bg Subunits of Pertussis Toxin-sensitive G Proteins Mediate A1 Adenosine Receptor Agonist-induced Activation of Phospholipase C in Collaboration with Thyrotropin A NOVEL STIMULATORY MECHANISM THROUGH THE CROSS-TALK OF TWO TYPES OF RECEPTORS*

COS-7 cells were transiently transfected with human thyrotropin receptor and dog A1 adenosine receptor cDNAs. An A1 agonist, N -(L-2-phenylisopropyl) adenosine (PIA), which is ineffective alone, enhanced the thyrotropin (TSH)-induced inositol phosphate production, reflecting phospholipase C (PLC) activation, but inhibited the TSH-induced cAMP accumulation, reflecting adenylyl cyclase inhibition. These PIA-induced actions were completely inhibited by pertussis toxin (PTX) treatment. Moreover, in the cells expressing a PTX-insensitive mutant of Gi2a or Gi3a, in which a glycine residue was substituted for a cysteine residue to be ADP-ribosylated by PTX, at the fourth position of the C terminus, PIA effectively exerted both stimulatory and inhibitory effects on the TSH-induced actions although the cells were treated with the toxin. Overexpression of the bg subunits of the G proteins enhanced the TSHinduced inositol phosphate production without any significant effect on the cAMP response; under these conditions, PIA did not further increase the elevated inositol phosphate response to TSH. On the contrary, overexpression of a constitutively active mutant of Gi2a, in which the guanosine triphosphatase activity is lost, inhibited the TSH-induced cAMP accumulation but hardly affected the inositol phosphate response; under these conditions, PIA never exerted further inhibitory effects on the cAMP response to TSH. In contrast to the case of the TSH-induced inositol phosphate response, the response to a constitutively active G11a mutant was not appreciably affected, and that to NaF was rather inhibited by PIA and overexpression of the bg subunits. Taken together, these results suggest that a single type of PTX-sensitive G protein mediates the A1 adenosine receptor-linked modulation of two signaling pathways in collaboration with an activated thyrotropin receptor; a subunits of the PTX-sensitive G proteins mediate the inhibitory action on adenylyl cyclase, and the bg subunits mediate the stimulatory action on PLC. In the case of the latter stimulatory action on PLC, the bg subunits may not directly activate PLC. The possible mechanism by which bg subunits enhance the TSH-induced PLC activation is discussed.

The activation of heterotrimeric guanine nucleotide-binding proteins (G proteins) 1 is involved in the stimulation of a variety of signaling pathways by hormone, neurotransmitter, and sensory receptors with seven transmembrane domains (1). Agonist-bound receptors activate G proteins by stimulating the exchange of GDP for GTP on ␣ subunits in a trimeric form of the proteins, which, in turn, accelerates dissociation of the ␤␥ subunits from the ␣ subunits (2). In early studies, only the ␣ subunits were studied as transducers, but now both the ␣ and ␤␥ subunits are recognized to be involved in the regulation of various effector systems such as adenylyl cyclase (AC), phospholipase C (PLC), or ion channels (3,4). At least 16, 5, and 6 species of the ␣, ␤, and ␥ subunits, respectively, have been identified in molecular cloning studies (5,6). Furthermore, several isoforms of each subunit may be expressed within the same cell type (7)(8)(9). Thus, it would be reasonable to assume that a variety of G-protein-dependent actions are executed by each different molecular species of the G proteins or their subunits. In this context, identification of molecular species that participate in certain signaling systems is necessary for understanding its mechanism.
