Stimulation of cAMP Synthesis by Gi-coupled Receptors upon Ablation of Distinct Gαi Protein Expression

The three Gαi subunits were independently depleted from rat pituitary GH4C1 cells by stable transfection of each Gαi antisense rat cDNA construct. Depletion of any Gαi subunit eliminated receptor-induced inhibition of basal cAMP production, indicating that all Gαi subunits are required for this response. By contrast, receptor-mediated inhibition of vasoactive intestinal peptide (VIP)-stimulated cAMP production was blocked by selective depletions for responses induced by the transfected serotonin 1A (5-HT1A) (Gαi2 or Gαi3) or endogenous muscarinic-M4 (Gαi1 or Gαi2) receptors. Strikingly, receptor activation in Gαi1-depleted clones (for the 5-HT1A receptor) or Gαi3-depleted clones (for the muscarinic receptor) induced a pertussis toxin-sensitive increase in basal cAMP production, whereas the inhibitory action on VIP-stimulated cAMP synthesis remained. Finally, in Gαi2-depleted clones, activation of 5-HT1A receptors increased VIP-stimulated cAMP synthesis. Thus, 5-HT1A and muscarinic M4 receptor may couple dominantly to Gαi1 and Gαi3, respectively, to inhibit cAMP production. Upon removal of these Gαi subunits to reduce inhibitory coupling, stimulatory receptor coupling is revealed that may involve Gβγ-induced activation of adenylyl cyclase II, a Gi-stimulated cyclase that is predominantly expressed in GH4C1 cells. Thus Gi-coupled receptor activation involves integration of both inhibitory and stimulatory outputs that can be modulated by specific changes in αi subunit expression level.

Heterotrimeric G proteins transduce signals generated by hormone receptors with seven transmembrane domain ␣-helices to various effectors such as AC, 1 phospholipase C, and ion channels (1)(2)(3). These G proteins are composed of a G␣ subunit and a tight complex of G␤ and G␥ subunits. The binding of agonist to receptors allows them to interact with G proteins, which subsequently accelerates the rate of dissociation of GDP and the binding of GTP to the G␣ subunits. Both GTP-liganded G␣ subunits and G␤␥ dimers then regulate downstream effector activities. The intrinsic GTPase activity of G␣ subunits hydrolyzes GTP to GDP to form G␣-GDP, which then associates with G␤␥ dimers to inactivate the complex.
The cyclic AMP-forming enzyme, adenylyl cyclase, is one of the ubiquitous effectors that is regulated by G protein-coupled receptors. The pathways of G s -coupled receptor-induced stimulation of AC have been well characterized (1,4). However, the inhibitory regulation of AC by G i -coupled receptors is far less clear, and the underlying mechanisms seem to be rather complex. For example, in several cell systems, e.g. mouse Ltk Ϫ , Rat-1, and NIH-3T3 fibroblast cells, inhibition of cAMP synthesis by G i -coupled receptors was only observed when AC was activated by forskolin or G s -coupled receptors (5)(6)(7). On the other hand, in neuronal and endocrine cells, G i -coupled receptors inhibited both unstimulated and stimulated AC activity (2,5,8). In addition, some G i -coupled receptors have even been reported to stimulate AC, e.g. ␣ 2 -adrenergic receptor in PC-12 pheocytochroma cells (7,9) or 5-HT1A receptors in hippocampus (10,11).
The specificity of distinct G i proteins in receptor-AC coupling remains incompletely understood. For example, using anti-G␣ i subunit antibodies to block receptor coupling, it has been reported that inhibition of AC by ␣ 2 -adrenergic receptors in platelet membranes is mediated by G␣ i2 (12) and that 5-HT1A receptor-induced inhibition of cAMP synthesis in HeLa cell membranes is preferentially mediated by G␣ i3 (13). This approach, however, is limited to cell-free preparations and depends on the specificity of the antibodies used. To assess the roles in receptor coupling of particular G proteins in whole cells, transfection of PTX-insensitive G␣ i mutant proteins has been used to rescue receptor-mediated signaling following PTX pretreatment. For example, the dopamine-D2S receptor appears to couple to G␣ i2 and G␣ i3 to mediate inhibition of forskolin and G s -stimulated AC, respectively (14,15). This approach depends on the specificity and functionality of the G␣ i mutants. We have used expression of antisense constructs to selectively deplete particular G proteins and assessed their contribution to receptor coupling (16 -18).
