of Health Grants DA 08863 and DA 00458 (to L.A.D.).
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10]
S. Cvejic, L.A. Devi, J. Biol. Chem. 272 (1997) 26959–26964. B.A. Jordan, L.A. Devi, Nature 399 (1999) 697–700. L.A. Devi, Trends Pharm. Sci. 22 (2001) 532–537. F.-Y. Zeng, J. Wess, J. Biol. Chem. 274 (1999) 19487–19497. I. Gomes, B.A. Jordan, A. Gupta, N. Trapaidze, V. Nagy, L.A. Devi, J. Neurosci. 20 (2000) RC1108. B.A. Jordan, N. Trapaidze, I. Gomes, R. Nivarthi, L.A. Devi, Proc. Natl. Acad Sci. USA 98 (2001) 343–348. H. Hastrup, A. Karlin, J.A. Javitch, Proc. Natl. Acad. Sci. USA 98 (2001) 10055–10060. B. Jordan, L.A. Devi, B. J. Anaesth. 81 (1998) 12–19. S. Angers, A. Salahpour, M. Bouvier, Life Sci. 68 (2001) 2243– 2250. I. Gomes, B.A. Jordan, A. Gupta, C. Rios, N. Trapaidze, L.A. Devi, J. Mol. Med. 79 (2001) 226–242.
Oligomerization of opioid receptors Ivone Gomes, Julija Filipovska, Bryen A. Jordan,1 and Lakshmi A. Devi* Department of Pharmacology, New York University School of Medicine, MSB 408, 550 First Avenue, New York, NY 10016, USA Accepted 4 June 2002
Abstract Opioid receptors belong to the family of G-protein-coupled receptors characterized by their seven transmembrane domains. The activation of these receptors by agonists such as morphine and endogenous opioid peptides leads to the activation of inhibitory Gproteins followed by a decrease in the levels of intracellular cAMP. Opioid receptor activation is also associated with the opening of Kþ channels and the inhibition of Ca2þ channels. A number of investigations, prior to the development of opioid receptor cDNAs, suggested that opioid receptor types interacted with each other. Early pharmacological studies provided evidence for the probable interaction between opioid receptors. More recent studies using receptor selective antagonists, antisense oligonucleotides, or animals lacking opioid receptors further suggested that interactions between opioid receptor types could modulate their activity. We examined opioid receptor interactions using biochemical, biophysical, and pharmacological techniques. We used differential epitope tagging and selective immunoisolation of receptor complexes to demonstrate homotypic and heterotypic interactions between opioid receptor types. We also used the proximity-based bioluminescence resonance energy transfer assay to explore opioid receptor–receptor interactions in living cells. In this article we describe the biochemical and biophysical methods involved in the detection of receptor dimers. We also address some of the concerns and suggest precautions to be taken in studies examining receptor–receptor interactions. Ó 2002 Elsevier Science (USA). All rights reserved. Keywords: Immunoprecipitation; Western blotting; Crosslinking; Bioluminescence resonance energy transfer
1. Biochemical techniques to study opioid receptor oligomerization To demonstrate receptor–receptor interactions using biochemical methods different epitope-tagged receptors (Flag- and myc-tagged versions of the same receptor) are expressed in heterologous cells. After membrane solubilization myc-tagged receptors are immunoprecipitated using anti-myc polyclonal antibodies; the immunoprecipitates are subjected to separation by SDS–polyacrylamide gel electrophoresis (SDS–PAGE). The associated receptors in the complex are visualized by Western blot with anti-Flag monoclonal antibodies. A signal is detected in the blots only if there is an association between the myc- and Flag-tagged receptors.
*
Corresponding author. Fax: +212-263-7133. E-mail address: [email protected] (L.A. Devi). 1 Present address: Department of Biochemistry, NYU School of Medicine, 550 First Avenue, New York, NY 10016, USA.