In thyroid cells, thyrotropin (TSH) activates PLC through the G q /G 11 protein as well as AC through the G s protein, resulting in mobilization of Ca 2ϩ and accumulation of cAMP in the cells (10). We have shown that in rat FRTL-5 thyroid cells, adenosine and its derivatives such as phenylisopropyl adenosine (PIA) inhibited TSH-induced AC activation and, in contrast, enhanced TSH-induced PLC activation and subsequent Ca 2ϩ mobilization through the A 1 type receptor (A 1 R) and PTX-sensitive G protein (11). This PTX-sensitive G proteinmediated PLC activation is not restricted to the cross-talk between the TSH and adenosine signaling mechanisms. In FRTL-5 thyroid cells, PIA, through A 1 R, also enhanced PLC activation induced by ␣ 1 -adrenergic receptor agonists (12) and P 2 -purinergic receptor agonists (13,14) as well as TSH. Likewise, A 1 R agonists enhance PLC activation induced by IgE in RBL2H3 cells and those induced by ATP and bradykinin in the smooth muscle cell line (15)(16)(17). All of these A 1 R-mediated actions were abolished by PTX treatment. In addition, this cross-talk is not specific to the A 1 R agonist; in NG108 -15 cells, enkephalin, somatostatin, ␣ 2 -adrenergic agonist, and carbachol have also been shown to enhance P 2 receptor agonist-or bradykinin-induced PLC activation and Ca 2ϩ mobilization, in a PTX-sensitive manner (18 -20). In a previous study, we also showed that the PTX-sensitive G protein-mediated modulation of the TSH and muscarinic acetylcholine actions by adenosine are reconstituted by expressing both the TSH receptor (TSHR) and A 1 R in COS-7 cells (21) and both the m3 muscarinergic acetylcholine receptor and A 1 R in CHO cells (22), respectively. Thus, our findings in concert with those of others have suggested the presence of a universal cross-talk mechanism mediated by a PTX-sensitive G protein(s) between AC-inhibitory and PLC-stimulatory signaling mechanisms resulting in the enhancement of PLC activation.
Two receptors involved in this cross-talk regulation of PLC are characterized as one type of receptor that couples to PTXsensitive G i /G o proteins whose stimulation inhibits AC but exhibits only a small or undetectable effect on PLC, and the other type of receptor, the so-called Ca 2ϩ -mobilizing receptor, couples to G q /G 11 proteins whose stimulation leads to activation of PLC. Thus, AC-inhibiting receptor agonists, through G i /G o proteins, permissively or synergistically enhance Ca 2ϩmobilizing receptor agonist-induced PLC activation. The molecular mechanism by which the respective receptor agonists induce stimulation of each signaling pathway leading to AC inhibition and PLC activation through G i /G o and G q /G 11 , respectively, has been well characterized (6). However, it has not been well elucidated how a single type of AC-inhibiting receptor simultaneously links to two signaling pathways with the aid of the PTX-sensitive G protein(s).
In this study, we aimed to further define the role of PTXsensitive G proteins in the cross-talk phenomena in COS-7 cells where TSHR and A 1 R were expressed. The specific objectives were to determine (a) whether a single type of PTX-sensitive G protein mediates the modulation of the two signaling pathways and (b) how the G protein participates in the two signaling pathways. In the reconstituted cross-talk systems constructed by recombinant receptors and manipulated G-protein pools in the cells, we found that at least a single molecular species of G i 2 or G i 3 can mediate the modulation of two signaling pathways; enhancement of PLC is mediated by ␤␥ subunits, and inhibition of AC is mediated by a single species of the G i ␣ subunit. Furthermore, our results suggest that the primary target of the ␤␥ subunits may not be PLC itself but may instead be located upstream of the enzyme in the signaling pathway, possibly at the level of the ␣ subunits of the G q /G 11 proteins.

EXPERIMENTAL PROCEDURES
Materials-PIA was purchased from Sigma; staurosporin was from Kyowa Medex Co. (Tokyo, Japan); and myo-[2-3 H]inositol (23.0 Ci/ mmol) was from NEN Life Science Products. Human TSHR cDNA in the pSVL expression vector (23) and dog A 1 R (24) cDNA were generously provided by Dr. G. Vassart (Université Libre de Bruxelles, Belgium), bovine G␤1 cDNA by Dr. M. I. Simon (California Institute of Technology), bovine G␥2 cDNA by Dr. T. Nukada (Tokyo Institute of Psychiatry, Tokyo, Japan), and pCDL-SR␣ 296 vector (25) by Dr. Y. Takebe (National Institute of Health, Tokyo, Japan). Rabbit antiserum specific to G i 2␣ was generously provided by Dr. Y. Kanaho (Tokyo Institute of Technology, Yokohama, Japan), and the ␤ subunit of the G protein was supplied by Dr. T. Katada (Tokyo University, Tokyo, Japan). The radioimmunoassay of cAMP used a Yamasa cAMP assay kit, which was a gift from Yamasa Shoyu Co. (Choshi, Chiba, Japan). The sources of all other reagents were the same as described previously (11,14,26).
