Constitutive Gi2-dependent Activation of Adenylyl Cyclase Type II by the 5-HT1A Receptor

The 5-HT1A receptor is implicated in depression and anxiety. This receptor couples to Gi proteins to inhibit adenylyl cyclase (AC) activity but can stimulate AC in tissues (e.g. hippocampus) that express ACII. The role of ACII in receptor-mediated stimulation of cAMP formation was examined in HEK-293 cells transfected with the 5-HT1A receptor, which mediated inhibition of basal and Gs-induced cAMP formation in the absence of ACII. In cells cotransfected with 5-HT1A receptor and ACII plasmids, 5-HT1A agonists induced a 1.5-fold increase in cAMP level. Cotransfection of 5-HT1A receptor, ACII, and Gαi2, but not Gαi1, Gαi3, or Gαo, resulted in an agonist-independent 6-fold increase in the basal cAMP level, suggesting that Gi2 preferentially coupled the receptor to ACII. The 5-HT1B receptor also constitutively activated ACII. Constitutive activity of the 5-HT1A receptor was blocked by pertussis toxin and the Gβγ antagonist, βCT, suggesting an important role for Gβγ-mediated activation of ACII. The Thr-149 → Ala mutation in the second intracellular domain of the 5-HT1A receptor disrupted Gβγ-selective activation of ACII. Spontaneous 5-HT1A receptor activity was partially attenuated by 5-HT1A receptor partial agonists with anxiolytic activity (e.g. buspirone and flesinoxan) but was not altered by full agonists or antagonists. Thus, anxiolytic activity may involve inhibition of spontaneous 5-HT1A receptor activity.

The 5-HT1A receptor functions as an inhibitory somatodendritic autoreceptor on serotonergic neurons of the raphe nuclei and as a post-synaptic receptor in a variety of serotonergic targets (1)(2)(3)(4). A number of partial agonists of the 5-HT1A receptor have been shown to synergize with serotonin reuptake blockers in treatment of depression (1, 4 -7). The antidepressant action of these compounds appears to involve both agonistmediated down-regulation of presynaptic 5-HT1A receptors and activation of the post-synaptic receptors (that are resistant to desensitization) to result in enhanced serotonergic neuro-transmission. By contrast, the anxiolytic actions of 5-HT1A partial agonists may involve inhibition of serotonergic activity by enhancing autoreceptor activation and reducing post-synaptic receptor action. Homozygous null 5-HT1A receptor mutant mice display enhanced serotonin release and anxiety-associated behaviors (8 -10), consistent with the idea that hyperactivity of the serotonin system by reduction in 5-HT1A autoreceptor activity results in anxiety disorders.
The 5-HT1A receptor mediates inhibitory signaling by coupling to pertussis toxin-sensitive (PTX-sensitive) 1 G proteins (G i and G o ) to mediate a variety of intracellular changes including inhibition of cAMP accumulation, activation of potassium channels, and inactivation of calcium channels (2,3). However, in hippocampal membrane preparations, 5-HT1A receptor activation stimulates adenylyl cyclase activity (11)(12)(13), an action that can be blocked by the specific antagonist WAY100,135 (11). Activation of the G i -coupled 5-HT1A or GABA-B receptors potentiated the electrophysiological actions of the G s -coupled ␤-adrenergic receptor in hippocampal CA1 neurons (14). The ACII protein has been localized to the dendrites and cell bodies of neurons of the hippocampus (15). This subtype of adenylyl cyclase is known to be conditionally stimulated by G␤␥ subunits derived from activation of G i /G o proteins (16,17). These data suggest that the 5-HT1A receptor may couple to, or potentiate G s -mediated actions in the central nervous system by activation of G␤␥-regulated adenylyl cyclases like ACII.
