点击显示 收起
Department of Pharmacology and Clinical Pharmacology (.U., .S.., K.S., H.O.O.) and Molecular Biology and Technology Research and Development Unit (H.O.O.), Ankara University Faculty of Medicine, Ankara, Turkey
Department of Pharmacology, Istituto Superiore di Sanite? Rome, Italy (P.M., C.A.)
Abstract
2-Adrenoceptor-mediated activation of Gs and adenylyl cyclase or other receptor-mediated G protein activations is believed to occur by receptor-catalyzed replacement of GDP with GTP on the -subunit of the G protein. Here we showed that a 2-adrenoceptor-Gs system, heterologously expressed in cyc- or human embryonic kidney (HEK)-293 cells, can be activated in the presence of GDP or its phosphorylation-resistant analog, guanosine 5'-O-(2-thiodiphosphate) (GDPS). The potency and maximal ability of GDP to activate Gs and adenylyl cyclase were identical to those of GTP. GDP-mediated activation of adenylyl cyclase, similar to that mediated by GTP, was concentration-dependent, required high magnesium concentrations, was inhibited by inverse agonists, and was correlated with the efficacy of receptor ligands used to stimulate the receptor. UDP did not block the GDP-mediated activation, although it completely blocked the formation of a small amount of GTP (5% GDP) from GDP. Moreover, the activation of Gs in the presence of GDP was insensitive to cholera toxin treatment of the cells, whereas that observed in the presence of GTP was amplified by the treatment, which showed that the activation observed in the presence of GDP was not mediated by GTP. Therefore, we concluded that GDP itself could mediate -adrenoceptor-induced activation of Gs-adenylyl cyclase system as much as GTP. We discuss the results in the context of the current paradigm of receptor-mediated G protein activation and propose an additional mode of activation for 2-adrenoceptor-Gs adenylyl cyclase system where nucleotide exchange is not necessary and GDP and GTP play identical roles in receptor-induced Gs protein activation.
Heterotrimeric G proteins and their cognate heptahelical membrane receptors constitute the largest family of transmembrane signaling proteins. According to the accepted model of G protein activation, the GTP-bound form of G protein is active, whereas the GDP-bound form is inactive and the receptor-mediated exchange of GDP with GTP on the nucleotide-binding site is necessary and sufficient to initiate the activation process. In turn, deactivation is achieved by the hydrolysis of bound GTP to GDP. Fine-tuning of the rates of the elementary steps involved in this activation-deactivation cycle determines the average activity of the G protein (i.e., the amount of accumulated G-GTP), thus the strength of the signal transmitted to the inside of the cell (for review see Gilman, 1987; Hamm, 1998). This scenario implies that GTP is the natural activator of the G proteins, whereas GDP is their universal inhibitor, and that GTPase reaction is the off-switch for the G protein-mediated signal transduction. Therefore, the presence of excess GDP is expected to inhibit receptor-induced G protein activation, at least partly, by competing with GTP and the potency of this inhibitory effect should be correlated with the ability of the receptor's ligand to displace GDP from G. Viewed from a more rigorous standpoint, the sensitivity of receptor-mediated G protein activation to the presence of GDP should be inversely correlated with the ability of the ligand to reduce the affinity of GDP for G. Based on this obvious inference, we designed experiments to compare GDP-displacing abilities of different 2-adrenoceptor (-AR) ligands in native or artificially fused -AR-Gs system by evaluating the sensitivity of ligand-induced adenylyl cyclase activity (which is well known to be mediated by Gs protein) to the presence of GDP. To our surprise, however, neither in native nor in fused system expressed in S49 cyc- lymphoma cells (will be referred to as cyc- throughout the text), GDP inhibited ligand-induced adenylyl cyclase activity, despite the fact that even full agonists were not able to reduce the binding affinity of GDP to zero, as assessed by competition-binding assays ( Uur and H. O. Onaran, unpublished observation).
A number of experiments in the literature show that Gs (or Gi) protein can indeed be activated in the presence of GDP (Iyengar and Birnbaumer, 1979; Schneyer et al., 1984; Harding and Harris, 1986; Quist et al., 1992; Piacentini et al., 1996; Seifert et al., 1998, Hatley et al., 2003; Lutz et al., 2002). These reports include observations dating from late 1970s to the present day. However, the activation in the presence of GDP is considered as either inconsequential or artifactual (i.e., explained by the conversion of GDP to GTP by nucleotide diphosphate kinase activity during the experiments) (Kimura and Schimada, 1983; Kikkawa et al., 1990). Thus, the true nature and significance of GDP-mediated activation of the G protein have never been fully understood. The possibility that G proteins can be activated in the presence of GDP may call for a need for a revision of the established paradigm of the activation of G proteins summarized above and may shed light on the understanding of the molecular mechanism of receptor-mediated activation of G proteins. Therefore, we further investigated and characterized the -AR-mediated adenylyl cyclase activation observed in the presence of GDP.
Materials and Methods
Plasmid Constructs. The cDNA encoding the long form of rat Gs was in the pCDNA3.1(+) vector (Invitrogen, Carlsbad, CA). Construction of cDNA encoding the fusion proteins of human -AR or -opioid receptor and the long form of rat Gs is described elsewhere (Molinari et al., 2003).
