Characterization of δ, κ, and μ Human Opioid Receptors Overexpressed in Baculovirus-infected Insect Cells*

The cDNAs encoding human δ (hDOR), κ (hKOR) and μ (hMOR) opioid receptors were cloned in the baculovirusAutographa californica (AcMNPV) under the control of the polyhedrin promoter with or without an amino-terminal hexahistidine tag. Expression levels were optimized in Spodoptera frugiperda (Sf9) cells and were in the following order hMOR > hDOR > hKOR. The receptors bound antagonists with affinity values similar to those published previously for the receptors expressed in mammalian cells. They also retained selectivity toward specific antagonists. The three receptors bound peptidic agonists with low affinity, suggesting that they might not be functionally coupled to intracellular effectors. Introduction of an amino-terminal hexahistidine tag decreased the levels of expression markedly. Only hMOR-his was expressed at a level allowing binding study, but no difference could be detected in the affinities of both agonists and antagonists compared with the nontagged protein. hMOR expression was also optimized in High Five cells leading to a further increase in protein production. The pharmacological profile was similar to the one obtained when the receptor was expressed in Sf9 cells. Our results show that the baculovirus expression system is suitable for large scale production of human opioid receptors.

Opioid receptors and endogenous opioid peptides form a neuromodulatory system that plays a major role in the control of nociceptive pathways. The opioid system is not only a key element for pain perception but also modulates affective behavior as well as neuroendocrine physiology and controls autonomic functions such as respiration, blood pressure, thermoregulation, and gastrointestinal motility. It affects locomotor activity and could be involved in learning and memory. The receptors are otherwise targets for exogenous narcotic drugs, a major class of drugs of abuse.
Genes coding for ␦, , and opioid receptor subtypes have now been identified and isolated from different vertebrates. Primary sequence analysis indicated that opioid receptors belong to the G-coupled receptor family whose structure shows a seven-transmembrane domain topology (for review, see Refs. 1 and 2). Engineering of protein chimeras together with generation of point mutants led to a better identification of the receptor regions interacting specifically with different ligands and/or involved in the signal transduction pathway. However, almost no structural data are available but only a few models based on rhodopsin and bacteriorhodopsin structures (2)(3)(4)(5). Recently a new insight was brought by a model drawn without template for the ␦ receptor (6). The baculovirus expression system is extremely efficient to produce large amounts of mammalian proteins. Post-translational modifications are identical to those observed in mammalian cells with the exception of glycosylation, which is mostly of the high mannose type. Several G-coupled receptors, some of human origin, have already been expressed successfully under a functional state including adrenergic; ␥-aminobutyric acid, type A; cholecystokinin; and different muscarinic, serotonin, or dopamine subtypes. The protein is produced in the range of 5-30 pmol/mg protein, and 1 ϫ 10 6 -2 ϫ 10 6 receptor sites are generally present at the cell surface (for review, see Refs. 7 and 8). Furthermore, the protein production can be scaled up fairly easily. The baculovirus expression system may therefore be appropriate for production of opioid receptors in amounts that would allow purification to homogeneity and characterization on a functional and structural basis.
We have cloned the cDNAs encoding full-length ␦, , and opioid receptor subtypes in the genome of the baculovirus Autographa californica (AcMNPV) 1 under the control of the polyhedrin promoter. In addition, all three receptors were cloned with an amino-terminal hexahistidine tag under the control of the same promoter to help subsequent purification. The protein expression was optimized for all six constructs, and a pharmacological profile was drawn for each mature receptor using whole cell binding. SHSY-5Y (9) and subcloned under the control of the polyhedrin promoter in the EcoRI and BamHI sites of pAcY2 (10).

Materials-[
The cDNA encoding the human opioid receptor (hKOR) was isolated from human placenta (11) and subcloned under the control of the polyhedrin promoter in the NotI site of pVL1393 (Pharmingen, San Diego).
The cDNA encoding the human opioid receptor (hMOR) was isolated from human brain (12) and subcloned under the control of the polyhedrin promoter in the EagI and SmaI sites of pVL1392 (Pharmingen).
hDOR, hKOR, and hMOR were also cloned under the control of the polyhedrin promoter as fusion proteins with an amino-terminal hexahistidine tag in pAcSG-His-NT vectors (Pharmingen): hDOR was inserted in the EcoRI site of pAc-His-NTB, hKOR in the NotI site of pAcSG-His-NTC, and hMOR as a EagI-SmaI fragment in pAcSG-His-NTA. All constructs were sequenced to control in-frame insertion.