1.1. Experimental protocols 1.1.1. Transient transfection and crosslinking We have typically used opioid receptors that have been tagged at the N termini with Flag, myc, or HA epitope tags. Human embryonic kidney 293 (HEK-293) or COS cells are used for the transient expression of these receptors since these cells are easy to grow, easy to transfect, express heterologous proteins at fairly high levels, and do not normally express opioid receptors. To transiently express these receptors we routinely use calcium phosphate-mediated transfection. In this method a calcium phosphate–DNA precipitate is formed by slowly mixing Hepes-buffered saline with a solution containing calcium chloride and the DNA to be transfected. This precipitate adheres to the surface of the cells to be transfected and ‘‘glycerol shock’’ increases the amount of DNA absorbed by the cells. HEK-293 or COS cells (American Type Culture Collection, Manassas, VA) are grown in the growth medium [Dulbecco’s modified Eagle’s medium (DMEM,
1046-2023/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved. PII: S 1 0 4 6 - 2 0 2 3 ( 0 2 ) 0 0 0 9 4 - 4
I. Gomes et al. / Methods 27 (2002) 358–365
Gibco-BRL, Gaithersburg, MD) containing 10% fetal bovine serum (FBS; Gibco-BRL) and 1% penicillin– streptomycin (Gibco-BRL)]. Cells from a confluent plate are split in a 1:5 ratio into 10-cm dishes (BectonDickinson, Falcon) the day before transfection. Care must be taken to ensure that the cells are well separated from each other. On the day of the transfection medium is replaced with fresh growth medium preferably 2 h prior to transfection. The calcium phosphate–DNA precipitate is generated as follows: 500 ll of 2 Hepesbuffered saline (HeBS: 16.4 g NaCl, 11.9 g Hepes acid, 0.21 g Na2 HPO4 ), the pH of which has been adjusted to exactly 7.05 with 5 M NaOH, is filter sterilized and placed in a 15-ml sterile centrifuge tube. In another tube 50 ll of filter sterilized 2.5 M calcium chloride is mixed with 10–50 lg of DNA (ethanol precipitated to sterilize) and sterile distilled water is added to a final volume of 500 ll. The DNA/CaCl2 solution is added dropwise to the HeBS solution with continuous stirring and mixed for another 5 s. The precipitate is allowed to form for 20 min at room temperature; it should have a fine sandy appearance when viewed under the microscope. The precipitate is distributed evenly (dropwise) over the cells and gently mixed with the medium. The cells are incubated for 4 h or overnight at 37 °C and the medium is removed. To ‘‘glycerol shock’’ 2 ml of 10% sterile glycerol (in culture medium) is added to the cells. Following a short incubation (3 min) at room temperature 5 ml of PBS is added and mixed, and the solution is removed. The cells are washed twice with 5 ml of PBS and replaced with complete medium. The cells are collected for analysis after 48–72 h. For crosslinking analysis, cells are chilled to 4 °C and incubated with crosslinkers in 2% dimethyl sulfoxide (DMSO) in phosphate-buffered saline (PBS) at 4 °C; 5 mM dithiobis(succinimidyl propionate) (DSP, Pierce, Rockford, IL) for 30–60 min, 5 mM bis[b-(4-azidosalicylaminido)ethyl]disulfide (BASED, Pierce) for 2–15 min, or 5 mM N-5-azido-2-nitrobenzoyloxysuccinimide (ANB-NOS, Pierce) for 1–3 min. The reaction is terminated by incubation with 50 mM Tris–Cl, pH 7.4, for 15 min. The cells are washed with ice-cold PBS prior to solubilization of membranes (described below). When examining the effect of agonists on the level of dimers, cells are incubated with either different doses of ligands for 10 or 30 min or with the same dose of the ligand for different periods in growth medium at 37 °C. At the end of the incubation period, cells are washed with cold PBS and chilled to 4 °C prior to crosslinking and/or solubilization. 1.1.2. Determining the level of opioid receptor expression To determine the level of expression of opioid receptors in the transiently transfected cells we use radioligand binding assays. A universal opioid antagonist, [3 H]diprenorphine, is used to determine the total level of
359
opioid receptors and selective radiolabeled ligands (e.g., DAMGO for l, Deltorphin II or DPDPE for d and U69593 for j receptors) are used to determine the expression levels of specific types of receptors. Typically, 5 105 transfected cells are suspended in 100 ll of 50 mM Tris–Cl pH 7.4. Then 100 ll of [3 H]diprenorphine in 50 mM Tris–Cl, pH 7.4, is added at different concentrations (0.1–10 nM) to determine Bmax (maximal amount of specific binding). Nonspecific binding is determined in the presence of unlabeled diprenorphine (1 lM final concentration). Cells are incubated with ligands for 1 h at 37 °C. The bound radiolabeled ligands are collected on Whatman (Clifton, NJ) GF-B filters, and filters are washed three times with ice-cold 50 mM Tris–Cl, pH 7.4. The bound radioactivity is determined following an overnight incubation of filters in scintillation fluid. The Bmax and IC50 values are determined from displacement curves using GraphPad Prism 2.0. The protein concentration is estimated using bovine serum albumin and BCA assay reagent (Pierce). 