Cell Culture and Transfection. We used the following cell systems throughout the present experiments: 1) native S49 cells; 2) cyc- cells transfected with GsL;3)cyc- cells transfected with -AR-Gs fusion protein; 4) HEK-293 cells transfected with -AR; 5) HEK-293 cells transfected with -AR-Gs fusion protein; and 6) HEK-293 cells transfected with -opioid receptor-Gs fusion protein. All of the cell lines were cultured in Dulbecco's modified Eagle's medium with 10% fetal calf serum in standard conditions. cyc- cells were transfected by electroporation essentially as described previously (Gonzales et al., 1992). HEK-293 cells were transfected with FuGENE 6 (Roche, Mannheim, Germany). Stable clones were selected in the culture medium containing 600 to 700 e蘥/ml G418 as described previously (Ugur et al., 2003).
Cholera Toxin Treatment. cyc- cells stably transfected with cDNA encoding Gs were incubated with holotoxin (at a final concentration of 1 e蘥/ml) under standard cell culture conditions for 2 h, which led to a considerable increase in basal intracellular cAMP accumulation. We did not check the completeness of the ADP-ribosylation, because the amplification of the GTP-mediated response in the cell membranes was the only requirement for the purpose of the present experiments (see Results).
Adenylyl Cyclase Assays. Adenylyl cyclase activity was measured essentially as described previously (Uur and Onaran, 1997). In brief, membranes were mixed in 96-well plates (0.25-2 e蘥 of protein/well) with 75 e蘬 of the indicated additives, and then the assay was initiated by adding 25 e蘬 of adenylyl cyclase assay buffer (unless indicated otherwise, final concentrations were 50 mM Tris-HCl, pH 7.4, 100 mM KCl, 10 mM MgCl2, 250 e ATP, and 1 mM isobutyl-methylxanthine) and terminated by adding 100 e蘬 of 0.2 N HCl. Before adding the ATP-containing adenylyl cyclase buffer, cell membranes and relevant additives were incubated together for 1 to 2 min. Preincubation time with GTPS was 15 min, because our kinetic experiments showed that adenylyl cyclase activation in the presence of GTPS reached a peak in 15 min. All of the incubations were at 37°C, and total assay time was 5 min. Time-dependent cAMP accumulation was determined by sampling the reaction mixture at the indicated time points. In this case, the reaction was started by adding membranes to the rest of the mixture. The amount of accumulated cAMP was measured by radioimmunoassay in all of the experiments as described previously (Ugur and Onaran, 1997). Control experiments showed that different nucleotides (GMP, GDP, GTP, ATP, etc.) when used at the same concentrations as in the experiments did not interfere with the radioimmunoassay determination.
We used an indirect approach to measure GDP-displacing ability of receptor ligands based on the fact that AlF can activate the G protein only in its GDP-bound form. We incubated cell membranes with relevant ligands for 5 min in the absence of guanine nucleotides to let GDP dissociate from Gs and then determined cyclase activity after adding NaF + AlCl3 as described above. We interpreted ligand-dependent "inhibition" of AlF-induced cyclase activity as the ligand's ability to displace GDP from Gs. As a normalizing control, we made the same experiment in the presence of GDP during incubation with ligands.
Membrane Preparation and Other Procedures. Homogenized cells (obtained by passing the cell suspension 10 to 15 times through a 26-gauge syringe tip in a hypotonic buffer) were centrifuged at 400g for 5 min, and the supernatant was repelleted at 100,000g. The pellet was washed once again, and the final pellet was resuspended in a buffer containing 50 mM Tris-HCl, pH 7.4, 10 mM MgCl2, protease mixture, and 25% (w/v) sucrose and stored at -70°C. Protein concentrations were determined by the Bradford assay (Bradford, 1976). Radioligand-binding assays and immunoblotting were performed by using standard methods. Radioligand-binding assays were performed in the binding buffer containing 50 mM Tris-HCl, pH 7.4, 100 mM KCl, and 10 mM MgCl2 in a final volume of 250 e蘬 containing 1 to 2 e蘥 of membrane protein at 27°C for 3 h. Reaction was stopped by filtration through Whatman GF/B filters by using a cell harvester (Skatron Instruments, Lier, Norway). For immunoblotting, 5 e蘥 of membrane protein was separated by SDS-polyacrylamide gel electrophoresis and electrotransferred to nitrocellulose or polyvinylidene difluoride paper [Bio-Rad (Hercules, CA) or Millipore (Billerica, MA)]. The proteins were detected with the polyclonal affinity-purified RM antibody (1 e蘥/ml), which recognizes the carboxyl-terminal decapeptide of Gs, and enhanced chemiluminescence (Amersham Biosciences, Piscataway, NJ) as described by the manufacturer. Horseradish peroxidase-conjugated goat anti-rabbit antibodies (developed in the Molecular Biology and Technology Research and Development Unit, Ankara University, Ankara, Turkey) were used to visualize the relevant bands.
Chromatographic separation of nucleotides was achieved by a continuous NaCl and pH gradient in 5 mM phosphate buffer using an HPLC system equipped with a UV-absorption detector (Jasco, Tokyo, Japan) and a weakly basic anion-exchanger column (Protein-Pak DEAE-5PW; Waters, Milford, MA). Fractions were collected in a volume of 500 e蘬, and radioactivity in the fractions was determined in a scintillation counter (Wallac MicroBeta 1450 Trilux; Millipore Corp.).
Results
Effect of GDP on Agonist-Induced Activation of Adenylyl Cyclase. We stably expressed fusion protein of human -AR and the long form of rat Gs in cyc- (which does not express Gs intrinsically) or in HEK-293 cells. Expression of the fusion protein in the selected clones was roughly 4 and 40 pmol/mg of membrane protein in cyc- and HEK-293 cells, respectively, as assessed by 125I-iodocyanopindolol saturation binding. Integrity of the fusion protein in the cell membranes was assessed by immunoblotting (data not shown).
