Simultaneous stimulation of GABA and beta adrenergic receptors stabilizes isotypes of activated adenylyl cyclase heterocomplex

Background We investigated how the synthesis of cAMP, stimulated by isoproterenol acting through β-adrenoreceptors and Gs, is strongly amplified by simultaneous incubation with baclofen. Baclofen is an agonist of δ-aminobutyric acid type B receptors [GABAB], known to inhibit adenylyl cyclase via Gi. Because these agents have opposite effects on cAMP levels, the unexpected increase in cAMP synthesis when they are applied simultaneously has been intensively investigated. From previous reports, it appears that cyclase type II contributes most significantly to this phenomenon. Results We found that simultaneous application of isoproterenol and baclofen specifically influences the association/dissociation of molecules involved in the induction and termination of cyclase activity. Beta/gamma from [GABA]B receptor-coupled Gi has a higher affinity for adenylyl cyclase isoform(s) when these isoforms are co-associated with Gs. Our data also suggest that, when beta/gamma and Gαs are associated with adenylyl cyclase isoform(s), beta/gamma from [GABA]B receptor-coupled Gi retards the GTPase activity of Gαs from adrenergic receptor. These reciprocal regulations of subunits of the adenylyl cyclase complex might be responsible for the drastic increase of cAMP synthesis in response to the simultaneous signals. Conclusions Simultaneous signals arriving at a particular synapse converge on molecular detectors of coincidence and trigger specific biochemical events. We hypothesize that this phenomenon comes from the complex molecular architectures involved, including scaffolding proteins that make reciprocal interactions between associated molecules possible. The biochemistry of simultaneous signaling is addressed as a key to synaptic function.


Background
The defensive withdrawal reflex of Aplysia affords a model of classical conditioning. This involves a conditioned stimulus (touching the siphon) and an unconditioned stimulus (shocking the tail), and has been interpreted in terms of the integration of two separate simultaneous signals in the same motor neuron: Ca ++ and Gαs-GTP [1,2]. Genetic analysis in Drosophila showed that a mutant of learning and memory, rutabaga, has a defect in the gene coding for an isoform of adenylyl cyclase (type II), which is regulated by both Ca++ and Gαs [3,4]. This mutant shows low and unregulated levels of cAMP. Paradoxically, elevation of cAMP levels in dnc mutants (defect in the cAMP phosphodiesterase gene) causes an identical learning and memory deficit. The strongest phenotype comes from an allele that increases the cAMP level to as much as eight times normal [4][5][6]. Moreover, the double mutant dnc/rut has a normal cAMP level but still shows a learning and memory deficit, which suggests that these behavioral defects depend on failure to regulate cAMP levels [7]. Studies of these molecules in mutants and/or transgenic animals have led to an accumulation of interesting data in behavioral analysis of memory, learning and associative competence [4]. Moreover and surprisingly, it has been demonstrated that a high level of cAMP also follows the simultaneous activation of Gi by GABA and Gs by isoproterenol in rat neurons, whose combined (opposing) effects were expected to restore the basal cAMP level [2,[8][9][10]. Biochemical analysis has revealed the role of adenylyl cyclase type II in this phenomenon, which is highly active when complexed with both beta/gamma and Gαs. Interestingly, beta/gamma in such complexes comes from the Gi component [2,11,12]. Six beta and 13 gamma isoforms and a large family of cyclase isoforms have been cloned [13][14][15][16][17][18], which suggests that only specific combinations might be involved in coincidental signaling. Molecules that act as detectors of coincidental signals allow us to understand how the topology of the synaptic network influences its biochemistry. Our analysis of the molecular components involved in simultaneous GABA/isoproterenol signaling has been facilitated by previously reported data on the role of beta4/gamma2 and beta1/gamma2 in the activation of cyclase type II [19,20]. In the present study we investigated whether beta/gamma might influence the GTPase activity of Gαs when beta/gamma, Gαs and cyclase are associated in a complex form and conversely, whether Gαs pre-associated with a cyclase, might increase the affinity of beta/gamma for this complex.