1.1.3. Lysis of cells and solubilization of membranes A number of different buffer combinations can be used to solubilize proteins from membranes. There are two important considerations in the choice of solubilization buffer: one is the efficient release of the protein and the other is retention of its physical and functional properties. The conditions used for solubilization should be as harsh as possible to ensure the quantitative release of the protein of interest while avoiding solubilization of background proteins and as gentle as possible so as to retain most of its functional properties including recognition of the receptor by the antibody. Variables that can drastically affect the solubilization of proteins are salt concentration, type of detergent, presence of bivalent cations, and pH. In addition, many extraction procedures can cause the release of proteases into the solubilization buffer. Since protease digestion can become a problem during some of the functional analyses care should be taken to minimize its effects. This is done by maintaining the samples at 4 °C as much as possible and supplementing the solubilization buffer with protease inhibitors. We routinely use a protease inhibitor cocktail (Sigma Chemical, St. Louis, MO) in our solubilization and immunoprecipitation buffers. In addition, 100 mM iodoacetamide, a capping agent that S-carboxymethylates cysteines, is used in all stages of sample preparation to prevent the artifactual associations by free sulfhydryl groups that could be exposed during solubilization. For the isolation of receptor–receptor interacting complexes we have used different types of solubilization buffers: (i) Buffer G (50 mM Tris–Cl, pH 7.4, containing 300 mM NaCl, 1% Triton X-100, 10% glycerol, 1.5 mM MgCl2 , and 1 mM CaCl2 ); (ii) RIPA buffer (50 mM Tris–Cl, pH 8, containing 150 mM NaCl, 1% Nonidet
360
I. Gomes et al. / Methods 27 (2002) 358–365
P-40 (NP-40), 0.5% deoxycholate, 0.1% sodium dodecyl sulfate (SDS), and 1 mM CaCl2 ); (iii) NP-40 solubilization buffer (10 mM Tris–Cl, pH 8, containing 1% NP-40, 150 mM NaCl, 1 mM EDTA, 10% glycerol, and 1 mM CaCl2 ); (iv) CHAPS buffer (0.5% in 50 mM Tris–Cl, pH 7.4; (v) dodecyl maltoside solubilization buffer (0.5% dodecyl b-maltoside in 50 mM Tris–Cl, pH 7). To solubilize receptors from whole cells the following protocol is used: Approximately 48–72 h after transfection cells are washed twice with 5 ml of PBS at room temperature. The dish is placed on ice and 1.5 ml of prechilled solubilization buffer containing protease inhibitors (15 ll), 1 mM phenylmethylsulfonyl fluoride (PMSF), and 10 mM iodoacetamide is added to each plate. The cells are collected using a rubber policeman into an Epperdorf tube. Following incubation for 60 min at 4 °C in a rocking shaker the extract is subjected to centrifugation at 15,000g for 20 min at 4 °C and the supernatant is transferred to a fresh Eppendorf tube. The protein concentration is estimated as described above. 1.1.4. Immunoprecipitation During immunoprecipitation, tagged receptors are isolated from the mixture of proteins in the detergentsolubilized cell lysate/membranes by means of a specific antibody directed against the tag on one of the receptors. The antibody in the immunocomplex is then allowed to bind to protein A–Sepharose beads and the unbound proteins are removed by washing the beads; this leaves the antibody–receptor complex bound to the beads. The immunoprecipitated material bound to the protein A beads is then subjected to further analysis by size fractionation using SDS–polyacrylamide gel electrophoresis (SDS–PAGE) and visualization of the second receptor in the complex using antibody directed against its epitope tag. Polyclonal antibodies are commonly used for immunoprecipitation because they generally bind to multiple sites on the antigen and therefore have a greater avidity for the antigen. They have another advantage in that the interaction is stronger, leading to very stable antigen–antibody–protein A complexes, which could be subjected to relatively harsh washing procedures. However, the disadvantage to using these antibodies is a possible higher background due to binding to nonspecific proteins. This could be prevented by preclearing the cell lysates with normal serum (rabbit serum) and/or protein A beads. A detailed protocol for the immunoprecipitation of epitope-tagged opioid receptors is as follows: Approximately 150 lg of the solubilized protein is placed in an Eppendorf tube, about 1 lg (5 ll) of c-myc polyclonal antibody (200 lg=ml, Santa Cruz Biotechnology) and 12 ll of protease inhibitor cocktail (Sigma Chemical) are added, and the volume is adjusted to 1.2 ml with cold solubilization buffer. The mixture is incubated overnight