The aim of the present study was to evaluate the contribution of the three known G␣ i subunits (19) in relaying inhibitory signals from 5-HT1A and muscarinic M4 receptors to AC in intact GH4C1 rat pituitary cells. We analyzed G i protein sub-type specificity in receptor-effector coupling by stably introducing distinct full-length rat G␣ i antisense constructs into GH4ZD10 cells (GH4C1 cells transfected with the rat 5-HT1A receptor (5)). This approach produced a specific block of the gene expression of these proteins (18). Characterization of these different G␣ i -deficient antisense clones indicates that G␣ i proteins specifically link receptors to inhibition of cAMP synthesis but not to closure of calcium channels and that the combined presense of all three G␣ i subunits is essential for receptor-mediated inhibition of unstimulated but not of VIP receptor-stimulated cAMP synthesis. Strikingly, upon depletion of distinct G␣ i subunits, the G i -coupled 5-HT1A and muscarinic M4 receptors switched to stimulate AC activity.

EXPERIMENTAL PROCEDURES
Materials-5-HT, carbachol, VIP, PTX, and isobutylmethylxanthine were purchased from Sigma. Hygromycin B was from Calbiochem (La Jolla, CA). BayK-8644 was from Research Biochemicals Inc. (Natick, MA). Fura 2-AM was from Molecular Probes (Eugene, OR). PTX was from List Biological Laboratories (Campbell, CA). Rat G protein G␣ i1 , G␣ i2 , and G␣ i3 subunits and AC type II cDNAs were gifts of Dr. R. Reed. G protein G␣ i subunit antibodies were kindly donated by Dr. D.

Manning.
Cell Culture-All cells were grown as monolayer in Ham's F-10 medium with 8% fetal bovine serum at 37°C in a humidified atmosphere with 5% CO 2 . Media were changed 12-24 h prior to experimentation.
Preparations of the Antisense Clones-The 2.0-kb EcoRI-EcoRI G␣ i1 cDNA fragment, the 1.8-kb EcoRI-EcoRI G␣ i2 cDNA fragment, or the 3.1-kb EcoRI-EcoRI G␣ i3 cDNA fragment (13) containing the full coding sequences and 0.5 to 0.8 kb of 5Ј and 3Ј noncoding regions were excised using EcoRI. The cDNA fragments were ligated into pcDNA (Invitrogen) in the reverse orientation with respect to the cytomegalovirus promoter, resulting in G␣ i1 , G␣ i2 , or G␣ i3 antisense expression vectors. The constructs were confirmed by restriction enzyme analysis and by DNA sequencing. A modified transfection procedure was used; 300 -500 g of each G␣ i antisense construct was cotransfected separately with 30 g of pY-3 hygromycin B resistance plasmid into G418-resistant GH4ZD10 cells by standard calcium phosphate co-precipitation protocol (20). The selection was initiated after 24 h by adding 150 g/ml hygromycin B into the culture medium to select the clones with expression of the antisense RNAs and to allow the clones to adapt the cytotoxicity of hygromycin B. After 2 weeks, the concentration of hygromycin B was raised to 400 g/ml to select the clones that have the highest resistance to hygromycin B and that most likely express the highest levels of the antisense RNAs. Isolated clones were propagated, and total RNA was prepared from them and screened by reverse transcription-PCR analysis using a pair of oligonucleotides specific for each G␣ i cDNA (G␣ i1 , 5Ј-ACCAGACGAGTACTTATA-3Ј and 5Ј-TAGTCTGTGCAACGTTTA-3Ј; G␣ i2 , 5Ј-CACTACCTGTGAGGAAGA-3Ј and 5Ј-ACTCCTCCAGACA-TAGG-3Ј; G␣ i3 , 5Ј-TGATATCAAATCTAGGGC-3Ј and 5Ј-TAGAACGC-ATTCCCAGAT-3Ј) to identify positive clones. Total RNA from GH4ZD10 or the different G␣ i antisense clones was reverse-transcribed after annealing the primer (sense or antisense) to RNA by heating the two together at 90°C and then chilling on ice. One-tenth of the reversetranscription mixture was used for PCR amplification. 