Specificity of receptor-G protein interactions. Discrimination of Gi subtypes by the D2 dopamine receptor in a reconstituted system.

The selectivity of D2 dopamine receptor-guanine nucleotide-binding protein (G protein) coupling was studied by reconstitution techniques utilizing purified D2 dopamine receptors from bovine anterior pituitary and resolved G proteins from bovine brain, bovine pituitary, and human erythrocyte. Titration of a fixed receptor concentration with varying G protein concentrations revealed two aspects of receptor-G protein coupling. First, Gi2 appeared to couple selectively with the D2 receptor with approximately 10-fold higher affinity than any other tested Gi subtype. Second, the G proteins differed in the maximal receptor-mediated agonist stimulation of the intrinsic GTPase activity. Gi2 appeared to be maximally stimulated by agonist-receptor complex with turnover numbers of approximately 2 min-1. The other Gi subtypes, Gi1 and Gi3, could be only partially activated, resulting in maximal rates of GTPase of approximately 1 min-1. Agonist-stimulated GTPase activity was not detected in preparations containing Go from bovine brain. The differences in maximal agonist-stimulated GTPase rates observed among the Gi subtypes could be explained by differences in agonist-promoted guanyl nucleotide exchange. Both guanosine 5'-3-O-(thio)triphosphate (GTP gamma S) binding and GDP release parameters were enhanced 2-fold for the Gi2 subtype over the other Gi subtypes. These results suggest that even though several types of pertussis toxin substrate may exist in most tissues, a receptor may interact discretely with G proteins, thereby dictating signal transduction mechanisms.

First, Giz appeared to couple selectively with the Dz receptor with -lo-fold higher affinity than any other tested Gi subtype.
Second, the G proteins differed in the maximal receptor-mediated agonist stimulation of the intrinsic GTPase activity. Gi2 appeared to be maximally stimulated by agonist-receptor complex with turnover numbers of -2 min-'. The other Gi subtypes, Gil and Gi3, could be only partially activated, resulting in maximal rates of GTPase of -1 min-'. Agonist-stimulated GTPase activity was not detected in preparations containing G, from bovine brain. The differences in maximal agonist-stimulated GTPase rates observed among the Gi subtypes could be explained by differences in agonist-promoted guanyl nucleotide exchange. Both guanosine 5'-3-O-(thio)triphosphate (GTPrS) binding and GDP release parameters were enhanced a-fold for the Giz subtype over the other Gi subtypes. These results suggest that even though several types of pertussis toxin substrate may exist in most tissues, a receptor may interact discretely with G proteins, thereby dictating signal transduction mechanisms.
In the pituitary, the major consequence of dopaminergic action is the inhibition of prolactin release from the anterior lobe. The action of dopamine in this tissue and other target tissues is presumably mediated through several transmembrane signaling pathways. The D2 dopamine receptor mediates inhibition of adenylyl cyclase in striatum and the anterior and neurointermediate lobes of pituitary (Giannatasio et al., 1981;McDonald et al., 1984;Cote et al., 1982 1983, 1986Simmonds and Strange, 1985;Enjalbert et al., 1986;Journot et al., 1987) and Ca*+ influx (Enjalbert et al., 1988;Schofield, 1983;Login et al., 1988aLogin et al., , 1988b. More recently, the stimulation of this receptor has been shown to activate K' channels in isolated primary lactotroph cells and striatal neurons (Castelletti et al., 1989;Margaroli et al., 1987;Vallar et al., 1988;Freedman and Weight, 1988). All of the observed effects of dopamine on various signaling systems are pertussis toxin sensitive, implicating the involvement of the Gi/G, family of proteins.' At present, three forms of Gi which are distinct gene products have been identified by molecular biology techniques (Itoh et al., 1986;Nukada et al., 1986;Jones and Reed, 1987;Van Meurs et al., 1987;Michel et al., 1986;Bray et al., 1987). Several forms of Gi are present in virtually every cell or tissue studied to date. Thus, due to the multiplicity and ubiquity of these proteins, the question has arisen of whether the specificity of receptor-effector coupling resides at the receptor-G protein level, the G protein-effector, or both. We have shown previously that partial purification of the pituitary D2 dopamine receptor by affinity chromatography results in co-purification of the receptor with a predominant pertussis toxin substrate of M, 40,000 ((Y subunit) (Senogles et al., 1987). These results suggested a certain selectivity between the interaction of the receptor and G proteins. In order to examine this question of specificity of signal transduction further, we reconstituted purified receptor and various resolved G proteins in phospholipid vesicles. The present studies reveal that the DZ dopamine receptor is able to discriminate G proteins in the Gi/G, family by two distinct mechanisms. The apparent affinity of the agonist-receptor complex for a given G protein is one such mechanism. The other is the ability of the receptor to stimulate (activate) a particular Gi subtype. Since these two mechanisms are evident in the interaction of G protein with the D2 dopamine receptor as well as rhodopsin, they may represent general mechanisms dictating the specificity of signal transduction.  (Senogles et al., 1986(Senogles et al., , 1987.