We measured stimulated adenylyl cyclase activity as a readout for Gs activation in cell membranes. Isoproterenol, a -AR agonist, activated adenylyl cyclase in a concentration-dependent manner in the presence but not in the absence of 1 e GTP in membranes prepared from cyc- or HEK-293 cells expressing the fusion protein. Untransfected control cell membranes exhibited no agonist-mediated stimulation of cyclase activity. Considering that GTP is the activating nucleotide for G proteins and that G protein activation is achieved by receptor-induced exchange of GDP with GTP, the presence of excess GDP would be expected to inhibit -AR-induced cyclase activation by competing with GTP. However, excess amounts of GDP had no effect on the adenylyl cyclase activation induced by isoproterenol in the presence of GTP; potency and maximal level of isoproterenol-induced adenylyl cyclase activities obtained in the absence or presence of 500 e GDP were indistinguishable (Fig. 1, A and B) as assessed by the means of F statistics based on the evaluation of extraresidual variance due to parameter sharing in a family of nonlinear regressions (DeLean et al., 1978). Moreover, GDP (in the absence of added GTP) activated adenylyl cyclase in cyc- membranes expressing -AR-Gs fusion protein in the presence of 100 e isoproterenol (with a maximal response not different from that mediated by GTP under identical conditions) (Fig. 1C). Indeed, the latter observation is consistent with the first one in the sense that GDP must mediate G protein activation as efficiently as GTP; otherwise, it would inhibit the adenylyl cyclase activation in the presence of GTP (see also below).
A full comparison of GDP- or GTP-mediated cyclase activation in HEK-293 cell membranes expressing the fusion protein is given in Fig. 2A, where similar potency for GDP and GTP in supporting cyclase activity is evident in the presence or absence of agonist. The same pattern of activation was observed in 1) HEK-293 cell membranes that heterologously express comparable amount of (nonfused) human -AR (Fig. 2B); 2) cyc- cell membranes that express Gs heterologously but -AR endogenously (Fig. 2C); or 3) wildtype S49 cell membranes that express both Gs and -AR (400 fmol/mg) endogenously (Fig. 2D). This showed that the similarity of GDP and GTP in supporting Gs activation was not a peculiarity of the fusion protein or artificial overexpression of the receptor or Gs protein. Note that the absolute value of overall signal and the relative magnitude of basal cyclase activity (in the absence of agonist but in the presence of GTP or GDP) diminished as the anticipated contribution of the receptor decreased (i.e., fusion > overexpression > normal expression) but that the general pattern remained almost the same (Fig. 2, compare A-D). Unlike GDP, GMP showed no activating effect in these experimental systems (data not shown).
In all instances, the concentration-response curves for the guanine nucleotides were bell-shaped with a descending arm starting from 10 e. Such a bell-shaped tendency was also evident in GTPS-induced activation of adenylyl cyclase in HEK-293 membranes that express -AR-Gs fusion protein (Fig. 2E). High concentrations of the nucleotides inhibited both the nucleotide-independent background (see below) and forskolin-induced cyclase activity to the undetectable level (data not shown), suggesting that the effect might be on the cyclase molecule rather than on Gs. However, we did not investigate further the possible causes of this inhibitory effect seen at high nucleotide concentrations.
We reproduced the same results with GDP preparations obtained from three different sources [Sigma-Aldrich (St. Louis, MO), Roche Applied Science (Indianapolis, IN), or Valeant Pharmaceuticals (Costa Mesa, CA)], HPLC and NMR (P, C, and proton) analyses of which showed no significant contamination with GTP or with any other organic chemicals (results not shown) that may explain the observed similarity of GDP and GTP in supporting Gs-mediated adenylyl cyclase activation.
Adenylyl cyclase activation observed in the presence of GDP or GTP (with or without agonist) was magnesium-dependent, and its sensitivity to [Mg2+] was the same for GDP and GTP (Fig. 3A), whereas the activation was invariant with respect to the salt (NaCl or KCl) used in the assay buffer (Fig. 3B). The only difference between the presence of potassium and sodium was that the inhibitory effect of inverse agonist (±)-1-[2,3-(dihydro-7-methyl-1H-indene-4-yl)oxy]-3-[(1-methylethyl)amino]-2-butanol (ICI-118,551) on the basal cyclase activity was more pronounced in the presence of K+. A more detailed analysis of the relationship between guanine nucleotide-mediated cyclase activation and ligand efficacy is given below.
