Regulation of Hormone-Receptor Coupling to Adenylyl Cyclase EFFECTS OF GTP AND GDP*

GDP and GTP regulation of receptor-mediated stim- ulation of adenylyl cyclases in membranes of 549 murine lymphoma cells (S49), NS-20 murine neuroblas- toma cells (NS-20), rabbit corpora lutea (CL), and turkey erythrocytes were studied under assay conditions which minimized conversion of added GTP to GDP and of added GDP to GTP. Hormonal stimulation in all systems required guanine nucleotide addition. In the presence of GTP, adenylyl cyclase activity in 549, NS-20, and CL was stimulated respectively by isoproterenol and prostaglandin El (PGEI), by PGEl and the aden- osine analog, phenylisopropyladenosine, and by PGEl and isoproterenol, with the first of the listed stimulants eliciting higher activities than the second. Activity in turkey erythrocyte membranes was stimulated by iso- proterenol. GDP was partially effective in promoting hormonal stimulation, being able to sustain stimulation by isoproterenol and PGEl in S49 cell membranes and by PGEl in CL membranes. In NS-20 membranes, both GDP and guanosine-S’-O-(2-thiodiphosphate) (GDPPS) were inhibitory on basal activity, yet promoted limited but significant stimulation by PGE1.

GDP and GTP regulation of receptor-mediated stimulation of adenylyl cyclases in membranes of 549 murine lymphoma cells (S49), NS-20 murine neuroblastoma cells (NS-20), rabbit corpora lutea (CL), and turkey erythrocytes were studied under assay conditions which minimized conversion of added GTP to GDP and of added GDP to GTP. Hormonal stimulation in all systems required guanine nucleotide addition. In the presence of GTP, adenylyl cyclase activity in 549, NS-20, and CL was stimulated respectively by isoproterenol and prostaglandin El (PGEI), by PGEl and the adenosine analog, phenylisopropyladenosine, and by PGEl and isoproterenol, with the first of the listed stimulants eliciting higher activities than the second. Activity in turkey erythrocyte membranes was stimulated by isoproterenol. GDP was partially effective in promoting hormonal stimulation, being able to sustain stimulation by isoproterenol and PGEl in S49 cell membranes and by PGEl in CL membranes. In NS-20 membranes, both GDP and guanosine-S'-O-(2-thiodiphosphate) (GDPPS) were inhibitory on basal activity, yet promoted limited but significant stimulation by PGE1. In turkey erythrocytes, stimulation by isoproterenol could not be elicited with GDP or GDPBS. Thus, although less effective than GTP in promoting hormonal stimulation of several adenylyl cyclase systems, GDP was clearly not inactive. Concentration effect curves for active hormone in the presence of GDP had higher apparent K, values than in the presence of GTP.
In spite of differences between the effects of GTP and GDP on hormonal stimulation of adenylyl cyclase activities, GTP and GDP affected equally well isoproterenol binding, regardless of whether or not its receptor could be shown to stimulate adenylyl cyclase in the presence of GDP. Determination of transphosphorylation of GDP to GTP showed that at saturating concentrations, the proportion of GDP converted to GTP is negligible and unaffected by hormonal stimulation. Concentrations giving 60% inhibition were determined for GTP-and GDP-mediated inhibition of guanyl-6'-yl imidodiphoephate stimulation in the absence and presence of stimulatory hormones. In all four systems studied, GTP and GDP interacted with about equal potency * Supported in part by National Institutes of Health Grant AM-19318. This is the fourth paper in a series exploring the characteristics of hormone-receptor coupling to adenylyl cyclase. Ref. 28 is the preceding paper in the series. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ship AM-06066. and hormonal stimulation was not accompanied by a selective decrease in affinity for GDP.
One way to explain all of the results obtained is to view hormonally sensitive adenylyl cyclase systems as two-state enzymes whose activities are regulated by GTP and GDP through an allosteric site related to the catalytic moiety, and receptors as entities that are inactive and hence unable to couple unless occupied by hormones and activated by any guanine nucleotide through a distinct receptor-related process.
Some of us have reported that the liver plasma membranebound adenylyl cyclase can be stimulated by glucagon to near maximal levels in the presence of GDP (1). This GDP-mediated stimulation occurred under conditions where transphosphorylation to GTP was less than 1%, and where activity stimulated by GMP-P(NH)P' was fully blocked. Ki values for GTP-and GDP-mediated inhibition of GMP-P(NH)P-stimdated activity were similar, as were apparent K, values for GTP-and GDP-mediated activation of glucagon-stimulated activity. Glucagon-stimulated activity in the presence of GDP was about 80 and 60% of that observed in the presence of GTP and GMP-P(NH)P, respectively. In contrast to hormone-stimdated activity, basal activity was unaffected by GDP. These results suggested that while the GTPase putatively associated with membrane-bound adenylyl cyclases (2-6) may play a key role in the deactivation of adenylyl cyclase, hormonal stimulation of adenylyl cyclase may occur independently of its existence. Indeed, it would appear that adenylyl cyclase is under two interacting types of regulatory controls: one consisting of a nucleotide-dependent regulation of stimulation by the hormone-occupied receptor, and the other of a hormoneindependent regulation by the GTPase postulated to be associated with the guanine nucleotide-binding component of adenylyl cyclases.