Purification of 02 Dopamine Receptor from Anterior Pituitary
The DQ receptor was purified as described below. Briefly, the solubilization and application to the affinity chromatography were carried out as described (Senogles et al., 1986) with the following changes.
Once absorbed on the affinity chromatography matrix, receptor preparations were washed with 2 bed volumes of 100 pM Gpp(NH)p in the affinity chromatography wash buffer (0.1% digitonin, 50 mM Tris-HCl, pH 7.2, at 25 "C, 100 mM NaCl, 10 mM EDTA, 10 mM EGTA, 5 pg/ml each of leupeptin, pepstatin, aprotinin, and 100 PM phenylmethylsulfonyl fluoride). This wash step was necessary to remove the endogenously associated G protein that copurifies with the anterior pituitary receptor (Senogles et al., 1987). The receptor was eluted and further purified on Datura stramonium agglutinin lectin as described previously (Senogles et al., 1988 Litman (1982). The rhodopsin was eluted from concanavalin A and stored at -135 "C in a buffer containing 0.1 mM methyl a-mannoside, 50 mM Tris-HCl, pH 7.2,l mM CaC&, and 0.3% octyl @-n-glucoside.

Purification of G Proteins
Brain-The purification of G,l, G,z, and G, from bovine brain was performed by modifications of both Katada et al. (1987) and Mumby et al. (1988 Inc. fast protein liquid chromatography system. Transducin-Transducin was purified as described by Gierschik et al. (1984).
The identity of the various G proteins was confirmed by the relative migration on SDS-polyacrylamide gel electrophoresis and Western blotting with subtype-specific antibodies (see "Results").

Reconstitution of G Proteins and Dz Dopamine Receptor
The co-reconstitution of receptor and G proteins was performed as follows.
DP dopamine receptor, usually 2-5 pmol (250 pl), in a buffer of 50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 1 mM EDTA, 0.05% digitonin, and 1 PM haloperidol, was incubated with 50 ~1 of 5 mg/ml soybean phosphatidylcholine, 50 ~1 of 20 mg/ml BSA, and buffer (50 mM Tris-HCl, pH 7.4) to yield a tinal volume of 500 ~1. G proteins (l-50 pmol) were added to this solution, and the entire mixture was incubated on ice for -1 h. The concentration of CHAPS in the reconstitution mixture was adjusted to achieve a uniform concentration with the various preparations of G proteins. After incubation on ice, the mixture was applied to a l-ml Kxtractigel column preequilibrated with 50 mM Tris-HCl. 100 mM NaCl. 1 me/ml BSA. The first 300 ~1 of the column elution was discarded, anzthe next 500 ~1 of phospholipid vesicles was collected and used for assays.

Reconstitution of G Proteins and Rhadopsin
Rhodopsin (usually 20-30 pmol/50 ~1) was incubated with 50 ~1 of 20 me/ml BSA. 50 ~1 of 5 me/ml sovbean nhosnhatidvlcholine. and buffer' (50 mM Tris:HCl, pH-7.4, lob mM NaCI, 1 mM EDTA) to a final volume of 500 ~1. G proteins (l-20 pmol) were added to the solution and incubated for 1 h on ice. The entire reconstitution was done under dark room conditions as much as possible, prior to assay. The samples were applied to Extractigel columns as described above.
[cI-~~P]GDP Release-GDP release was performed essentially as described by Brandt and Ross (1985 Protein was quantitated by the method of Bradford (1976) using BSA as the protein standard.