Plasmid Construction-The PTX-insensitive mutant G i 2␣(C352G) expression plasmid was obtained by polymerase chain reaction mutagenesis using wild type rat G i 2␣ cDNA (27) as a template. The 5Ј-primer, 5Ј-gggaattcCCACCATGGGCTGCACCGTGAG-3Ј, contains the EcoRI site, Kozak sequence (CCACC) (28), and the first six amino acids of the G i 2␣. The 3Ј-primer, 5Ј-gggcggccgcTCAGAAGAGGCCGC-CGTCCTTCAGCGTTGTTCTTGATGATGACGT-3Ј, contains the NotI site, a stop codon, and 14 amino acids in the C-terminal region of G i 2␣, except for GCC (underlined), which encodes glycine instead of cysteine at position 352 from the N terminus. The PTX-insensitive mutant G i 3␣(C351G) expression plasmid was obtained by polymerase chain reaction mutagenesis using wild type rat G i 3␣ cDNA (29) as a template. The 5Ј-primer, 5Ј-gggaattcCCACCATGGGCTGCACGTTGAGCGCCG-AGGAC-3Ј, contains the EcoRI site, Kozak sequence (CCACC), and the first nine amino acids of G i 3␣. The 3Ј-primer, 5Ј-gggcggccgcTCAGTA-AAGCCCGCCTTCCTTTAAGTTGTT-3Ј, contains the NotI site, a stop codon, and nine amino acids in the C-terminal region of G i 3␣, except for GCC (underlined), which encodes glycine instead of cysteine at position 351 from the N terminus. The amplified G i 2␣(C352G) or G i 3␣(C351G) was digested with EcoRI and NotI and then inserted into the EcoRI/NotI site of the pcDNA/AMP expression plasmids (Invitrogen, CA). The constitutively active mutant G 11 ␣(Q209L) expression plasmid was obtained by polymerase chain reaction mutagenesis using wild type mouse G 11 ␣ cDNA from mouse S49 lymphoma cells as a template. The 5Ј-primer, 5Ј-tagcaagcttcatATGACTCTGGAGTCCATGATGGC-3Ј, contains the HindIII site and the first eight amino acids of G 11 ␣. The 3Ј-primer, 5Ј-caatggatccACTTCCTGCGCTCTGACCTCAGGCCTCCC-3Ј, contains the BamHI site and 14 amino acids including the leucine codon (underlined) instead of the glutamine codon at position 209 from the N terminus. The amplified fragment was digested with HindIII and BamHI and then inserted into the corresponding region of the wild G 11 ␣ cDNA. The cDNA was subcloned into the HindIII site of the pCMV5 expression plasmid (30) in the right orientation. The constitutively active mutant G i 2␣(Q205L) was obtained by replacing the glutamine codon at position 205 from the N terminus to the leucine codon, using a uracil-containing single-stranded DNA of the rat G i 2␣ as a template. DNA sequences of the mutants were confirmed by dideoxynucleotide sequencing (31). The G␤1 and G␥2 cDNAs were subcloned into pCMV5 as described previously (32).
Assay of Inositol Phosphate and cAMP-The [ 3 H]inositol-labeled cells were washed twice with Hepes-buffered medium, which consisted of 20 mM Hepes (pH 7.5), 134 mM NaCl, 4.7 mM KCl, 1.2 mM KH 2 PO 4 , 1.2 mM MgSO 4 , 2 mM CaCl 2 , 2.5 mM NaHCO 3 , 5 mM glucose, and 0.1% (w/v) bovine serum albumin (fraction V). The cells were then incubated for 30 min at 37°C with the same medium containing 10 mM LiCl and 0.5 units/ml adenosine deaminase, 0.1 mM Ro20 -1724, and 1 g/ml staurosporin, and the agents were tested in a final volume of 0.6 ml. The reaction was terminated by the addition of 0.1 ml of 1 N HCl. The contents of cAMP and 3 H-labeled inositol phosphate were measured in the same HCl extract. Cyclic AMP was measured by radioimmunoassay. [ 3 H]Inositol mono-, di-, and trisphosphates were separated from inositol and glycerophosphoinositol on Dowex 1-X8 (formate form, Bio-Rad) columns as described previously (11). The radioactivity of the trichloroacetic acid (5%)-insoluble fraction was measured as the total radioactivity incorporated into the cellular inositol lipids. Where indicated, the results were normalized to 10 5 cpm of the total radioactivity.
