A Dominant Negative GαsMutant That Prevents Thyroid-stimulating Hormone Receptor Activation of cAMP Production and Inositol 1,4,5-Trisphosphate Turnover

A Ser to Asn mutation at position 54 of the α subunit of Gs (designated N54-αs) was characterized after transient expression of it with various components of the receptor-adenylyl cyclase pathway in COS-1, COS-7, and HEK 293 cells. Previous studies of the N54-αs mutant revealed that it has a conditional dominant negative phenotype that prevents hormone-stimulated increases in cAMP without interfering with the regulation of basal cAMP levels (Cleator, J. H., Mehta, N. D., Kurtz, D. K., Hildebrandt, J. D. (1999) FEBS Lett. 243, 205–208). Experiments reported here were conducted to localize the mechanism of the dominant negative effect of the mutant. Competition studies conducted with activated αs* (Q212L) showed that the N54 mutant did not work down-stream by blocking the interaction of endogenous αs with adenylyl cyclase. The co-expression of wild type or N54-αs along with the thyroid-stimulating hormone (TSH) receptor and adenylyl cyclase isotypes differing with respect to βγ stimulation (AC II or AC III) revealed that the phenotype of the mutant is not dependent upon the presence of adenylyl cyclase isoforms regulated by βγ. These studies ruled out a downstream site of action of the mutant. To investigate an upstream site of action, N54-αs was co-expressed with either the TSH receptor that activates both αs and αq or with the α1B-adrenergic receptor that activates only αq. N54-αs failed to inhibit α1B-adrenergic receptor stimulation of inositol 1,4,5-trisphosphate production but did inhibit TSH stimulation of inositol 1,4,5-trisphosphate. These results show that Gs and Gq compete for activation by the TSH receptor. They also indicate that the N54 protein has a dominant negative phenotype by blocking upstream receptor interactions with normal G proteins. This phenotype is different from that seen in analogous mutants of other G protein α subunits and suggests that either regulation or protein-protein interactions differ among G protein α subunits.

Heterotrimeric GTP-binding proteins or G proteins transmit signals from receptors on the surface of the cell to various intracellular effector enzymes. G proteins have traditionally been defined by their ␣ subunits, which consist of four families: ␣ s ; ␣ i/o ; ␣ q/11 ; and ␣ 15/16 . One of the best-characterized ␣ subunits, ␣ s , stimulates adenylyl cyclase in response to receptormediated hormone stimulation (1)(2)(3)(4). Inactive G s consists of a GDP-bound ␣ subunit and a ␤␥ dimer. Upon receptor activation, the ␣ s ⅐GTP complex dissociates from the receptor and its ␤␥ dimer and then stimulates adenylyl cyclase to increase cAMP levels in the cell. Certain isotypes of adenylyl cyclase (e.g. AC II and AC IV) are also stimulated by ␤␥ in the presence of activated ␣ s (5,6), whereas other isotypes of adenylyl cyclase are inhibited by ␤␥ (e.g. AC I). Deactivation occurs when the ␣ subunit hydrolyzes its bound GTP to GDP, allowing it to reassociate with ␤␥.