In contrast, Cassel, Selinger, and co-workers (4-9) have, on the basis of work with turkey erythrocyte adenylyl cyclase, proposed that hormonal stimulation involves an obigatory exchange of GTP (or GTP analog) for GDP and that the primary effect of the hormone-occupied receptor is the facilitation of this reaction (7). It has been suggested that such an exchange mechanism facilitating displacement of GDP by GTP (8-10) may be operative not only in the turkey erythrocyte system, but also in mammalian systems such as the rat liver adenylyl cyclase (9). According to this hypothesis, the data showing glucagon stimulation upon GDP addition could be accounted for by assuming that hormonal stimulation decreases GDP affinity and increases GTP affinity for the regulatory site, so as to allow the 1% of GTP formed by transphosphorylation from added GDP to act positively and become the actual nucleotide responsible for the effect seen.
Since the intrinsic mechanism by which hormones act would be different if it were one of altering the adenylyl cyclase-guanine nucleotide interaction as opposed to activation of adenylyl cyclase regardless of what nucleotide happens to be occupying it, the resolution as to whether or not hormone can activate adenylyl cyclases known to be occupied by GDP becomes of central importance in the elucidation of the mechanism of hormone and receptor action. To decide whether, in the liver system, glucagon alters the affinities for GDP and GTP in the manner indicated above, it would be necessary to determine their respective apparent K, values in the presence of glucagon. Technically, this is possible for GDP by evaluating the competitive (partial) inhibition of GMP-P(NH)P stimulation by GDP, because glucagon-stimulated activity in the presence of GDP is about 60% of that seen in the presence of GMP-P(NH)P. However, it is not possible to determine the apparent K, for GTP because, under the assay conditions used, the difference between glucagon-stimulated activity in the presence of GMP-P(NH)P and that in the presence of GTP (-15%) is too small to allow a meaningful assessment of GMP-P(NH)P uersus GTP competition.
To obtain independent information on the mechanism of hormone action, we explored the possibility that GDP would promote hormonal stimulation in other adenylyl cyclase systems. We also sought to gain information as to whether hormonal stimulation affects differentially GTP and GDP interaction in these systems. For these reasons, we studied the hormonal stimulation of adenylyl cyclases in membranes from S49 mouse lymphoma cells, NS-20 neuroblastoma cells, and rabbit corpora lutea, and compared some of the findings to results we obtained with turkey erythrocyte membranes. Findings with GDP were compared with findings with GTP, and degrees of interconversion between GTP and GDP were monitored to allow for proper interpretation of data.
The results indicate that hormones do not differentially affect the potencies of GDP and GTP, and that GDP is capable of promoting hormonal stimulation. This provides key information on the possible kinetic mechanism by which hormones may act. However, the overall picture is complex.
GDP promotes hormonal activation in some, but not all adenylyl cyclases tested. In systems where GDP supports activation by one hormone, it does not necessarily do so with respect to a second hormone affecting the same adenylyl cyclase.
The data are discussed in terms of a recently proposed twostate model of adenylyl cyclase and strongly suggest that the role of guanine nucleotide regulation of hormone action on adenylyl cyclases constitutes a GTP-mediated amplification and GDP-mediated dampening system whose degree of amplification and dampening varies with the receptor type as well as from tissue to tissue.
Membrane Preparation-Membrane particles from 549 mouse lymphoma cells were prepared according to Rosa et al. (15). S49 cells were grown in 6-liter spinner flasks in Dulbecco's modified Eagle's medium (Grand Island Biological Company, Garden City, NY) containing 10% heat-inactivated horse serum (Grand Island Biological Company) as described elsewhere (16).
Membrane particles from corpora lutea (19) were prepared from 7day pseudopregnant New Zealand white rabbits.
Untreated and isoproterenol plus GMP-treated turkey erythrocyte membranes were prepared and purified according to Abramowitz et al. (20).
Except for turkey erythrocyte membranes, which were diluted in the same medium, all other membranes were diluted at least 10-fold in H20 immediately prior to their use. Adenylyl Cyclase Assays-Unless indicated otherwise, adenylyl cyclase activities were determined by monitoring the conversion of [LX-~'P]ATP to [32P]cAMP in 5 min at 32.5% Incubations contained in a final volume of 50 4: 0.5 mM AMP-P(NH)P, 50 X IO6 cpm of [c~-~'P]ATP (specific activity, 1 X lo5 cpm/pmol), 5.0 n m MgClZ, 1 m~ EDTA, 1 mM [3H]cAMP (approximately 10, OOO cpm), 25 mM Tris-HC1, pH 7.5, and lOpl of either NS-20 neuroblastoma membranes (2 to 2.4 pg of protein) or corpus luteum membranes (1.5 to 1.8 pg of protein). Activities in 10 pl of S49 membrane suspension (1.6 to 1.8 pg of protein) were determined under the same conditions except that MgC12 was 10 mM and Tris-HC1 buffer was replaced with 25 KIM Hepes-Tris pH 8.0. Activities in 10-pl aliquots of isoproterenol plus GMP-treated turkey erythrocyte membranes (10 pg of protein) were determined also under the same conditions with the exception that ethylene glycol bis(P-aminoethyl ether)N,N,N',N"tetraacetic acid (EGTA) was substituted for EDTA and 10 mM KC1 was added. Other additions (hormones and guanine nucleotides) are denoted in the tables and legends to figures describing the experiments performed. Reactions were stopped by addition of 100 pl of a "stopping solution" containing 10 m~ CAMP, 40 mM ATP, 10 m~ AMP, 10 mM ADP, and 1% sodium dodecyl sulfate followed by cooling to 04OC until they were processed further. [32P]cAMP formed was isolated and quantitated by double chromatography over Dowex 50 and alumina according to a modification (21) of the method of Salomon et al. (22). This modifcation, which consists only of increasing the size of the Dowex 50 column by 50% and that of the alumina columns by 1002, leads to significantly lower blanks, especially when more than lo' cpm are used per assay, and allows for indefinite re-use of the Dowex columns and for re-use of the alumina columns for between 25 to 50 times.