50 l of PCR mixture contained 50 mM KCl, 25 mM Tris-HCl, pH 8.3, 2.5 mM MgCl 2 , 1 mg/ml bovine serum albumin, 0.2 mM each dCTP, dATP, dGTP, and dTTP, 1 unit Hot-Tub DNA polymerase (Amersham Pharmacia Biotech) and 3 M primer. The PCR cycle included denaturation for 1 min at 94°C, annealing for 1 min at 48 to 52°C (varied with different pairs of primers), and extension for 20 s at 72°C. For Western blotting, membranes (50 g protein/lane) prepared from the positive antisense clones were solubilized, electrophoresed, transferred to nitrocullose sheets, and probed with specific antibodies as described (16). Densitometric scanning of the blots was done on the Scanmaster 3 densitometer (Howtek, Hudson, NH). The data were digitized, quantitated using the Masterscan analysis program (Scanalytric, Billerica, MA), and reconstructed as Masterscan images presented in the figures. This analysis allows enhanced resolution of weak signals. Western blots were quantitated from the Masterscan data. Optical densities of equal area samples from each band of interest were subtracted from background for that lane and normalized (ϫ100) to the control samples to obtain the percentage of control.
Measurement of [Ca 2ϩ ] i -Measurement of [Ca 2ϩ ] i was performed as described previously (5). In brief, cells were harvested by incubation in calcium-free HBSS containing 5 mM EDTA. The cells were washed once with HBSS (118 mM NaCl, 4.6 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 10 mM D-glucose and 20 mM HEPES, pH 7.2) and then incubated for 30 min at 37°C in the presence of 2 M Fura 2-AM. The cells were then diluted to 10 ml, centrifuged, washed twice with HBSS, resuspended in 2 ml HBSS, and finally placed in a fluorescence cuvette. Change in fluorescence ratio was recorded on a Perkin-Elmer (Buckinghamshire, UK) LS-50 spectrofluorometer and analyzed by computer, based on a K D of 227 nM for the Fura 2-calcium complex. Calibration of R max was performed by addition of 0.1% Triton X-100 and 20 mM Tris base, and calibration of R min was performed by addition of 10 mM EGTA. All experimental compounds were added directly to the cuvette from 200fold concentrated test solutions.
cAMP Assay-Measurement of cAMP was performed as described previously (5). In brief, the cells were plated in six-well 35-mm dishes. After removal of the medium, the cells were preincubated in 2 ml/well of HBSS for 5-10 min at 37°C. The buffer was replaced by 1 ml of HBSS containing 100 M isobutylmethylxanthine, a cAMP phosphodiesterase inhibitor, and the incubation was continued for another 5 min. Then the various test compounds were added to the wells, and the cells were incubated at room temperature for 20 min. The buffer was collected for cAMP assay using a specific radioimmunoassay (ICN) as described before (5).

