Evidence for opioid receptor-mediated activation of the G-proteins, Go and Gi2, in membranes of neuroblastoma x glioma (NG108-15) hybrid cells.

In membranes of neuroblastoma x glioma (NG108-15) hybrid cells, the photoreactive GTP analog, [alpha-32P] GTP azidoanilide, was incorporated into 39-41-kDa proteins comigrating in urea-containing sodium dodecyl sulfate-polyacrylamide gels with immunologically identified G-protein alpha-subunits, i.e. a 39-kDa Go alpha-subunit, a 40-kDa Gi2 alpha-subunit, and a 41-kDa Gi alpha-subunit of an unknown subtype. The synthetic opioid, D-Ala2,D-Leu5-enkephalin (DADLE), stimulated photolabeling of the 39-41-kDa proteins. In the presence of GDP, which increased the ratio of agonist-stimulated to basal photolabeling, DADLE at a maximally effective concentration stimulated photolabeling of the 39- and the 40-kDa protein 2-3-fold. Somatostatin, adrenaline, and bradykinin were less potent than DADLE and, to varying degrees, stimulated photolabeling of the 40-kDa protein more than that of the 39-kDa protein. Prostaglandin E1 was inactive. The present data represent direct evidence for an activation of endogenous Go and Gi2 via opioid receptors and other receptors in the native membrane milieu.

Similar to neurons, neuroendocrine, and pituitary cells (8), NxG cells possess subtypes of Gi and Go (5,9). Activation of opioid receptors in NxG cells results in the inhibition of adenylylcyclase (10) and the inhibition of voltage-dependent Ca2+ channels (4, 11). In PTX-treated NxG cells, the opioidinduced inhibition of Ca2+ currents is efficiently restored by intracellular application of purified Go (4). Evidence for coupling of opioid receptors to Gi and Go comes from reconstitution experiments with the p-opioid receptor purified from rat brain and G-proteins purified from porcine and rat brain (12). While the data obtained in reconstitution experiments demonstrate the ability of activated opioid receptors to couple to exogenous Go or Gi, they do not necessarily imply that coupling to these G-proteins occurs in the native plasma membrane.
In order to identify endogenous G-proteins activated via opioid receptors, we studied in membranes of NxG cells the effects of the synthetic 6-opioid receptor agonist, DADLE, and, for comparison, of other receptor agonists on photolabeling of G-proteins with the photoreactive GTP analog, [a-32P]GTP azidoanilide.
The reaction was stopped by cooling the sample on ice. After centrifugation a t 12,000 X g for 5 min at 4 'C (for removal of unbound [cz-~'P]GTP azidoanilide), the obtained membrane pellets were resuspended in 60 pl of a modified GDP-free incubation buffer supplemented with 2 mM dithiothreitol. Membrane suspensions were then transferred onto Parafilm, on which they formed a droplet, and were irradiated for 10 s a t 4 "C with a 254 nm/150 watts/7 X 30-cm UV light source (Vilber Lourmat, Torcy, France). The distance of the light source from the samples was 3 cm. Thereafter, membrane suspensions were retransferred into the assay tubes, again centrifuged, and prepared for SDS-PAGE by adding sample buffer (15).
SDS-PAGE, Immunoblotting, and Antisera-SDS-PAGE (18), blotting of proteins onto nitrocellulose filters, and autoradiography of gels were performed as described (16) with the following modifications. The separating gels contained 8% (w/v) acrylamide and 4.3 M urea (17). In some instances, the 39-41-kDa regions of the dried gels were cut out and shaken in 1 ml of 30% (v/v) H202 for at least 1 h, and the incorporated radioactivity was counted after the addition of 5 ml of scintillant. Nitrocellulose filters were incubated with antisera generated against synthetic peptides corresponding to confined regions of G-protein a-subunits. Properties of the employed antisera and immunostaining of filters have been described elsewhere (9,13).
Miscellaneous, Reproducibility of Data-Protein was determined according to Lowry et al. (18). Cell culture, differentiation of NxG cells with dibutyryl CAMP, and preparation of membranes were performed as described (4,9). Autoradiograms were scanned with a laser densitometer (LKB 2202 Ultroscan). The experiments shown are representative for three or more independently performed exper-

When membranes of NxG cells were incubated with [(Y-~'P]
GTP azidoanilide and subsequently exposed to UV light, photolabeling of 39-, 40-, and 41-kDa proteins was observed (Figs. [2][3][4][5]. For identification of these proteins, membranes were photolabeled with [cx-~'P]GTP azidoanilide; subsequently, proteins were separated by SDS-PAGE and blotted onto nitrocellulose filters. The filters were first autoradiographed and then incubated with different antisera; filterbound antibodies were visualized by a color reaction (Fig. 1). The aeommon peptide antiserum, which recognizes the a-subunits of G,, Gi, and Go, detected three immunoreactive proteins of 39, 40, and 41 kDa, comigrating with the three photolabeled proteins. The a . peptide antiserum, which specifically recognizes the two forms of the Go a-subunit, reacted with a 39-kDa protein. The ai(common) peptide antiserum, which recognizes the a-subunits of Gil, Gi2, and Gi3, reacted with two proteins of 40 and 41 kDa. A 40-kDa protein was also detected by the L~I~ peptide antiserum which only recognizes the Gi2 asubunit. The 41-kDa Gi a-subunit was not detected by antisera which recognize the a-subunit of Gil or Gi3.' Thus the photolabeled proteins of 39, 40, and 41 kDa apparently correspond to a-subunits of Go, Gi2, and an unidentified Gi subtype, respectively.