Results are expressed as cpm X of C3'P]cAMP formed during the 5-min incubations per mg of membrane protein added. Since, unless indicated otherwise, the assays were carried out a t a specific radioactivity of total adenine nucleotides (AMP-P(NH)P plus ATP)

Receptor Coupling to Adenylyl Cyclases by GDP
of lo4 counts/5 min/pmol, the numerical values reported are also equivalent to the picomoles of CAMP formed/min/mg of protein if it is assumed that the accumulation rates were constant with time and both ATP and AMP-P(NH)P serve as substrates with equal efficacy. Test for Maintenance of Nucleotide Levels Throughout Incubations-To assess the proportion of initially added [a-3zP]ATP remaining as such at the end of incubation for adenylyl cyclase activity, 0.5-to 1.0-pl aliquots of the stopped and cooled reaction mixtures were spotted upon polyethyleneimine-cellulose plates. After chromatography (ascending with 1 M LiCl as developing solvent), the areas containing ATP, ADP, and AMP were localized under UV light and cut out, and the content of 32P was determined by liquid scintillation counting (19). Results are expressed as the percentage of initially added [a-32P]ATP that co-chromatographs with ATP.
To test for maintenance of GTP and GDP and for their interconversions, parallel incubations were carried out in which either were substituted for the labeled ATP and reactions were stopped with 100 p1 of solution containing 10 m~ GTP, 10 mM GDP, 10 mM GMP, and 1% sodium dodecyl sulfate followed by immediate cooling to 04°C. Distribution of radioactivity between the various nucleotides was determined by thin layer chromatography of 1-to 2 4 aliquots of the stopped reaction mixtures on polyethyleneimine-cellulose plates using 1 M LiCl as developing solvent. The percentage of initially added [a-32P]GTP or [a-32P]GDP migrating as GTP and GDP were calculated after determination of radioactivity co-chromatographing with GTP, GDP, and GMP by liquid scintillation counting.
lZ5I-HYP Binding Assays-The binding reactions were carried out in a final volume of 500 pl containing 0.1 nM lZ5I-HYP (120,000 cpm/ assay tube), 2 mM MgCl2,l.O m~ EDTA, 0.1% bovine serum albumin, 0.1 m~ ascorbic acid, and 25 mM of the indicated buffer. When present, the concentrations of guanine nucleotides were 0.5 mM. Incubations were carried out in glass test tubes (12 X 75 m m ) at 325°C. After 30 min, the binding reactions were stopped by addition of 1.5 ml of ice-cold 0.1% bovine y-globulin (Schwarz/Mann, Catalogue No. 903004) in 0.1 M NaCI. All subsequent steps were carried out in the cold. After mixing, 1.0 ml of 20% polyethylene glycol (Fisher, Catalogue No. P156) was added to each tube; the contents of the tubes were mixed and centrifuged at 4°C for 10 min at 3,000 rpm. The supernatant fluids were discarded by careful aspiration, and the pellets were resuspended in 2.0 ml of 0.1 mM NaCl and reprecipitated by addition of 1.0 ml of 20% polyethylene glycol. After a second centrifugation for 10 min at 3,000 rpm and separation of the supernatant fluid by aspiration followed by careful wiping of the walls of the tubes with cotton-tipped applicators (Scientific Products, Catalogue No. A5002-1), the '251-HYP retained in the tubes on the membranes that had been co-precipitated with the y-globulin was determined by scintillation counting in a Searle Analytic Autogamma counter, model 1195.
Proteins were determined by the method of Lowry et al. (23) using bovine serum albumin as standard.

RESULTS
Effects of Guanine Nucleotides on Adenylyl Cyclase Activities in S49, NS-20, and Corpus Luteum Membranes-Activities were determined in the absence of nucleoside triphosphate-regenerating system using 0.5 mM AMP-P(NH)P and 10 p~ [cx-~'P]ATP as substrates. Under these conditions (Table I), GTP was found to stimulate basal activities in all three systems 2-to %fold. Addition of 100 PM GDP did not alter basal activities in S49 and corpus luteum membranes, and inhibited NS-20 adenylyl cyclase between 50 and 70%.
No hormonal stimulation could be elicited unless a guanine nucleotide was added.
In ,349 lymphoma membranes, GTP promoted coupling of isoproterenol and PGEl receptors, with PGEl-stimulated activity being 65 to 70% of that seen with isoproterenol. GDP addition also promoted coupling of both receptors, but with less efficiency: isoproterenol-and PGE1-stimulated activities were 60 and 20%, respectively, of that seen with isoproterenol in the presence of GTP.
In NS-20 neuroblastoma membranes, GTP promoted coupling of PGEl and adenosine receptors, with phenylisopro-  pyladenosine-stimulated activity being 35 to 45% of that seen with PGEI. GDP addition promoted coupling of the more active PGEl receptor, but not of adenosine receptor. PGEIstimulated activity sustained by GDP was between 50 and 75% of that seen with GTP.