Immunoblot Analysis-Crude plasma membranes and their cholate extracts were prepared as described previously (33,34). The cholate extract (25 g of protein) was resolved on SDS-12.5% polyacrylamide slab gel electrophoresis and then electrophoretically transferred to a Millipore Immobilon sheet (35). The expressions of G i 2␣(Q205L) and the ␤1 subunits in addition to endogenous G i 2␣ and the ␤ subunits were visualized by incubating the sheet with each specific rabbit antiserum, with alkaline phosphate-conjugated goat antibody against rabbit IgG and finally with 5-bromo-4-chloro-3-indoylphosphate and nitro blue tetrazolium as described previously (35).
Data Presentation-All experiments were performed in triplicate or quadruplicate with more than three different batches of cells. The results were presented as means Ϯ S.E. of three or four values in a representative experiment, unless otherwise stated.

A Single Type of PTX-sensitive G protein, G i 2 or G i 3, Mediates A 1 R-linked Stimulation of Two Signaling Pathways Leading to Inhibition of AC and Enhancement of PLC in Collabora-
tion with an Activated TSHR-In accordance with our previous study (21), when both TSHR and A 1 R were expressed in COS-7 cells, although PIA, an A 1 agonist, alone hardly affected the basal activities of PLC and AC, this A 1 agonist enhanced TSHstimulated PLC but inhibited the TSH-stimulated AC in a way similar to those in FRTL-5 thyroid cells (11) (Fig. 1, A and C). These PIA actions were completely abolished by a PTX treatment without an appreciable effect on intrinsic TSH actions (Figs. 1, B and D).
In COS cells, several types of PTX-sensitive G proteins are expressed (36 -38). To ascertain the role of a particular PTXsensitive G protein subtype in the bidirectional A 1 R agonist action, we planned to use the cells where only a species of PTX-insensitive and active G i ␣ mutant is expressed under the conditions where native G i /G o is inactivated by the PTX treatment. We constructed a mutant cDNA, G i 2␣(C352G), in which a glycine residue substitutes for a cysteine residue at position 352 of G i 2␣. COS-7 cells were transfected with this PTX-insensitive G i 2␣ cDNA together with cDNAs of the ␤1 and ␥2 sub-units of the G protein, expecting the expression of the PTXinsensitive G i 2(C352G) (Fig. 2). The transfection of these subunit-cDNAs did not appreciably affect the TSH-and PIAinduced IP responses (compare panels A of Figs. 1 and 2) and cAMP responses (compare panels C of Figs. 1 and 2) in the control cells. However, when the cells were treated with PTX, a clear difference with respect to PIA actions was observed; PIA was still effective at enhancing the TSH-induced PLC activation (compare panels B of Figs. 1 and 2) and inhibiting the TSH-induced AC activation (compare panels D of Figs. 1 and 2) in the cells transfected with these subunit cDNAs. Thus, G i 2 can couple to A 1 R and mediate the reduction of the TSHinduced AC activation as well as the enhancement of the TSHdependent PLC activation.
Another PTX-insensitive ␣ subunit of G i 3, G i 3␣(C351G), which was expressed in COS-7 cells, also coupled to A 1 R and modulated the TSH-induced actions even in the PTX-treated cells; the TSH (100 nM)-induced levels were 857 Ϯ 83 or 1162 Ϯ 45 dpm for IP production and 2.42 Ϯ 0.05 or 1.75 Ϯ 0.06 nmol/mg for cAMP accumulation in the absence or presence of PIA, respectively. In all of the cases using or not using either the G i 2 or G i 3 mutant, PIA alone exerted no significant effect on either the AC or PLC activities.
It has been reported that the C-terminal region of ␣ subunits of G proteins is important for their receptor recognition (39). Therefore, we examined whether the cysteine to glycine substitution affects the ability of PTX-sensitive G proteins to enhance the TSH-induced PLC activation. As shown in Fig. 3, no significant difference was detected between mock-and G i 2␣(C352G)-transfected cells in terms of their response to any PIA dose. This result indicates that G i 2␣(C352G) still retains the ability to associate with A 1 R in COS-7 cells in a way similar to endogenous PTX-sensitive G proteins. The response of G i 2␣(C352G) transfected cells to higher than 100 nM PIA was slightly stronger than those of the mock-transfected cells. However, this feature of the mutant transfected cells is not specific for the PIA effect on the TSH-induced PLC activation, because in the mutant G protein-transfected cells, PIA alone slightly stimulates PLC in the absence of TSH stimulation (Fig. 3).