G protein-coupled receptors (GPCRs) 1 activate G proteins by catalyzing GTP exchange on the ␣ subunit. Although some GPCRs activate only one G protein isotype (e.g. ␤-adrenergic receptor activation of ␣ s ), others are described as promiscuous and can activate multiple G protein isotypes (e.g. the receptor for TSH). The TSH receptor has been shown to couple to adenylyl cyclase and phospholipase C (7)(8)(9), in part because the receptor can couple to both G s and G q/11 (10). It has not been determined whether one population of TSH receptors exists that possesses the ability to couple both G s and G q/11 or whether multiple populations of receptor exist (e.g. arising from either splice variants, post-translational modifications, or compartmentalization) that could account for the ability of the TSH-R to activate both G s and G q/11 . The superfamily of GTP-binding proteins, which include the ␣ subunits of heterotrimeric G proteins and the monomeric GTP-binding proteins, share four conserved amino acid sequences that form the core of the GTP regulatory mechanism of these proteins (11)(12)(13). The consensus sequence of the first conserved region of the nucleotide binding site is GX 4 GK(S/T), the residues of which include Ser 17 in Ras and the equivalent Ser 54 in G␣ s . These residues are involved in binding Mg 2ϩ ion when it is complexed with nucleotide (14). Ras with an Asn mutation at this site (N17) has an increased preference for GDP over GTP and a dominant negative phenotype (15). The dominant negative phenotype of N17-Ras can be explained by its ability to sequester its own guanine nucleotide exchange factor (16,17), which is functionally analogous to a GPCR. An extension of the N17-Ras mutation to other members of the Ras superfamily, such as Ran (18), Rab (19), Ral (20), Rho (21), and Rac (22), also results in a dominant negative phenotype. G␣ s 1 The abbreviations used are: GPCR, G protein-coupled receptor; TSH, thyroid-stimulating hormone; PI, phosphatidylinositol; TSH-R, TSH receptor; ␣ s *, activated ␣ s Q212L mutant; VIP, vasoactive intestinal polypeptide; VIP-R, vasoactive intestinal polypeptide receptor; HEK, human embryonic kidney; GMP-P(NH)P, guanyl-5Ј-yl imidodiphosphate; AC I and AC II, adenylyl cyclases I and II; AR, adrenergic receptor; IP, inositol phosphate. with a Ser to Asn mutation at this site (N54) also has an increased preference for GDP over GTP (23) and a conditional dominant negative phenotype (24). Its phenotype is conditional, because it prevents hormone-stimulated increases in cAMP, even though it itself activates adenylyl cyclase downstream and increases basal cAMP levels (24).
The site analogous to Ser 54 of ␣ s in ␣ i and ␣ o is Ser 47 . Cys mutants of this site also have a dominant negative phenotype (25,26). However, the mechanism in this case appears to be different from that of the N17-Ras mutant. In particular, others have suggested that C47-␣ o and C47-␣ i interfere with hormone activation of G proteins by tightly binding ␤␥ dimers (25,26). These mutants are presumably not active themselves, because G protein subunit dissociation is required for activation (2). Further, their dominant negative phenotype is explained by their ability to sequester ␤␥ dimers away from other G protein ␣ subunits and by the fact that intact G protein heterotrimers facilitate greatly receptor interactions. Because Ras mutants and ␣ i/o mutants of this essential Ser residue involved in complexing Mg and nucleotide seem to have different mechanisms of action, we sought to determine the analogous mechanism for the N54-␣ s mutant. Surprisingly, the N54-␣ s mutant has a mechanism of action more similar to that of Ras than its more closely related ␣ i/o counterparts. These results suggest differences either in the mechanism of regulation of ␣ s and ␣ i proteins by magnesium and nucleotides or in the protein-protein interactions of these different G proteins.

Materials-[ 3 H]adenine, [ 3 H]inositol
, donkey anti-rabbit horseradish peroxidase-linked antibody, and ECL reagents were purchased from Amersham Biosciences. Bovine TSH (Lot-AFP-5555B) was obtained from the National Hormone and Pituitary Program, NIDDK, National Institutes of Health. All of the other materials were of the highest quality grade available.
Construction of Vectors-cDNAs encoding the long form of G␣ s and the N54 mutation of G␣ s in pMV7 (23) were subcloned into the HindIII site of the mammalian expression vector pcDNA3 (Invitrogen) driven by the cytomegalovirus promoter. The activated (Q212L) ␣ s mutant (␣ s *) in the mammalian expression vector pCr-cytomegalovirus was kindly provided by R. Iyengar. Types II and III adenylyl cyclase cDNAs (27,28), kindly provided by Dr. R. Reed, were subcloned from Bluescript KS between the HindIII and BamHI sites of pcDNA1 and into the EcoRI site of pcDNA3, respectively. The rat TSH-R cDNA inserted in the simian virus 40 promoter-driven expression vector, pSG5 (29), was kindly provided by Dr. L. Kohn. The rat vasoactive intestinal polypeptide receptor (VIP-R) cDNA (30) in the expression vector CDM8 under control of the cytomegalovirus promoter was a kind gift from Dr. S. Nagata. The ␣ 1B -adrenergic receptor in the expression vector pMT2 under control of the SV40 major late promoter was kindly provided by Dr. D. Perez.
Transient Transfections and cAMP Determination-COS-1 and HEK 293 cells grown in Dulbecco's modified Eagle's medium with 10% fetal bovine serum at 37°C were transfected with LipofectAMINE as previously described (24). Cyclic AMP was measured by use of the [ 3 H]adenine uptake assay as described previously (24).
