Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Opioid receptors belong to the seven transmembrane superfamily of heterotrimeric guanine nucleotide-binding protein (G protein)-coupled receptors (Uhl et al., 1993). The interaction of receptors and effectors has been described by Stickle and Barber (1991, 1996) in a random encounter-coupling model based on the collision-coupling model proposed by Tolkovsky and Levitzki (1978) for the activation of adenylyl cyclase by -adrenergic receptors in turkey erythrocytes. According to the model receptors and G proteins diffuse freely at the cell membrane such that agonist-activated receptors act as mobile catalysts for the activation of G proteins. Assumptions of the model are that receptors have access to numerous G proteins, but G protein inactivation is independent of receptor activity. This can be represented by the following simplified equation derived from Tolkovsky and Levitzki (1978):
where A is agonist, R is receptor, G represents heterotrimeric G protein, and K_D the agonist dissociation constant. The rate constants k₄ and k₅ are rapid compared with k₃, which becomes rate limiting. Thus, the intermediate ARG never accumulates and constitutes only a small fraction of the total receptor and G protein. The rate of G protein activation is proportional to the intrinsic collision frequency between receptor and G protein multiplied by the probability that a collision involves an agonist bound receptor and so is proportional to the concentration of agonist-bound receptors. On the other hand individual G proteins are accessible to more than one receptor and so the number of G proteins that can be activated should be independent of receptor concentration. This model is supported by observations from several G_s- and G_i-coupled receptors in membranes from erythrocytes (Pike and Lefkowitz, 1981) and adipocytes (Murayama and Ui, 1984).

In contrast, in membranes from C6 rat glioma cells expressing the µ-opioid receptor (C6 µ), the rate of agonist-stimulated [³⁵S]guanosine-5'-O-(3-thio)triphosphate ([³⁵S]GTPS) binding, a measure of G protein activation, is independent of either receptor or G protein concentration (Remmers et al., 2000). Additionally the encounter-coupling model predicts that "cross talk" should be seen between receptor types that activate the same G protein subtype. However, Graeser and Neubig (1993) failed to find evidence for interactions between 2, m4, and -opioid receptors in membranes from NG108-15 cells. Together, these findings lend support for an organization of receptors and G proteins (Neubig, 1998). On the other hand the inability to confirm the encounter-coupling model may reflect an artifact of the membrane preparation and homogenization process. G proteins may become isolated on membrane vesicles that contain few receptors, artificially limiting the accessibility of G proteins.

The aim of the present study was to develop a permeabilized cell system to evaluate µ-receptor-G protein interactions using the binding of the nonhydrolyzable GTP analog [³⁵S]GTPS as the measure of G protein activation. This assay has been shown previously to provide quantitative and reproducible measurements of G protein activation in membrane preparations by nonopioid (Lorenzen et al., 1993; Tian et al., 1994) and opioid agonists (Traynor and Nahorski, 1995; Emmerson et al., 1996). The assay has conceptual validity in that the active state of the G protein has been defined as the GTP-bound species (Gilman, 1987; Birnbaumer et al., 1990). The [³⁵S]GTPS binding assay cannot be used to evaluate receptor-driven G protein activation in intact cells because [³⁵S]GTPS is unable to cross cell membranes. However, the [³⁵S]GTPS assay has been used successfully in digitonin-permeabilized HL-60, human erythroleukemia cells, and human embryonic kidney cells to show receptor-G protein coupling (Wieland et al., 1995). Digitonin binds to cholesterol in eukaryotic plasma membranes, creating pores that are permeable to ions and proteins (Bittner and Holz, 1988).