␣ Subunits of PTX-sensitive G Proteins Are Responsible for the Inhibition of the AC and ␤␥ Subunits for the Enhancement of PLC-We next examined which subunits, i.e. the ␣ subunit or ␤␥ subunits of the PTX-sensitive G protein, mediate each A 1 R-mediated signaling. To clarify this point, we transfected mutant DNAs encoding GTPase-deficient G i 2␣(Q205L), G␤1, G␥2, or a combination of them. G i 2␣(Q205L) is mutated by substituting a leucine residue for a glutamine residue at position 205. This mutant is lacking in GTPase activity and thereby constitutively stimulates an effector enzyme (40). Expression of the mutated ␣ and ␤ subunits was verified by immunoblotting using specific antisera against the G i 2␣ (Fig. 4A) or ␤ (Fig. 4B) subunit. Consistent with previous results (41), no significant increase in the expression of the ␤ subunit was detected when this subunit cDNA was transfected alone. A ␤ subunit might be unstable in an intracellular environment unless it forms a complex with a ␥ subunit. We measured cAMP accumulation and IP production in the absence or presence of TSH in the cells overexpressing these ␣ or ␤␥ subunits. As shown in Fig. 4D, TSH-induced cAMP accumulation was significantly inhibited by the expression of G i 2␣(Q205L), whereas the transfection of ␤ and/or ␥ subunit cDNAs hardly affected the cAMP response. Conversely, the overexpression of ␤␥ subunits stimulated the TSH-induced PLC activation, whereas the expression of G i 2␣(Q205L) did not appreciably influence it (Fig. 4C). These results suggest the potential roles of the ␣ and ␤␥ subunits for the inhibition of the TSH-induced AC activation and the enhancement of the TSH-induced PLC activation, respectively.
To determine whether these G protein subunits actually mediate the A 1 R agonist-induced modulation of TSH-induced actions, we examined the effects of PIA on the dose-dependent TSH activation of AC and PLC in the cells overexpressing G i 2␣(Q205L) or ␤␥ subunits (Fig. 5). As previously shown (Ref. 21 and Figs. 1 and 2), PIA enhanced the dose-dependent TSH activation of IP production (Fig. 5A) and conversely inhibited the TSH stimulation of cAMP accumulation (Fig. 5D). When the ␤␥ subunits were overexpressed, similar to the results shown in Fig. 4, the level of TSH-induced PLC (Fig. 5B) but not cAMP accumulation (Fig. 5E) was increased from that of the control cells at any TSH dose. Under these conditions, the addition of PIA together with TSH did not change the PLC level obtained by TSH alone but reduced the cAMP accumulation induced by the hormone. On the other hand, when the constitutively active mutant of G i 2␣ was expressed in the cells, the level of the TSH-induced accumulation of cAMP, but not of PLC activation, was lower than that in the control cells. In this case, the further addition of PIA did not change the level of TSHinduced cAMP accumulation but induced a further increase in the PLC level from that obtained by TSH alone. Thus, the overexpressed ␤␥ subunit complex mimicked the stimulatory action of PIA on the TSH-induced PLC, while the constitutively active ␣ subunit mimicked the inhibitory action of PIA on the TSH-induced AC.