Recombinant Baculovirus Isolation and Amplification-hDOR recombinant baculoviruses were selected in yeast according to a method we used previously (13) and originally described by Patel et al. (10). All other recombinant viruses were generated by cotransfection in Spodoptera frugiperda (Sf9) cells with the appropriate transfer vector and Baculogold DNA (Pharmingen) using calcium phosphate coprecipitation.
Single viral clones were isolated by plaque assay (except for hDOR) and amplified three times according to O'Reilly et al. (14). Viral titers were determined by end point dilution and calculated according to Reed and Muench (15).
Insect Cell Culture Conditions-Sf9 cells were grown and maintained in TC100 medium supplemented with 10% fetal calf serum (Life Technologies, Inc.) and Trichoplusia ni (BTI-TN-5B1-4 or High Five ) cells in serum-free medium Express Five (Life Technologies, Inc.). Both cell lines were cultivated either in monolayers or in suspension (50 -200 ml) at 27°C at 100 rpm (14). Cell viability was determined by the trypan blue exclusion method.
Infections at the required multiplicity of infection (m.o.i.) were performed either in monolayers according to O'Reilly et al. (14) or in suspension at a cell density of 1.6 10 6 cells/ml for Sf9 cells and 2.5 10 6 cells/ml for High Five cells unless otherwise stated. Cells were harvested either periodically to determine the time course of the expression maximum or at the peak of production (48 -64 h postinfection) for saturation and competition experiments.
Binding Assay-Cells were harvested, centrifuged at 4°C, washed with ice-cold PBS containing 0.32 M sucrose, and resuspended in icecold 50 mM Tris, pH 7.4, 1 mM EDTA, 0.32 M sucrose, conditions known to prevent receptor internalization (16). Binding assays were performed on 10 6 cells (Sf9) or 10 5 cells (High Five) after a 30-min incubation at 18°C with the appropriate ligands. Samples were filtered on GF/B filters treated with 0.1% polyethylenimine and washed twice with icecold 50 mM Tris, pH 7.4, on a Brandell cell harvester. In all saturation experiments naloxone was used at 2 ϫ 10 Ϫ6 M to determine nonspecific binding.
Fluorescence Microscopy-Sf9 cells expressing either hMOR or hMOR with the amino-terminal hexahistidine tag (hMOR-his) were harvested 64 h postinfection, washed with PBS, 0.32 M sucrose, and resuspended in Eppendorf tubes at 2 ϫ 10 8 cells/ml. Fixation with 4% paraformaldehyde in PBS for 10 min was followed by three washes with PBS, 0.1% Tween 20 and incubation overnight at 4°C in PBS, 0.1% Tween 20, 1% bovine serum albumin with a monoclonal antibody directed toward the histidine tag (Dianova, Federal Republic of Germany) at a 1:10 dilution. The cells were then washed three times with PBS, 0.1% Tween 20 and incubated for 4 h at room temperature in PBS, 0.1% Tween 20, 1% bovine serum albumin with a secondary anti-mouse antibody coupled to fluorescein isothiocyanate (Jackson ImmunoResearch Laboratories, Inc.) at a 1:200 dilution. After another wash with PBS, 0.1% Tween 20, nuclei were stained with 4,6-diamidino-2-phenylindole (Sigma), and cells were washed again. Fluorescence was observed using an Axioplan Zeiss microscope.