Immunoblotting-Samples were prepared by washing cells in phosphate-buffered saline followed by addition of Laemmli sample buffer containing 10% ␤-mercaptoethanol (31). Proteins were separated by SDS-polyacrylamide gel electrophoresis on 11% polyacrylamide gels by the procedure of Laemmli (31) and transferred to nitrocellulose using a Bio-Rad semi-dry transfer apparatus. Proteins were incubated with anti-␣ s antibody, ASC, produced according to the protocol of Spiegel and co-workers (32). Bands were visualized by ECL.
Inositol Phosphate Determination-Inositol phosphate production was measured by use of the [ 3 H]inositol uptake assay (33)(34)(35)(36). Cells were labeled with 2 Ci/ml myo-[ 3 H]inositol ϳ24 h prior to treatment with hormone. Cells were washed once with Dulbecco's modified Eagle's medium followed by incubation in Dulbecco's modified Eagle's medium/20 mM NaHepes (or Hanks' balanced salt solution where indicated) containing 10 mM LiCl for 10 min prior to hormone treatment. Cells were incubated for 40 min at 35°C in the presence or absence of the hormone. The reaction was terminated by the addition of 0.75 ml of ice-cold 10 mM formic acid to the cells. 0.75 ml of each reaction (1 well of a 24-well dish) was diluted in 3 ml of 20 mM ammonium hydroxide (pH of final solution, 8.0 -9.0) and applied to regenerated Dowex anionexchange columns (AG1-X8 resin, Bio-Rad 140-1444) followed by the addition of 1 ml of H 2 0. The combined sample and water wash was collected as the [

RESULTS
Previous characterization of the N54 mutant revealed that it possessed a conditional dominant negative phenotype with respect to the TSH-R and the ␤-adrenergic receptor (24). The nature of this phenotype is shown in Fig. 1A where the rat VIP-R (30) was co-expressed in COS-1 cells with either wild type or N54-␣ s . The expression of ␣ s resulted in elevated basal cAMP levels with the mutant being slightly more effective than wild type. Co-expression with wild type ␣ s was substantially more effective at increasing VIP-stimulated cAMP levels than co-expression with the N54-␣ s mutant. When basal cAMP levels were subtracted from those in the presence of VIP (Fig. 1A, right graph), N54-␣ s was seen to actually decrease VIP-stimulated cAMP levels. These results were similar to what is seen with the TSH receptor and the ␤-adrenergic receptors (Fig. 1B). Interestingly, whereas the expression of wild type ␣ s increased FIG. 1. Conditional dominant negative phenotype of N54-␣ s . A, effect of co-expression of VIP-R with either wild type ␣ s or N54-␣ s in COS-1 cells on basal and VIP-stimulated cAMP levels. Cells were transfected with 0.2 g of VIP-R cDNA with 0.6 g of either wild type (wt) or N54-␣ s cDNA. Cells were stimulated with 0.1 nM VIP. The graph on the right was generated by subtracting basal cAMP levels from the VIPstimulated cAMP levels. B, effect of wild type and N54-␣ s expression on hormone-stimulated cAMP levels. The graph was generated by subtracting wild type ␣ s and N54-␣ s basal cAMP levels from hormonestimulated cAMP levels and expressing them as percent of control (i.e. in the absence of ␣ s expression). Isoproterenol data represent experiments conducted in both COS and HEK 293 cells utilizing the endogenous ␤-adrenergic receptor present (n ϭ 5). TSH data represent experiments conducted in both COS and HEK 293 cells exogenously expressing the TSH-R (n ϭ 5). VIP data represent experiments conducted in both COS and HEK 293 cells exogenously expressing the VIP-R (n ϭ 3). Data are expressed as mean Ϯ S.E. Con, control.
hormone-stimulated levels, N54-␣ s decreased both TSH and VIP-stimulated cAMP levels by ϳ50%. Based on dose-response curves for the expression of wild type and mutant ␣ s (data not shown and Ref, 24), this appears to be a maximal effect of the mutant. The mutant also decreased stimulation of cAMP levels through the endogenous ␤-adrenergic receptor but to a lesser degree than for the transfected receptors. The lower inhibition seen in this case probably reflects the relative transfection efficiency, and stimulation through this receptor is probably blunted by ϳ50% as well. Fig. 1 and our previous results (24) indicate that N54-␣ s interferes with receptor stimulation of adenylyl cyclase. The N17-Ras homolog of N54-␣ s interferes with Ras signaling by binding unproductively to its upstream regulator (16,17,37). Signaling by G proteins is more complex than for the monomeric G proteins, and hence, there are more possible sites of action for N54-␣ s . In fact, the mutants of the cognate site in ␣ o and ␣ i (S47C) appear to have a dominant negative phenotype because they bind tightly to ␤␥ (25,26). Therefore, we sought to determine whether N54-␣ s had a mechanism analogous to its related ␣ i/o proteins or one similar to the more distantly related Ras protein.