Simultaneous signaling determines the scale of cAMP synthesis
We used isolated neuronal plasma membranes instead of membranes of transfected cells in order to preserve the molecular topology and protein localisation. We thought that the complex scaffolding proteins, which link molecules attached to plasma membranes, might be absent and/or different in cell transfection systems. The isolated membranes were treated with isoproterenol (a beta adrenergic receptor agonist) and/or baclofen (a [GABA]B receptor agonist). Simultaneous binding produced 340 +/-55 pmol/ min/mg of cAMP, compared to 105 +/-30 pmol/ min/mg and 15 +/-10 pmol/min/mg respectively for isoproterenol or baclofen alone (Fig. 1). We also observed that when the membranes were preincubated with antagonists of [GABA]b (saclofen) or beta adrenergic (propanolol) receptors before the simultaneous incubation of agonists (iso+bac), the levels of cAMP generated by isoproterenol or baclofen respectively were restored (Fig. 1).

The affinity of beta/gamma subunits for cyclase isoform(s) depends on co-association with Gαs
After solubilizing the membranes with nonionic detergent, we performed a 'pull down' experiment with Forskolin-agarose to isolate the cyclase complex. Lubrol is known to inhibit G protein GTPase activity, which allowed us to 'freeze' the G protein activated state while we isolated the cyclase complex. ADP-ribosylation by cholera toxin inhibits Gαs GTPase. ADP-ribosylation by pertussis toxin uncouples receptors and Gi [36,37]. We Determination of adenylyl cyclase activity from isolated rat brain plasma membrane: measure of cAMP synthesis by iso-prenolol plus baclofen Figure 1 Determination of adenylyl cyclase activity from isolated rat brain plasma membrane: measure of cAMP synthesis by isoprenolol plus baclofen Membranes were assayed in presence of the following drugs (50 nM): isoproterenol + baclofen (iso+bac), isoproterenol (iso), baclofen (bac). A control membrane is included. Membranes were also incubated first with saclofen (sac) or propanolol (prop) (100 nM) for 10 min, then with iso + bac as above (50 nM). cAMP levels were analyzed using a RIA procedure (NEN kit). The bars show the averages of three determinations +/-SE. Asterisks represent values statistically different from the control: ** P < 0.005 * P < 0.01 cAMP pmol/min/mg 0 100 200 300 400 500 ** * * iso+bac iso bac control sac prop + iso+bac measured the amount of radioactive ADP-ribosylated Gαs in the cyclase complex isolated with Forskolin-agarose.
Isotopic ADP ribosylation was not performed quantitatively (in conditions that eliminate "back ribosylations"). We observed that these optimal conditions saturated the signal.Gαs from inactivated receptors represents a huge amount compared to Gαs activated by isoproterenol. This interference might overwhelm the specific signal induced by isoproterenol if a probable artefactual slow exchange GDP/GTP occurs at silent Gαs. We carried out therefore the radioactive ADP ribosylation for a short time and a limited concentration of reagent in order to label preferentially the isoproterenol activated Gαs. This experimental design allowed us to compare relative levels of labeling as an index when membranes were treated with different drugs. The quantification of radioactivity was normalized with the quantity of protein present in each individual 'pull down'.
Simultaneous incubation with baclofen + isoproterenol significantly increased the amount of Gαs bound to isolated cyclase heterocomplex(s), compared to the amount bound after isoproterenol treatment alone (Fig. 2a). We analyzed also the kinetics of Gαs/cyclase association induced by drugs at low concentration on a time scale of 10 min, by measuring the amount of ADP radiolabel in the isolated complex. Beta/gamma subunits released upon activation of the [GABA]b receptors by baclofen increased the amount of Gαs bound to cyclase (300 cpm +/-80 and 470 cpm +/-50 at 1 and 5 min respectively for iso + bac, compared to 180 cpm +/-50 and 270 cpm +/-70 at 1 and 5 min for iso alone) ( fig. 2b).