Further data support the assumption that the photolabeled proteins in the 40-kDa region represent a-subunits of PTXsensitive G-proteins. Modification of G-protein a-subunits by PTX leads to a decreased mobility of G-protein a-subunits in urea-containing SDS-polyacrylamide gels (17). If membranes were prepared from PTX-treated cells, the by far major portion of photolabeled proteins showed a decreased mobility (-2 kDa) accompanied by a corresponding decrease in the mobility of G-protein a-subunits recognized by the various antisera (not shown).
The 6-opioid receptor agonist, DADLE, stimulated photolabeling of the 39-41-kDa proteins. In an incubation medium devoid of GDP, the stimulatory effect of DADLE was rather small (Fig. 2). Since the nucleotide decreased photolabeling of 39-41-kDa proteins more in the absence of DADLE than in its presence, it increased the portion of photolabeling sensitive to the peptide. The present data are highly consistent with those reported by Florio and Sternweis (19). By measuring GDP binding to Go in a reconstituted system, they provided evidence that unstimulated G-proteins possess a relatively high affinity for GDP, whereas activated G-proteins prefer GTP (or a GTP analog). This may explain the different sensitivities of photolabeling toward GDP in the absence and presence of receptor agonists (see also Ref. 13).
Photolabeling of 39-41-kDa proteins increased with the time elapsed after addition of [a-"P]GTP azidoanilide to the reaction mixture (Fig. 3). DADLE stimulated photolabeling at all incubation times (ranging from 0.5 to 12 min). At incubation times of 0.5-1.5 min, the 39-and 40-kDa protein incorporated approximately equal amounts of radioactivity in the absence of DADLE; in the presence of DADLE, the 39-kDa protein was the preferentially photolabeled protein. At incubation times of 3-6 min, the 39-or 40-kDa protein was preferentially photolabeled, depending on whether or not DADLE was present in the reaction mixture. Only at long incubation times (12 min) was the 40-kDa protein the preferentially photolabeled protein, irrespective of the absence and presence of DADLE. The findings indicate that at incubation times of 0.5-6 min, DADLE promotes nucleotide exchange on the 39-kDa protein more than that on the 40-kDa protein. Since [a-"PIGTP azidoanilide is a poorly hydrolyzable GTP analog (20), [cY-~'P]GTP azidoanilide-liganded Gproteins (ie. activated G-proteins) will accumulate, in contrast to the intact cell. Thus the photolabeling pattern ob- served after prolonged incubation reflects the accumulation of activated G-proteins, irrespective of the kinetics of receptor/G-protein interaction. Yatani and Brown (21) provided evidence that membrane-confined p-adrenoceptor/G.-mediated stimulation of cardiac Ca2+ channels requires less than a second. Therefore, photolabeling patterns observed after short incubation intervals are more likely to reflect the receptor/G-protein interaction in vivo.
In further experiments, an incubation time of 3 min was chosen since it (i) allowed sufficient incorporation of radioactivity and (ii) the ratio of agonist-stimulated to basal photolabeling was satisfactory. Under the conditions described under "Experimental Procedures,'' DADLE dose dependently stimulated photolabeling of 39-41-kDa proteins (Fig. 4). Densitometric evaluation of autoradiograms revealed that DA-DLE was half-maximally and maximally active at concentrations of 10-100 nM and 1 pM, respectively (not shown). Similar concentrations were reported for maximal and halfmaximal effects of DADLE on GTPase and adenylylcyclase activity in membranes of NxG cells (22). Employed at a maximally effective concentration, DADLE increased the incorporation of radioactivity into the 39-41-kDa proteins about 2-3-fold (see Fig. 2-3). Photolabeling of the 39-kDa protein was slightly more stimulated than that of the 40-kDa protein. A quantitative evaluation of the stimulatory effect of DADLE on photolabeling of the 41-kDa protein was not possible. DADLE did not stimulate photolabeling of proteins modified by PTX (not shown). The stimulatory effect of DADLE was abolished by the opioid receptor antagonist, naloxone (see Fig. 4), indicating that the effect of DADLE was due to activation of opioid receptors.