Optimization of Nontagged Recombinant Opioid Receptor Expression Levels in Sf9
Cells-Insect cells were infected with recombinant baculoviruses encoding nontagged opioid receptors under the control of the polyhedrin promoter. Expression levels were optimized for each recombinant virus using different initial Sf9 cell densities (ranging from 0.5 to 2 ϫ 10 6 cells/ml) and various m.o.i. (ranging from 0.5 to 5). Samples were analyzed for receptor expression 24, 48, 56, and 72 h postinfection. The three opioid receptor subtypes were best expressed when infection was performed at a cell density of 1 ϫ 10 6 cells/ml with m.o.i. ϭ 2 for hMOR and m.o.i. ϭ 1 for hDOR and hKOR (Fig. 1). Similar profiles were observed for the three receptors with a maximum of expression reached 48 h postinfection and stable for the next 24 h. However, the maximal expression level was strongly subtype-dependent despite the high amino acid sequence homology between the receptors. Scatchard analysis was performed on data from whole cell binding experiments with [ 3 H]diprenorphine, a widely used nonspecific antagonist. The deduced B max values allowed comparison of the expression levels of mature receptors present at the cell surface (Table I). hMOR was best expressed at levels approaching 1 nmol/liter of culture, which corresponds on average to about 500,000 receptor sites at the cell surface. hDOR was about four times less expressed, and a 10-fold lower protein expression was observed for hKOR. It was hypothesized previously that the cloning strategy could influence the protein expression level. According to O'Reilly et al. (14), the starting codon of the gene to be expressed should be located out of frame and fewer than 100 base pairs downstream from the modified original ATG of the polyhedrin gene. These two requirements could be fulfilled for hMOR and hDOR but not for hKOR for which cloning into the restriction site NotI was 87 base pairs downstream but in-frame with the modified ATG. Another construct was then made in which the gene encoding hKOR was introduced in the unique XbaI restriction site of pVL1393. The starting codon of the gene to be expressed this time was out of frame but 119 base pairs after the modified ATG. About 60,000 -120,000 receptor sites were detected at the cell surface, which represents a 2-fold increase (data not shown). These results suggest that factors governing protein expression levels are complex and cannot be easily predicted.
Pharmacological Properties of Nontagged Recombinant Human Opioid Receptors-Saturation analysis of [ 3 H]diprenorphine binding on recombinant ␦, , and receptors showed that the measured K d values were similar to (hDOR, hKOR) or slightly higher (hMOR) than values published for receptors expressed in COS cells (Table II).
The use of specific antagonists in competition binding experiments provided information on the subtype selectivity of the three recombinant proteins (Table III). The antagonists naloxonazine, naltrindole, and nor-BNI bound preferentially to hMOR, hDOR, and hKOR, respectively, which is in agreement with their , ␦, and selectivities described previously in other cell backgrounds (17)(18)(19)(20). The K i values are in good agreement with former estimates for native and/or recombinant receptors expressed in mammalian cells. However, the relative selectivity observed in our case is less pronounced than reported previously (see "Discussion").
Competition binding experiments were performed using potent and subtype-selective agonists (Table IV). K i values for both DAMGO and dermorphin were 2 orders of magnitude higher than those reported for hMOR expressed in COS cells. This suggests that the receptor expressed in Sf9 cells is in a low affinity binding state and may therefore not be coupled to intracellular effectors. Similarly, deltorphin II bound with low affinity to hDOR. However, BW 373U86 bound to hDOR with very high affinity, and the K i value for DPDPE was similar to values reported in mammalian cells for apparently coupled receptors. K i values were also determined for hKOR. The K i for U50488H was in the nanomolar range in agreement with values obtained previously for the receptor expressed in COS cells, whereas the K i value for dynorphin A was 2 orders of magnitude higher than expected (see "Discussion").

Effect of the Amino-terminal Histidine Tag on Receptor Expression Levels and Ligand Binding Properties-Introduction
of an amino-terminal histidine tag modified significantly the receptor expression level at the cell surface. The B max value of hMOR-his is reduced about 5-fold compared with wild type hMOR receptor ( Fig. 1 and Table I). hDOR-his expression could still be detected but at a level too weak to allow any pharmacological characterization, and no hKOR-his receptor sites could even be detected on intact cells. These results strongly suggest that introduction of six positive charges at the amino terminus of the receptor interfered with membrane insertion and/or proper folding within the lipid bilayer. Interestingly, the pharmacological profile of hMOR-his seemed indistinguishable from that of hMOR (Tables II, III, and IV).
Immunolocalization of hMOR-his-Immunofluorescence experiments were performed on Sf9 cells infected with recombinant viruses encoding either hMOR or hMOR-his using a monoclonal antibody directed against the hexahistidine sequence. As expected, only background staining was observed with cells expressing hMOR or cells expressing hMOR-his incubated with the secondary antibody alone (Fig. 2, a and c). The fluorescence was detected mainly at the plasma membrane of Sf9 cells expressing hMOR-his, suggesting that the mature protein reaches the cell surface. Staining was also visible inside the permeabilized cells, indicating that part of the receptor is still trapped likely within the endoplasmic reticulum and Golgi compartments (Fig. 2b).