We tried also to quantify the ratio of molar amounts Gαs/ beta attached to affinity chromatography when membranes were incubated with drugs. Analysis by densitometry of immunostained bands was not carried out because the relation quantity/staining was not linear. We proceeded also as follows: proteins isolated with forskolinagarose were dissociated with 1 M NaCl and derivatized with the reagent: N-hydroxysuccinimido biotin. The biotinyl proteins were then precipitated with avidin-agarose. These complex were incubated with an antibody against Gαs or Gbeta then a rabbit antibody-alkaline phosphatase in order to quantify these molecules. Precise quantity of Gαs and Gbeta (few µg of immunoprecipitated material) were submitted to the same protocol of biotinylation (data not shown). Unfortunately this method gave us variable data to some extent. We decide therefore to shift to the Bolton Hunter iodination method that has the advantage to be faster and more accurate. The labeled material attached to affinity chromatography was submitted to gel electrophoresis, then we measured the radioactivity in 35 and 45 Kda bands of dried electrophoresis gels although these iodinated bands might include unrelated proteins and don't discriminate between isoforms of Gα. We obtained the following ratio: isoproterenol + baclofen, Gα/ Gbeta = 1.8+/-0.5, isoproterenol, Gα/Gbeta= 5+/-2.2, baclofen,Gα /Gbeta = 1.1+/-0.2, isoproterenol + pertussis toxin, Gα/Gbeta= 15+/-4). Again, this method don't discriminate Gαs from Gαi. The amount of cyclase(s) was >10 fold to the amount of Ga and/or Gbeta. The quantification of Bolton Hunter labeled bands were normalized with the quantity of protein present in each individual 'pull down'.
A representative colloidal blue stained gel of attached material to affinity chromatography is shown ( fig. 2c). We see two bands at 45 and 35 kDa, corresponding to the Gαs and Gbeta subunits respectively. The 35 kDa band vanished when membranes were treated with pertussis toxin and a strong doublet appeared when membranes were incubated with iso+bac. Moreover, the specificity of the 'pull down' procedure was assessed with membranes incubated with GDP versus GTPyS+iso+bac ( fig. 2c). We see clearly the absence of the doublet with the GDP treated membranes and the non specific proteins attached to agarose. We did also competition experiment with samples pre-incubated with free Forskolin at 10 nM and 10 µM. We observed that two bands (45 and 35 kDa) disappeared while the non specific binding was unchanged (data non shown).
Western blot analysis using an anti-beta antibody showed that [GABA]b receptor activation induced the formation of the cyclase-beta/gamma complex, detected in the 'pull down' material attached to Forskolin affinity beads ( fig.  3a). Treatment of the membranes by pertussis toxin largely eliminated the formation of this complex. This suggests that beta/gamma bound to the cyclase(s) complex mostly comes from the Gi component ( fig. 3a). Kinetic studies on a time scale of 10 min show that baclofen is more efficient than isoproterenol in generating beta/gamma-cyclase(s) association ( fig. 3b). This was confirmed by quantifying the radioactivity incorporated in the 35 kDa band using Bolton Hunter reagent (1000 cpm+/-250 and 750 cpm+/-150 for bac and iso respectively) ( fig. 3c). We observed again that membranes treated with pertussis toxin and then incubated with the drugs gave a basal level, similar to control membranes ( fig. 3c).
We carried out a time course analysis under different experimental conditions to determine how beta faded out in the 'pull down' material at 4°C after the samples were treated with an excess of GDP and Mg. The complex was more abundant and more durable when baclofen and isoproterenol were used simultaneously than when baclofen alone was used ( Fig. 4a and 4b). To analyze further how activated Gαs might influence the affinity of beta/gamma Analysis of cholera toxin [ 32 γP]-ADP-ribosylated Gαs bound to cyclase   for the cyclase(s) molecule(s), we incubated the membranes with both baclofen and isoproterenol in the presence of GTPyS. The results showed that the beta subunit was present in the 'pull down' material for longer (Fig. 4b).