Independent Depletion of Distinct G␣ i Subunits from
GH4ZD10 Cells-Reverse transcription-PCR and Western blot analysis indicated the presence of all three known G␣ i subunits in GH4C1 cells (Fig. 1), although G␣ i1 may be the least abundant based on the weakness of the signal. The three G␣ i subunits are highly homologous in their coding regions but exhibit a clear variability in their 5Ј-and 3Ј-noncoding sequences (19). To specifically block the protein expression of distinct G␣ i subunits, the antisense constructs were chosen to include the full coding sequences and, in addition, 500 -800 base pairs of the 5Ј-and 3Ј-untranslated sequences (18). Selection for stably transfected clones exhibiting the highest levels of antisense RNA expression was accomplished by raising temporarily the concentration of hygromycin B to 400 g/ml. The expression of antisense RNA was detected by reverse transcription-PCR analysis, using a pair of oligonucleotides specific for each G␣ i subunit (not shown). The extent of depletion of distinct G␣ i subunits was verified by Western blot analysis, using antibodies specific for each G␣ i subunit (22,23). It was found that G␣ i1 and G␣ i2 subunits were virtually eliminated in clones Gi1ZD-3 and Gi1ZD-5 (Fig. 1A) and Gi2ZD-4 and Gi2ZD-5 (Fig. 1B), respectively, whereas G␣ i3 subunits were largely depleted in clones Gi3ZD-3 and Gi3ZD-4 (Fig. 1C). To examine the extent of cross-hybridization of different G␣ i antisense RNAs with sense RNAs of other G␣ i subunits, the membranes prepared from different G␣ i antisense clones were also probed with antibodies specific for other G␣ i and G␣ o subunits. In G␣ i3 -depleted cells, the amount of G␣ i1 was similar to the control cells (Fig. 1D), whereas in G␣ i1 -and G␣ i3 -depleted clones the amount of G␣ i2 or G␣ o was unchanged (data not shown). Thus, stable transfection of G␣ i antisense constructs can specifically eliminate their cognate G␣ i proteins without major alterations in the amount of other G␣ i subunits. (5), it was found that agonist activation of the transfected 5-HT1A receptor inhibits both cAMP accumulation and calcium entry in GH4ZD10 cells. Similarly, activation of endogenously expressed muscarinic M4 receptors (24) in GH4ZD10 cells also inhibited both cAMP accumulation and calcium entry (Fig. 2, upper panel, and Table I). These receptor actions were sensitive to PTX treatment (Fig. 2, lower panel), suggesting the involvement of G i /G o proteins. Muscarinic M4 receptors exhibited greater efficacies than 5-HT1A receptors ( Fig. 2 and Table I), although the number of endogenously expressed muscarinic M4 receptors in GH4ZD10 cells is 2-3fold lower than that of the transfected 5-HT1A receptors (5). Similarly distinct efficacies were observed for agonist-stimulated GTPase activity by these two receptors in membrane preparations (data not shown), suggesting that in this cellular system, muscarinic M4 receptors couple more effectively to G proteins than do 5-HT1A receptors.

5-HT1A and Muscarinic M4 Receptor-mediated Inhibition of cAMP Synthesis and Calcium Entry in GH4ZD10 Cells-In previous studies
Influence  GH4ZD10, 5-HT, and carbachol inhibited both basal and VIPstimulated cAMP formation (Fig. 2). In contrast to calcium entry, receptor-mediated inhibition of cAMP synthesis in the different G␣ i -depleted clones was clearly and selectively altered. Specifically, in G␣ i1 -depleted clones, Gi1ZD-3, both 5-HT and carbachol failed to inhibit basal cAMP production (Fig. 3A). Surprisingly, 5-HT induced a 5-fold increase in basal cAMP accumulation in these cells, an action almost as efficacious as that of VIP. The 5-HT-induced stimulation of basal cAMP ac-cumulation was blocked by PTX pretreatment, suggesting mediation by G i /G o proteins. This novel stimulatory action of 5-HT on basal cAMP formation was also observed in another G␣ i1 depleted clone, Gi1ZD-5 (Table II). In the same G␣ i1 -depleted cells, however, 5-HT inhibited VIP-stimulated cAMP accumulation by some 45%, similar to the results obtained in GH4ZD10 cells. These data indicate that the presence of G␣ i1 subunits is not essential for the 5-HT1A receptor-mediated inhibition of G s -stimulated cAMP accumulation. In contrast to 5-HT1A receptor action, in both Gi1ZD-3 and Gi1ZD-5 clones, carbachol-induced inhibition of VIP-stimulated cAMP accumulation was blocked (Fig. 3A). As observed in G␣ i1 -depleted clones, the ability of 5-HT1A and muscarinic M4 receptors to inhibit basal cAMP synthesis was also ablated in G␣ i2 -depleted clones, Gi2ZD-4 (Fig. 3B). In addition, both receptors failed to inhibit VIP-stimulated cAMP synthesis. Activation of 5-HT1A receptors potentiated slightly VIP-stimulated cAMP accumulation by 35%, an action sensitive to PTX treatment.