on a rocker at 4 °C. Protein A beads (Sigma) are equilibrated in the solubilization buffer and 100 ll of a 50% slurry (protein A in buffer) is added to each tube. Following incubation at 4 °C on a rocker for 2 h the mixture is subjected to centrifugation at 14,000 rpm for 1 min at 4 °C. The beads are washed three times with 500 ll of solubilization buffer containing protease inhibitors (supernatant removed from the immunoprecipitate with the help of an insulin syringe). Sample buffer (2X) with or without 50 mM dithiothreitol (DTT) is added and incubated at 60 °C for 15 min before subjecting the sample to SDS–PAGE and Western blotting (see below). 1.1.5. Cell surface labeling and immunoprecipitation HEK cells transiently expressing epitope-tagged receptors are washed with ice-cold PBS and incubated with 5 lg=ml anti-Flag antibody for 2 h at 4 °C. Cells are washed and lysed with the solubilization buffer (Buffer G) containing protease inhibitor cocktail for 1 h at 4 °C. The lysate is incubated with 15 ll of protein A–Sepharose 4B beads for 16 h at 4 °C. The immunoprecipitates (containing cell surface receptors) are collected, washed twice with Buffer G, and eluted with 30 ll of nonreducing sample buffer. Ten microliters of the eluate are subjected to SDS–PAGE/Western blotting analysis with anti-myc antibody as described below. 1.1.6. Western blot analysis This technique, when used in combination with immunoprecipitation, is a powerful tool that can be used for the detection of the second receptor in the complex and to study the specific interactions between proteins. Proteins separated by gel electrophoresis are transferred to nitrocellulose membranes and the membrane is incubated with a known protein or protein mixture to eliminate nonspecific binding of proteins to the membrane. The specific receptor is then localized using a primary antibody against the unique epitope tag on the second receptor followed by a secondary antibody conjugated to horseradish peroxidase. The sensitivity of the Western blotting procedure depends on the type of detection method used and the amount of antigen. Also, it is important that the antibodies can recognize denatured epitopes since sample preparation and electrophoresis conditions are likely to denature the epitope. One of the major advantages in using monoclonal antibodies for immunoblotting lies in the specificity since they bind to only one epitope and thus they provide an elegant tool for the identification of a particular region of the antigen. The major disadvantage is the possibility of cross-reactivity with other closely related proteins since the epitope usually comprises only four or five amino acids. The detailed protocol used for the visualization of receptor–receptor complexes is as described below: Approximately 70 ll of 2 sample buffer (120 mM
I. Gomes et al. / Methods 27 (2002) 358–365
Tris–Cl, pH 6.8, containing 4% SDS, 20% glycerol and 0.002% bromphenol blue with or without 50 mM DTT) is added to the immunoprecipitates (described in the previous section). Urea (8 M) could be added to the sample buffer and/or to the gel to disrupt oligomeric arrays and sharpen the specific bands. The samples are incubated at 60 °C for 15 min (it is important not to subject the complexes to higher temperatures, especially boiling, since this may lead to nonspecific aggregation of proteins). The samples are briefly cooled and about 10–15 ll is subjected to SDS–PAGE on 8% gels at 90 V until the dye front reaches the bottom of the gel. Urea gels are preferred when greater separation of protein is sought. The separated proteins are transferred to Protran nitrocellulose membranes (Schleicher & Schuell, Keene, NH) in a transfer buffer (composed of 25 mM Tris, 192 mM glycine, 20% methanol) for 16 h at 30 V. The membranes are briefly rinsed in TBS (50 mM Tris– Cl, pH 7.4, containing 150 mM NaCl), stained for 1 min with Ponceau S (Sigma) to visualize the bands, and washed with 20 ml TTBS (TBS containing 0.05% Tween 20) to remove Ponceau S stain. Following overnight incubation at 4 °C with 25 ml of 5% nonfat dried milk in TTBS to block nonspecific binding the membranes are rinsed in TTBS and incubated with 25 ml anti-Flag monoclonal antibody [1:500; 8 lg=ml in TTBS containing 5% protease-free bovine serum albumin and 0.01% sodium azide, (Sigma)] for 2 h at room temperature in a shaker. The membranes are washed six times (5 min each) with 20–25 ml TTBS and incubated with 25 ml anti-mouse IgG conjugated to horseradish peroxidase (1:5000 dilution in 5% nonfat dried milk in TTBS, Vector Laboratories, Burlingame, CA) for 1–2 h at room temperature in a shaker. Following incubation, the membranes are subjected to six washes (5 min each) with 20–25 ml TTBS. Approximately 8 ml of SuperSignal Chemiluminiscent Substrate (Pierce) is added. Following an incubation of 5 min, the membranes are exposed to X-ray films (Eastman Kodak, Rochester NY) for different periods to obtain multiple exposures that differ in the intensity of signal.