Baclofen/isoproterenol was also used simultaneously to stimulate cholera toxin ADP-ribosylated membranes. Again we observed that the presence of beta in the 'pull down' lasts for longer, compared to the experiment without cholera toxin treatment ( fig. 4b).
Cross-linking of proteins in the 'pull down' material gave similar results ( fig. 5a). The molecular weight of the bands correspond to cyclase/beta/Gαs, beta/Gαs and beta. We observed that GTPyS re-enforces the effect of iso+bac. These data were confirmed by quantifying the radioactivity incorporated in the 35 kDa band by Bolton Hunter reagent at 1 and 10 min after a large excess of GDP and Mg were added ( fig. 5b). We obtained for iso + bac: 620 cpm+/ -250 and 220 cpm+/-170 respectively at 1 and 10 min; iso + bac + GTPyS: 830 cpm+/-180 and 520+/-45 respectively at 1 and 10 min;iso + bac + CTx gave similar results; GTPyS alone: 380 cpm+/-190 and 330 cpm+/-50 respectively at 1 and 10 min. Membranes treated with cholera toxin gave higher counts at 1 or 10 min compared to membranes incubated with isoproterenol (50 +/-30 and 250 +/-50 at 10 min for iso and CTx respectively). Altogether, these results suggest that beta/gamma from Gi coupled to [GABA]B receptors has an affinity for adenylyl cyclase(s) that depends on co-association of Gαs with the complex.

The GTPase activity of Gαs bound to the cyclase(s) complex is decreased by co-association with beta/gamma subunits
Conversely, we checked whether co-association of beta/ gamma in the cyclase(s) complex isolated in the 'pull down' material, might influence the Gαs GTPase activity. We determined the fractional occupancy of GTP in the catalytic site of Gαs and the rate of GTP hydrolysis. We used [ 32 γP]-GTP bound to ADP-ribosylated Gαs (cholera toxin treated) as an internal control. Simultaneous treatment by GABA and isopreterenol decreased the rate of the cleavage of the γ-phosphate compared to isopreterenol or GABA alone after 10 min at 0°C (0.3+/-0,1 compared to 0.05+/ -0,05 for isopreterenol and 0.05 +/-0.1 for GABA) (Fig. 6a). Moreover, the decrease of [ 32 γP]-GTP bound to the cyclase complex in a single turnover experiment at 0°C was significantly less marked when the membranes were incubated with both agents ( fig. 6b). At 5 min, iso + bac gave 0.25 +/-0.15 compared to 0.05 for iso or bac alone. To further characterize the effect of simultaneous iso + bac effect on GTPase activity, we examined the amount of [ 32 γP]-GTP in the 'pull down' material, 5 min at 0°C after the samples were complemented with Mg ( fig. 7a). This amount was compared with the corresponding amount at t = 0 without Mg. Simultaneous incubation resulted in a level of radioactivity (1300 cpm +/-300) considerably greater than the sum of the counts obtained with iso and bac alone (150 cpm+/-105 and 200 cpm+/-120 respectively). As expected, the remaining amount was greater when the membranes were treated with cholera toxin (3500 cpm+/-550 for iso+bac+CTx and 2600 cpm+/-250 for CTx alone).