Here we show that following treatment of C6 µ cells with digitonin the [³⁵S]GTPS binding assay can be used to assess G protein activation by µ-opioid agonists in a permeabilized cell preparation. This system can be used to show potency and efficacy differences between µ-opioid agonists. Moreover, the system provides further evidence for a higher degree of organization for the activation of G proteins by µ-opioid receptors than can be accounted for by a random encounter-coupling model.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals and Drugs. [³H]Diprenorphine (specific activity 2.15 TBq/mmol) and [³⁵S]GTPS (specific activity 46.25 TBq/mmol) were purchased from PerkinElmer Life Science Products (Boston, MA). [D-Ala²,N-Me-Phe⁴,Gly⁵-ol]-enkephalin (DAMGO) was obtained from Sigma (St. Louis, MO), naloxone was from DuPont (Wilmington, DE), morphine sulfate was from Mallinckrodt (St. Louis, MO), and trypan blue was from Matheson Coleman and Bell (Norwood, OH). Digitonin and analytical grade biochemicals were from Sigma. Fetal bovine serum, Geneticin, and Dulbecco's medium were purchased from Life Technologies (Gaithersburg, MD).

Cell Culture. C6 rat glioma cells stably transfected with a rat µ-opioid receptor [C6 µ; Lee et al., 1999] were grown under 5% CO₂ in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. Stock flasks were maintained in the presence of 1 mg/ml Geneticin to select for the presence of the transfected plasmid, which codes for both the µ-receptor and antibiotic resistance. Cells used for experiments were split from the stock flasks and grown to confluence in the absence of Geneticin without significant loss in receptor density.

-Funaltrexamine (-FNA) Treatment. Plates of cells were incubated for 1 h in serum-free medium in the presence or absence of 10 nM -FNA. Cells were then washed with serum-free medium four times to remove unbound -FNA, and immediately harvested. Following collection and permeabilization, both control and -FNA-treated cells were divided into two aliquots. [³⁵S]GTPS binding was measured in one aliquot, while the other aliquot was used to determine receptor concentration.

Cell Permeabilization. Cells were collected from plates using lifting buffer (5.6 mM glucose, 5 mM KCl, 5 mM HEPES, 137 mM NaCl, 1 mM EGTA, pH 7.4) and washed with KGEH buffer (1.39 M potassium glutamate, 40 mM MgCl₂, 100 mM EGTA, 300 mM HEPES, pH 7.4). Cells were then incubated for approximately 5 min at 37°C in KGEH buffer with 20 µM digitonin, as described by Bittner and Holz (1988). Cells were tested for permeabilization using trypan blue and were considered permeabilized when 20% of cells excluded the dye. Cells were then washed two additional times in KGEH buffer. A hemacytometer (American Optical Corporation, Buffalo, NY) was used to count the cells.

[³⁵S]GTPS Binding Assay. Permeabilized cells were incubated for 60 min unless otherwise specified at 25°C with 50 pM [³⁵S]GTPS, in the absence or presence of varying concentrations of agonist, in binding buffer (final concentration: 500 µM dithiothreitol, 500 µM EDTA, 18.5 mM MgCl₂, 50 mM NaCl, 23.5 mM Tris, 556 mM potassium glutamate, 40 mM EGTA, 120 mM HEPES, pH 7.4) containing 50 µM GDP in a final assay volume of 400 µl. A similar level of GDP was found to be optimal in membrane preparations from these cells (Remmers et al., 2000). The reaction was terminated by the addition of 2 ml of ice-cold washing buffer (50 mM Tris, 5 mM MgCl₂, 100 mM NaCl) and the contents of the tubes were rapidly filtered through glass fiber filters (no. 32; Scheicher & Schuell, Keene, NH). The tubes and filters were rinsed with 2 ml of washing buffer an additional three times. Filters were then placed in scintillation vials containing 4 ml of scintillation cocktail for liquid scintillation counting. Saturation binding followed the same procedure with [³⁵S]GTPS concentration varying from 5 pM to 100 nM, in the absence or presence of 100 µM DAMGO.

Receptor Binding Assay. Saturation binding experiments on permeabilized cells were performed using varying concentrations of [³H]diprenorphine in 1 ml of binding buffer under conditions identical to [³⁵S]GTPS binding assays. To study the effect of digitonin on ligand binding displacement assays were performed. Membranes from digitonin-treated and untreated cells were incubated with 0.2 nM [³H]diprenorphine and varying concentrations of DAMGO (0.3-300 nM) in Tris-HCl buffer (50 mM, pH 7.4) at 25°C for 1 h. In all cases nonspecific binding was determined from samples that contained 10 µM naloxone. Incubations were stopped by rapid filtration and radioactivity retained on the filters determined by liquid scintillation counting as described above.