Co-expression of Wild Type or N54-␣ s with Activated ␣ s in COS-7 Cells-One possible mechanism of action of a dominant negative effect of N54-␣ s would be for it to bind to adenylyl cyclase, thus blocking stimulation by wild type ␣ s . Although at first thought this mechanism might seem inconsistent with the fact that N54-␣ s has inherent basal activity (24), this is not necessarily true. For example, N54-␣ s could have a lower intrinsic efficacy for activating adenylyl cyclase than does wild type ␣ s . Thus, its binding to adenylyl cyclase could activate the enzyme to a low level but preclude its interaction with an activated wild type ␣ s with greater efficacy and generated by receptor activation of G s . In fact, our previous data could be consistent with the idea that N54-␣ s does have less efficacy than does a persistently activated ␣ s protein (24). To test whether this idea is probable, we determined whether N54-␣ s could effectively compete with an activated ␣ s (Q212L-designated ␣ s *) for stimulation of adenylyl cyclase. To establish conditions for this experiment, increasing amounts of ␣ s * cDNA were transfected with a fixed amount of TSH-R cDNA into COS-1 cells and cAMP levels were measured in the absence or presence of TSH ( Fig. 2A). As the amount of ␣ s * cDNA increased, the difference between hormone-stimulated and basal cAMP levels decreased. At 450 ng of ␣ s * transfected with the TSH-R, there was little hormone-stimulated activity, suggesting that the adenylyl cyclase in these cells was near maximally stimulated by the activated ␣ s mutant ( Fig. 2A).
To examine whether the N54 mutant could inhibit activated ␣ s * from stimulating adenylyl cyclase, ␣ s * cDNA (450 ng) was co-transfected alone or with an equal amount of either wild type or N54-␣ s (Fig. 2B). As seen in Fig. 2B, a representative experiment conducted in COS-1 cells, the expression of ␣ s * increased basal cAMP levels to a greater extent than did wild type or N54-␣ s . The co-expression of N54-␣ s with ␣ s * failed to appreciably inhibit ␣ s * from stimulating adenylyl cyclase under basal conditions or under hormone stimulation (Fig. 2B, compare ␣ s * to ␣ s * ϩ N54-␣ s ). A summary of experiments conducted in COS-1 cells is shown in Fig. 2C. Control immunoblots showed that mutant and wild type ␣ s were actually expressed at ϳ10 times the level of the activated ␣ s * protein in these experiments (Fig. 2D). Thus, at least at a 10-fold excess, N54-␣ s does not block the stimulation of adenylyl cyclase by ␣ s *. These studies did not support the idea that N54-␣ s blocks the ability of wild type ␣ s to stimulate adenylyl cyclase.
Interestingly, it was observed that transfection of ␣ s * in COS-1 cells resulted in an increase in the endogenous long form of ␣ s (Fig. 2D). Although this is a potentially interesting result, we have not yet followed it up. Because it was not clear how this induction affected the results of these experiments, we conducted the same experiment in HEK 293 cells, which do not show an increase in the expression of endogenous long form of ␣ s upon expression of ␣ s * (Fig. 2D). The effect of transfection of ␣ s * alone compared with ␣ s * plus N54-␣ s on basal and TSHstimulated cAMP levels in HEK 293 cells demonstrated that N54-␣ s did not prevent ␣ s * from stimulating adenylyl cyclase in these cells either (Fig. 2C).