Recently, we have shown that modulation of G protein subunits by expression of antisense G␣ i1 subunit cDNA can switch inhibitory coupling of the 5-HT1A receptor to mediate a PTXsensitive stimulation of cAMP in GH4 pituitary cells, which express ACII endogenously (18). We hypothesized that in G␣ i1depleted GH4 cells, the 5-HT1A receptor may activate ACII via release of G␤␥ subunits derived from other G i /G o proteins. Using a cotransfection approach, we have identified a specific requirement for ACII in coupling of the 5-HT1A receptor to enhance AC activity. Furthermore, cotransfection of 5-HT1A receptor, ACII, and G␣ i2 specifically, resulted in an agonistindependent increase in cAMP that was not inhibited by receptor antagonists. Interestingly, 5-HT1A partial agonists but not full agonists or antagonists inhibited spontaneous coupling of the receptor to stimulate cAMP accumulation. Thus, part of the anti-anxiety activity of 5-HT1A ligands may be to silence the spontaneous activity of the 5-HT1A receptor.
cAMP Assay-24 h after plating cells in 6-well dishes, the cells were rinsed with DMEM and incubated at 37°C in 1 ml/well DMEM, 0.01% bovine serum albumin, 20 mM Hepes (pH 7.0), 100 M isobutylmethylxanthine for 15-20 min. Experimental compounds were added to triplicate wells as indicated. Media were recovered and centrifuged at 13,000 ϫ g (30 s) to remove floating cells, and the supernatant was recovered and stored at Ϫ20°C. Attached cells were extracted using reporter lysis buffer (Promega), centrifuged for 10 min at 4°C, and stored at Ϫ20°C. Media were assayed for cAMP by specific RIA, as described previously. Cell extracts were assayed for ␤-galactosidase activity to monitor transfection efficiency. Statistical analyses of the data were done using the GraphPad Prism software, and groups were compared using unpaired t test with Welsh's correction to determine significance (p Ͻ 0.05).
␤-Galactosidase Assay-The transfected cells were rinsed with PBS, resuspended in 200 l of reporter lysis buffer, incubated for 15 min at room temperature, scraped, frozen, and thawed to complete cell lysis. The lysates were centrifuged (14,000 rpm, 20 s), and the supernatant was recovered for measurement of ␤-galactosidase activity. Equal volumes (30 l each) of cell extract and 0.3 mM 4-methylumbelliferyl-␤-Dgalactoside substrate in 15 mM Tris (pH 8.8) were mixed gently, incubated in the dark at 37°C for 30 min, and the reaction was terminated upon addition of 50 l of Stop solution (300 mM glycine, 15 mM EDTA (pH 11.2)). The sample was transferred to 2 ml of Z buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl, 1 mM MgSO 4 ), and fluorescence was measured at EX ϭ 350 nm, EM ϭ 450 nm on a Perkin-Elmer LS-50 spectrofluorometer (Buckinghamshire, United Kingdom). The variation of ␤-galactosidase activity between wells observed within a transfection was Ͻ6%, and between transfections was Ͻ7%.
X-Galactosidase Staining-Cells were fixed in 2% formaldehyde, 0.2% glutaraldehyde in PBS for 10 min and rinsed 2-3 times with PBS and stained overnight using the X-galactosidase solution (5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, 2 mM MgCl 2 , 1 mg/ml Xgal in PBS) in a humidified container at 37°C. The proportion of stained cells was quantitated from representative photographic frames (2 photos/well). The transfection efficiency attained was 36.5 Ϯ 5.4% (mean Ϯ S.E.), and the variation in transfection efficiency between different transfections was 7.5% by this method.
Ligand Binding Assay-Membranes were prepared from cotransfected HEK-293 cells as described (23) and stored as pellets at Ϫ80°C. Membrane pellets were resuspended in TME (75 mM Tris, pH 7.4, 12.5 mM MgCl 2 , 1 mM EDTA), aliquotted (100 g in 50 l) in tubes containing 200 l of TME and [ 3 H]DPAT (20 nM) or [ 3 H]5-HT (50 nM) and incubated at room temperature for 40 min. The binding reaction was applied to Whatman GF/C filters under vacuum and rinsed three times with 3 ml of ice-cold 50 mM Tris (pH 7.4). The filters were placed in 3 ml of scintillation fluid (Ultima-Gold, Packard) and counted for 2 min using the direct-disintegrations/min program of the Tricarb TR2100 counter (Packard). 5-HT (10 M) was used to define nonspecific binding.