Data Analysis. The GraphPad Prism computer program (San Diego, CA) was used to perform linear and nonlinear regression analysis of the data. Concentration-response curves for [³⁵S]GTPS binding were fitted to a sigmoidal curve with a Hill coefficient of 1 and baseline fixed at 0% stimulation. Saturation binding data were analyzed using a one-site saturation binding equation, and time course experiments were fit to a one-phase exponential association curve. The antagonist affinity of naloxone (K_e) value for naloxone was determined from the rightward shift produced in the concentration-response curve for DAMGO using the formula K_e = [antagonist]/(dose ratio  1). Dose ratio is the ratio of the EC₅₀ for an agonist in the presence and absence of the antagonist (Kosterlitz and Watt, 1968).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

To permit [³⁵S]GTPS binding to be used as a measure of G protein activation, digitonin (20 µM) was used to permeabilize the C6 µ cells, allowing [³⁵S]GTPS entry into the cell. Permeabilized cells appeared otherwise unchanged and were the same shape and size as control cells. Digitonin-treated cells showed a basal level of [³⁵S]GTPS binding that was not seen in nonpermeabilized cells. (Fig. 1). The µ-agonist DAMGO produced a concentration-dependent increase in [³⁵S]GTPS binding in digitonin-permeabilized cells, while DAMGO stimulation of [³⁵S]GTPS binding was not observed in nonpermeabilized cells (Fig. 1). Addition of 20 nM naloxone produced a 5-fold parallel rightward shift in the DAMGO concentration-response curve (Fig. 2), yielding a calculated K_e (affinity) value for naloxone of 5 nM, consistent with its affinity for the µ-receptor (Alt et al., 1998).

View larger version (14K): [in this window] [in a new window]   Fig. 1.   Effect of digitonin treatment of C6 µ cells on [35S]GTPS binding. Concentration-response curves were determined for DAMGO in control cells () and cells treated with 20 µM digitonin (), as described under Materials and Methods. Shown are the mean and S.E. (bars) from three independent experiments.

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Effect of naloxone on DAMGO-stimulated [35S]GTPS binding in C6 µ cells permeabilized with digitonin. Permeabilized cells were incubated with DAMGO in the absence () or presence () of 20 nM naloxone. Shown are the mean and S.E. (bars) from three independent experiments. Addition of 20 nM naloxone produced a 0.70 ± 0.05 log rightward shift in the DAMGO concentration-response curve.

To ascertain whether digitonin treatment altered the ability of µ-receptors to bind agonist, membranes were prepared from both untreated cells and cells treated with digitonin. DAMGO displacement of [³H]diprenorphine was the same in both sets of membranes. DAMGO displaced [³H]diprenorphine with an IC₅₀ of 12 nM (95% CI, 5-29 nM) in membranes from control cells versus 14 nM (95% CI, 5-38 nM) in membranes from digitonin-permeabilized cells.

The ability of several well studied µ-opioids to stimulate [³⁵S]GTPS binding in the permeabilized C6 µ cell preparation was measured (Table 1). Relative µ-agonist efficacy was DAMGO > fentanyl  morphine > buprenorphine. Nalbuphine showed no efficacy in this system. Agonists exhibited potency in rank order: buprenorphine (EC₅₀ = 2 nM) > morphine > fentanyl > DAMGO (EC₅₀ = 224 nM), although the 95% CIs for morphine, DAMGO, and fentanyl overlapped.

View this table: [in this window] [in a new window]   TABLE 1 Intrinsic activity and potency of µ opioid ligands Intrinsic activity is defined as the maximum stimulation of [35S]GTPS binding produced relative to DAMGO and is given as mean ± S.E. EC50 values given are the geometric mean and the values in parentheses represent 95% confidence intervals. Data represent three experiments for each drug, carried out in duplicate.