Co-expression of Wild type or N54-␣ s with Adenylyl Cyclase II and III in COS-1 Cells-Another possible mechanism for the dominant negative effect of N54-␣ s might involve its interference with downstream effects of ␤␥. This could result from N54-␣ s irreversibly binding to ␤␥, as has been suggested to explain the phenotype of the analogous mutation in G␣ i (25). To determine whether the mutant interferes with the ␤␥ regulation of downstream effectors, experiments were conducted with co-expressed adenylyl cyclase isotypes (AC II or AC III) that differ with respect to ␤␥ regulation. AC II can be activated directly by ␤␥ dimers in the presence of activated ␣ s , whereas AC III is not regulated directly by ␤␥ (5, 6).
AC II or AC III were co-expressed with the TSH-R and with ␣ s (wild type or N54 mutant) in COS cells. Co-expression of AC II with the TSH-R increased TSH stimulation of cAMP accumulation 5-fold compared with transfection of the TSH-R alone (Fig. 3), whereas co-expression of AC III with the TSH-R only increased cAMP accumulation 3-fold in response to TSH. It is not clear whether these differences are due to differences in the level of expression of these two adenylyl cyclase isoforms, to differences in their inherent activities, or to differences in their regulation. Nevertheless, the differences in the inherent regulation of these two enzymes allowed us to test whether they differed also in their response to the presence of the ␣ s mutants. When wild type ␣ s was transfected with AC II, TSHstimulated cAMP accumulation was decreased. This is consistent with the idea that TSH maximally stimulated AC II by a dual mechanism involving both ␣ s and ␤␥ and that expressed ␣ s inhibited this maximal stimulation by binding ␤␥. In contrast, the expression of wild type ␣ s with AC III increased cAMP accumulation in response to TSH, compatible with the idea that ␣ s is rate-limiting for activation of AC III under these conditions. However, regardless of the response to wild type ␣ s , transfection of the N54-␣ s mutant with either AC II or AC III decreased cAMP accumulation in response to TSH (Fig. 3,  insets). This finding suggests that the effects of N54-␣ s on receptor regulation of cAMP are independent of the properties of the downstream adenylyl cyclase isoform producing the cAMP. This makes it unlikely that the mutant inhibits TSH stimulation of adenylyl cyclase by simply sequestering ␤␥ and preventing the effects of ␤␥ on downstream effectors. This idea is further supported by the fact that the mutant can inhibit TSH stimulation of adenylyl cyclase in HEK 293 cells, which reportedly lack an adenylyl cyclase that is stimulated by ␤␥ (38).

Effect of Wild type ␣ k or N54-␣ s on Signaling to Phospholipase-␤ through the ␣ 1B -Adrenergic Receptor in COS-7
Cells-In the case of the Ser 47 mutant of ␣ i , the analogous site to Ser 54 of ␣ s , it interferes not only with receptor regulation of ␣ i -mediated signaling but also with signaling mediated through other G proteins, even for receptors that do not regulate G i (25). This is one argument that the mutant works by sequestering ␤␥ and preventing upstream receptor heterotrimer coupling. The ␤-adrenergic receptor, the TSH receptor, and the VIP receptor all couple to G s . Presumably, a dominant negative effect of ␣ s mediated by its binding to receptor would not directly interfere with signaling through receptors that do not couple to G s . Therefore, we tested whether N54-␣ s blocked ␣ 1B -adrenergic receptor (␣ 1B -AR) signaling to phospholipase C␤ through G q . The ␣ 1B -adrenergic receptor is not known to couple to G s . COS-7 cells were used for these experiments, because they reportedly lack a phospholipase C␤ isoform that is stimulated by ␤␥ (39). The ␣ 1B -AR when expressed in COS-7 cells (Fig. 4A) had an EC 50 for phenylephrine of ϳ1 M (Fig.  4B). Increasing the expression of wild type ␣ s slightly decreased the ability of the ␣ 1B -AR to increase inositol phosphate levels (Fig. 4C). In contrast, increasing the expression of N54-␣ s had no effect on the ability of the ␣ 1B -AR to stimulate phospholipase C (Fig. 4C). The expression of ␣ s * failed to alter phenylephrine stimulation of PI turnover in COS-7 cells (Fig.  4D), indicating that the effects of wild type ␣ s cannot be accounted for by elevations in cAMP levels. Although these data are complicated by small effects of wild type ␣ s protein on signaling through phospholipase (discussed below), they clearly demonstrate that N54-␣ s does not block ␣ 1B -AR stimulation of phospholipase C through G q . This result supports the idea that N54-␣ s does not nonspecifically interfere with receptor regulation of downstream processes by sequestering ␤␥ dimers necessary for receptor G protein coupling.