In corpus luteum membranes, GTP promoted coupling of PGEl and isoproterenol receptors, with isoproterenol-stimulated activity being about 50% of that seen with PGEI. GDP addition promoted coupling only to a minor degree of the more effective receptor. PGEl-stimulated activity with GDP was 12 to 20% of that seen with GTP.
As shown in Table 11, stimulation of S49 cell and NS-20 neuroblastoma adenylyl cyclases by isoproterenol and PGEI and by PGEl and phenylisopropyladenosine, respectively, are not additive, indicating that in each of these membranes, a single adenylyl cyclase system is stimulated by more than one hormone receptor. Combinations of PGEl and isoproterenol elicited a partially additive response in corpus luteum membranes, a finding that suggests involvement of more than one adenylyl cyclase system. Neither corpus luteum nor S49 lymphoma membranes were contaminated by significant amounts of catecholamine-like substances, as evidenced by a lack of inhibition of basal activity by the P-adrenergic receptor blocker propranolol (data not shown). Similarly, the basal activity of NS-20 neuroblastoma membranes was unaffected by 1 international unit/ml of adenosine deaminase, a concentration sufficient to abolish the stimulatory effect of 100 pM added adenosine (data not shown). It is therefore concluded that: (a) stimulation of basal activities by GTP is not due to a potentiation of catecholamine action in S49 and corpus luteum membranes, or due to a potentiation of adenosine action in NS-20 membranes; and ( b ) the partial inhibition of basal activity by GDP (and as will be shown below, by GDPPS) observed in NS-20 membranes does not result from blocking the stimulatory action of adenosine-like material that may be present.
Incubation Conditions and Maintenance of Added NUcleotides-The conditions used for assaying adenylyl cyclases in the experiments of Table I were chosen so as to minimize PGEl a When present, PGE, was 10 pg/ml, isoproterenol was M, and PIA was M. Each of these concentrations elicits a maximal response in the systems tested.
Adenylyl cyclase activities were assayed under conditions described under "Experimental Procedures" except that 0.1 mM [~x-~*P]ATP (2,000 cpm/pmol) was used as substrate, and the nucleoside triphosphate-regenerating system consisting of 20 nud creatine phosphate, 0.2 m g / d of creatine phosphokinase, and 0.02 mg/rnl of myokinase was included. Incubations were for 10 min at 32.5"C.
Values are means f S.D. of triplicate determinations.
conversion of GTP to GDP and of GDP to GTP. This was accomplished, in part, by using 0.5 mM AMP-P(NH)P as the bulk of substrate (4, 24), by reducing membrane concentrations (and hence nucleotide-metabolizing activities) as much as possible, and by carrying out incubations for only 5 min.
Under the conditions chosen (see under "Experimental Procedures"), more than 80% of the [LY-~'P]ATP initially added as well as more than 75% of initially added GTP are found at the end of the incubation period still to co-chromatograph upon thin layer chromatography on polyethyleneimine-cellulose with ATP (not shown) and with GTP (left panels of Fig. l), respectively. In fact, nucleoside triphosphate levels were maintained over a wide variation of concentrations ranging from 0.01 to 100 ~L M and above (Fig. 1). Incubations were also carried out over a wide range of GDP concentrations in the absence and the presence of stimulatory hormones and the chromatographic behavior of trace amounts of initially added [cY-~'P]GDP was monitored. As shown in the right panels of Fig. 1, S49, NS-20, and corpus luteum membranes have a transphosphorylation mechanism. This system, which converted GDP to GTP, was found to be of low capacity and high affinity, such that at concentrations in excess of 1 PM added GDP, the proportion of GDP converted to GTP decreased substantially. This decrease in the percentage of GDP appearing as GTP at the end of incubations was not only observed up to 100 p~, as shown in Fig. 1, but continued beyond that and up to 1000 p added GDP, the highest concentration tested (not shown). When incubations with [a-32P]GTP or [c~-~*P]GDP were performed without membranes, 90 to 92% of the added guanine nucleotide chromatographed as the added nucleotide. No attempt was made to subtract these background distributions in the experiments shown in Fig. 1.
The transphosphorylation reactions of GDP to GTP do not appear to be under hormonal influence. We did not observe any change or decrease in the GTP formed from GDP at any of the tested concentrations of GDP (0.01 to 1000 PM) due to the addition of hormone (not shown).
From these data, it appears that if it is assumed that the effectiveness of interaction of (i.e. K,,, values for) GTP and GDP are approximately equal (see below for actual tests), then it is possible to obtain concentration effect curves for GTP that are relatively unbiased by conversion to GDP. However, since depending on the membranes tested, up to 75% of the initially added GDP can be converted to GTP (see lower right panel of Fig. l), it is impossible under the conditions of assay used by us to obtain meaningful concentration effect curves for GDP. In fact, at low concentrations of added GDP, mostly effects of GTP would be observed in some instances, and true effects of GDP would be seen only at rather high Concentrations of added GDP.
Lack of Hormone Effect on the Relative Potency with which GTP and GDP Interact with Adenylyl Cyclase-We determined the relative effectiveness with which GTP and GDP interact with the hormone-responsive adenylyl cyclase systems of S49, NS-20, and corpus luteum membranes, both under control and hormonally stimulated conditions.