To examine whether the rate and maximal level of G protein activation are dependent upon receptor concentration, C6 µ cells were treated with the irreversible opioid antagonist -FNA prior to harvest and digitonin treatment. Cells pretreated for 1 h with 10 nM -FNA were found to have 4.3 ± 0.5 × 10⁴ receptors/cell, compared with 1.96 ± 0.38 × 10⁵ receptors/cell for control cells, i.e., a reduction of 77%, as measured by [³H]diprenorphine binding (Fig. 3). The determined affinity of [³H]diprenorphine was identical in control and -FNA cells (K_d = 0.33 ± 0.04 nM for both control and -FNA-treated cells). Cells treated with -FNA exhibited 29 ± 10% of the DAMGO stimulation of [³⁵S]GTPS binding seen in control cells, with no change in the t_1/2 (18 ± 3 min for -FNA-treated cells versus 20 ± 5 min for control; Fig. 4). Increasing the DAMGO concentration to 1 mM did not increase the level of [³⁵S]GTPS bound. -FNA had no effect on basal [³⁵S]GTPS binding, or on [³⁵S]GTPS binding in the absence of GDP (data not shown), indicating that -FNA does not directly interfere with the binding of guanine nucleotide to G proteins.

View larger version (19K): [in this window] [in a new window]   Fig. 3.   Effect of -FNA treatment on receptor levels in C6 µ cells. a, saturation binding of [3H]diprenorphine to permeabilized cells () or cells treated with 10 nM -FNA prior to permeabilization (). b, Scatchard analysis of the saturation binding data. Cells treated with -FNA retain 23 ± 3% of control [3H]diprenorphine binding. Data from one of five independent experiments are shown. Experiments were performed on different passages of cells, showing a consistent receptor number across passages. Each individual experiment with control and -FNA-treated cells was performed in parallel from cells of the same passage number.

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Time course of DAMGO-stimulated [35S]GTPS binding in permeabilized C6 µ cells. DAMGO (10 µM)-stimulated [35S]GTPS binding was measured in permeabilized cells () and cells treated with 10 nM -FNA prior to permeabilization (). -FNA-treated cells are able to produce only 29 ± 10% of the [35S]GTPS binding stimulation seen in control cells. Shown are mean values and S.E. (bars) from five experiments.

It is possible that the ability of DAMGO to activate G protein is reduced during the assay due either to desensitization or a general deterioration of the system. Preincubation of permeabilized cells at 25°C with 100 µM DAMGO for up to 2 h before addition of [³⁵S]GTPS (for 10 min) did not change the level of DAMGO-stimulated [³⁵S]GTPS binding (226 ± 24 and 202 ± 5% stimulation over basal binding after a 30-min or a 2-h preincubation, respectively).

To determine the number of µ-opioid receptor-accessible G proteins in the digitonin-permeabilized cells, binding of [³⁵S]GTPS at varying concentrations was performed in the presence and absence of 100 µM DAMGO; the nonstimulated [³⁵S]GTPS binding was subtracted from the DAMGO-stimulated value for each point to give a saturation binding isotherm (Fig. 5). Since the dissociation of GTPS from G proteins is not readily reversible (Higashijima et al., 1987), [³⁵S]GTPS binding does not represent equilibrium and it is inappropriate to make affinity (K_d) estimates from these data. The B_max, however, should not be affected. The total number of DAMGO-stimulated G proteins was determined to be 7 ± 2 × 10⁵/cell. This provides for approximately four G proteins per receptor in the permeabilized cell.

View larger version (14K): [in this window] [in a new window]   Fig. 5.   DAMGO-stimulated [35S]GTPS binding. Saturation binding of [35S]GTPS was performed in the presence and absence of 100 µM DAMGO. Nonstimulated [35S]GTPS binding was subtracted from the DAMGO-stimulated value at each point to determine the number of G proteins activated by the µ-opioid receptor. Data from one of three independent experiments are shown.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

This study was designed to evaluate permeabilized whole C6 µ cells as a model for µ-opioid agonist activation of G proteins. Once established the model was used to test the accuracy of the collision-coupling theory of receptor-effector interaction, in which G proteins are activated by agonist-bound receptors that make contact with the G protein by random diffusion (Tolkovsky and Levitzki, 1978; Stickle and Barber, 1991).