Co-expression of Wild Type or N54-␣ s with the TSH-R in COS-7 Cells: Effect on Inositol Phosphate Levels-Another possible mechanism for the dominant negative action of the N54 mutant could be by its upstream non-productive interaction (sequestration) with receptor. If this mechanism was responsible, we might hypothesize that for receptors that activate more than one kind of G protein, N54-␣ s would not only block recep-

FIG. 3. Co-expression of different adenylyl cyclase isotypes with the TSH-R with either wild type or N54-␣ s in COS-7 cells.
All of the cells were transfected with 0.2 g of TSH-R. Where indicated, cells were transfected with either 0.6 g of AC II or AC III and with either 0.8 g of wild type ␣ s or N54-␣ s cDNA. Cells were stimulated with 3 nM TSH. Values represent the mean ϩ S.E. of an experiment conducted in triplicate. Inset graphs were generated by subtracting basal from TSH-stimulated cAMP levels. These data are typical of three experiments with similar results. V, vector.
tor stimulation of cAMP through G s but also signaling by that receptor through other G proteins. To investigate this possibility, we have taken advantage of the ability of the TSH-R to activate inositol phosphate production through G q/11 (7)(8)(9). Experiments were conducted in COS-7 cells, because they are thought to lack a phospholipase C␤ isoform stimulated by ␤␥ (39). COS-7 cells transfected with a fixed amount of TSH-R cDNA were co-transfected with increasing amounts of either wild type or N54-␣ s cDNA. Basal inositol phosphate accumulation was unaltered with increasing expression of either wild type or mutant ␣ s (Fig. 5A). However, in the presence of TSH, N54-␣ s was clearly more potent than wild type ␣ s at inhibiting hormone-stimulated PI turnover. Interestingly, the maximal inhibition of TSH-stimulated PI turnover by N54-␣ s was 50%, similar to the maximal effect of N54-␣ s on inhibition of TSH and VIP-stimulated cAMP levels. To rule out the possibility that the effects of ␣ s were mediated by cAMP, the TSH-R was co-expressed with ␣ s *, which increases basal cAMP levels to a greater extent than does either wild type or N54-␣ s . Co-expression of ␣ s * failed to inhibit the TSH-R from increasing IP levels in response to hormone (Fig. 5D). This eliminates the possibility that elevated basal cAMP could account for the decrease in TSH-induced IP production upon co-expression of either wild type or N54-␣ s . These data support the idea that N54-␣ s binds the TSH-R non-productively, blocking endogenous ␣ q/11 from being activated.

DISCUSSION
The N54-␣ s mutant was the first ␣ s mutation tested for its potential to have a dominant negative phenotype (23,24). Its design was based upon the previous description of the phenotype of the N17 mutation in Ras (15,40), a phenotype present in a large number of small G proteins with this substitution (18 -22). Despite this strong conservation of phenotype among the small G proteins and the conservation of the site of mutation to that in the heterotrimeric G protein ␣ subunits as well, it has been difficult to understand the phenotype of the ␣ s mutant. In part, this is because the N54-␣ s mutant has inherent basal activity (23), which the N17-Ras mutant does not (15), and this has compromised acceptance of the mutant as a negative regulator of adenylyl cyclase. Instead, we have more appropriately characterized the phenotype of N54-␣ s as a conditional dominant negative, because it is negative with respect to hormone stimulation but not with respect to adenylyl cyclase regulation (24). However, those studies did not disclose the mechanism of the dominant negative effect on hormone stimulation. This is a significant issue because a related analogous C47 or N47 mutation of ␣ o or ␣ i suggests a dominant negative phenotype of these proteins based upon their increased binding to ␤␥ dimers as opposed to increased binding to the upstream receptor (25,26,41). Inability to resolve these issues has compromised the usefulness of the ␣ s mutant as an experimental reagent compared with that of these mutations in the small G proteins. Here we resolve the phenotype of the N54-␣ s mutant and show that its dominant negative phenotype is different from that of the analogous ␣ i/o mutants. The differences in ␣ s and ␣ i/o described here suggest differences in the regulation of these proteins by guanine nucleotides and magnesium and/or the associated protein-protein interactions.