GDP-Mediated Adenylyl Cyclase Activation Does Not Result from Conversion to GTP. Early observations that Gs or Gi could be activated in the presence of GDP (Iyengar and Birnbaumer, 1979; Schneyer et al., 1984; Harding and Harris, 1986; Quist et al., 1992; Piacentini et al., 1996; Lutz et al., 2002) have mostly been attributed to the conversion of GDP to GTP during the experiments (Kimura and Schimada, 1983; Kikkawa et al., 1990; Kowluru et al., 1996). Such a conversion must be nearly complete and almost instantaneous to be an explanation for the present results, because GTP and GDP supported the cyclase activation with almost identical potencies in the presence or absence of agonist. This implies that equal concentrations of GTP and GDP lead to an almost equal level of adenylyl cyclase activation throughout the concentration-response curves (see Figs. 2 and 6). Under the assumption that only GTP is the activating nucleotide, the observation of a given level of adenylyl cyclase activity with equal concentrations of GDP and GTP is possible only when the conversion of GDP to GTP is complete in the bulk solution or near the nucleotide-binding site (or even at the nucleotide-binding site) of Gs. The following control experiments showed that the conversion in the bulk solution cannot explain the results. 1) HPLC analysis of nucleotides before and after incubation with HEK-293 cell membranes under identical conditions used in cyclase assays showed no obvious loss of GDP (100 e), which ruled out a bulk conversion of GDP to GTP in solution (data not shown). 2) Using 32P-labeled GDP, we detected 5 or 21% conversion of GDP to GTP in 5 or 30 min, respectively. UDP, which is known to inhibit nucleotide diphosphate kinase activities (Kimura and Schimada, 1983) or local transphosphorylation of GDP on the nucleotide-binding site of G (Kikkawa et al., 1990; Kowluru et al., 1996), totally blocked the conversion in 5 min (which was the maximum incubation time used in the present assays) (Fig. 4A) but did not affect GDP- (or GTP-) mediated activation of adenylyl cyclase significantly (Fig. 4, B and C). Moreover, in S49 (or in cyc-) membranes, the conversion of GDP to GTP in the same conditions was even less (1.5% of initial GDP) than that observed in HEK-293 cells (data not shown), whereas the pattern of GDP- or GTP-mediated activation was still the same. 3) GDPS, a phosphorylation-resistant analog of GDP (Eckstein et al., 1979; Kowluru et al., 1996), was able to mediate adenylyl cyclase activation as much as GTP or GDP (Fig. 4D). 4) In the absence of ATP but in the presence of 500 e App(NH)p, a substrate for adenylyl cyclase but not a phosphate donor, GDP was still able to support receptor-induced cyclase activation (Fig. 4E). Finally, further washing the membranes to remove loosely attached membrane proteins or contaminating nucleotides did not change the activation pattern observed in the presence of GDP (data not shown).
These findings clearly show that a bulk conversion of GDP to GTP cannot explain the present results. However, a locally complete in situ conversion still remains a possibility that can explain the results. This has always been a problem for obvious technical difficulties in measuring in situ concentration of GTP formed from GDP. Therefore, we tried to get around this problem by designing the following two experiments. The logic of the first experiment is as follows. If Gs (and adenylyl cyclase) activation were mediated only by GTP, then any finite-rate process that converts GDP to GTP would be expected to cause a lag that would make the kinetics of adenylyl cyclase activation different depending on whether the experiment was started in the presence of GDP or GTP. Therefore, we evaluated time-dependent cAMP accumulation as an integral measure of time-dependent adenylyl cyclase activity in the presence or absence of isoproterenol and in the presence of GDP or GTP in HEK-293 membranes expressing -AR-Gs fusion protein. Results are shown in Fig. 5A; the process of cAMP accumulation was identical for GTP and GDP and did not depart from linearity in either case during 5 min of observation, and steady-state adenylyl cyclase activity was achieved almost instantaneously with the addition of either nucleotide. This implies either that the local conversion from GDP to GTP does not occur and binding of GDP directly activates Gs or that the rate of the conversion is so fast that the formation of GTP is quasi-instantaneous and no GDP can stay in the vicinity of Gs, which is also an interesting possibility. To distinguish between these possibilities, we made the second experiment. We used cholera toxin (CTX)-treated membranes as a reporter of the type of nucleotide locally available to Gs. In cyc- cells expressing the long form of Gs, cyclase activity was 4 times higher in CTX-treated membranes than in the control membranes in the presence but not in the absence of added GTP. Therefore, the comparison of the reactivities of CTX-treated and untreated membranes could provide information on the presence of GTP available to Gs. In other words, adenylyl cyclase activity of CTX-treated membranes could be used as a bioassay in which Gs itself could report the presence of available local GTP in its microenvironment. In this bioassay, we saw no such "report from Gs" (i.e., signal amplification by CTX) in a wide range of GDP concentrations (Fig. 5C), whereas the amplification was obvious in the presence of added GTP (Fig. 5D). If the activity observed in the presence of GDP were actually mediated by local conversion into GTP, it should also have been amplified by CTX treatment to the same extent (unless we postulate that ADP-ribosyl-Gs can tell the difference between externally added or locally formed GTP). This inevitably means that added GDP did not provide appreciable amounts of GTP to Gs in the present experimental system. Therefore, we concluded that the effect of GDP could not result from its conversion to GTP on the G-binding site or in its vicinity by any mechanism whatsoever. Note that isoproterenol-induced adenylyl cyclase activation in the presence of GDP was still evident in CTX-treated membranes (Fig. 5B).
Receptor Dependence of Cyclase Activation in the Presence of GDP or GTP. The following results showed that the cyclase activity observed in the presence of GDP (or GTP) was strictly receptor-dependent. 1) ICI-118,551, an inverse agonist that inhibits basal activity of -AR (Chidiac et al., 1994; Samama et al., 1994) suppressed the cyclase activity observed in the presence of GDP (or GTP) (Fig. 6A). 2) No nucleotide-mediated cyclase activity was observed when Gs was fused to -opioid receptor instead of -AR (Fig. 6B). Note that -opioid receptor is not actually coupled to Gs. 3) The ability of different -AR ligands to induce "GDP-mediated activation" was correlated with their ability to induce "GTP-mediated activation" (i.e., with their efficacies that are well known to be receptor-dependent) (Fig. 6C).