Receptor Coupling to Adenylyl Cyclases by GDP
Although it would be possible to determine apparent K , values for the effects of GTP in the absence and the presence of hormonal stimulation, such a determination is not possible for GDP, because it does not affect basal activity in either S49 or corpus luteum membranes. Further, due to the extent of transphosphorylation at low concentrations of added GDP, hormonal stimulation observed at these concentrations would be heavily biased by formation of GTP. We therefore compared interactions of GTP and GDP by testing for their effectiveness in inhibiting stimulation by GMP-P(NH)P. Table I11 presents ICso values for GTP-and GDP-mediated inhibition of GMP-P(NH)P stimulation of ,5349, NS-20, and corpus luteum membrane adenylyl cyclases, as assayed in the absence of hormones and under conditions of membrane concentration, AMP-P(NH)P, and ATP identical with those used in the experiments shown in Table I and Fig. 1. The results indicate that in the absence of hormone, GDP and GTP are about equipotent in all three systems, Addition of hormones in the presence of GMP-P(NH)P resulted in activities such that although IC50 values could be obtained for GDP using the same incubations conditions, it was not possible to determine a competitive effect of GTP. So we determined IC50 values for GDP in the absence and presence of hormone under the conditions used in the experiments of Fig.  1 and Table I, and used different conditions that allowed us to measure IC50 values for GTP both in the absence and presence of hormone in S49 cell membranes and corpus luteum membranes.
The results of one set of these experiments is shown in Fig.  2. It can be seen that addition of stimulatory hormone did not alter the potency of GDP in any of the systems studied (right panels of Fig. 2). Neither did hormonal stimulation affect significantly the potencies with which either the S49 cell system or the corpus luteum system interact with GTP (top and bottom left panels of Figure 2). We were unable to determine the effect, if any, of hormonal stimulation on the potency of GTP in NS-20 membranes (middle left panel of Fig. 2 ) , for we were faced with a situation similar to that seen in liver membranes in which hormone-stimulated activities in the presence of GMP-P(NH)P were too close to those seen in presence of GTP to allow us to study a competitive interaction.
The experiments shown in Fig. 2 were repeated several times and showed unequivocally that hormonal stimulation has no selective effect on the ICso value with which GTP interacts with the S49 and corpus luteum adenylyl cyclases when compared to those with which GDP interacts with the same systems. Thus, GDP and not GTP formed during the incubation, was responsible for the stimulation of hormonal response in S49 membranes and for the small but significant prostaglandin response in corpus luteum membranes. The finding that both control and hormone-stimulated activities were constant from 16 to 500 ~L M added GDP (Fig. 2 ) is consistent with this contention, since over this range of con-

4.0
a ICoo, concentration of GTP or GDP required to obtain 50% inhibition of the stimulation of activity due to 20 p~ GMP-P(NH)P in the assays. These values were obtained from concentration effect curves for GDP and GTP between 0 and 200 p~ assayed in the absence and presence of GMP-P(NH)P as described under "Experimental Procedures." centration, the percentage of GTP formed is constantly decreasing. If hormonal responses were absolutely dependent on GTP, then one should have observed a continuous decrease in hormonal stimulation with increasing concentration of added GDP, since GDP is a competitive ligand at the guanine hpM PGE, ' €

FIG. 2. Inhibition of GMP-P(NH)P stimulation of adenylyl cyclase activities by GTP (lefl panels) and GDP (right p a n e l s )
as determined in the absence (0) and presence ( . ) of hormonal stimulation. Incubations with GTP were for 10 min at 325°C using 0.2 m~ ATP (2,000 cpm/pmol) in the presence of a nucleoside triphosphate-regenerating system (RS) consisting of 20 mM creatine phosphate, 0.2 mg/ml of creatine phosphokinase, 0.02 m g / d of myokinase, 10 PM GMP-P(NH)P, and either 5.3 pg/assay of S49 membranes, 4.2 pg/assay of NS-20 membranes, or 2.8 &assay of corpus luteum membranes. Incubations with GDP were for 5 min at 32.5OC using 0.5 m~ AMP-P(NH)P, 10 p~ [cx-~'P]ATP (10' cpm/ pmol), no nucleoside triphosphate-regenerating system, 20 mM GMP-P(NH)P, and either 1.8 pg/assay of S49 cell membranes, 2.4 pg/assay of NS-20 neuroblastoma membranes, or 1.8 pg of corpus luteum membranes. When present, isoproterenol was M and PGEI was 10 p g / d . Activities obtained in the absence of GMP-P(NH)P are shown as open symbols: 0, no hormone addition; 0, plus isoproterenol (top panels) or PGE, (middle and bottompanels). For the rest of the conditions, see the figure and under "Experimental Procedures." Vertical lines and numbers next to them denote ICs0 values, i.e., the concentrations of added GTP or GDP at which 50% inhibition was obtained. nucleotide regulatory site($, both in the absence and in the presence of hormone, as shown in the right panels of Fig. 2. Although an ICso value for the interaction of GTP with the NS-20 neuroblastoma system could not be obtained in the presence of hormone, the following two arguments are in favor of GDP being a positive, though less effective, mediator of hormone stimulation in this system as well: 1) as shown above for S49 and corpus luteum membranes, prostaglandin stimulation did not decrease when the added GDP was increased from 16 to 500 PM; and 2) GDPPS an analog of GDP, was also effective in promoting PGEi stimulation (Table IV). It has

Effect of GDP and GDPPS addition on basal and PGE, stimulation of adenylyl cyclase activity in NS-20 murine neuroblastoma membranes
Nucleotide addition Adenylyl cyclase activities" Basal Activity due to been argued (9) that GDPPS may act in some systems as a "GTP-like'' nucleotide. However, the data shown in Table IV indicate that GDPPS behaves as an analog of GDP inasmuch as both inhibit basal activity. A GTP-like effect of GDPPS would have led to an increase in basal activity.