The N54-␣ s mutant could inhibit receptor stimulation of adenylyl cyclase by at least four different mechanisms. The mutant could 1) bind receptor non-productively, preventing receptor activation of wild type ␣ s , 2) bind adenylyl cyclase non-productively, blocking the activation of the enzyme by activated wild type ␣ s , 3) sequester ␤␥ released upon activation of wild type G s , preventing stimulation of adenylyl cyclase isoforms activated by ␤␥, or 4) sequester ␤␥ released upon activation of endogenous G proteins, preventing their re-association with endogenous ␣ subunits to form heterotrimers required for receptor recognition. The two downstream possibilities do not seem likely. The failure of the N54 mutant to appreciably inhibit ␣ s * from activating adenylyl cyclase, even when expressed in vast excess (Fig. 2), makes unlikely the possibility that the mutant works by binding to adenylyl cyclase non-productively. In addition, the ability of the N54 mutant to inhibit TSH stimulation of adenylyl cyclase III, which is not activated by ␤␥, suggests that the N54 mutant does not function simply by sequestering ␤␥ and preventing it from activating adenylyl cyclase. Finally, this mechanism could not account for the ability of the mutant to block TSH stimulation of inositol 1,4,5-trisphosphate production (Fig. 5).
N54-␣ s inhibition of TSH stimulation of cAMP accumulation and PI turnover in COS-7 cells suggests that the mutant binds non-productively to the TSH-R, preventing activation of G s and G q/11 . The possibility that N54 non-specifically inhibits PI turnover in COS-7 cells is ruled out because N54-␣ s fails to prevent ␣ 1B receptor-mediated activation of PI turnover. The mechanism characterized here for N54-␣ s is analogous to the proposed mechanism of action of the N17-Ras mutant, which most probably works upstream by sequestering the guanine nucleotide exchange factor (16,17,37). The Ras-guanine nucleotide exchange factor is functionally analogous to the G proteincoupled receptor in that it catalyzes the exchange of GDP for GTP on Ras. This conclusion is also compatible with previous studies supporting the idea that N54-␣ s does interact with receptors because receptor activation enhances the preference of the mutant for GDP (23).
There are important similarities and important differences in the phenotypes of the N54-␣ s mutant and its ␣ o and ␣ i counterparts. Although not characterized as extensively as the N54-␣ s mutant (23,24), the C48-␣ o (26), C47-␣ i (25), and the N47-␣ i (41) mutants suggest a pattern of biochemical changes similar to those originally described for the N17-Ras mutant (15). These include decreased affinity for activating guanine nucleotides, altered preference for GDP over GTP (or its analogs), and decreased sensitivity to magnesium ion. These are all compatible with the role of this conserved Ser residue in complexing Mg 2ϩ and the ␥ phosphate of GTP as disclosed in the x-ray structures of GTP-binding proteins (42)(43)(44). Outwardly, they all too have a similar phenotype as dominant negative inhibitors of signaling events. Importantly, however, there are real differences in the mechanism of the dominant negative phenotypes of the ␣ s and ␣ i/o mutants. The ␣ i/o mutants are nonspecific inhibitors of G protein signaling, acting both on receptors coupled to G i/o and on receptors coupled to other G proteins such as G q (25). This pattern was interpreted to be consistent with the ␣ i/o mutants irreversibly complexing ␤␥ dimers (25), although the effects on receptor interactions as well could probably not be ruled out. In contrast, the ␣ s mutant is shown here to be a direct inhibitor of signaling specifically through G s -coupled receptors, probably mediated primarily by its nonproductive binding to the receptor itself. Interestingly, the G q ␣ subfamily may be different still, because the ␣ q equivalent to N54-␣ s does not display a dominant negative phenotype at all (cited in Ref. 25). The different phenotypes of ␣ s , ␣ i , and ␣ q with the mutations in this conserved Ser suggest differences secondary to their regulation by magnesium and/or guanine nucleotides and influences the ability of the proteins to interact with other proteins. Probably, these differences affect not only the interactions of the mutants, but they also translate into differences in the consequences of regulation of the wild type proteins as well.