The present results, when considered as a whole, suggest that receptor-mediated activation of Gs (and thus adenylyl cyclase) is not necessarily achieved by receptor-mediated exchange of GDP with GTP. However, the ability of -receptor ligands to induce cyclase activation in the presence of GDP (or GTP) was still found to be correlated with their "GDP-displacing" ability in the present experimental system (Fig. 7) (see Discussion).
Two apparently different components of adenylyl cyclase activity can be distinguished by close inspection of Figs. 1, 2, and 6; one that depends on the presence of guanine nucleotides and the activity of the cognate receptor (-AR) and another that does not. The latter activity can be observed in the absence of guanine nucleotides and is insensitive to the presence of agonist or inverse agonist, despite the ability of agonists to displace GDP from the G protein (see Fig. 7) and the ability of inverse agonists to inhibit nucleotide-dependent adenylyl cyclase activity (Fig. 6). This invariant background was 200 pmol/mg/min in the -AR-Gs system. Its magnitude was apparently related with whether the Gs was fused to a receptor or not, regardless of the identity of the fused receptor (compare Fig. 2, A-C, with Fig. 6B). We did not further investigate this activity and concentrated only on the nucleotide- and receptor-dependent one.
Discussion
In the present study, we showed that -AR, Gs, and adenylyl cyclase system can be activated in the presence of GDP and that this activation could not be attributed to the conversion of GDP to GTP. The following is a brief discussion of the observed phenomenon and its implications that are apparently not concordant with the current paradigm of G protein-mediated signal transduction. By current paradigm, we mean the basic idea that the GDP-bound form of the G protein is inactive, that receptor stimulation "catalyzes" the exchange of GDP with GTP on the G protein, yielding the active GTP-bound state, and that GTPase activity of G is the turn-off mechanism for the activation (for review see Gilman, 1987; Hamm, 1998).
Experimental foundations of the current paradigm for G protein activation can be roughly summarized as follows. 1) G subunits bind GDP or GTP (or their analogs) and hydrolyze the bound GTP to GDP in the presence of magnesium (Cassel and Selinger, 1978; Higashijima et al., 1987). Receptor-induced G protein activation requires the presence of GTP (Rodbell et al., 1974; Hanoune et al., 1975). 2) Purified G proteins, when reconstituted with their effectors in the GDP-bound form, lack the ability to modify effector activity, whereas nonhydrolyzable analogs of GTP [GTPS or Gpp(NH)p] are able to activate the G protein (Graziano et al., 1989). Likewise, nonhydrolyzable GTP analogs can permanently activate G proteins in the cell membrane, even in the absence of receptor stimulation (Schramm and Rodbell, 1975). 3) Steady-state rate of GTPase is limited by the dissociation rate of GDP, which holds the G protein in the GDP-bound form. Receptor activation increases the rate of GDP dissociation without affecting the intrinsic rate of GTPase and thus leads to the exchange of GTP for GDP in the presence of GTP (Higashijima et al., 1987). This is observed as an increased rate of steady-state GTPase (Brandt et al., 1983; Cerione et al., 1984), which is correlated with the efficacy of the ligand used to stimulate the receptor (Costa and Hertz, 1989). 4) Inhibition of the GTPase activity by CTX-mediated ADP ribosylation of G (Johnson et al., 1978) or by mutations in the G subunit (Graziano and Gilman, 1989) increases the constitutive activity of the G protein in the presence of GTP, whereas mutations or regulating factors, such as regulators of G protein signaling, that increase the GTPase activity inhibit G protein activation (Watson et al., 1996). 5) Fluoroaluminate ion (AlF-4), which imitates the -phosphate of GTP in the nucleotide-binding pocket of G protein, can activate the GDP-bound G protein in purified or native systems (Bigay et al., 1985; Higashijima et al., 1991). Finally, 6) the crystal structures of G proteins have revealed that there are common conformational changes in the active G subunit (bound with GTPS or GDP-AlF4) compared with its GDP-bound inactive form (Noel et al., 1993; Coleman et al., 1994; Lambright et al., 1994).
On the other hand, the present observations can be summarized as follows. 1) Adenylyl cyclase could be activated in the presence of GDP in a concentration-dependent manner. 2) GDP had almost similar potency with GTP in this activation. 3) GDP-mediated activation of adenylyl cyclase, similar to GTP-mediated activation, required the receptor activity; minimizing receptor signal by using inverse agonists or by fusing "wrong" receptor (-opioid) to Gs prevented nucleotide-mediated activation. 4) This activation could still be observed when GDPS was used instead of GDP or in the presence of App(NH)p (instead of ATP) or UDP (which fully blocked the minimal conversion of GDP to GTP in the present experimental system). When considered collectively, the present results strongly suggest that the observed activation cannot be attributed to the conversion of GDP to GTP by any mechanism whatsoever (see Results). Among all of the control experiments, the one in which we used CTX-treated membranes as a bioassay tool (see Results for the logic) constitutes the most critical evidence for this conclusion.
We used two different cell systems (HEK-293 and S49 or its cyc- variant) to rule out the possibility that the present phenomenon was a peculiarity of a particular cell line that hosts receptor, G protein, and adenylyl cyclase. Moreover, the present phenomenon was observable in the following expression combinations: 1) S49 cells, a native system that expresses Gs and -AR endogenously (Fig. 2D); 2) cyc- cells that express -AR endogenously but Gs heterologously (Fig. 2C); 3) HEK-293 cells that express -AR heterologously but Gs endogenously (Fig. 2B); or 4) HEK-293 or cyc- cells that express -AR-Gs fusion construct heterologously (Figs. 1 and 2A) in which the patterns of activation by GDP or GTP with or without agonist were all the same. This also showed that the observed phenomenon did not depend on whether the Gs protein or -AR was overexpressed by heterologous transfection or whether they were both fused together. Therefore, we confidently used the -AR-Gs fusion construct to maximize receptor signal in the relevant experiments.