In G␣ i3 -depleted clone Gi3ZD-3, activation of 5-HT1A or muscarinic M4 receptors did not result in inhibition of basal cAMP accumulation (Fig. 3C). Curiously, carbachol induced a 3-fold increase in cAMP production in these antisense clones. This stimulation was prevented by PTX pretreatment of the cells. In the same G␣ i3 -depleted clones, however, the inhibitory action of muscarinic M4 receptors on VIP-stimulated cAMP synthesis remained largely unaltered. In Gi3ZD-4 cells, another G␣ i3 depleted clone, carbachol-induced inhibition of VIPstimulated cAMP accumulation was also essentially unaltered (Table II). By contrast, 5-HT1A receptor-induced inhibition of VIP-stimulated cAMP accumulation was completely blocked in G␣ i3 -depleted cells.
Subtypes of AC Expressed in GH4C1 Cells-The possible mechanism of G i -mediated stimulation of cAMP production upon depletion of specific G␣ i subunits was addressed. Certain AC subtypes (types II, IV, and VII) have been demonstrated biochemically to be conditionally stimulated by G␤␥ subunits.

TABLE II
Receptor-mediated inhibition of cAMP accumulation in G␣ i -depleted cells The level of cAMP (pmol/dish) was determined as described in the legend to Fig. 2 in parental GH4ZD10 cells (data from Fig. 2  In addition AC II is known to mediate G i -induced stimulation of cAMP levels when co-transfected with specific G i -coupled receptors, such as the ␣ 2 -adrenergic receptor (26). We examined the RNA expression of the most extensively characterized subtypes, AC types I-VI (4), using reverse transcription-PCR analysis at different concentrations of cDNA (Table III). In rat brain each subtype was present, although type I was weakly expressed (25). The rank order of expression of adenylyl cyclases in GH4C1 cells was II ϭ VI Ͼ III Ͼ Ͼ (I, IV, and V). Of particular interest was the predominant expression in GH4C1 cells of AC type II, which was detected as a major species of 6.7 kb by Northern blot analysis of poly(A) ϩ RNA from GH4C1 cells (Fig. 4). By contrast, AC type II RNA was undetectable in various fibroblast cell lines (Ltk Ϫ and Balb/c-3T3), adrenocortical Y1, DDT1-MF2 smooth muscle, or PC-12 pheocytochroma cells (data not shown and Refs. 15 and 25). Thus, AC II is abundantly expressed in pituitary GH4C1 cells and brain tissue, permitting the possibility for receptor coupling to G i -dependent stimulation of cAMP accumulation in these tissues. DISCUSSION Using stable transfection of distinct G␣ i full-length antisense constructs, we were able to specifically deplete the protein expression of individual G␣ i subunits from GH4C1 pituitary cells. It was found that knocking out of any of the three G␣ i subunits specifically altered 5-HT1A and muscarinic M4 receptor-mediated inhibition of AC but not the receptor-induced inhibition of calcium entry. The latter was achieved by specific ablation of ␣ o subunits from GH4C1 cells (17). These data confirm that G␣ i subunits mediate inhibition of AC but not closure of calcium channels. Similar specificities have been reported for the coupling of different receptors to voltage-dependent calcium channels by ␣ oA or ␣ oB proteins but not G␣ i proteins in GH3 cells (27)(28)(29).