RESULTS AND DISCUSSION
Inhibitory Coupling of the 5-HT1A Receptor-The 5-HT1A receptor couples to inhibition of cAMP accumulation in most cell types. Inhibition of cAMP formation by 5-HT1A receptor activation was examined in HEK-293 cells transiently cotransfected with the G s -coupled dopamine-D1 receptor (20) and the 5-HT1A receptor (Fig. 1). In three experiments, dopamine induced a 7.6 Ϯ 0.9-fold increase in cAMP level that was inhibited by 49 Ϯ 3% upon activation of 5-HT1A receptors with DPAT. DPAT-mediated inhibition of dopamine-induced cAMP accumulation was reversed by the 5-HT1A receptor antagonist spiperone. Thus, the 5-HT1A receptor is coupled to inhibition of G s -mediated cAMP accumulation in HEK-293 cells.
5-HT1A Receptor Coupling to ACII-We addressed whether the 5-HT1A receptor was coupled to enhancement of cAMP levels upon cotransfection with an expression plasmid encoding ACII (Fig. 2). In HEK-293 cells transfected transiently with the rat 5-HT1A receptor cDNA alone ( Fig. 2A, first lane), addition of the selective 5-HT1A agonist DPAT (1 M) inhibited basal cAMP accumulation by 40%, as observed in other cell lines (24). By contrast, in cells cotransfected with 5-HT1A receptor and ACII cDNAs, DPAT induced a 1.5-fold increase in cAMP accumulation ( Fig. 2A, second lane; Table I), which was blocked by pretreatment with 50 ng/ml PTX (data not shown). DPAT had no effect in non-transfected cells or cells transfected with ACII alone. Thus, the 5-HT1A receptor couples to endogenous G proteins to stimulate adenylyl cyclase activity upon cotransfection with ACII, as observed for other G i -coupled receptor subtypes (16,(25)(26)(27). The 5-HT1A receptor may also weakly activate G␣ s , because activation of ACII requires both G␤␥ release from G i and minimal G␣ s activity (17). However in the absence of transfected ACII, DPAT-induced inhibition of basal cAMP is observed (Table I) indicating that the 5-HT1A receptor couples via G␣ i to inhibit the activity of endogenous AC subtypes in HEK-293 cells.
G i2 -dependent Constitutive 5-HT1A Receptor Activation-Upon cotransfection of 5-HT1A receptor, G␣ i2 , and ACII cDNAs, a pronounced agonist-independent increase in cAMP level was observed that was not altered by addition of the agonist, DPAT ( Fig. 2A, Table I). Although constitutive activity of the 5-HT1A receptor/G protein complex has been indirectly suggested by the spontaneous binding of GTP␥S in membranes from cells transfected with 5-HT1A receptors (28 -30), the present results provide the first direct evidence of agonist-independent coupling of the 5-HT1A receptor in the absence of GTP␥S. Constitutive enhancement of basal cAMP levels required cotransfection of 5-HT1A receptor, G␣ i2 , as well as of ACII cDNA (Fig. 2) and was blocked by Ͼ80% following pretreatment with 50 ng/ml PTX (Fig. 2B), further indicating the participation of G i /G o proteins. By contrast, little increase in cAMP was ob-served upon cotransfection with G␣ o .
The G protein specificity of agonist-independent stimulation of cAMP levels by the 5-HT1A receptor was examined further by co-transfecting equal amounts of 5-HT1A receptor, ACII, and either G␣ i1 , G␣ i2 , G␣ i3 , or G␣ o cDNAs in HEK-293 cells ( Table I). Cotransfection of receptor and ACII with G␣ i2 , but not G␣ i3 , G␣ i1 , or G␣ o , resulted in a 6-fold increase in basal cAMP levels compared with cells transfected with receptor and ACII (16% versus 100%). In multiple experiments, the rank order of G protein preference (G i2 Ͼ G i3 , G o , G i1 ) is similar to that identified for G protein interaction with the bacterially expressed 5-HT1A receptor protein (31). Thus, the extent of constitutive activation of the 5-HT1A receptor appears to reflect the extent of coupling between receptor and G protein and is most dependent on the G protein subtype G i2 .