361
receptor interaction does not depend on a high level of receptor expression [1,2]. Another concern with the immunoprecipitation studies is the possibility of artifactual receptor aggregation during solubilization/immunoprecitipation procedures due to the inherent hydrophobic nature of GPCRs; this is addressed with a number of controls. One of the important controls is to mix cells individually expressing one or the other tagged receptor. These mixed cells are then subjected to the same solubilization and immunoprecipitation conditions as the cells expressing both epitope-tagged receptors. Under these conditions if the dimers are observed only in cells coexpressing both receptors (and not in the mixed cells) this would imply that the association between proteins is not the result of artifactual aggregation. Finally, a variety of agents can be used to study the biochemical properties of the receptor oligomers. (a) Detergents that differ in their physical properties have been used to help address a concern with detergent-induced association [2]. We have previously used a variety of combinations of different detergents to study the hydrophobic nature of opioid receptor oligomers [3]. (b) Capping agents that carboxymethylate free sulfhydryls (on cysteines) that are exposed during solubilization of membranes have been used to examine nonselective associations [4]. We have used iodoacetamide in solubilization and immunoprecipitation buffers, especially following treatment of cells with reducing agents, to address the involvement of disulfide bonds in the associations between receptors [2,5,6]. (c) Crosslinking agents have been used to examine detergent sensitive associations; these allow for stringent conditions of washing during the isolation of receptor complexes. A variety of crosslinking agents that differ in their functional properties have been used to investigate the nature of associations of a number of cell surface proteins including GPCRs [1,7]. Therefore, when appropriate controls are used immunoprecipitation can provide valuable information on receptor–receptor associations.
1.2. Concerns and controls
2. Biophysical techniques to monitor opioid receptor oligomerization in living cells
A concern with the biochemical studies is the level of expression of receptors in transiently transfected cells that can reach ‘‘nonphysiological’’ levels. We have addressed this issue by generating stable cell lines that express the interacting receptors at 1:1 ratio and at various levels, i.e., low (100,000 receptors/cell), medium (500,000 receptors/cell), and high (2,000,000 receptors/cell); the receptor numbers are determined by binding assays. In all three types of cell lines we observed coimmunoprecipitation of the receptors, albeit to varying extents, supporting the notion that the receptor–
As discussed above, biochemical studies have been used to demonstrate that opioid receptors can physically associate with each other to form dimers/oligomers [3,8]. Furthermore, pharmacological analysis has suggested that these complexes may play an important role in ligand binding, signaling, and/or internalization of opioid receptors on ligand binding [3,8]. Although crucial in establishing the concept of dimerization for opioid receptors and G-protein-coupled receptors in general, biochemical studies do not allow examination of receptor–receptor interactions or their modulation by
362
I. Gomes et al. / Methods 27 (2002) 358–365
ligands in live cells. Therefore we have recently employed bioluminescence resonance energy transfer (BRET), a proximity-based biophysical method, to examine receptor–receptor associations under physiological conditions in living cells. This assay is based on a naturally occurring phenomenon in which the energy generated by a luminescent donor, Renilla luciferase (Rluc) interacting with its substrate coelenterazine, is transferred to a fluorescence acceptor, a mutant form of green fluorescent protein (GFP). The energy transfer can occur only if the donor and the acceptor are in close ), thus allowing examination proximity (less than 100 A of close interactions between the donor- and acceptortagged opioid receptors (Fig. 1). 2.1. Experimental protocols 2.1.1. Preparation of tagged receptors We chose to tag opioid receptors with Rluc as the donor and mutant forms of GFP (yellow fluorescent protein, YFP, or enhanced green fluorescent protein,