Identity and Purity of Various G Protein Preparations
G proteins with (Y subunits of 41, 40, and 39 kDa were purified to apparent homogeneity from bovine brain. A 40-kDa G protein was purified from bovine anterior pituitary, and a 41-kDa G protein was purified from human erythrocyte. These resolved proteins were subjected to SDS-polyacrylamide gel electrophoresis, and the Coomassie Blue stain of this gel is shown in Fig. lA. The identity and apparent purity of the G proteins were confirmed by Western blotting with G, subtype-specific antipeptide antibodies characterized previously (Goldsmith et al. 1988a, 198813) (Fig. 1B). The 41-kDa protein purified from brain was identified on the basis of reactivity with LD antisera to be G,l. Both of the 40-kDa proteins from brain and pituitary were confirmed to be Gi2 on the basis of reactivity with LE3 antisera. The Il-kDa protein from human erythrocyte was identified as Gi3 on the basis of cross-reactivity with SQ antisera. A 1-pg sample of each G protein was electrophoresed on a 10% SDS-polyacrylamide gel electrophoresis. Panel A shows the Coomassie Blue staining pattern of this gel. Panel B, a 500-ng (upper) and 5-gg (lower) sample of G proteins were electrophoresed and transferred to nitrocellulose. The lane containing either 5 rg or 500 ng of protein was aligned on a slot-blot apparatus and nrobed with the four soecific antisera: slot 1. LE (Gi, snecific): slot 2. LD (G,, specific); slot 3, GC (G, specific), and slot 434 (Gis specific): The blot was developed with goat anti-rabbit conjugated alkaline phosphatase. Shown on each side are the molecular masses of the (Y subunits (41,40,39 kDa) and of the fl subunit (35 kDa).
brain was confirmed to be G, by cross-reactivity with GC antibody. The immunoblots were performed with two protein concentrations: 500 ng (upper) and 5 pg (lower) for each G protein. The lower protein concentration is clearly detectable with our system and demonstrates that the preparations are >90% homogeneous with regard to subtype since no bands other than the major bands are detected at the higher protein concentration. The specific activity (data not shown) of these preparations (9-10 nmol of [35S]GTPyS binding/mg of holoprotein) indicates that all of the proteins are >90% functional, as defined by their ability to bind GTP-+S (theoretical specific activity, -11 nmol/mg for the holoprotein) and to serve as substrates for pertussis toxin (data not shown).

Characterization of the D2 Dopamine Receptor-G Protein Coupling in Phospholipid Vesicles
The ability to detect agonist-dependent coupling between receptor and G protein depends greatly on the reconstitution and assay conditions. For example, addition of most exogenous detergents to promote micelle formation such as octyl p-D-glucoside, Lubrol PX, or cholate to the purified Dz receptor from anterior pituitary prevented subsequent detection of agonist-dependent activation of the G protein after reconstitution. For this reason, the latter steps of the brain and pituitary G protein purification were carried out with CHAPS as the detergent, as CHAPS can be used in the reconstitution protocol without interference.
The agonist dependence of G protein activation was highly dependent on the free metal concentration. Fig. 2  To assess the relative affinities of various G proteins for the Dz dopamine receptor, a fixed receptor concentration was reconstituted with increasing amounts of G proteins, and the G protein-receptor interactions were monitored by GTPase and [%]GTPyS binding. Fig. 3 shows the titration of D2 dopamine receptor with Gi2 isolated from bovine anterior pituitary. When G protein alone was reconstituted into phospholipid vesicles, GTPase activity for a fixed time was essentially linear with increasing concentrations of G protein in the vesicles (open circles). However, when the Dz receptor (-75 fmol/reconstitution) was co-reconstituted with G protein and assayed in the presence (closed triangles) or absence (open triangles) of agonist, a deviation from linearity was observed. At low concentrations of Gi2, GTPase activity in the presence of receptor was identical to that of the G protein alone. However, with increasing Giz concentrations, the activity of the vesicles containing receptor and G protein was significantly higher than the vesicles with G protein alone. Thus, there appeared to be significant interaction of receptor and G protein at higher G protein/receptor ratios which was not agonist driven. These data were transformed to show the -fold stimulation by agonist of the GTPase turnover number at each of the G protein concentrations (Fig. 3, inset). Stimulation by agonist is biphasic over the range of G protein concentrations used in the experiments due to the lack of agonist-dependent activation at high G protein/receptor ratios. This pattern was consistently observed with all Gi subtype proteins used in this study except G,, which did not interact with the Dz receptor (Fig. 4).