␤␥ Subunits May Not Directly Activate PLC-The direct in- hibitory action of an ␣ subunit of the G i protein on activated AC has been reported by several groups (40,42,43), and this mechanism may account for the A 1 R-mediated inhibition of AC in the present COS-7 cell system. As with the analogy of the ␣ subunit, it might be possible that the ␤␥ subunit complex directly interacts with the activated PLC and further stimulates the activated enzyme. If this is the case, PLC activated by any means should be further stimulated by the overexpressed ␤␥ subunit complex even in the absence of TSH. Since the TSH-induced activation of PLC is mediated by the TSHR activation of G q /G 11 (10), we planned to provide activated G q /G 11 to the cells to directly activate PLC without TSH. For this purpose, the cells were transfected with a constitutively active G 11 ␣ mutant, G 11 ␣(Q209L), which lacks GTPase activity (44) (Fig. 6), or challenged with NaF, a nonselective G protein activator (Fig. 7A). In both cells where PLC was activated by the active G q /G 11 or NaF, the level of inositol phosphate production was as high as that in the cells stimulated by TSH (Figs. 6A and 7A). However, in contrast to the TSH-induced PLC activation that was enhanced by the overexpression of the ␤␥ subunits (Figs. 6A and 7), the G 11 ␣(Q209L)-induced activation was hardly affected by the overexpressed ␤␥ subunits (Fig.  6A). An increasing amount of G 11 ␣(Q209L) plasmid for transfection was associated with a further increase in the inositol phosphate production (Fig. 6, A and B), and these activities were further stimulated by TSH (Fig. 6B). This eliminates the possibility that failure of the ␤␥ subunits to enhance the G 11 ␣(Q209L)-induced PLC activation is due to the saturation of the enzyme activity. Likewise, the NaF-induced PLC activation was not enhanced but rather was inhibited by the overexpression of the ␤␥ subunits (Fig. 7, A and B, compare first columns). These results support the idea that the target of the ␤␥ subunits is not PLC itself.
As mentioned above, in this experiment, we noticed that the NaF-induced action was slightly but significantly (p Ͻ 0.05) inhibited by the overexpressed ␤␥ subunit complexes. This was somewhat unexpected because we expected the positive role for the ␤␥ subunit complexes in the PLC activation. In relation to this, we found that this NaF-induced PLC activation is inhibited by PIA (Fig. 7A). This inhibitory effect of PIA on the NaF-induced PLC activation was significantly reduced in the ␤␥ overexpressed cell, in parallel with the reduction in the stimulatory effect of PIA on TSH action (Fig. 7B). This suggests that the PIA inhibition of the NaF-induced action is also mediated by the ␤␥ subunits. One might wonder why PIA can inhibit NaF-induced PLC activation, because it might be expected that PTX-sensitive G proteins would also be stimulated by a nonselective G protein activator, NaF. This may be simply explained if, in our COS-7 cell system, NaF is an activator for G q /G 11 but not for G i /G o ; NaF hardly influenced the TSHinduced cAMP accumulation but PIA inhibited it even in the presence of NaF to the same extent as in its absence (data not shown). DISCUSSION In the present study, we constructed two PTX-insensitive mutant DNAs of G i 2␣ and G i 3␣, both with a mutation in the sequence encoding the PTX-sensitive domain of the G proteins. By using the cells expressing each one of these mutant G i ␣ subunits, we were able to analyze the functions of a single subtype species of G i in the cells where all of the native wild type G i /G o proteins are inactivated by the PTX treatment. We demonstrated that a single type of G protein, either G i 2 or G i 3, mediates both the inhibition of TSH-induced AC activation and the enhancement of TSH-induced PLC activation. A similar strategy had been chosen by other groups to identify a particular PTX-sensitive G protein subtype to inhibit AC through dopamine D2 receptor isoforms (45) and to regulate multiple effectors including AC, PLC, and phospholipase A 2 through the m2 muscarinic acetylcholine receptor (46).
On the other hand, one might consider that the molecular mutation affects its activity and hence results in a misleading conclusion, because, in this mutant ␣ subunit, a glycine residue substitutes for a cysteine residue near the C terminus, where the region interacting with receptors is located (39). However, the ability of PIA, an A 1 R agonist, to modulate the TSH-induced actions in the cells expressing the mutant G i 2␣ subunit in the PTX-treated cells was comparable with that in the control cells only transfected with A 1 R and TSHR without the toxin treatment (Fig. 3). This suggests that the change in the affinity of the mutant G i 2 for A 1 R, even if occurring, is negligible. During the coupling of the D2 dopamine short form receptor to G i 2, no detectable difference in affinity has been observed between the native and the PTX-insensitive mutant G i 2 (45). Thus, the application of this technique is reasonable for the evaluation of the role of a particular G protein in living cells.