EGFP) as the acceptor. Coelenterazine h (Molecular Probes) is used as an Rluc substrate. The emission maximum (470 nm) of coelenterazine h oxidized by Rluc is sufficient to excite YFP at one of its two excitation peaks (470 and 514 nm). The emission maximum of YFP at 530 nm allows good separation of light emissions from the donor and the acceptor. This increases the signal-to-noise ratio at the acceptor emission wavelength and thus increases the ability to detect BRET signal, i.e., receptor–receptor interactions with lower levels of protein expression. EGFP that has an excitation maximum at 488 nm can also be used as an acceptor. However, this acceptor molecule emits maximally at 509 nm, which partially overlaps with the emission spectrum of oxidized coelenterazine h. Thus, only very strong or close interactions can be detected due to the low signal-tonoise ratio at this wavelength (compare Figs. 2A and B). The donor Rluc or acceptor YFP is genetically fused to the C termini of opioid receptors to allow for membrane expression of the fusion proteins. A set of vectors for Rluc fusion (pRluc-N1, -N2, and -N3) can be ob-
Fig. 1. Principle of bioluminescence resonance energy transfer assay. Upper panels: Schematic representation of seven transmembrane opioid receptors fused to Renilla luciferase (Rluc) and yellow fluorescent protein (YFP). When they are far apart the light from the luminescent donor Rluc cannot excite the fluorescent acceptor YFP (A). If two differentially tagged receptors interact and are brought close together the acceptor is excited and emits light at 530 nm (B). Lower panels: Typical spectra obtained in the absence (A) and in the presence (B) of receptor–receptor interactions with a peak at 470 nm (A) and peaks at 470 and 530 nm (B).
I. Gomes et al. / Methods 27 (2002) 358–365
363
Fig. 2. Intensity and resolution of the BRET assay depend on the type of acceptor. Magenta curve: Light emission spectrum with cells expressing d opioid receptors fused to Renilla luciferase (dluc) and yellow fluorescent protein (dYFP) (A) or enhanced green fluorescent protein (dEGFP) (B). The dark blue curve represents the spectrum obtained with cells expressing dLuc alone.
tained from Packard Bioscience and the YFP fusion vector (pEYFP-N1) from Clontech. To minimize the possibility that Rluc or YFP will interfere with the normal functioning of the opioid receptors the entire sequence of the opioid receptor’s C terminus is preserved. Only the stop codon is mutated and fused inframe with the donor or acceptor molecule. A small intervening sequence between the C terminus of the opioid receptor and Rluc or YFP of 6–10 amino acids is introduced to allow for relative flexibility of the donor and acceptor. To maximize the interactions between opioid receptor molecules tagged with donor and acceptor tags and consequently the BRET signal, we chose HEK-293 cells that do not express endogenous opioid receptors and have high levels of trimeric G-proteins. This allows proper coupling of ectopically expressed opioid receptors to the cell’s signaling machinery. A transfection method that maximizes the fraction of cells cotransfected with the tagged opioid receptors should be used. In our case that is the calcium phosphate transfection. The level of expression of the tagged opioid receptors influences the ability to detect a BRET signal and also the specificity of interaction. This is determined by binding of radiolabeled ligands (described previously). A balance between high levels of expression required for BRET signal detection and relatively low levels of expression, as seen under physiological conditions, that allow specific interactions should be sought. This is achieved by using different amounts and ratios of Rluc and YFP-tagged receptors. Specificity of the BRET signal could be confirmed by competition with specific (opioid receptors) and nonspecific (other seven-transmembrane proteins) untagged competitors whose level of expression can be estimated by ligand binding assays.