Specificity of G Protein-D2 Dopamine Receptor Coupling
Using Gi Subtypes The pituitary D2 dopamine receptor was co-reconstituted with Gil, Giz, and G, from bovine brain; Giz from bovine pituitary; and Gi3 from human erythrocyte. The data were The data shown were obtained as in Fig. 3 and transformed to give -fold agonist stimulation of GTPase activity. The data are a compilation of four to seven experiments for each G protein. The symbols represent: Gil (0), Giz (m), Gi3 (A), G, (v). The basal rates of GTPase for all of the Gi subtypes were 0.17-0.21 mol of Pi released/min/mol of G protein.
transformed as described for Fig. 3 and are shown in Fig. 4. The data obtained with Giz from either the brain or pituitary source were indistinguishable in terms of this study. Reconstitution experiments using Giz (closed squures) showed agonist activation at lower G protein/receptor ratios (maximal stimulation at ratios of 3-5) than any other G protein tested (maximal at G protein/receptor ratios of 25-30). Gi3 from human erythrocyte (open triangles) and Gil from brain (open circles) were maximally activated at approximately a IO-fold higher G protein/receptor ratio than Giz. The stimulation of the molar turnover number by agonist was -lo-fold with any of the Gi2 preparations.
Experiments with the Gil and Gi3 subtypes showed only 4-5-fold stimulation of turnover numbers by agonist. No observable agonist-stimulated change in turnover number was observed in co-reconstitution experiments with G, (closed inverted triangles). Several preparations of G, were tested with similar results. The addition of GDP to the incubation or reconstitution did not confer agonist stimulation to the G, preparations. However, these preparations of G, were active, as judged by stimulation by mastoparan (Higashijima et al, 1988), as -IO-fold stimulation of GTPase activity resulted with 100 PM mastoparan (data not shown).

Comparison of G-Protein-Rhodopsin
Coupling Using Transducin and Gi Subtypes To assess whether the pattern of coupling observed with the Dz dopamine receptor was unique to this receptor, similar experiments were performed using rhodopsin purified from bovine rod outer segments. The data from titration of a fixed rhodopsin concentration with increasing amounts of various G proteins are shown in Fig. 5. Because of the inherent difficulty in obtaining truly inactive rhodopsin (absence of light), the data generated with vesicles containing the G protein alone were used as the basal rate. Since this parameter is linear with increasing G protein, the transformed data do not show the biphasic pattern observed with the Dz receptor. Transducin is maximally stimulated by light-activated rho-by guest on March 20, 2020 http://www.jbc.org/ Downloaded from dopsin at low G protein/rhodopsin ratios (G protein/receptor ratio of Z-3). Gil, Giz, and G, appear to be -5-&fold lower in affinity for rhodopsin compared with transducin, as their maximal activation occurs at G protein/receptor ratios of -10. Rhodopsin was able to stimulate the turnover number for transducin -IO-fold, which is approximately 3-fold more than any of the other Gi subtypes. Thus, the same patterns The Dt dopamine receptor (1 pmol) was co-reconstituted with -5 nmol of G;2 or -20 nmol of Gi,. of specificity can be evidenced for rhodopsin/transducin as were obtained for the Dz dopamine receptor/Gin.