One of the important findings of the present study is that although the three different G␣ i subunits all participate in receptor-mediated inhibition of AC, each G␣ i subunit apparently plays a distinct role in this signal transduction process. First, the contemporaneous expression of all three G␣ i subunits appeared to be essential for both 5-HT1A and muscarinic M4 receptor-mediated inhibition of unstimulated cAMP synthesis (Table III). Depletion of any of the three G␣ i subunits led to the inability of both receptors to inhibit basal cAMP synthesis in GH4C1 cells. This is consistent with our previous observations in GH4C1 cells that inhibition of basal cAMP level by dopamine-D2S, dopamine-D2L, and somatostatin receptors was blocked upon depletion of G␣ i2 (17). In cell types in which two or fewer types of G␣ i subunits are expressed, such as fibroblast Rat-1, Chinese hamster ovary, or JEG-3 choriocarcinoma cells, G i -coupled receptors inhibit stimulated cAMP production but do not inhibit basal cAMP accumulation (5-7, 30, 31). This supports the hypothesis that G i -coupled receptors require all three G␣ i subunits for inhibition of basal cAMP accumulation. Second, different receptors link to different G␣ i subunits to inhibit G s -stimulated cAMP synthesis. For example, 5-HT1A receptors couple to G␣ i2 and G␣ i3 subunits to suppress cAMP accumulation stimulated by the G s -coupled VIP receptor, because depletion of either subunit blocks this response. By contrast, muscarinic M4 receptors apparently couple to G␣ i1 and G␣ i2 subunits for inhibition of VIP-stimulated cAMP formation. Interestingly, each receptor interacts with and activates more than one type of G i proteins to inhibit G s -stimulated cAMP synthesis. Using immunoprecipitation with specific G protein antibodies and cholera toxin-catalyzed labeling of G i proteins, it has been shown that G i -coupled receptors may simultaneously activate more than one G protein and that different receptors apparently exhibit different preferences to different G proteins (6,7,31,32).
A third major conclusion to be drawn from the data presented is that a receptor may dominantly link to different G␣ i subunits depending on whether G s is engaged in stimulation of AC. For example, depletion of G␣ i1 and G␣ i3 subunits led 5-HT1A and muscarinic M4 receptors, respectively, to stimulate basal cAMP synthesis. These depletions had no effect on inhibition of VIP-stimulated AC activity by the same receptors. FIG. 4. RNA expression of adenylyl cyclase type II in GH4C1 cells. Poly(A) ϩ RNA (5 g) was subjected to Northern blot analysis under high stringency conditions using rat AC type II cDNA as a probe (21). The migration of RNA molecular mass markers (in kb) is indicated. A major RNA species of 6.7 kb in size was detected with a minor species at 4.5 kb.

TABLE III Expression of different adenylyl cyclase subtypes in GH4C1 cells
The PCR was performed for AC I-VI with 0.1, 0.5, and 1.0 g of cDNA synthesized from GH4C1 poly(A) ϩ RNA or rat brain RNA as indicated. Data were obtained from at least two independent experiments for each condition. Specific primers for each subtype of AC amplified only a single product of the predicted size. Plus (ϩ) and minus (Ϫ) signs indicate the presence and the absence of the specific product on ethidium bromide stained gels, respectively, and Ϯ represents a weakly detectable product.  These results suggest that different G proteins regulate inhibition of basal (primarily G i1 ) and G s -stimulated (primarily G i2 ) cAMP synthesis by the 5-HT1A receptor in GH4C1 cells. Results from transfections of PTX-insensitive G proteins in Ltk Ϫ cells indicate that the D2S receptor couples to G i2 to inhibit forskolin-induced cAMP accumulation but uses G i3 to inhibit G s -stimulated action (15). Taken together, these results are consistent with a state-dependent inhibition of adenylyl cyclase by specific G␣ i subunits.