Addition of the full agonist DPAT resulted in a small increase in cAMP level in cells not transfected with exogenous G proteins (Control , Table I), as observed above. DPAT induced no change or a small decrease in cAMP levels in cells cotransfected with G␣ i1 or G␣ i2 . In contrast, DPAT inhibited cAMP levels by 30% in cells transfected with G␣ o ( Table I), suggesting that this G protein mediates inhibitory receptor coupling in cells transfected with ACII. A slight DPAT-induced increase in cAMP level was observed with G␣ i3 but did not achieve significance. Thus, the spontaneous activity of the 5-HT1A receptor was specific for the presence of G␣ i2 and was insensitive to the full agonist DPAT.
Expression of 5-HT1A Receptor, ACII, and G Proteins-The relative inactivity of G␣ o (or G␣ i1 , G␣ i3 ) to mediate constitutive coupling of the 5-HT1A receptor could be because of inefficient expression of the receptor, ACII, or the G protein upon cotransfection in HEK-293 cells. The level of 5-HT1A receptors expressed in G o -and G i2 -transfected cells was not significantly different as detected by specific binding of [ 3 H]DPAT (Table II). Because the ACII subtype is regulated by activation of PKC, functional expression of ACII was detected indirectly by measuring the change in cAMP level induced by the PKC activator, TPA. In non-transfected HEK-293 cells, no response to TPA was observed, consistent with a lack of ACII immunoreactivity observed in these cells (15). In cells cotransfected with plasmids for 5-HT1A receptor, ACII and either G␣ o or G␣ i2 , a strong response to TPA (5-to 6-fold) was detected that did not differ between G␣ o or G␣ i2 (Table II). This result suggests that robust and equivalent expression of ACII was obtained in both transfections.
Although transfection efficiencies were equivalent for each G protein tested (35%, see "Experimental Procedures"), it re-FIG. 2. Constitutive 5-HT1A receptor activity in the presence of G i2 and ACII. HEK-293 cells were transfected transiently by calcium phosphate coprecipitation with 12 g each of the indicated plasmids, and cAMP production was measured. Transfection efficiency was monitored by cotransfection of ␤-galactosidase plasmid and quantitated as described (see "Experimental Procedures"). DPAT (1 M) was added to 5-HT1A-transfected cells acutely during the assay. Data are presented as mean Ϯ S.E. of triplicate samples for experiments that were repeated at least three times (see Table I). A, 5-HT1A receptor-mediated stimulation of ACII. Cotransfection of plasmids encoding the 5-HT1A receptor (5-HT1A), G␣ i2 (␣i2), or ACII led to enhancement of cAMP levels. B, PTX induced uncoupling of the 5-HT1A receptor from G i2 /ACII. Requirement for 5-HT1A, G␣ i2 , and ACII. Various plasmids, including G␣ i2 (␣i2) and G␣ oA (␣o) were transfected in equal amounts (12 g each). PTX (50 ng/ml) was present overnight in the indicated samples.  Fig. 2 in which HEK-293 cells were cotransfected with expression plasmids for 5-HT1A receptor, ACII, ␤-galactosidase, and either no G protein (Control), or G␣ o , G␣ i1 , G␣ i2 , or G␣ i3 as indicated. The level of basal cAMP (normalized to the level of cAMP in G␣ i2 -transfected cells) or DPAT (1 M)-induced change in cAMP (expressed as percent basal) is presented as the mean Ϯ S.E. of the number of independent experiments indicated in parentheses. Data were compared by unpaired t-test with ***p Ͻ 0.001 compared to control for basal cAMP; and *p Ͻ 0.05; **p Ͻ 0.01 compared to untreated for DPAT-induced cAMP.  (Fig. 3). In two independent transfections, protein extracts from non-transfected HEK-293 cells displayed an undetectable level of G␣ o immunostaining, whereas samples from cells transfected with G␣ o cDNA expression plasmid displayed robust expression of G␣ o that was proportional to the amount of protein/well (Fig. 3). By contrast, endogenous G␣ i proteins were detected in both non-transfected and transfected cells. In cells transfected with p3-G␣ i2 , a 4.8 Ϯ 0.75-fold (n ϭ 4) increase in G i2 protein level relative to control was observed. A smaller 2.0 Ϯ 0.2-fold control increase was detected in G␣ i2 -transfected cells using the less selective anti-G i -common antibody which also detected endogenous G i proteins in