[³⁵S]GTPS cannot pass through the cell membrane. Treatment of C6 µ cells with digitonin resulted in permeablization of the cell and allowed the well characterized [³⁵S]GTPS binding assay (Traynor and Nahorski, 1995; Wieland et al., 1995; Emmerson et al., 1996) to be used as a measure of opioid agonism. Agonist stimulation of [³⁵S]GTPS binding in this system was concentration-dependent and antagonized in a competitive manner by naloxone. The potency (EC₅₀) values for opioid agonists were somewhat higher than seen using a conventional [³⁵S]GTPS assay with membranes from these cells (Lee et al., 1999), and were similar for morphine, DAMGO, and fentanyl. However, the EC₅₀ values are not dissimilar to values obtained under more stringent conditions where morphine and DAMGO have similar potencies (Alt et al., 1998), or in rat thalamic membranes (Selley et al., 1997). Relative efficacy decreased in the same rank order as would be predicted from studies in membrane preparations (Emmerson et al., 1996; Lee et al., 1999), namely, DAMGO > fentanyl > morphine > buprenorphine > nalbuphine. It is worth noting that the efficacy requirements for full agonism in this system are very high, resulting in a wider separation between full and partial agonists than has been previously observed in membranes (Emmerson et al., 1996). However, the ability of peptidic µ-agonists such as DAMGO to produce a greater response than nonpeptide agonists has been previously reported in membranes from C6 µ cells under conditions that increase efficacy requirements for agonism (Alt et al., 1998) and is presumably a reflection of the intrinsic efficacies of the µ-agonists (Emmerson et al., 1996).

Previous experiments using membranes from C6 µ cells have suggested some form of organized association of receptors and G protein (Remmers et al., 2000), as opposed to a random-encounter model. However, inferences that may be drawn from these experiments are limited because they may reflect an artifact of the membrane system; G proteins may be artificially compartmentalized by the limited size of the resulting membrane fragments. Indeed, these results appear to be incompatible with findings in erythrocyte membranes (Pike and Lefkowitz, 1981) and rat adipocyte membranes (Murayama and Ui, 1984) that different G_s-coupled receptors share a common pool of G_s. Similarly, in hamster adipocyte membranes inhibitory G proteins appear to communicate between different types of G_i-coupled receptors (Murayama and Ui, 1984) and in membranes from transfected COS-7 cells opioid and cannabinoid receptors access the same pool of G (Shapiro et al., 2000). On the other hand, opioid and cannabinoid receptors in membranes from neuroblastoma cells access different G proteins (Shapiro et al., 2000), and in SH-SY5Y cells there is evidence that µ- and -opioid receptors show a preference for different inhibitory G subtypes (Laugwitz et al., 1993).

To further study these divergent findings we have used permeabilized C6 µ cells, where free access of receptors to G protein across the whole cell membrane should be possible. In these cells we find support for an organization of receptors and G protein. As stated in the introduction the collision-coupling model predicts a reduction in receptor concentration would decrease the likelihood of a random encounter and should therefore decrease the rate of G protein activation, but should not affect the maximum number of G proteins activated. Treatment of cells with -FNA prior to permeabilization reduced the receptor levels to approximately one-quarter of the receptor concentration of control cells as measured by [³H]diprenorphine binding and also reduced the maximum level of DAMGO-mediated G protein activation commensurate with this reduction in receptor concentration. This demonstrates that receptors remaining after -FNA treatment are not able to access all available G proteins, and so access must be restricted in some way. The rate (t_1/2) for the binding of [³⁵S]GTPS in permeabilized cells was the same in control and -FNA-pretreated cells. Thus, the interaction of those receptors remaining after -FNA treatment with G proteins to which they have access is not altered. These findings are not compatible with the hypothesis that interaction of opioid µ-receptors with G protein occurs by random encounter across the permeabilized C6 µ cell membrane, but suggest a model in which there is some form of organization of receptors and G proteins.