The data can be summarized by stating that while GTP and GDP are almost equipotent in the presence and absence of hormone, the extent of coupling of hormone-occupied receptors to adenylyl cyclases is much greater in the presence of GTP than GDP. GDP couples only the effective receptors and fails to do so for receptors that show rather small responses in the presence of GTP.
Inhibition of the Turkey Erythrocyte Adenylyl Cyclase System by GDP: Lack of Effect of Isoproterenol-Cassel and Selinger (8) described that in contrast to "untreated" turkey erythrocyte membranes, the adenylyl cyclase system in membranes subjected to previous treatment with isoproterenol and GMP is stimulated by GMP-P(NH)P. The "appearance" or "unmasking" of the GMP-P(NH)P effect was ascribed to a "clearing" of the enzyme system of GDP during the isoproter-

Receptor Coupling to Adenylyl Cyclases by GDP
enol plus GMP treatment. We investigated the effect of isoproterenol on the potency with which GDP inhibits GMP-P(NH)P stimulation in treated turkey erythrocyte membranes and found that isoproterenol stimulation is associated with a slight increase in the overall affinity of the system for GDP , (Fig. 3). As shown, at lo00 p~ added GDP, this nucleotide did not sustain hormonal stimulation of the turkey erythrocyte adenylyl cyclase system (Fig. 3). Similarly, no hormone effect was seen in the presence of 1 to 100 ~L M GDPpS (not shown). Effect of Type of Nucleotide Used on the Concentration Effect Curve for Hormonal Stimulation- Fig. 4 shows representative experiments in which the apparent K, for activation of three adenylyl cyclase systems and the extent of stimulation obtained were determined as a function of the type of guanine nucleotide added. In agreement with results shown in Table  I and Fig. 2, GTP and GDP promoted coupling of PGEI receptor to the corpus luteum and the NS-20 neuroblastoma adenylyl cyclase systems, and of isoproterenol receptor to the S49 cell system. In all the cases, when maximal receptormediated stimulation was lower (presence of GDP), the apparent K , was higher.
Effect of GDP on Isoproterenol Binding to S49 cell, Corpus Luteum, and Turkey Erythrocyte Membranes-lZ5I-HYP bound specifically to membranes containing ,&adrenergic receptors. Binding was prevented by simultaneous addition of propranolol or isoproterenol and showed stereoselectivity. Between 10-and 100-fold higher concentrations of (+) isomers than of (-) isomers were required to effect 50% inhibition of '251-HYP binding (not shown). The effect of GTP and GDP on inhibition of lZ5I-HYP by varying concentrations of (-)isoproterenol was tested on S49, corpus luteum, and turkey erythrocyte membranes. As shown in Fig. 5 for S49 and corpus luteum membranes, and in Fig. 6 for untreated and isoproterenol plus GMP-treated turkey erythrocyte membranes, both GTP and GDP (at 500 PM) caused an increase in the ICSO value for isoproterenol. The effect was more marked in S49 cell membranes than in corpus luteum membranes, and was approximately equal in untreated and treated erythrocyte membranes. The guanine nucleotide effect on isoproterenol binding was mimicked by both the GTP analog GMP-P(NH)P and the GDP analog GDPBS (not shown for S49 and corpus luteum membranes, but shown for turkey erythrocyte membranes in Fig. 6). Transphosphorylation of GDP to GTP under the conditions of incubation used for the binding experiments of Figs. 5 and 6 was unaffected by isoproterenol addition ( M), and was less than 0.5% with S49 and turkey erythrocyte membranes and less than 1% with corpus luteum membranes (not shown).

DISCUSSION
The Interactions of GDP with Adenylyl Cyclases: Mediation of Hormonal Stimulation-Since the discovery of a hormone-stimulated GTPase thought to be intimately associated with the regulation of adenylyl cyclase (2-9, 25), it has been postulated that GTP occupancy of the regulatory site would be required for further hormonal stimulation of adenylyl cyclases. This hypothesis has gained credence since Cassel and Selinger (8) reported that in turkey erythrocyte membranes, isoproterenol increases the off-rate of bound GDP, thus facilitating the binding of GTP and subsequent activation of adenylyl cyclase. Such a mechanism, if generally applicable, would preclude the possibility of hormonal stimulation in the presence of excess GDP, provided that in the presence of hormone, GDP interacts with the system.
The data reported here indicate that: 1) GTP and GDP are approximately equipotent in their interaction with adenylyl cyclases; 2) in none of the systems examined, including that of turkey erythrocytes, did hormonal stimulation result in a signifcant decrease in the apparent potency of GDP; and 3) in the two systems where the potency of GTP in the presence and absence of hormone has been susceptible to examination, no major difference is observed due to the presence of hormone. It therefore seems safe to conclude that hormonal effects seen in the presence of a known 10-fold excess of GDP over GTP are occurring predominantly due to GDP. The data show that GDP is clearly not inactive in all cases, even though it is less effective than is GTP in promoting coupling of hormone receptors to adenylyl cyclases. Further, GDP sustains stimulation only by rather effective receptors (see below). Thus, we have defined situations where hormone-receptor coupling is mediated truly by GDP and not by contaminating GTP. Since there appears to be no experimental basis for assuming that hormonal stimulation results in a selective alteration of potency of either GDP or GTP, we feel that the coupling of glucagon receptors to liver adenylyl cyclase observed by some of us earlier under circumstances of only minimal transphosphorylation was indeed due to GDP and not to the less than 1% contaminating GTP (1).