non-transfected cells. In summary, there was an equivalent induction of G␣ o and G␣ i2 proteins upon separate cotransfection with 5-HT1A receptor and ACII plasmids. Thus, the inactivity of G␣ o to mediate constitutive coupling of the 5-HT1A receptor to ACII was not because of inefficient expression of the G␣ o protein, 5-HT1A receptor, or ACII. Concentration-dependence for G␣ i2 Plasmid-The concentration dependence of p3-G␣ i2 cDNA for 5-HT1A receptor-mediated activation of ACII and inhibition of PGE1-stimulated cAMP levels was examined in parallel (Fig. 4). Upon cotransfection with 5-HT1A receptor and ACII plasmids, the p3-G␣ i2 plasmid induced a concentration-dependent enhancement of cAMP levels, with an EC 50 value of 1.7 g of G␣ i2 cDNA (Fig.  4A). The response appeared to saturate at a DNA concentration that was approximately equimolar for 5-HT1A receptor, G protein, and ACII plasmid (i.e. approximately 15 g). For comparison, the concentration dependence of G i2 cDNA for 5-HT1A receptor-mediated inhibition cAMP levels was examined. PGE1 induced a 3-fold increase in cAMP above the constitutive level of cAMP in HEK-293 cells transfected with 5-HT1A, ACII, and G␣ i2 (Fig. 4B). Increasing amounts of G␣ i2 inhibited PGE1mediated enhancement of cAMP levels only at the highest plasmid concentration (50 g). Thus, the requirement for G␣ i2 cDNA was ten-fold greater for agonist-independent inhibition of PGE1-stimulated cAMP than for activation of ACII by the 5-HT1A receptor. The concentration dependence on G␣ i2 cDNA suggests that the extent of constitutive activation of the 5-HT1A receptor depends on the content of G i2 and on the signaling pathway (i.e., stimulation versus inhibition of adenylyl cyclase) examined.
Role of the i2 Domain in 5-HT1A Receptor Coupling to ACII-To identify the receptor domain implicated in constitutive activation of ACII by the 5-HT1A receptor, we utilized a point mutant (5-HT1A-i2, Fig. 5) in the second intracellular loop (T149A) of the 5-HT1A receptor that couples to G␣ i -mediated inhibition of cAMP, but not to various G␤␥-mediated pathways such as activation of PLC␤2 or inactivation of calcium channels (23). The 5-HT1A-i3 is a receptor mutant with three point mutations that retains coupling to PLC-mediated calcium mobilization (32). When cotransfected with G␣ i2 and ACII, the 5-HT1A-i2 mutant failed to increase cAMP over the level obtained in the presence of PTX. By contrast, the 5-HT1A-i3 mutant displayed agonist-independent enhancement of cAMP levels similar to the wild-type 5-HT1A receptor (Fig. 5). The levels of receptor expressed were similar for each clone although the value for the 5-HT1A-i2 clone was slightly lower; this may reflect the lower affinity of this mutant for [ 3 H]DPAT (23). Thus, the second intracellular domain of the 5-HT1A receptor appears to dictate coupling to G␤␥ subunits resulting in activation of ACII, as well as other G␤␥-regulated effectors (33). Because the 5-HT1B receptor is structurally homologous to the 5-HT1A receptors, we tested whether this receptor subtype mediated agonist-independent activity (Fig. 5). Upon cotransfection with ACII and G␣ i2 , the 5-HT1B receptor mediated a 5-fold increase in cAMP levels that was inhibited by PTX. The smaller effect of the 5-HT1B receptor may reflect a lower constitutive activity compared with the 5-HT1A receptor because the levels of receptor expressed were similar. Other receptor Cells were cotransfected with 5-HT1A receptor, ACII, ␤-galactosidase and either G o or G i2 expression plasmids as indicated. ␤-Galactosidase activity was measured as an index of overall transfection efficiency and was equivalent between the transfections. For measurement of 5-HT1A receptor levels, membranes were prepared and subjected to ligand binding with [ 3 H]DPAT (20 nM), with 10 M 5-HT added to determine non-specific binding. Data are presented as mean Ϯ S.E. of three independent experiments. In separate transfections, the -fold induction over basal cAMP following addition of TPA (100 nM) was measured as an index of ACII content. In non-transfected cells, TPA did not alter significantly the basal cAMP level (0.82 Ϯ 0.20 -fold basal). Data represent the mean Ϯ S.E. of three independent experiments. subtypes like the