Although the current paradigm that attributes G protein activation to the GTP bound form is an inevitable and firm inference made from the above-mentioned facts, it obviously does not explain the present observations made in the -AR-Gs adenylyl cyclase system. Therefore, we propose an additional (but not necessarily alternative) mechanism as a plausible explanation for the present data. Activated -AR can convert Gs to a state that is active independently of the identity of the bound nucleotide (i.e., GDP or GTP); in other words, the active receptor, when bound to Gs, can transmit a conformational signal to Gs, which results in a Gs state that can fruitfully interact with adenylyl cyclase, even when it is bound with GDP (see Fig. 8 for a schematic representation of the idea). Inactivation, in turn, is simply achieved upon termination of the "activating" signal from the receptor (e.g., upon dissociation of agonist from receptor). The existence of such a state that is equally active in its GDP-bound form under the influence of the receptor implicitly means that the GDP-GTP exchange on Gs is not necessary for 2-receptor-induced activation. In fact, we were unable to witness any advantage of GTP over GDP in supporting receptor-mediated adenylyl cyclase activity; the two nucleotides behaved almost identically in all respects (save the CTX effect, but see below) (Figs. 2, 3, 4, 5, 6). When the basal activity of the receptor was suppressed by inverse agonists or by modifying the receptor expression qualitatively or quantitatively, GTP-mediated activation was affected as much as the GDP-mediated one. This suggests that, in Gs, GTP, unlike GTPS (or Gpp(NH)p), may not be an activating nucleotide by itself but, similar to GDP, requires the intervention of the receptor for its activating potency to become apparent. However, the binding of a nucleotide (either GDP or GTP) seems to be necessary for Gs to adopt this hypothetical state because no receptor-induced cyclase activation was observed in the absence of the nucleotides (see Figs. 2 and 6).
An independent experimental support for the latter hypothesis that GTP-Gs may not be active in the absence of receptor intervention can be found in the functional behavior of an interesting mutant of Gs where Gln170 in the helical domain was changed to Ala. This mutation resulted in a 4-fold increase in GDP dissociation rate without affecting the intrinsic GTPase or its coupling to receptor and effector (for details, see Warner and Weinstein, 1999). Hence, it is reasonable to expect that, in the presence of GTP, the GTP-bound form of the G protein accumulates more in Q170A-mutant than in the wild type, which should lead to a higher cyclase activity, if the GTP-bound form of Gs were active on its own. However, it was not so experimentally. The cyclase-activating ability of the mutant and wild type was exactly the same in the presence of GTP alone. These results are consistent with the idea that GTP binding to Gs is not a sufficient condition for activating Gs, but an additional activating signal from the receptor is required. Indeed, by using environmentally sensitive fluorophore-labeled guanine nucleotides, previous studies (Remmers et al., 1994; Remmers and Neubig, 1996) have shown that, compared with GTPS, the conformational effect of GTP on purified Go protein is partial (independently of the GTPase) and that mastoparan, a receptor mimicking wasp venom peptide, can induce activation not only by facilitating the dissociation of GDP but also by inducing a conformational change on GTP-bound Go (i.e., by converting inactive GTP-Go to active GTP-Go as the authors have stated), which suggests that the above hypothesis may also apply to Go protein. However, different receptor-G protein-effector combinations should be studied in direct functional assays to generalize the hypothesis. Hence, at present, it should be limited to -AR-Gs-cyclase system.
The mechanism we proposed here, based on the hypothesis that guanine-nucleotide exchange is neither required nor sufficient for Gs activation, seems to be the most plausible explanation for the present observations. Other explanations such as GDP-induced disinhibition of Gi protein (Piacentini et al., 1996; Lutz et al., 2002) or conversion of GDP to GTP do not explain the present results as a whole. However, there are still two compelling arguments that suggest that the guanine nucleotide exchange is a requirement for G activation and that GTP is the activating nucleotide. One is the strong correlation between the ability of ligand to induce GDP dissociation and ligand-induced activation of adenylyl cyclase. Such a correlation is also shown here under the same experimental conditions in which GDP activated adenylyl cyclase as efficiently as GTP (Fig. 6D). Thus, this correlation should not necessarily prove a causal link between receptor-induced GDP dissociation and activation but implies that these two processes (i.e., receptor-induced activation and increased rate of GDP dissociation) might be two different facets of the active conformation of the G protein. In other words, they may occur simultaneously as the protein enters into the active state, rather than being mechanistically dependent from one another. The second comes from the observations that mutations and modifications causing reduction or enhancement of GTPase activity also result, respectively, in enhancement (Graziano and Gilman, 1989) or inhibition (Watson et al., 1996; Warner and Weinstein, 1999) of GTP-mediated G protein activation. This suggests that the GTP-bound form of G is the active state and supports the idea that hydrolysis of bound GTP to GDP is the turn-off mechanism of activation. However, we show here (as others did under different experimental systems) that GDP does not turn off but actually sustains receptor-mediated activation of cyclase via Gs. We also show that ADP-ribosylation by CTX, a modification that diminishes GTPase activity, does not change GDP-mediated activation but dramatically amplifies that induced by GTP. According to the classic paradigm, toxin-induced reduction of intrinsic GTPase activity enhances the lifetime of the GTP-bound form of G and hence produces constitutive activation. However, it is just as likely that the primary effect of ADP-ribosylation (or modification of GTPase activity in general) may be to trigger an active state of GTP-bound Gs in the absence of receptor stimulus. Thus, we may speculate that the GTPase incompetence of the Gs protein may be conformationally (rather than causally) linked to its ability to assume a GTP-bound active state in the absence of receptor stimulus. In other words, there may not be a causal relationship between the inhibition of GTPase and enhanced ability of GTP to activate Gs protein. These two processes may simply be two different aspects of the same conformational perturbation imparted by the factors that eventually interfere with the GTPase activity. Such a perturbation can be achieved either by modifying the protein (i.e., ADP-ribosylation or mutagenesis) or the nucleotide chemistry [e.g., GTPS or Gpp(NH)p].