Discussion
Adenylyl cyclases contain two conserved homologous cytoplasmic domains (C1 and C2) that form the catalytic core of the enzyme [21]. Forskolin appears to induce the assembly of these two domains by interacting with the catalytic cleft [21][22][23]. The affinity between C1 and C2 is also facilitated by Gαs binding. These data have been confirmed by the finding that Forskolin and Gαs stimulate synergistically the cyclase activity [22]. In the presence of Gαs the affinity of Forskolin for the dimer C1/C2 is high (K d = 0.1 µM), which suggests a stronger affinity for the complete cyclase molecule [23]. The site of interaction of the cyclase (type I or II) for beta/gamma has been located in the C1 b region. This site is independent of the Gαs and Forskolin interaction domains [24]. These findings argue in favor of isolating the cyclase(s) and its associated proteins using Forskolin-agarose affinity chromatography although this procedure enriches indistinctly the different isoforms of the cyclase family. The authors who described this method [25] reported that the complex could be dissociated with high salt concentrations in order to purify the cyclase(s). Our 'pull down' material is heterogeneous because this isolation procedure does not discriminate between the different isotypes of cyclase. Furthermore, Gαs recognizes all the isoforms of adenylate cyclase whereas the association Gbeta/cyclase is limited to two isotypes (type I and II) which share similar features. Consequently, this mismatch makes the comparison of Gαs/ Gbeta precipitated in the cyclase 'pull down' difficult. Inconstant data were observed in experiments of kinetics where both molecules were quantified on the same gel.
On the other hand, Gαs-GTP in the Gαs/cyclase(s) complex might have a Kcat for the hydrolysis of bound GTP different from one isoform of cyclase to another. The heterogeneity of cyclases in the 'pull down' makes the analysis of co-associated Gαs uncertain, specially if we aim to parallel the dynamics of association/dissociation of Gbeta and Gαs. Furthermore, our analysis might be hampered by variable elution/retention of components from the affinity column and the yield of their isolation from crude membranes. These limitations are difficult to assess.
On the other hand, the drastic increase of cAMP synthesis by simultaneous Gi/Gs activation finely analyzed in rat hippocampus [26] is known to be attributable to the type II isoform [2]. Type I is also regulated by beta/gamma and Gαs but here beta/gamma is inhibitory. High concentrations of beta/gamma do not eliminate the cyclase type I activity induced by Gαs, but stabilize it at intermediate levels, which suggests that the conformation of cyclase type I is rearranged by association with beta/gamma [11]. The two isoforms show similar affinities for beta/gamma (half-maximal effect about 5 nM) [11]. We postulate therefore that the molecular events described in this report mostly highlight the relevance of cyclase type II, despite isotype I contamination. We might also speculate that the intra-complex regulatory mechanisms described here have the same characteristics for both isotypes, although they induce opposite effects on cyclase activity.
Furthermore, we carried out our experiments with intact membranes, in order to preserve the architecture of the multiple associated proteins that might be relevant to the biochemistry involved in coincidental phenomenon. For example, regulators of G proteins (RGS) interact mostly with the Gα subunit of Gi/o/q by accelerating their intrinsic GTPase activity, which alters the amplitude of the effect of a stimulated receptor [27]. Each member of this large family of genes (16 isotypes have been described) displays differential selectivity for these G proteins [27]. Interestingly, RGS 14 and 12 are unique in that they inhibit the guanine nucleotide exchange of Gi in addition to their GAP (GTPase activating protein) activity. A recent report shows that PKA phosphorylation of RGS 14 increases the inhibition of nucleotide exchange 3-fold, thereby increasing the binding Gi/GDP and consequently limiting the interactions of Gi with effectors [28]. RGS 14 is expressed in brain [29], which might be of interest in relation to our work. By increasing PKA activity, high cAMP concentrations should inhibit more Gi molecules enhancing the effect of the simultaneous stimulation (Gαs and Gbeta) of cyclase type II. After a while beta/gamma would then be captured by Gi/GDP, ending the simultaneous stimulation of cyclase type II by a retro-control loop.