dopamine-D1 (G s coupled) or the dopamine-D2S (G i /G o -coupled) did not increase cAMP in the absence of agonist (data not shown). Previous studies have indicated that the dopamine-D2S receptor does not couple to G s in the presence of ACII (16), suggesting that the activation of ACII by 5-HT1A receptors does not result from basal activation of G s by other receptors endogenous to HEK-293 cells. This supports the hypothesis that the 5-HT1A receptor does couple to Gs (weakly) in the HEK-293 cells, despite the inability to detect biochemically an interaction between the receptor and G s (28, 31, 34 -36). Although 5-HT1A and 5-HT1B receptors inhibit cAMP synthesis in most tissues and cell lines, these results indicate that, in cells where ACII is predominant, these receptors spontaneously couple to stimulate cAMP accumulation.
Inhibition of Constitutive Activity of the 5-HT1A Receptor-The activity of a variety of ligands and protein inhibitors to modulate 5-HT1A receptor coupling to G i2 and ACII was examined and compared with the action of PTX, which entirely uncouples the receptor (Fig. 6). Because ACII is conditionally activated by G␤␥ subunits from G i /G o proteins (17), the importance of G␤␥ subunits was examined by cotransfection of an equal amount of expression plasmid containing the carboxylterminal domain of GRK2 (␤CT), a G␤␥ antagonist (37,38). ␤CT inhibited spontaneous coupling of the 5-HT1A receptor with maximal inhibition equivalent to that achieved with PTX. This suggests that the 5-HT1A receptor couples spontaneously to the G i2 heterotrimer to release G␤␥ subunits, resulting in a ␤CT-sensitive activation of ACII. Transfection of G␣ i2 apparently augments the activity of endogenous G␤␥ subunits to couple to ACII in the presence of the 5-HT1A receptor, presumably by increasing the level of G i2 heterotrimer. Co-regulation of G␣ and G␤␥ subunit protein levels has been observed in G␣ i2 nullizygous mice that are depleted of G␣ i2 and specific G␤␥ subunits (39).
The activity of 5-HT1A agonists and antagonists was tested in cells transfected with 5-HT1A receptor, G i2 , and ACII (Fig.  6). Ligands for the 5-HT1A receptor have been previously characterized as full or partial agonists or neutral antagonists (40 -43). Several partial agonists (e.g. buspirone, BMY-7378) of the 5-HT1A receptor possess anti-anxiety activity (5, 44 -46), and 5-HT1A agonists and antagonists potentiate the anti-depressant response to serotonin reuptake blockers (1, 4, 6, 7). Full agonists (e.g., DPAT, 5-HT) and antagonists (spiperone, (ϩ)WAY100,135, NAN-190, pindolol) lacked or had minimal inhibitory activity compared with control. Lack of inhibition by antagonists suggests that G i2 -dependent constitutive activation of 5-HT1A receptor is indeed agonist-independent and not mediated by endogenous 5-HT. Spiperone has been characterized as an inverse agonist that induces a small (30%) decrease in 5-HT1A receptor-induced GTP␥S binding to Chinese hamster ovary cell membranes (29). The lack of effect of spiperone or (ϩ)WAY100,135 to inhibit spontaneous coupling of the 5-HT1A receptor was not due to their inactivity in the HEK cells because both spiperone (Fig. 1) and (ϩ)WAY100,135 (data not shown) blocked 5-HT1A-mediated inhibition of cAMP formation. Determination of inverse agonism in membrane preparations in the presence of the high affinity G protein ligand GTP␥S may not reflect modulation of receptor-G protein coupling to specific effectors in intact cells which contain GTP/ GDP. The lack of inverse agonist activity of spiperone in coupling to G i2 /ACII suggests that as observed for agonists, inverse agonists may show greater efficacy for some effectors but not others (47)(48)(49). Depending on the effector studied, agonists of the 5-HT1A receptor can have different efficacies of coupling. A "tight" coupling of the 5-HT1A receptor to inhibition of cAMP is observed, and all agonists mediate some response; however, for calcium mobilization or pH change, a "loose" coupling is apparent in which some agonists are ineffective (40,50). Based on our results, spiperone induced a conformation of the 5-HT1A receptor that permits coupling to G i2 /ACII but inhibits G␣ i -mediated inhibition of G s -stimulated adenylyl cyclase activity.