Expression of hMOR in High Five Cells-Recombinant viruses encoding hMOR were also used to infect High Five cells, another insect cell line that was recently reported to be suitable for large scale protein production (21). Expression was optimized using different initial cell densities (ranging from 1 ϫ 10 6 to 3.5 ϫ 10 6 cells/ml) and various m.o.i. (ranging from 2 to 16). Expression was maximum 48 h postinfection at m.o.i. ϭ 4 and an initial cell density of 2.5 ϫ 10 6 cells/ml. Data from whole cell binding experiments with [ 3 H]diprenorphine were used for Scatchard analysis. B max values were between 0.9 and 1.7 nmol/liter of culture corresponding in average to 5-10 ϫ 10 5 receptor sites at the cell surface (Fig. 3). Those values represent a 2-fold improvement over expression in Sf9 cells.
The pharmacological profile of the recombinant receptor expressed in High Five cells was also determined.     (Fig. 3). hMOR expressed in High Five cells retained its selectivity for the antagonist naloxonazine, whereas K i values for the agonists DAMGO and dermorphin suggested that the receptor was in a low affinity binding state as also inferred in Sf9 cells (Table V). DISCUSSION We used the baculovirus expression system to overexpress ␦, , and human opioid receptors in insect cells. We achieved production of protein amounts close to 1 nmol/liter of culture, which are comparable to yields published for other G-coupled receptors (for review, see Refs. 7 and 8) and are also similar to those obtained for recombinant opioid receptors in mammalian cells (9,11). The three receptor subtypes were expressed at clearly distinct levels in Sf9 cells, although they share high sequence homology (up to 80% identity in the transmembrane domain). Such a variation has already been observed with other highly homologous G-coupled receptors either from different species origin such as ␤ 2 -adrenergic or between different subtypes such as muscarinic ones (see Refs. 7 and 8).
Previous reports suggested that a portion of recombinant membrane proteins, including G-coupled receptors, produced in Sf9 cells does not reach the cell surface but remains in the endoplasmic reticulum and Golgi compartments because of possible saturation of the translocation machinery of the insect cell (22)(23)(24)(25). This hypothesis is also supported by our immunofluorescence experiments on permeabilized Sf9 cells expressing hMOR-his in which antibody labeling was detected not only at the cell surface but also inside the cell. Our B max values deduced from whole cell binding experiments only represent the receptor sites located at the cell surface. Therefore, we likely underestimate the real amount of properly folded receptors able to bind ligands and subsequently the number of receptors suitable for purification and further studies.
High Five cells were recently described to achieve higher protein production when compared with Sf9 cells (21,26). Using whole cell binding experiments with [ 3 H]diprenorphine we could detect on a cell-to-cell basis approximately twice as much hMOR in High Five cells as in Sf9 cells. One possible explanation could be the larger size of High Five cells. Moreover, we were able to grow the latter at a higher cell density (5-6 ϫ 10 6 against 3 ϫ 10 6 cells/ml for Sf9 cells) leading to a further increase in protein production. High Five cells were also recently reported as more effective in complex glycosylation than Sf9 cells which can only synthesize high mannose type sugars (27). All of this favors High Five cells for opioid receptor large scale production.
Introduction of a hexahistidine tag at the amino terminus decreased opioid receptor expression levels markedly. In fact, the addition of six positive charges at the amino terminus may impair the so-called positive inside rule originally described for Escherichia coli membrane proteins and later applied to Gcoupled receptors by Wallin and von Heijne (28). However, no modification of the binding properties could be detected. This is in agreement with previous works reporting that introduction of a tag at the amino terminus of a G-coupled receptor does not modify its binding characteristics (29 -32) and that histidine tags do not generally interfere with protein function (33)(34)(35). Moreover, the amino-terminal part of hMOR does not seem to be important for ligand recognition since deletion of the first 64 amino acids of hMOR did not affect the receptor binding properties (36). Pharmacological characterization of the opioid receptors expressed in Sf9 and High Five cells revealed affinities for antagonists very similar to those reported for native and/or recombinant receptors expressed in mammalian cells (for review, see Refs. 1, 2, 20, and 37). Our work also allowed us to compare in a single cell type the affinity of the three receptor subtypes for each tested antagonist. hMOR, hDOR, and hKOR bound preferentially naloxonazine, naltrindole, and nor-BNI, respectively. Our data confirm that nor-BNI is -specific (19) with ␦: and : ratios greater than 100. Surprisingly, naltrindole, reported as ␦-specific (18), showed ratios two to four times lower than expected (:␦ ϭ 23 and :␦ ϭ 45); but the most striking result is the weak specificity (: ϭ 7 and ␦: ϭ 6) for the previously described -specific antagonist naloxonazine (17). However, interpretation of our finding is difficult because of the paucity of available information. Despite numerous data collected using mammalian cells and/or brain tissue, only very few comparisons were performed on a single cell type using cloned receptors from the same species origin.