The above conclusions imply that interaction of GTP (or a GTP analog) with adenylyl cyclase is not an obligatory requirement for hormonal stimulation and that increased nucleotide exchange rates observed (or assumed) to occur upon hormone addition are a consequence of the stimulatory effect of active hormone receptor on the guanine nucleotide-binding site as opposed to being the cause for stimulation of adenylyl cyclase by GTP (20). The fact that GTP is not an obligatory requirement for hormonal stimulation is further underscored by the finding that in the NS-20 neuroblastoma system (Table  IV), the liver glucagon-sensitive adenylyl cyclase (9) and the catecholamine-stimulated parotid adenylyl cylcase (9) are hormonally stimulated in the presence of the GDP analog GDPPS. This analog was shown not to be susceptible to transphosphorylation (9).
Differential Effects of Guanine Nucleotide on Binding and Coupling: An Interpretation-There seems to be no apparent correlation between the capacity of a nucleotide to affect binding and promote positive coupling. GTP and GDP affect binding equally in corpus luteum, S49, and turkey erythrocyte membranes. Both nucleotides are equipotent in interacting with the guanine nucleotide-binding site on the regulatory component as measured by adenylyl cyclase activities. However, GDP promotes coupling of isoproterenol receptor to adenylyl cyclase in S49 cell membranes but not in corpus luteum or turkey erythrocyte membranes. We have recently proposed a two-state model (20,26-28) based on the hysteretic enzyme concept (29-31) that satisfactorily accounts for the known features of GTP as well as GDP regulation of hormone stimulation of several mammalian adenylyl cyclases, primarily that from rat liver but also those of rabbit corpus luteum, kitten heart, S49 cells (281, and turkey erythrocytes (20). We find that the data presented here can also be satisfactorily explained by a two-state model for the guanine nucleotidemodulated adenylyl cyclase which is further regulated by active hormone -receptor complex. It has been proposed that an active hormone. receptor complex stimulates adenylyl cyclase by increasing the rate of transition of the basic enzyme system' from an inactive E" state to its active E' state. Further, it was proposed and shown by simulations that if the guanine nucleotide regulatory site of the system is occupied by a guanine nucleotide which itself increases the rate of * The basic enzyme system referred to here is the complex between the catalytic subunit responsible for the cyclization reaction and the guanine nucleotide-binding component identified by Ross et al. (36) to be essential for expression of magnesium-dependent cyclizing activity.
transition from the inactive to the active form, this would result in amplification of the hormonal response (28). If a nucleotide-sensitized hormone -receptor complex is considered to be the active state of the receptor, i.e. one that can stimulate adenylyl cyclase, it would be predicted that potency of hormonal stimulation depends on the type of nucleotide that occupies the guanine nucleotide site on the regulatory component associated with the catalytic subunit. Such an argument tacitly includes the presence of two distinct guanine nucleotide effects: one affecting hormone-receptor interaction, rendering the hormone -receptor complex active, and another regulating the catalytic moiety of adenylyl cyclase, such as has been proposed on kinetic (32, 33) and biochemical (34) grounds for the rat liver system.
While we have no evidence in any of the systems studied here that there are two structurally different guanine nucleotide sites involved in hormonal stimulation of adenylyl cyclase activity, there are evidences that guanine nucleotide-binding sites behave differently, depending on whether adenylyl cyclase regulation or receptor regulation is examined. It has been shown that, in liver membranes preactivated with GMP-P(NH)P and then washed, a subsequent stimulatory effect of glucagon requires the co-addition of further guanine nucleotide, be it GMP-P(NH)P itself, GTP, or GDP (1, 33). Ross et al. (15) reported that while the effect of GMP-P(NH)P on S49 cell membrane adenylyl cyclase is of the apparently irreversible type being resistant to washing, that on isoproterenol binding is readily reversible. Strosberg and co-workers (35) have recently shown that pre-exposure of turkey erythrocyte membranes to GMP-P(NH)P, known to lead to "persistent" wash-resistant activation of the adenylyl cyclase in these membranes, does not result in a persistent effect in terms of protecting against isoproterenol-mediated sensitization of a N-ethylmaleimide-affected "SH group whose alkylation results in a decrease in measurable binding sites. Involvement of interacting nucleotide sites is also suggested by Cassel and Selinger's data (8), which showed that dissociation of [3H]-GDP from turkey erythrocyte membranes under the influence of isoproterenol is stimulated by free guanine nucleotide addition. Although Cassel and Selinger's data may indicate the existence of two nucleotide sites related in such a manner that in the presence of hormone, occupancy of one may result in "tight" binding and occupancy of the other may result in increased nucleotide exchange rates at both sites (homotropic negative cooperativity), there are currently no data to indicate that such interactions occur in the absence of hormone. The data in this communication do not address the question of homotropic interactions between guanine nucleotide-binding sites. Further experimentation is required to determine whether such interactions do exist. Regardless of whether two sites are involved, it seems clear that at least two distinct guanine nucleotide effects are involved in the overall guanine nucleotide regulation of receptor-mediated stimulation of adenylyl cyclase activity. If so, the following points can be made. 1) Based on the data obtained, we shall ascribe an efficacy or intrinsic activity factor3 to the active hormone .receptor complex. This factor increases with increasing capacity of the active receptor to stimulate the basic adenylyl cyclase system. By doing so, we can assign orders of efficacy to the catecholamine, prostaglandin, and adenosine receptors as seen in the presence of GDP in the systems studied here. Thus, under the assay conditions used in the S49 system, the isoproterenol receptor has higher intrinsic activity than does the PGE, We do not as yet understand the reasons for this differential efficacy. It may be related to the total number of receptors and/or to the capacity of hormone.receptor complex to interact with other component(s) of the system.