The partial agonists buspirone, BMY-7378, and especially flesinoxan, all demonstrated inhibitory activity, which was maximal at 1 or 10 M. By comparison, pretreatment with PTX or cotransfection with ␤CT induced a greater reduction in cAMP levels, reflecting a more complete uncoupling of the receptor by these agents. The partial agonism of these ligands to incompletely stimulate the inactive 5-HT1A receptor, and to inhibit partially the constitutively active receptor, suggests that the partial agonists promote an incompletely active conformation of the 5-HT1A receptor upon binding to the receptor. Furthermore, there appears to be a correlation between the clinical effectiveness of these compounds as anti-anxiety drugs  6. Inhibition of 5-HT1A constitutive activity by uncoupling agents and partial agonists. All cells were cotransfected with equal amounts of 5-HT1A receptor, G␣ i2 , and ACII plasmids and subjected to cAMP assay in the presence of the indicated compounds except for control (no additions), ϩPTX (overnight pretreatment with 50 ng/ml PTX), and ␤CT (cotransfected with addition of an equal amount of ␤CT plasmid (21)). The following compounds were used: (ϩ)WAY100,135 (WAY); Flesinoxan (Fles.); BMY-7378 (BMY); buspirone (Busp.); spiperone (Spip.); NAN-190 (NAN); pindolol, (Pind.). The data are expressed as percent control, with addition of 0.1% Me 2 SO as the vehicle control for spiperone, NAN-190, and pindolol. The data represent mean Ϯ S.E. of 3-5 independent assays with statistical significance compared with control as indicated: *p Ͻ 0.05; **p Ͻ 0.01. and their ability to reduce coupling of the 5-HT1A receptor to ACII.
It has been proposed that anxiety results in part from a hyperactivity of the serotonin system and that anxiolytic 5-HT1A partial agonists inhibit serotonergic neurotransmission by activating the presynaptic autoreceptor (51); hence the knockout mice that lack 5-HT1A receptors display both serotonergic hyperactivity and anxiety-related behavior (8 -10). Our results suggest that in addition to partial activation of the presynaptic 5-HT1A autoreceptor to inhibit raphe firing, antianxiety agents may also inhibit constitutive, as well as 5-HTinduced activation of postsynaptic 5-HT1A receptors. It is tempting to speculate that 5-HT1A receptor ligands that inhibit constitutive receptor activity may have potentially greater therapeutic efficacy as anti-anxiety or antidepressant agents than ligands that lack this activity. The cotransfection paradigm presented here for detecting inhibitory activity at the 5-HT1A receptor may provide an effective screening method for identifying anti-anxiety agents.
In summary, we have identified stimulatory coupling of the 5-HT1A receptor to adenylyl cyclase in the presence of ACII that becomes agonist-independent upon co-transfection with G␣ i2 . Coupling to ACII required mobilization of G␤␥ subunits and was mediated by a putative G␤␥ coupling domain in the i2 loop of the 5-HT1A receptor. Partial 5-HT1A receptor agonists that have anti-anxiety properties inhibited spontaneous receptor coupling to ACII, whereas full agonists or antagonists lacked this activity; thus, inhibition of constitutive 5-HT1A receptor activity may contribute to the therapeutic actions of these compounds.