In conclusion, the present results show that -AR activates Gs (and thus adenylyl cyclase) in the presence of excess GDP, which implies that the activation can occur without the involvement of a nucleotide exchange on the Gs subunit. The hypothesis that the Gs protein can assume an active conformation under the influence of -AR, regardless of the identity of the bound nucleotide (GTP or GDP), seems to be a plausible explanation for the present results. However, it should be kept in mind that the observed phenomena may be a peculiarity of -AR, Gs (long form), and adenylyl cyclase system and its generalization to other receptor and G protein systems obviously requires further studies. We believe that it is worth subjecting this hypothesis to further experimental testing, which may eventually help understand the molecular mechanism(s) of receptor-mediated activation of Gs or G proteins in general.
Acknowledgements
We thank Dr. Inci Akahin (TUBITAK, Ankara Test and Analysis Laboratories, Ankara, Turkey) for performing NMR analysis of the nucleotide preparations and Dr. Tommaso Costa (ISS, Rome, Italy) for critical suggestions.
References
Bigay J, Deterre P, Pfister C, and Chabre M (1985) Fluoroaluminates activate transducin-GDP by mimicking the gamma-phosphate of GTP in its binding site. FEBS Lett 191: 181-185.
Bradford MM (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 72: 248-254.
Brandt DR, Asano T, Pedersen SE, and Ross EM (1983) Reconstitution of catecholamine-stimulated guanosinetriphosphatase activity. Biochemistry 22: 4357-4362.
Cassel D and Selinger Z (1978) Mechanism of adenylate cyclase activation through the beta-adrenergic receptor: catecholamine-induced displacement of bound GDP by GTP. Proc Natl Acad Sci USA 75: 4155-4159.
Cerione RA, Codina J, Benovic JL, Lewkowitz RJ, Birnbaumer L, and Caron MG (1984) The mammalian beta 2-adrenergic receptor: reconstitution of functional interactions between pure receptor and pure stimulatory nucleotide binding protein of the adenylate cyclase system. Biochemistry 23: 4519-4525.
Chidiac P, Hebert TE, Valiquette M, Dennis M, and Bouvier M (1994) Inverse agonist activity of -adrenergic antagonists. Mol Pharmacol 45: 490-499.
Coleman DE, Lee E, Mixon MB, Linder ME, Berghuis AM, Gilman AG, and Sprang SE (1994) Crystallization and preliminary crystallographic studies of Gi alpha 1 and mutants of Gi alpha 1 in the GTP and GDP-bound states. J Mol Biol 238: 630-634.
Costa T and Hertz A (1989) Antagonists with negative intrinsic activity at delta opioid receptors coupled to GTP-binding proteins. Proc Natl Acad Sci USA 86: 7321-7325.
DeLean A, Munson PJ, and Rodbard D (1978) Simultaneous analysis of families of sigmoidal curves: application to bioassay, radioligand assay, and physiological dose response curves. Am J Physiol 235: E97-E102.
Eckstein F, Cassel D, Levkowitz H, Lowe M, and Selinger Z (1979) Guanosine 5'-O-(2-thiodiphosphate). An inhibitor of adenylate cyclase stimulation by guanine nucleotides and fluoride ions. J Biol Chem 254: 9829-9834.
Gilman AG (1987) G proteins: transducers of receptor-generated signals. Ann Rev Biochem 56: 615-649.
Gonzales JM, O'Donnell JK, Stadel JM, Sweet RW, and Molinoff PB (1992) Down-regulation of beta-adrenergic receptors by pindolol in Gs alpha-transfected S49 cyc-murine lymphoma cells. J Neurochem 58: 1093-1103.
Graziano MP and Gilman AG (1989) Synthesis in Escherichia coli of GTPase-deficient mutants of Gs alpha. J Biol Chem 264: 15475-15482.
Graziano MP, Freissmuth M, and Gilman AG (1989) Expression of Gs alpha in Escherichia coli. Purification and properties of two forms of the protein. J Biol Chem 264: 409-418.
Hamm HE (1998) The many faces of G protein signaling. J Biol Chem 273: 669-672.
Hanoune J, Lacombe ML, and Pecker F (1975) The epinephrine-sensitive adenylate cyclase of rat liver plasma membranes. Role of guanyl nucleotides. J Biol Chem 250: 4569-4574.
Harding SE and Harris P (1986) GDP activates rabbit heart adenylate cyclase but does not support stimulation by isoproterenol: a re-appraisal of the control mechanism. J Mol Cell Cardiol 18: 793-806.
Hatley ME, Lockless SW, Gibson SK, Gilman AG, and Ranganathan R (2003) Allosteric determinants in guanine nucleotide-binding proteins. Proc Natl Acad Sci USA 100: 14445-14450.