In this report, we have investigated whether beta/gamma bound to cyclase(s) might influence the rate of GTP hydrolysis of Gαs, or conversely whether Gαs bound to cyclase(s) might modify its affinity for beta/gamma. We found that the GTPase activity of Gαs coupled to the cyclase(s) complex decreases when beta adrenergic and GABA receptors are simultaneously stimulated. We also demonstrated that the affinity of beta/gamma (from Gi) for the cyclase(s) is increased when the latter is associated with GTP-Gαs. Our data strongly suggest that simultaneous application of stimuli, which individually exert opposing effects on the levels of second messengers, triggers specific kinetics of protein association and enzyme catalysis. This highlights the paradoxically high level of cAMP obtained by simultaneous GABA/beta adrenergic receptor activation. Adenylate cyclase was reported to function as a GTPase activating protein for Gαs [30]. Our data support the idea that beta/gamma from [GABA]b receptor-coupled Gi retards the GTPase activity of adrenergic receptor-coupled Gαs when these molecules are complexed with adenylyl cyclase isoforms.

Conclusions
Cyclases are not the only molecules to be regulated 'à la carte' by simultaneous signaling. Various molecules in different tissues correspond to the definition of 'detectors of simultaneity': for example, muscle phosphorylase kinase, a tetramer constituted of a catalytic subunit, two cAMP dependent regulatory subunits and calmodulin. Two stimuli (Ca++ and cAMP) synergistically stimulate phosphorylase kinase activity [31]. Furthermore, some molecules fine-tune their activities, depending on the local context, in 'reading' the temporal order of their regulatory signals. For example, phosphatase I is associated with a potent protein inhibitor: this inhibitor has a site that is phosphorylated by caseine kinase I. This phosphorylation protects an adjacent phosphorylated site from calcineurin and consequently these two phospho sites in the inhibitor controls the activity of the phosphatase [32]. Another example is the endothelial nitric oxide synthase (eNOS). This enzyme has two major phosphorylation sites, which have opposite functions. One, phosphorylated by PKA, stimulates the enzyme activity, the other, phosphorylated by PKC, inhibits it. Interestingly, PKA and/or PKC phosphorylation of one site induces dephosphorylation of the other. Two phosphatases associated with eNOS have one or the other phosphorylated site as specific substrate. When one phosphatase is active the other is inactive [33]. All these coordinated controls display different modes used by key molecules acting as sensors of multiple converging second messengers. Specific molecular events induced by simultaneous signals and/or sequentially ordered signals imply interactions of components in a large complex. This seems the essential paradigm of synaptic function.

Reagents
Western blotting was carried out with a polyclonal anti-

Preparation of plasma membrane of isolated nerve terminals (synaptosomes)
Synaptosomes were prepared by homogenization of rat brain (Sprague Dawley, 150 g) in 310 mM sucrose, 10 mM Hepes (pH 7.4), 1 mM EDTA. The homogenate was centrifuged for 5 min at 900 g max at 4°C. The supernatants were combined and centrifuged at 11 000 g max for 15 min.
Pellets were re-suspended and this synaptosome-enriched fraction was layered on to each of two discontinuous gradients (12% (w/v), 9% and 6% Ficoll) and centrifuged for 30 min at 75,000 g max according to the protocol described elsewhere [34]. The intermediate band (within the 9% layer) was resuspended in HEPES 5 mM (pH 7.4), EDTA 1 mM for 1 h in ice and the plasma membrane fraction released by osmotic shock was re-centrifuged for 5 min at 10,000 g max . The isolated plasma membranes were incubated at 30°C for 15 min with anti proteases (PMSF: 1 µg/ ml, leupeptin: 1 µg/ml and aprotinin: 1 µg/ml) in order to wash out endogenous neuromediators, then aliquoted and frozen at -70°C.