2.1.2. Receptor expression and the BRET assay The protocol for transient transfection of HEK cells is essentially as described in Section 1.1. The amounts and ratio of plasmid vectors expressing Rluc and YFP tagged opioid receptors are optimized in each individual case; this differs for different combinations of opioid receptors. The ratio becomes very important when examining the formation of heterodimers between opioid receptors that are also known to homodimerize. We typically use 0:25 lg of each recombinant plasmid DNA and incubate cells with the DNA–calcium phosphate precipitate for 16–18 h. The cells are collected 36–72 h posttransfection and used for analysis. The level of receptor expression is determined using the radioligand binding assay as described earlier (Section 1.1). For BRET assay, about 36 h after transfection cells are detached from plates by incubation with PBS containing 1 mM EDTA for 2 min, collected by centrifugation, washed with PBS, and resuspended to 1–2 106 cells=ml with PBS containing 1 mM EDTA. Cells (2 ml) are placed in a cuvette, coelenterazine is added to 5 lM final concentration, and the suspension is quickly mixed with a pipette. Light emission is immediately monitored from 420 to 590 nm at 5-nm intervals for 0.5 s using a FluoroMax-2 spectrometer (Jobin Yvon-Spex Instruments S.A.) with closed excitation slit. A peak at about 470 nm generated on the addition of coelenterazine is detected in all samples expressing luciferase-tagged opioid receptors (Fig. 2). A second peak around 530 nm is detected in samples where Rluc-tagged receptors are coexpressed with YFP-tagged receptors (Fig. 2A) and at 509 nm with EGFP-tagged receptors (Fig. 2B). Alternatively, instead of monitoring light emission for the entire spectrum, the peak emissions at 470 and 530 nm (if YFP is used as an acceptor) or at
364
I. Gomes et al. / Methods 27 (2002) 358–365
509 nm (if EGFP is used as an acceptor) are monitored. In this case a spectrophotometer (BRET counter) with appropriate filters can be used to measure light emissions. In cases where light emission is monitored from 470 to 530 nm, the BRET signal can be represented as the difference between the area under the emission spectrum of the cotransfected cells (Fig. 2) and that of the luciferase-tagged opioid receptor transfected alone (Fig. 2). In other cases where light emission is monitored only at 470 and 530 nm the BRET signal is represented as BRET ratio. This is calculated as the ratio of the emission at 530 nm to the emission at 470 nm obtained when Rluc-tagged receptors and YFP-tagged receptors are coexpressed, standardized to the ratio when Rluc-tagged receptor is expressed alone [4]. 2.2. Concerns and controls The BRET assay can give very valuable information about the proximity and likelihood of interactions of different proteins in the context of live cells. Therefore, this assay excludes the possible effect of different detergents and membrane preparations on protein–protein interactions. However, certain concerns should be addressed because it relies on transient transfections and the ectopic expression of proteins from strong viral promoters. One important issue that is always being considered when examining protein–protein interactions in transiently transfected cells is the level of expression of the proteins of interest. Very often the level of protein expression that generates the best signal-to-noise ratio for the assay is not necessarily its ‘‘physiological level’’ of expression. To optimize the signal-to-noise (BRET to luciferase) ratio as well as the specificity of interactions a
variety of transfection protocols and expression conditions that allow for optimal expression of fusion proteins should be considered. The level of expression should be quantiated by independent methods (ligand binding assay). To maximize BRET signal, i.e., donor– acceptor associations as compared with donor–donor and acceptor–acceptor combinations, we have varied the levels of Rluc to GFP/YFP fusion proteins. As demonstrated in Fig. 3A in the case of d–d interactions, increasing the level of the acceptor dEGFP gives a stronger BRET signal (Fig. 3A). The increase was significant when a 1:3 instead of a 1:1 donor:acceptor ratio was tested (compare red and magenta curves in Fig. 3A). This could be due to the increased probability of donor– acceptor association when increasing the concentration of one of the components. Further increase in the acceptor molecules does not significantly increase the BRET signal (compare magenta and green curves in Fig. 3A). It is possible that increasing acceptor expression leads to increased acceptor–acceptor associations (that do not contribute to BRET signal). It is also possible that increasing the protein expression leads to nonspecific interactions driven by mass action. To examine if the total level of protein expressed affected the BRET signal we varied the level of receptor expression while keeping the donor:acceptor ratios constant (Fig. 3B). Our results show that under these conditions the BRET signal does not vary even with a five fold increase in the level of expression; the level of receptor expression varied from 200 to 1000 fmol/mg of protein (determined by radioligand binding) in these studies. In Fig. 3B a 1:3 donor:acceptor ratio is presented; however, similar results were obtained with a 1:1 ratio (data not shown). Another important consideration especially when examining receptor heterodimerization is the differences
Fig. 3. Determination of optimal conditions for specific BRET signal. Homotypic association of d (A) and l (B) opioid receptors monitored by BRET in live cells. (A) The level of BRET signal with varying ratios of donor and acceptor fusion proteins: red (1:1), magenta (1:3), and green (1:4). (B) The level of BRET signal while varying the total amount lluc þ lEGFP DNA while keeping the ratio constant (1:3): magenta (4 lg), and green (20 lg). The dark blue curve represents the spectrum obtained with cells expressing lluc alone. Numbers correspond to micrograms of each DNA used for transfection.