Receptor Coupling to
Adenylyl Cyclases by GDP receptor (Table I), and in the NS-20 neuroblastoma system, the PGEl receptor has a higher intrinsic activity than does the adenosine receptor (Table I). The difference between the two receptors in the NS-20 system is more marked than in the S49 cell system, and in the corpus luteum, the PGEl receptor has higher efficacy than does the isoproterenol receptor (Table I).
2) In all three systems, hormone-stimulated activities were less in the presence of GDP than GTP. One explanation for this finding is that although both GTP and GDP alter receptor behavior equally, they act differently on the basic adenylyl cyclase system where GTP promotes activation by stabilizing the active conformation, but where GDP either does not modify basal activity (S49, corpus luteum) or actually promotes inhibition by shifting the equilibrium between the active and inactive conformations towards the inactive form in a manner that is barely detectable (S49, corpus luteum) or clearly evident NS-20 neuroblastoma, turkey erythrocytes).
3) If the extent of transition from the inactive to active form of the basic adenylyl cyclase induced by the active receptor is directly proportional to the efficacy factor of the receptor, it follows from points 1 and 2 above that: ( a ) adenylyl cyclase should be stimulable by an active hormone .receptor complex with or without a nucleotide bound to the regulatory site associated with the catalytic moiety; (b) the stimulation of the enzyme would be facilitated by a nucleotide that by itself tends to activate (such as GTP; see Fig. 4); (c) the stimulation of the enzyme could occur but to a lesser extent also in the presence of an inhibitory nucleotide, provided that the receptor acting on the system has a sufficiently high efficacy so as to overcome the nucleotide effect (e.g. less than optimal stimulation of S49 by isoproterenol and NS-20 and corpus luteum by PGE,); and (d) stimulation of the enzyme in the presence of inhibitory nucleotide would not be obtained if inhibition by the nucleotide is extensive, such as seen in the turkey erythrocyte adenylyl cyclase, where basal activity is more than 99% inhibited by GDP (cf Ref. 20), or if the enzyme is coupled to a low efficacy receptor even if the receptor is active, such as observed for the catecholamine-sensitive adenylyl cyclase of corpus luteum adenylyl cyclase, where no stimulation by isoproterenol is seen with GDP even though the binding effect is obtained and receptor is presumably active.
Thus, in the adenylyl cyclase systems studied here, as well as in the rat liver adenylyl cyclase, the recently proposed hypothesis (26-28) that adenylyl cyclase behaves as a hysteretic two-state enzyme system in which the state transitions are regulated independently by the active hormone .receptor complex and guanine nucleotide (GTP-stimulating and GDPinhibiting to various degrees) seems to be a reasonable one. In addition, the data and reasonings presented here indicate that receptors are affected and ultimately activated by the concerted interactions with guanine nucleotides and hormones. We have previously shown that a two-state guanine nucleotide-regulated adenylyl cyclase, upon which hormones act, serves as a signal amplifier when the guanine nucleotide site on the regulatory moiety associated with the catalytic component is occupied by a stimulatory nucleotide, and as a signal dampener if the regulatory moiety is occupied by an inhibitory nucleotide. As shown by the data of Fig. 4 and corroborated by simulations (28), amplification and dampening results in corresponding changes in both sensitivity and magnitude of responses. Although the data shown here demonstrate that under defined conditions of assay, hormones activate some adenylyl cyclases in the presence of GDP, it should be emphasised that in vivo, the relative GTP:GDP ratios are such that it is most likely that the system is occupied by GTP.
Hence, GTP should be considered the main intracellular physiological effector of the system. However, dampening effects due to GDP, such as shown here, may become important if 1) a GTPase such as proposed by Cassel and Selinger (8) is associated with the adenylyl cyclase, and 2) local compartmentalization phenomena exist in the environment (membranes) that surround the nucleotide-binding components. There exist no data at the present time to support or dispute the second suggestion. Clearly, it remains to be demonstrated whether hormonal regulation of adenylyl cyclase in the intact cells is under the influence of GDP.

O n the Interpretation of the Data: Some Limitations-
Models and interpretations are not always unique. There may, therefore, be alternative explanations for the data presented above. An alternative explanation would have to account, just as our interpretation does, for the following findings: 1) stimulation of the rate of activation of GMP-P(NH)P by hormones independent of increases of nucleotide exchange rates in some but not all systems; 2) GDP-mediated hormonal stimulation in some adenylyl cyclases, even though both GTP and GDP act equally well at the level of hormone-receptor interaction; 3) inhibitory guanine nucleotide mediation of positive coupling of a nucleotide-dependent, hormone-occupied receptor; 4) a concentration effect curve for hormones obtained in the presence of a guanine nucleotide that stimulates basal adenylyl cyclase activity more effectively lies to the left of that obtained in the presence of a less effectively stimulating guanine nucleotide (28); and 5) signal-dampening consequences observed upon substituting GDP for GTP.
In spite of being able to account for all of the above experimental findings, the model set forth by us is limited, for it is based solely on kinetic arguments. Other approaches, especially biochemical ones, should prove useful in elucidating the molecular mechanism of how hormone receptors couple to adenylyl cyclase.