Effect of Receptor-G Protein Coupling on Guunine Nudeotide Exchange
The functional differences in coupling with the Dz dopamine receptor observed among Gi subtypes in Fig. 4 were of two kinds: (a) affinity of the receptor for the G protein; and (b) ability of the agonist-receptor complex to stimulate the intrinsic GTPase activity. The GTPase cycle can be thought of as two kinetic processes, as reviewed recently by Freissmuth et al. (1989). The basal GTPase activity is influenced by the rates of the hydrolytic step of GTP cleavage and the rate of release of the product, GDP. Much evidence (reviewed by Gilman, 1987) has suggested that the rate-limiting step for basal GTPase in the cycle is the release of GDP, which is l/ 10 as fast as the hydrolytic step. We chose to investigate the guanine nucleotide exchange reactions in order to explain the  Dz dopamine receptor co-reconstituted with Giz, Gil, Gi3, or G,. The ratios of G protein to receptor were chosen to be the same ratio as those that gave maximal agonist-stimulated values from Fig. 4. The [35S]GTPyS binding to Gil, Giz, and Gi3 was increased in the presence of agonist. G, displayed no agonist-stimulated binding, which was consistent with the observations in Fig. 4. Table I

DISCUSSION
Previous studies of receptor G protein coupling have revealed some striking differences in the patterns of coupling. For example, the /3-adrenergic receptor couples to G, more efficiently than either Gi, G,, or transducin (Cerione et al., 1985), as evidenced by agonist stimulation of GTPase activity. By the same criteria, the cuz-adrenergic receptor appears to couple more efficiently with either Gi -G, > transducin > > G, (Cerione et al., 1986). However, studies that have attempted to delineate receptor-G protein coupling further have failed to show receptor-mediated specificity within the Gi or G. isoforms. Reconstitution studies with crude DP dopamine receptors (O'Hara et al., 1988), muscarinic receptors (Haga et al., 1985(Haga et al., , 1986Kurose et al., 1986), or partially purified prostaglandin E1 receptors (Negishi et al., 1988) have shown no detectable differences in receptor-G protein coupling with G, or any Gi. Studies exploring G protein-effector coupling have also failed to show selectivity. Recent work with the four isoforms of G. have revealed no detectable differences in coupling directly to Ca*' channel or adenylyl cyclase (Mattera et al., 1989). Three forms of Gi (Gil, Gi2, Giz) have been shown to activate the atria1 K' channel with approximately equal potency (Yatani et al., 1988). These studies suggest that G proteins may be multifunctional in terms of receptors and effecters. However, in the cell, some level of discrimination is required since receptors do not randomly activate all signal transduction pathways. The experiments reported here utilized a reconstitution system, using the purified Dz dopamine receptor from bovine anterior pituitary and G proteins purified from several sources, to examine questions of receptor-G protein coupling. This paper documents several aspects of receptor-G protein coupling which may give insights into selectivity and specificity of these interactions.
Two aspects of G protein-receptor interactions appeared to distinguish the various preparations of G proteins. There was a striking difference in apparent affinity between the receptor and the various subtypes of the inhibitory G proteins, as well as a difference in the ability of the agonist-receptor complex to stimulate the intrinsic GTPase activity of the G protein (efficacy).
The Da dopamine receptor appeared to couple with -lOfold higher affinity to Giz, as evidenced by lower G protein/ receptor ratios for achieving maximum coupling. The source of Gi2 was not critical, as brain and pituitary proteins were indistinguishable in all of the functional assays. These data indicating Gi2 couples with apparently higher affinity are in agreement with previous studies from this laboratory (Senogles et al., 1987). The experiments utilizing rhodopsin also supported the observation of apparent differences in affinity for receptor-G protein coupling. Transducin clearly couples with higher affinity to rhodopsin than other G protein preparations, and rhodopsin varies in its ability to stimulate the GTPase activity of these proteins. This agrees with previous observations by Cerione et al. (1985). The DZ dopamine receptor also appeared to activate the various Gi subtypes to different extents, as evidenced by stimulation of the intrinsic GTPase.
Agonist-stimulated GTPase activity obtained with Gi2 yielded a turnover number of -2 mol/min/mol of G protein, which is close to the upper catalytic limit (Gilman, 1987). However, Gil and Gi3 were only able to achieve rates of -0.8-1.0 mol/min/mol of G protein under agonist-stimulated conditions. The maximal rates of GTPase under high [Me] conditions were similar for all of the G protein preparations.