The three opioid receptor subtypes share high homology in the intracellular loops and proximal carboxyl-terminal region, suggesting that they are coupled to the same G␣ protein classes (1). Experimental evidence points to interactions with G␣ subunits from the G i or G o types mainly (for review, see Ref. 38). Reconstitution in lipid vesicles showed that receptors purified from bovine striatal membrane were able to bind G i,1 , G i,2 , G o,A and G o,B individually or as a mixture (39), whereas reconstituted receptors could bind to G i,1 and G i,2 and not G o (40). In Chinese hamster ovary cells the three opioid receptor subtypes interact with G i,2 , G i,3 , G o,2 and an unidentified G␣ subunit (41)(42)(43). In these cells coupling seems to be a spontaneous process that is not induced by receptor overexpression since it is not dependent on receptor density (42). In insect cells, however, only a G o -like protein but no G i could be detected so far (44 -48). Still, functional coupling to endogenous G proteins was depicted for receptors known to be coupled to pertussissensitive G proteins such as human serotonin 5HT 1A (46) and 5HT 1B (45). Similarly, Vasudevan et al. (49) reported that the rat muscarinic acetylcholine receptor subtype m 3 activates potassium channels, and Oker-Blom et al. found (50) that the human ␣ 2 C-C4 adrenergic receptor inhibits forskolin-stimu-lated adenylyl cyclase in Sf9 cells. These observations suggest that receptors coupled to G o and/or G i may interact with Sf9 endogenous G␣ subunits.
The binding of agonists is known to be dependent on the coupling state of the receptor. We determined binding affinities of well known highly potent agonists to get information on the possible coupling to endogenous G proteins in insect cells (for review, see Refs. 1, 20, and 37). In competition binding assays the tested peptidic agonists showed affinities 2 orders of magnitude lower than those described for the opioid receptors expressed in COS cells. This suggests that the opioid receptors would not be well coupled to intracellular effectors in insect cells. It is however of interest to notice that the K i value for DPDPE is still compatible with values reported for rodent DOR on membrane preparation (16.4 nM) (51) or expressed in COS cells (58 nM) (52). Also, the alkaloids BW 373U86 and U50488H bound to hDOR and hKOR, respectively, with affinity values in the nanomolar range. The extremely high affinity observed for BW 373U86 is likely independent of hDOR coupling since BW 373U86 was described as a potent agonist endowed with the unique property of being insensitive to distinct receptor affinity states (53). Interpretation of the high K i value of U50488H remains on the contrary ambiguous. The binding efficacy of this alkaloid might also be little dependent on receptor coupling as was the case for BW 373U86. An alternative hypothesis would be that coupling of the receptor to endogenous G proteins might induce a receptor high affinity conformation recognized by alkaloid agonists exclusively.
While preparing this manuscript expression of rodent opioid receptors in insect cells was reported (54). Using homologous competition binding assays, agonist binding affinities in the nanomolar range were measured on High Five membrane preparations. However, this method allows detection of high affinity state receptors only (55). In our case preliminary results on High Five membrane preparations indicated that the agonists DAMGO and dermorphin bind to hMOR with affinity in the range of 100 -200 nM (data not shown) in agreement with our data using intact cells.
In conclusion, we were able to overproduce hMOR at a level that will allow purification of protein amounts sufficient for structural analysis by biophysical and biochemical means including two-dimensional crystallization. Still further increases of hDOR and hKOR expression are necessary to reach levels at least comparable to that of hMOR. The recombinant opioid receptors showed a pharmacological profile similar to that described for native and/or opioid receptors expressed in mammalian cells. Introduction of an amino-terminal hexahistidine tag did not modify this profile but reduced the amount of receptor sites present at the cell surface. We are currently investigating the influence of a carboxyl-terminal histidine tag on hMOR expression levels and binding properties. Recombinant opioid receptors do not seem to be functionally coupled to intracellular components likely because of the different content  in endogenous G␣ subunits present in insect cells compared with mammalian cells. This heterologous background will allow characterization of the specific interactions of opioid receptors with defined G␣ proteins by coexpression of both partners in the insect host cell.