Higashijima T, Ferguson KM, Smigel MD, and Gilman AG (1987) The effect of GTP and Mg2+ on the GTPase activity and the fluorescent properties of Go. J Biol Chem 262: 757-761.
Higashijima T, Graziano MP, Suga H, Kainosho M, and Gilman AG (1991) 19F and 31P NMR spectroscopy of G protein alpha subunits. Mechanism of activation by Al3+ and F-. J Biol Chem 266: 3396-3401.
Iyengar R and Birnbaumer L (1979) Coupling of the glucagon receptor to adenylyl cyclase by GDP: evidence for two levels of regulation of adenylyl cyclase. Proc Natl Acad Sci USA 76: 3189-3193.
Johnson GL, Kaslow HR, and Bourne HR (1978) Reconstitution of cholera toxin-activated adenylate cyclase. Proc Natl Acad Sci USA 75: 3113-3117.
Kikkawa S, Takahashi K, Takahashi K, Shimada N, Ui M, Kimura N, and Katada T (1990) Conversion of GDP into GTP by nucleoside diphosphate kinase on the GTP-binding proteins. J Biol Chem 265: 21536-21540.
Kimura N and Schimada N (1983) GDP does not mediate but rather inhibits hormonal signal to adenylate cyclase. J Biol Chem 258: 2278-2283.
Kowluru A, Seavey SE, Rhodes CJ, and Metz SA (1996) A novel regulatory mechanism for trimeric GTP-binding proteins in the membrane and secretory granule fractions of human and rodent beta cells. Biochem J 313: 97-108.
Lambright DG, Noel JP, Hamm HE, and Sigler PB (1994) Structural determinants for activation of the alpha-subunit of a heterotrimeric G protein. Nature (Lond) 369: 621-628.
Lutz S, Baltus D, Jakobs KH, and Niroomand F (2002) Spontaneous release of GDP from Gi proteins and inhibition of adenylyl cyclase in cardiac sarcolemmal membranes. Naunyn-Schmiedeberg's Arch Pharmacol 365: 50-55.
Molinari P, Ambrosio C, Riitano D, Sbarccia M, Gre?MC, and Costa T (2003) Promiscuous coupling at receptor-G fusion proteins. The receptor of one covalent complex interacts with the -subunit of another. J Biol Chem 278: 15778-15788.
Noel JP, Hamm HE, and Sigler PB (1993) The 2.2 crystal structure of transducin-alpha complexed with GTP gamma S. Nature (Lond) 366: 654-663.
Piacentini L, Mura R, Jakobs KH, and Niroomand F (1996) 8. Stable GDP analog-induced inactivation of Gi proteins promotes cardiac adenylyl cyclase inhibition by guanosine 5'-(beta gamma-imino)triphosphate and muscarinic acetylcholine receptor. Biochem Biophys Acta 1282: 11-16.
Quist E, Powell P, and Vasan R (1992) Guanylnucleotide specificity for muscarinic receptor inhibitory coupling to cardiac adenylate cyclase. Mol Pharmacol 41: 177-184.
Remmers AE and Neubig RR (1996) Partial G protein activation by fluorescent guanine nucleotide analogs. Evidence for a triphosphate-bound but inactive state. J Biol Chem 271: 4791-4797.
Remmers AE, Posner R, and Neubig RR (1994) Fluorescent guanine nucleotide analogs and G protein activation. J Biol Chem 269: 13771-13778.
Rodbell M, Lin MC, and Salomon Y (1974) Evidence for interdependent action of glucagon and nucleotides on the hepatic adenylate cyclase system. J Biol Chem 249: 59-65.
Samama P, Pei G, Costa T, Cotecchia S, and Lefkowitz RJ (1994) Negative antagonists promote an inactive conformation of the 2-adrenergic receptor. Mol Pharmacol 45: 390-394.
Schneyer CR, Pineyro MA, Kirkland JL, and Gregerman RI (1984) Stimulation of human fat cell adenylate cyclase by GDP and guanosine 5'-O-(2-thiodiphosphate). J Biol Chem 259: 7038-7044.
Schramm M and Rodbell M (1975) A persistent active state of the adenylate cyclase system produced by the combined actions of isoproterenol and guanylyl imidodiphosphate in frog erythrocyte membranes. J Biol Chem 250: 2232-2237.
Seifert R, Wenzel-Seifert K, Lee TW, Gether U, Sarders-Bush E, and Kobilka BK (1998) Different effects of Gs splice variants on 2-adrenoreceptor-mediated signaling. The 2-adrenoreceptor coupled to the long splice variant of Gs has properties of a constitutively active receptor. J Biol Chem 273: 5109-5116.
Uur O and Onaran HO (1997) Allosteric equilibrium model explains steady-state coupling of beta-adrenergic receptors to adenylate cyclase in turkey erythrocyte membranes. Biochem J 323: 765-776.
Uur O, Onaran HO, and Jones TLZ (2003) Partial rescue of functional interactions of a nonpalmitoylated mutant of the G-protein G alpha s by fusion to the beta-adrenergic receptor. Biochemistry 42: 2607-2615.
Warner DR and Weinstein LS (1999) A mutation in the heterotrimeric stimulatory guanine nucleotide binding protein alpha-subunit with impaired receptor-mediated activation because of elevated GTPase activity. Proc Natl Acad Sci USA 96: 4268-4272.
Watson N, Linder ME, Druey KM, Kehrl JH, and Blumer KJ (1996) RGS family members: GTPase-activating proteins for heterotrimeric G-protein alpha-subunits. Nature (Lond) 383: 172-175.