Assumptions made in the present analysis are that the system remains active over the entire 2-h duration of the time course experiments and the -FNA is exerting its effect only by inactivating receptors. No significant effect of a preincubation with the highly efficacious µ-agonist DAMGO was observed. Thus, there appeared to be neither receptor/G protein desensitization nor degredation of any components over the duration of these experiments. Also, -FNA had no detectable effect on the ability of G proteins to bind [³⁵S]GTPS under these conditions. A less testable assumption is that -FNA alkylates receptors in a purely random manner. It is conceivable that -FNA may selectively block µ-receptors that are in particular states (Franklin and Traynor, 1991) which may be more or less able to activate G proteins, and this would be a potential confound to these results. Additionally, the relationship to other endogenous receptors coupled to inhibitory G protein that might be expressed in C6 cells is unknown. For example, these cells have been reported to express µ- and -receptors (Bohn et al., 1998). However, we detect no opioid binding in these cells in the absence of transfection (Lee et al., 1999) and so interference with endogenous opioid receptors is not likely. Finally, the C6 µ cell is an artificial system, overexpressing µ-receptors. A high number of artificially expressed receptors might be expected to provide for promiscuity so the fact that a random collision model is not supported in this cell is strong evidence for an organizational model of receptor-G protein coupling.

In permeabilized C6 µ cells the ratio of µ-agonist-stimulated G proteins to µ-receptors was shown to be approximately 4:1. The observation that the reduction seen in maximal G protein activation closely matches the reduction in receptor number suggests that each receptor is associated with approximately four G proteins. The ratio of receptor to µ-agonist-stimulated G protein in C6 µ membranes (Remmers et al., 2000) is the same as the ratio determined in the permeabilized cells. Thus, since the cell homogenization and membrane preparation process does not disrupt this organization, the mechanics of compartmentalization must be very closely associated with the membrane. The current study provides no information as to what type of mechanism may link G proteins to individual receptors, but two possibilities present themselves. One is that receptors and G proteins may be held in physical proximity by large protein complexes. There is ample evidence that such complexes play a role in targeting G proteins to effector molecules (Choi et al., 1994; Leeuw et al., 1995; Whiteway et al., 1995) and G proteins could be targeted to receptors in a similar manner (Neubig, 1994, 1998). Another explanation is that receptors and G proteins are compartmentalized by cytoskeletal divisions of the cellular membrane (Edidin et al., 1991). With either of these situations a reduction in receptor number by alkylation with -FNA would result in a loss of the ability to activate those G proteins that were associated with the alkylated receptors. This would give a decrease in maximal [³⁵S]GTPS binding. However, since G proteins would be associated with a particular receptor and not require a random encounter to be activated the t_1/2 for activation of the G proteins associated with the remaining receptors would remain constant.

The contrasting findings in erythrocytes (Pike and Lefkowitz, 1981), adipocytes (Murayama and Ui, 1984), and COS-7 cells (Shapiro et al., 2000) of free access of different receptors to the same G protein pool can be rationalized on the basis of cellular differences in organization. Alternatively, compartmentalization of receptors and G protein may be highly complex and involve several different receptor classes that couple to a particular type of G protein. One possible mechanism for this would be through the formation receptor heterooligomers (Jordan and Devi, 1999; George et al., 2000).

In conclusion, the permeabilized C6 µ cell provides a convenient model for studying the activation of G proteins by µ-opioid agonists in a system that closely resembles the intact cell. The interaction of G proteins and activated µ-receptors in the permeabilized C6 µ cell does not support a random encounter-coupling model. Rather, it supports some organization of receptors and G protein. Since similar findings are observed in membrane preparations (Remmers et al., 2000), the mechanism of organization does not require an intact cell to maintain its integrity and must be very highly associated with the membrane.