Why mutation of the Ser of the first guanine nucleotide binding consensus sequence should have functionally different effects in different G␣ subunit proteins is not clear. The crystal structures of ␣ s (45) and ␣ s in complex with a soluble adenylyl cyclase (46) disclose no grossly apparent differences in the interactions of this Ser with bound guanine nucleotide, Mg 2ϩ , or close elements of the protein as compared with the ␣ i proteins. However, this is not completely diagnostic of differences in the proteins because it has also been suggested that known biochemical differences exist in effector interactions of the G␣ proteins and ␣/␤␥ interactions. For example, whereas ␣ i requires only micromolar levels of free Mg 2ϩ for binding of GMP-P(NH)P (a hydrolysis-resistant GTP analog), ␣ s requires millimolar levels of free Mg 2ϩ for the binding of GMP-P(NH)P (47,48). Thus, differences in the interactions of Ser of the first guanine nucleotide binding consensus sequence may lead to subtle effects mediated by changes in side groups not immediately obvious in crystal structures. Alternatively, it is significant that there are not yet crystal structures for the G s heterotrimer or for the GDP-liganded ␣ s protein. It may be that the real differences in the G␣ subunits are in their interactions with GDP rather than GTP. Perhaps the ␣ s and ␣ i/o proteins differ specifically in the role of Mg 2ϩ in their interactions with GDP and that this results in differences in the regulation of their interactions with other proteins such as the G protein ␤␥ dimer.
The ability of N54-␣ s to block both TSH-activation of cAMP accumulation and PI turnover has important implications regarding the TSH-R binding site for ␣ s and ␣ q/11 . Point mutations of the TSH-R have been identified that cause constitutive activation of cAMP but not of PI turnover (49 -51), which in the simplest sense suggests that the binding of ␣ s and ␣ q/11 are separate or recognize different confirmations of the protein. An Ala to Ile mutation at position 623 (located in the third intracellular loop) of the TSH-R produces constitutive activation of cAMP (51); however, the substitution of Ala 623 with either Lys or Glu results in the loss of TSH-induced PI turnover but not TSH-induced cAMP formation (9). This would seem to suggest that the binding of ␣ s and ␣ q/11 overlap to some degree. Our data support the idea that different G proteins compete for binding of activated receptor and that one receptor population can activate multiple G proteins.
Recently, a dominant negative mutant of G␣ s containing three sets of separate mutations that affect distinct functions of G␣ s has been characterized in HEK 293 cells (52). The mutant contains five substitutions in the ␣ 3 ␤ 5 loop of region, which increase receptor affinity, decrease receptor-mediated activation, and disrupt adenylyl cyclase activation (53), in addition to mutations G226A, which increases the affinity of ␣ for ␤␥ (54,55), and A366S, which decreases affinity for GDP (56). The triple mutant inhibited the luteinizing hormone and calcitonin receptors from activating adenylyl cyclase in addition to inhibiting the calcitonin receptor from activating PI turnover (52). Those data and ours presented here, using an entirely unrelated dominant negative mutation, suggest that individual receptors have access to multiple G proteins and that different G proteins can compete with each other for binding to the same receptor.
The ability of N54-␣ s to inhibit TSH stimulation of PI turnover by 50% (Fig. 5) is strikingly similar to the ability of N54-␣ s to inhibit TSH/VIP stimulation of cAMP accumulation by 50% (Fig. 1). This phenomenon is not just a coincidence of any specific set of conditions, because it is true of multiple pathways regulated by a single receptor (i.e. TSH-R regulation of cAMP or PI turnover), multiple receptors acting on the same pathway (TSH-R, VIP-R and, probably, ␤-adrenergic receptor), and receptors expressed in multiple cell types (HEK293 cells and COS-7 cells) and whether or not the mutant simultaneously affects basal signaling (for cAMP it does, but for PI turnover it does not). Thus, this phenomenon appears to represent a basic property of how N54-␣ s interacts with receptor systems. These results suggest that there is something about the mutant that precludes it from blocking 50% of the signaling through unmodified G proteins whether they are similar to or different from the mutant itself. We would suggest that this phenomenon relates to some fundamental property by which receptors interact with G proteins. It has been demonstrated that GPCRs (␤ 2 -adrenergic (57), ␦ opioid (58), and metabotropic glutamate (59) receptors) can form dimers, suggesting that dimerization is involved in the activation of GPCRs. In addition, it has been demonstrated the GABA B receptors are assembled as functional heterotrimeric complexes containing GABA B R1 and GABA B R2 (60 -62). Although no evidence exists that either the TSH-R or VIP-R form dimers, perhaps the 50% inhibition observed with TSH and VIP stimulation of signal transduction is related to dimerization and subsequent activation of these GPCRs.