In the present study, we mainly presented the results obtained from the cells expressing the G i 2 mutant because we observed no appreciable difference between the cells expressing the mutant G i 2 and those expressing the mutant G i 3. It had been previously shown, however, that in a liposome system constituted with purified bovine brain A 1 R and one of recombinant ␣ subunits of various PTX-sensitive G proteins, the affinity of G i 3 for A 1 R was higher than those of other G protein subtypes tested, including G i 1, G i 2, G i 3, and G o (47)(48)(49). This discrepancy may be explained by the saturation of transfected G i ␣ cDNAs in our case where the cells were transfected with cDNAs at doses needed for maximal responses. Even in the liposome experiments mentioned, the maximal response to G i 2 was the same as that to G i 3.
The present results showed that the expression of a constitutively active mutant of the G i 2␣ subunit caused an appreciable inhibition of the TSH-induced AC activation. This is consistent with the current idea of the G i action mechanism, such that the ␣ subunit of the PTX-sensitive G protein liberated by receptor activation directly inhibits AC, but the ␤␥ subunits act as either an inhibitor or activator of the enzyme depending on the type of enzyme isoform (42). In HEK293 cells transfected by cDNAs encoding the GTPase-deficient derivatives of G i 1␣, G i 2␣, or G i 3␣, AC is inhibited, which is consistent with our results with COS-7 cells (40). However, a previous paper has reported that the transfection of COS-7 cells with the same mutant G i 2␣ cDNA fails to inhibit AC (43). This discrepancy between the two COS-7 cell experiments might be due to the presence or absence of staurosporin, a potent protein C inhibitor, in the assay medium. Actually, under our experimental conditions, AC in COS-7 cells was not inhibited even by PIA, a typical inhibitory receptor agonist, unless staurosporin was added (21). Although the mechanism by which the suppression of protein kinase C discloses the G i ␣-induced inhibition of AC has not yet been elucidated, this phenomenon might be related to the ability of the enzyme to phosphorylate the G i ␣ (50).
The PTX-insensitive nature of the TSH-induced PLC activation suggests that the activation occurs on the ␤1 or ␤3 isoform of PLC through a PTX-insensitive G protein, G q or G 11 (38,51,52). The present study demonstrates the participation of ␤␥ subunits in the A 1 R-linked enhancement of the TSH-induced PLC activation. Recently, ␤␥ subunits of G i /G o have also been shown to activate PLC␤2 and ␤3 isoforms in a PTX-sensitive manner, using cell-free reconstituted systems and cells transfected with cDNAs (53,54). Therefore, one may presume that in our experimental system, the ␤␥ subunit of the G i /G o complex dissociated from the ␣ subunit upon A 1 R receptor stimulation directly activates the ␤2 or ␤3 isoform of PLC in COS-7 cells where at least the PLC␤3 expression has been reported (52). This mechanism, however, would not explain the permissive or synergistic nature of the AC-inhibiting receptor agonist-induced enhancement of PLC activation observed in the many cell systems including the present COS-7 cell system, where PIA alone induced only a small, if any, activation of PLC, but it markedly enhanced the enzyme in concert with Ca 2ϩ -mobiliz- FIG. 6. IP production in the cells overexpressing ␤␥ subunits and/or constitutively active mutant of G 11 ␣ In A, COS-7 cells were transfected with expression plasmids encoding G␤1 (40 g) and G␥2 (40 g) (ϩ) or pCMV5 vector (80 g) (Ϫ) together with TSHR (20 g) and A 1 R (20 g). In some experiments (fifth and sixth columns), 1 g of expression plasmid encoding G 11 ␣(Q209L) was also cotransfected. The cells were then cultured in medium containing [ 3 H]inositol for 2 days. The cells were incubated without (Control and G 11 ␣(Q209L)) or with 100 nM TSH (TSH) as described under "Experimental Procedures." In B, the cells were transfected with the indicated amount of G 11 ␣(Q209L) expression plasmid together with TSHR (20 g) and A 1 R (20 g). The total amount of DNA for each transfection was adjusted by pCMV5 vector. The cells were then cultured in medium containing ing receptor agonists such as TSH (Ref. 21 and Figs. 1 and 2).
The present study also excluded the possibility that PLC prestimulated by activated G q /G 11 interacts with the ␤␥ subunits, because PLC prestimulated by the expression of a constitutively active mutant of G 11 ␣ was not further stimulated by the overexpression of the ␤1 and ␥2 subunits. Furthermore, when PLC was prestimulated by NaF through the AlF 4 Ϫ -activated G protein, the enzyme was even inhibited by the A1R agonist or the ␤␥ overexpression.