Other receptor agonists were employed to compare their effects on photolabeling of membrane proteins with that of DADLE. Prostaglandin El, a potent stimulator of adenylylcyclase in membranes of NxG cells (23), had no effect on photolabeling of 39-41-kDa proteins (see Fig. 4). In contrast, somatostatin, adrenaline, and bradykinin, which activate pertussis toxin-sensitive G-proteins in membranes of NxG cells (10,(24)(25)(26), stimulated photolabeling of 39-41-kDa proteins (Fig. 5). None of these receptor agonists was as potent as DADLE. Somatostatin, adrenaline, and particularly brady- kinin showed a preference for the 40-kDa protein at all concentrations tested. Maximal stimulations of these agonists on photolabeling of the 40-kDa protein were comparable (about 1.5-fold). The differences among the employed receptor agonist were highly reproducible with different batches of membranes or [ (u-~*P] GTP azidoanilide. All receptor agonists stimulated photolabeling of the 41-kDa protein (see Figs. 3 and 4). However, photolabeling of the 41-kDa protein was too small to be well documented by densitometric scanning of autoradiograms (see Fig. 5).
Small amounts of [w3*P]GTP azidoanilide were also incorporated into a 32-kDa protein (see Fig. 4). Photolabeling of this protein was not affected by DADLE or GDP and occurred in the absence of M e . Its mobility in urea-containing SDSpolyacrylamide gels was not affected by PTX (not shown). In addition, photolabeled 32-kDa proteins in NxG and RINm5F Opioid Receptor-mediated Activation of Go and Gi2 cells (not shown) and in GH3 cells (13) migrated faster (by at least 3 kDa) than proteins with molecular masses of less than 39 kDa, which were recognized by various aeommon peptide antibodies. Thus, photolabeled 32-kDa proteins are apparently not related to G-protein a-subunits. Limited proteolysis of a 32-kDa protein photolabeled with [cx-~'P]GTP azidoanilide in membranes of rat cerebral cortex revealed that it was not related to G-protein a-subunits (20).

DISCUSSION
The present data indicate that the &opioid receptor couples to G-proteins of the Gi and Go families. Evidence for the activation of different types of G-proteins by a single receptor has been provided in previous reports. Studies with membranes on adenylylcylase regulation and on binding of receptor agonists suggest that 8-adrenoceptors can couple not only to G. but also to Gi (27,28). In membranes of HL-60 cells, formyl peptides stimulate the cholera toxin-catalyzed ADPribosylation of two Gi subtypes (29). In addition, data obtained in reconstituted systems suggest that 8-adrenoceptors, popioid receptor, and muscarinic receptors interact with multiple G-proteins (12,30,31).
The method described here may be advantageous if compared with other experimental approaches designed to study receptor/G-protein interactions. (i) In contrast to the agonistsensitive ADP-ribosylation catalyzed by cholera toxin, it does not require the absence of guanine nucleotides, the natural ligands of G-proteins. Instead, it is based on a physiological response of G-proteins to receptor-induced activation, i.e. the increase in the exchange of guanine nucleotides on the asubunits. (ii) It appears to be more generally applicable than agonist-sensitive ADP-ribosylation. In membranes of HL-60 cells, leukotriene B4 stimulates photolabeling of G-protein asubunits (13) but does not affect cholera toxin-catalyzed ADP-ribosylation (32). Since the affinities of [a-32P]GTP azidoanilide to G,, G,, and Gi are very similar (33), the method should be suitable for examining the interaction of receptors with at least these families of G-proteins. (iii) It does not depend on a reconstituted system in which the interaction of signal transduction components may lack specificity (see the Introduction). (iv) In contrast to methods based on the determination of enzyme activities, it allows identification of receptor-activated G-proteins.
The observed differences among receptor agonists on photolabeling of G-protein a-subunits may reflect differences in their ability to regulate certain effectors. DADLE, somatostatin, and adrenaline (in the order of decreasing maximal a~t i v i t y )~ inhibit voltage-dependent Ca2+ currents in NxG cells (4, 11,34) and stimulate photolabeling of the 39-kDa protein comigrating with a G, a-subunit. Bradykinin does not inhibit Ca'+ currents in NxG cells in a pertussis toxin-sensitive m~n n e r .~ Compared with its ability to stimulate photolabeling of the 40-kDa protein, bradykinin has little effect on photolabeling of the 39-kDa protein. Considering the ability of Go to reconstitute the inhibitory modulation of ea2+ currents in various PTX-treated neuronal cells (4, 35-37) and the finding that antibodies against the Go a-subunit attenuate the Ca'+ current inhibition by noradrenaline in NxG cells (34) and by dopamine in snail neurons (36), the present data J. Hescheler, R. Eckert, and W. Trautwein, unpublished experiments.
are consistent with the hypothesis that Go couples activated inhibitory receptors (e.g. the &opioid receptor) to neuronal voltage-dependent Ca2+ channels.
All the employed receptor agonists stimulate effectively photolabeling of the 40-kDa protein, apparently representing the Giz a-subunit. Previously published data indicate that all these agonists inhibit adenylylcyclase in membranes of NxG cells (10,22,(24)(25)(26). Thus the present data support the notion that a G-protein of the Gi family, possibly Gi2 (5), is involved in the receptor-mediated inhibition of adenylylcyclase.