I. Gomes et al. / Methods 27 (2002) 358–365
365
3. Summary
Fig. 4. Interaction (heterodimerization) of d and j opioid receptors. A BRET signal (peak at 509 nm) is detected when dluc is used as a donor and jEGFP as an acceptor (magenta curve) as well as when jluc is the donor and dEGFP the acceptor (green curve). The dark blue curve represents the spectrum obtained with cells expressing dluc alone. Similar results were obtained with jluc (not shown).
in the relative efficiency of expression of each of the fusion proteins. We have addressed this by using both combinations (jluc-dEGFP as well as dluc-jEGFP) of receptors. As shown in Fig. 4, association of j and d opioid receptors can be detected with either jRluc as donor and d-EGFP as acceptor or dRluc as donor and jEGFP as acceptor. These data further support our biochemical and pharmacological data that explored j–d receptor interactions [2]. To determine the specificity of the interactions between different opioid receptors fused to Rluc and YFP, a competition assay with untagged opioid receptors and other unrelated proteins is used [9]. The use of unrelated proteins that do not interact with opioid receptors under the same transfection conditions ideally serves as a negative control. Since such an experiment would involve simultaneous transfection of three different expression vectors the relative levels of expression of the proteins are difficult to control. The simplest controls universally used are the Rluc or YFP vectors (without the recombinant protein) in combination with Rluc- or YFP-tagged opioid receptors respectively.
We have used classic biochemical methods and modern biophysical methods to examine interaction of opioid receptors [10]. By coimmunoprecipitation studies we have shown that opioid receptors form homotypic as well as heterotypic associations in the absence of the agonist treatment. Results from BRET assays have extended these observations and demonstrated that the receptor–receptor association can be seen in live cells. This is very important since biochemical studies require solubilization of membranes with different detergents; this, in addition to disrupting the cellular environment of the receptors, may also affect their association of other regulatory proteins. BRET assays monitor interactions within the dynamic environment of the cell membrane, allowing us to assess modulation of these interactions by different ligands as a crucial aspect of the functioning of opioid receptors.
Acknowledgments This work was supported in part by National Institutes of Health Grants DA 08863 and DA 00458 (to L.A.D.).
References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10]
S. Cvejic, L.A. Devi, J. Biol. Chem. 272 (1997) 26959–26964. B.A. Jordan, L.A. Devi, Nature 399 (1999) 697–700. L.A. Devi, Trends Pharm. Sci. 22 (2001) 532–537. F.-Y. Zeng, J. Wess, J. Biol. Chem. 274 (1999) 19487–19497. I. Gomes, B.A. Jordan, A. Gupta, N. Trapaidze, V. Nagy, L.A. Devi, J. Neurosci. 20 (2000) RC1108. B.A. Jordan, N. Trapaidze, I. Gomes, R. Nivarthi, L.A. Devi, Proc. Natl. Acad Sci. USA 98 (2001) 343–348. H. Hastrup, A. Karlin, J.A. Javitch, Proc. Natl. Acad. Sci. USA 98 (2001) 10055–10060. B. Jordan, L.A. Devi, B. J. Anaesth. 81 (1998) 12–19. S. Angers, A. Salahpour, M. Bouvier, Life Sci. 68 (2001) 2243– 2250. I. Gomes, B.A. Jordan, A. Gupta, C. Rios, N. Trapaidze, L.A. Devi, J. Mol. Med. 79 (2001) 226–242.