This indicates that the differences observed in agonist-stimulated turnover numbers were not due to large intrinsic differences in GTPase activity but reflected a difference in receptor-G protein coupling. Interestingly, brain G, displayed no detectable interaction with the Dz dopamine receptor. Several preparations of G, from brain were used and documented to be active either by coupling to rhodopsin or activation by mastoparan. In our previous studies (Senogles et al., 1987), we documented that immunoreactive GCX proteins co-purified with the D2 dopamine receptor from pituitary. A minor component of the total GLY protein was recognized by a G,cu-specific antibody, whereas the major component reacted with an antibody that recognizes Gi subtypes. This observation would seemingly be at odds with our observation that brain G, does not functionally couple with the DP dopamine receptor from pituitary. However, several forms of G,cu have been documented (Goldsmith et al., 1988a), and the G,a species purified from brain and used in these studies may not correspond to the species detected in pituitary. The observed differences in maximal catalytic rate attained by Gi2 could be accounted for by enhanced guanine nucleotide exchange under conditions of agonist stimulation. Both the rate of GTP binding (as inferred by the rate of [35S]GTPyS binding) and GDP release from the Gis were stimulated by D2 dopamine receptor in the presence of agonist.
Reconstitution procedures utilizing receptor and G proteins purified by conventional chromatography have one major caveat: the purity of the various components. We have documented by protein staining and Western blotting the integrity of our G protein preparations. The differences observed in the reconstitution experiments, i.e. apparent differences in affinity or activation of G proteins, could not be accounted for by a low potential level of contamination (Fig. 1B). The results presented here have demonstrated that various forms of Gi are not equivalent with regard to receptor interactions and that specificity can be demonstrated using a reconstituted system. The significance of these findings, in terms of determining the specificity of signal transduction for a given receptor, remains to be explored further.
The D2 dopamine receptor in anterior pituitary signals through at least two distinct pathways. The DP receptor mediates inhibition of adenylyl cyclase, but this pathway does not account for all of the effects of dopamine on prolactin secretion. Dopamine, even in the presence of elevated CAMP, still mediates inhibition of prolactin release, indicating the existence of a CAMP-independent mechanism (Tam and Dannies, 1981). The Dz receptor has also been shown to cause activation of K' channels, and this effect through the K' channel may in turn regulate voltage-sensitive Ca2+ channels. These pathways may account for the CAMP-independent effects of dopamine on prolactin secretion (Margaroli et al., 1987). A question of interest is whether these effecters couple to the Dz receptor through one distinct type of G protein or a network of G proteins. Using patch-clamping techniques, no specificity has been observed among the native or recombinant Gi family for direct activation of the chick atria1 K' channel (Yatani et al., 1988). These data suggest that the specificity of coupling is not present at the level of G proteineffector or either too subtle to be observed in the experimental design used.
The receptor-G protein-coupling properties documented here appear to have two distinguishing characteristics that could account at least in part for biological specificity: the apparent affinity of the receptor-G protein complex, and the ability of the receptor-agonist complex to activate the intrinsic GTPase. In terms of signal transduction, a logical consequence of Da dopamine receptor activation may be a preferential activation of Giz by one or both mechanisms. The apparent differences in affinity between Gi2 and the other Gi subtypes in coupling with the DP dopamine receptor are of unknown significance since the ratios of G proteins vary greatly with tissue source. Thus, in some tissues, this aspect of affinity may solely dictate coupling specificity. The ability of the receptor to activate Gi proteins to varying degrees is another mechanism for dictating specificity of coupling. For example, in the system described, the Giz protein is activated maximally by D2 dopamine receptor. The amount of the activated subunit, and not the specific subtype, released by receptor activation may influence the choice of effector system. This specificity could be afforded by the existence in the membrane of a loosely associated multifunctional complex made up of receptor-G protein-effector similar to the many multienzyme complexes documented (Srere, 1987). It is possible that the true biochemical mechanisms underlying receptor-effector coupling may involve both of these characteristics. Other biochemical or architectural events may also influence the receptor-G protein-effector coupling specificity and allow for a preferential activation of selective signaling pathways. Ultimately, the question of receptor-G protein-effector specificity will have to be examined using molecular biology techniques. Constructing a cell that expresses receptor, effector, and a single functional G protein may ultimately provide the necessary insights into the levels of control which must exist to account for the specificity of receptor signaling. Goldsmith, P., Rossiter, K., Carter, A., Simonds, W., Unson, C. G., Vinitsky, R., and Spiegel, A. M. (1988b) J. Biol. Chem. 263, 6476-