A remaining possibility for the ␤␥ subunits to participate in the synergistic or permissive stimulation of PLC is their interaction with G q /G 11 . The ␤␥ subunits released by the action of A 1 R stimulation or overexpressed by transfection of their cDNAs could theoretically either enhance or inhibit the activity of PLC depending on their stoichiometry with the ␣ subunits of the G q /G 11 proteins (Fig. 8A). A similar model has already been presented for the explanation of the inhibitory or stimulatory effect of free ␤␥ subunits on the receptor-mediated PLC activation (55,56). If there is an excess of G q /G 11 ␣(GDP) over the ␤␥ subunits, the increase in this subunit complex could lead to the formation of a heterotrimeric form of G q /G 11 protein (G q / G 11 ␣(GDP) ␤␥) that can couple to TSHR, thereby potentiating the TSH-induced action (through mechanism a in Fig. 8A). On the other hand, if there is no excess of G q /G 11 ␣(GDP) over the ␤␥ subunits or if the formation of the heterotrimeric form of the G q /G 11 protein is not rate-limiting, the production of free ␤␥ subunits could have an inhibitory effect by binding to the active form of G q /G 11 (G q /G 11 ␣(GTP)) (through mechanism b in Fig.  8A). AlF 4 Ϫ , which is produced from the added F Ϫ ions and traces of Al 3ϩ ion, converts the G q /G 11 protein to the active form (G q / G 11 ␣(GDP-AlF 4 Ϫ ) in this case) by interacting with the inactive GDP forms of the monomeric as well as the heterotrimeric G proteins (57) (Fig. 8B). In this case, the formation of the active form of the G q /G 11 protein is rate-limiting regardless of the excess of the inactive form of G q /G 11 ␣(GDP) over the ␤␥ subunits. Thus, according to the model shown in Fig. 8B, the production of ␤␥ subunits would result in the inhibition of NaF-induced PLC activation by inhibiting the formation of the active form of the G q /G 11 protein. Actually, A 1 R stimulation and also overexpression of the ␤␥ subunits inhibited the NaFinduced PLC activation (Fig. 7). This inhibition also supports this model.
In addition to transfected COS-7 cells used in the present study, the permissive or synergistic stimulation by G i /G o proteins of the receptor-mediated PLC activation has been observed in a variety of cells including FRTL-5 thyroid cells, NG 108 -15 cells, and smooth muscle cells (see Introduction). In these systems as well, a similar stimulatory mechanism through the ␤␥ subunits (Fig. 8A) might account for the positive cross-talk as in the case of the present COS-7 cellular reconstituted system. A similar ␤␥ subunit action has also been observed by Harden and colleagues (55) in a cell-free reconstituted system using turkey erythrocyte membranes; the reconstituted ␤␥ subunits of the G proteins alone hardly affected the basal activity of PLC but enhanced the P 2 receptor-mediated activation of PLC. In this cell-free system, a large amount of ␤␥ subunits significantly inhibited the AlF 4 Ϫ -induced activation of the enzyme as in the case of the present cellular system. In some cases (e.g. in GH3 cells), however, AC-inhibiting receptor agonists inhibit, through the G i /G o proteins, the receptor-induced PLC activation (58). In this case, the inhibitory mechanism as described in Fig. 8A might operate.
In conclusion, using COS-7 cells transfected with cDNAs encoding a receptor and one of the G protein subunits, we have shown that a single type of PTX-sensitive G protein, G i 2 or G i 3, upon stimulation of A 1 R, can couple to PLC in a stimulatory fashion and to AC in an inhibitory fashion in collaboration with an activated TSHR. Within the subunits of the PTX-sensitive G proteins, the ␣ subunit seems to be involved in the inhibition of AC activity, while the ␤␥ subunits are involved in the enhancement of PLC. The cross-talk between the ␤␥ subunits derived from PTX-sensitive G proteins and the ␣ subunit of G proteins, which couples to TSHR (probably one of the G q /G 11 family members) might account for the permissive or synergistic activation of PLC by simultaneous stimulation of A 1 R and TSHR.