A surprising finding of the present study was the efficient stimulation of basal cAMP synthesis in G␣ i1 -and G␣ i3 -depleted antisense clones by agonist-activated 5-HT1A and muscarinic M4 receptors, respectively. The weaker stimulation by muscarinic compared with 5-HT1A receptors (3-fold versus 5-fold basal cAMP) may reflect the less complete depletion of G␣ i3 compared with G␣ i1 or weaker efficacy of the muscarinic receptor for this signaling pathway. This stimulation of cAMP synthesis by 5-HT or carbachol was sensitive to PTX treatment, suggesting the involvement of the remaining G i /G o proteins to induce direct stimulation of AC. Functional analysis of AC enzymes I-VI indicates that all of them are stimulated by G␣ s , and some of them (types I, V, and VI) have been shown to be inhibited by G␣ i proteins (4,(33)(34)(35)(36). Interestingly, specific G␤␥ dimer combinations potentiate G s stimulation of AC types II, IV, and VII while either inhibiting (type I) or having no effect on the other isoenzymes (36 -38). In cotransfection experiments, G i -coupled receptors have been shown to potentiate G sor protein kinase C-stimulated cAMP synthesis by AC type II, apparently by release of G␤␥ dimers (26,39). Our observations indicate that in the presence of multiple G␣ i subunits and AC subtypes, receptor-mediated inhibition is the dominant pathway. However, upon depletion of particular G␣ i subunits, the remaining PTX-sensitive G proteins stimulate cAMP levels. Because GH4C1 cells appear to express AC type II, the enhancement of cAMP by these receptors could be mediated by conditional activation of AC type II. We have identified G i2 as the major G protein that mediates coupling of the 5-HT1A receptor to AC type II upon cotransfection in HEK-293 cells (40). These results suggest that upon depletion of G i1 in GH4C1 cells, G␤␥ subunits associated with G i2 mediate positive coupling to AC type II resulting in enhanced cAMP production. Distinct G protein specificities of 5-HT1A or muscarinic M4 receptors to enhance cAMP levels may reflect the association of specific G protein complexes with each receptor. For example coupling of somatostatin and muscarinic M4 receptors in G omediated inhibition of calcium channels, a G␤␥-mediated response, is mediated by distinct combinations of ␣ o and G␤␥ subunits (29). Similarly, specific ␣ i and G␤␥ subunit combinations may mediate receptor-specific activation of AC II to confer G protein-selective enhancement of cAMP levels.
Another novel signal revealed by depletion of G␣ i subunits was that 5-HT1A receptors potentiate VIP-stimulated cAMP synthesis in G␣ i2 -deficient antisense clones in a PTX-sensitive manner, as observed for endogenously expressed somatostatin receptors in G␣ i2 -deficient antisense GH4C1 clones (17). Potentiation of VIP-stimulated adenylyl cyclase by 5-HT1A receptors in G␣ i2 -deficient antisense clones may be caused by G␤␥ dimers released from G i3 , which weakly couples this receptor to AC type II (40). It is interesting that G i2 is implicated in both 5-HT1A-mediated inhibition (rather than enhancement) of VIP-stimulated cAMP accumulation and in stimulation of basal cAMP. This suggests that G␣ i -mediated inhibition and G␤␥mediated activation of AC subtypes may be triggered simultaneously and that the outcome of receptor activation involves integration of both stimulatory and inhibitory actions.
In summary, stable transfection of distinct G␣ i subunit an-tisense constructs was used to eliminate their cognate proteins and to study G i protein subtype specificity of inhibitory receptor coupling to AC. Characterization of the different G␣ i antisense clones not only demonstrated G i protein specificity of receptor-mediated signal transduction pathways but also revealed unexpected signaling mechanisms. The biological and physiological significance of the novel stimulation of cAMP synthesis by G i -coupled receptors observed in distinct G␣ ideficient antisense clones is not yet clear. It has been reported, however, that in hippocampal membranes activation of 5-HT1A receptors can stimulate basal adenylyl cyclase activity while inhibiting forskolin-stimulated cAMP synthesis (10,11,41,42). Interestingly, AC type II is most abundant in regions of the brain (e.g. hippocampus CA1 area) in which 5-HT1A receptor activation appears to stimulate cAMP levels (43). Finally, it is known that the expression of G protein ␣ subunits is regulated by a variety of hormones and neurotransmitters (44 -47). Thus, the present results suggest that significant reduction of a G␣ subunit may not only affect G protein-coupled receptor signaling quantitatively but may qualitatively alter the receptor signaling phenotypes.