Isolation of the cyclase complex
Membranes (50 µg protein) were assayed as described in the figure legends, then solubilized in Triton X-100 1%, 150 mM NaCl, 10 mM Tris (pH 7.4), 1 mM EDTA. Samples were briefly centrifuged and the supernatant was diluted five times (final concentration of Triton X-100 = 0.2%) in 150 mM NaCl, 10 mM Tris (pH 7.4), 1 mM EDTA plus 0.5 mM Mg. Lubrol at these concentrations potently inhibits GTPase activity by preventing the release of hydrolysis products, as reported elsewhere [35]. We observed that Triton X-100 or Tween 20 (Sigma) gave similar effects. Forskolin-agarose (25 µl) was added to the extract and incubated for 10 min at 4°C. Samples were constantly re-suspended by tube rotation. Samples were then centrifuged briefly, the supernatants discarded and the pellets washed once with the same buffer. Molecules attached to affinity beads were separated by SDS-PAGE alongside pre-stained molecular weight markers (Sigma). The electrophoresis gels were stained with colloidal blue (Sigma).

ADP-ribosylation of Gs and Gi by Pertussis toxin and cholera toxin
ADP-ribosylation using pertussis toxin (Sigma) was carried out as indicated in the figure legends, in the presence of 1 mM EDTA, 2 mM DTT, 5 mM thymidine, 10 mM HEPES (pH 7.4), 25 µg/ml pertussis toxin and 10 µM NAD + , according to the protocol described elsewhere [36]. ADPribosylation was also carried out with 10 µg/ml cholera toxin as described elsewhere [37] with a slight modification: the endogenous cofactor for cholera toxin (ARF) was activated with only GTP, as indicated in the figures. For radioactive labeling by cholera toxin (10 µg/ml toxin), 10 µM [ 32 P] NAD + (50 000 cpm/pmol) were incubated with membranes in a buffer with 10 mM thymidine and 200 mM Nacl as indicated in figure legends. Radioactivity attached to the isolated cyclase complex was counted and normalized with the amount of proteins precipitated by the affinity chromatography as follows: an aliquot of each sample was incubated with 5 µl of trypsin-agarose (Sigma) for 30 min in 10 mM phosphate buffer pH = 8 and the released peptides were measured at 220 nm in 150 µl of the same buffer.
Cold ADP-ribosylation by cholera toxin was carried out with membranes 20 min at 30°C with 30 µg/ml toxin, 50 µM NAD + and the concentration of GTP indicated in figure legends. Cold ADP-ribosylation by pertussis toxin was carried out with membranes with 50 µg/ml toxin and 50 µM NAD + for 20 min at 30°C.

Analytical methods
The material bound to Forskolin-agarose was submitted to SDS acrylamide gel electrophoresis (12%) using prestained standards (Sigma). To monitor the accumulation of beta/gamma, the gels were blotted on to nitrocellulose for Western analysis using antibodies at the dilution indicated by Santa Cruz Biotech. The anti-beta used (T-20) is broadly reactive with beta 1, 2, 3 and 4. The blots were developed by alkaline phosphatase conjugated anti rabbit IgG (Sigma). For cross-linking experiment, the precipitate was incubated with glutaraldehyde 1% in 0.2 M Na carbonate pH = 8 for few minutes then the reaction was stopped with 10 µl of lysine solution (1 mM). Experimental conditions for the kinetic analysis are given in the figure legends. To quantify the beta subunit, the material bound to Forskolin-agarose was labeled with Bolton Hunter reagent (NEN) (20 000 cpm/samples) in 150 µl phosphate buffer 10 mM (pH 7.4) for few minutes, then molecules were separated by SDS polyacrylamide gel electrophoresis. Gels were dried and the bands around 35 kDa were excised and counted. Radiolabeled ADP-ribosylated Gαs was counted according to the same method. Counts were normalized with the amount of protein in each individual 'pull down'.
To quantify the molar amount of precipitated Gα and beta, the material attached to affinity chromatography was labeled with Bolton Hunter reagent as above. This material was first submitted to high salt treatment in 50 µl of 500 mM Nacl and 10 mM phosphate pH = 8 for 1 hour to dissociate proteins, then samples were diluted 3 times before the labeling. This material was precipitated by trichloracetic acid (5%), the pellet was neutralized with Tris buffer