Elaine Sanders-Bush and Hervé Canton

Two major receptor-linked signal transduction pathways exist: a multistep enzyme mediated pathway and a direct regulation of ion channels. Both require a guanine nucleotide triphosphate (GTP)-binding protein (G protein) to link the receptor to the effector molecule. The sequence of steps involved in enzyme-dependent biochemical signaling is: cell-surface receptor G protein effector enzyme second messenger protein kinase phosphoprotein. This multistep scheme applies to receptors linked to activation and inhibition of adenylate cyclase and to activation of phospholipase C. These enzyme-dependent pathways lead to amplification of cellular signals. Each step involves proteins that exist in multiple forms. For example, G proteins are composed of a, b, and g subunits, each of which exists in multiple isoforms. More than 15 a subunits and at least three b and g subunits have been identified, leading to a tremendous diversity in G proteins. Multiple isozymes of both adenylate cyclase and phospholipase C have been found, and more than one isoform is involved in signal transduction. In addition, protein kinase A, protein kinase C, and calcium/calmodulin-dependent kinase exist as multiple isoforms. A final level of complexity, which may serve to integrate various signals, is protein phosphorylation. The protein substrates are numerous, and it is impossible to list them. Key proteins regulated by phosphorylation include neurotransmitter and growth factor receptors, G proteins, protein kinases, protein phosphatases, ion channels, neurotransmitter synthetic and metabolic enzymes, transport molecules, and DNA transcription factors. This brief discussion serves to illustrate the multitude of cellular responses that may result when neurotransmitters such as 5-hydroxytryptamine (5-HT) activate receptors that couple to the adenylate cyclase and phospholipase C pathways. 5-HT receptors are numerous, and at least one receptor subtype is linked to each of the major signal transduction pathways (Table 1) (see also Molecular Biology of Serotonin Receptors: A Basis for Understanding and Addressing Brain Function and Serotonin Receptor Subtypes and Ligands).

Stimulation of adenylate cyclase was the first signal transduction pathway to be linked to 5-HT—in invertebrates as well as in immature rat brain. However, the specific receptor mediating activation of adenylate cyclase has only recently been identified. This receptor, named 5-HT4, was first characterized in mouse collicular neurons (21). It is also found in hippocampus and in peripheral tissues such as guinea-pig ileum, rat esophagus, and human atrium (12). In the past 2 years, a second receptor (5-HT6) has been conclusively shown to couple positively to adenylate cyclase by cloning and functional expression in cell lines (51). Abstracts of recent or upcoming meetings suggest that still other 5-HT receptor subtypes will be linked to activation of adenylate cyclase.

Proximal cellular events that result from an increase in cyclic 3¢,5¢-adenosine monophosphate (cAMP) include activation of protein kinase A, which, in turn, regulates the activity of cellular proteins by phosphorylation (Fig. 1). Electrophysiological studies in collicullar and hippocampal neurons suggest that one consequence of activation of 5-HT4 receptors is an inhibition of a voltage-dependent K+ current (6). cAMP was proposed as the mediator based on the following evidence: A cell-permeated analogue of cAMP and forskolin, a direct activator of adenylate cyclase, mimic the effects of 5-HT; a specific protein kinase A inhibitor blocks the effect of 5-HT; and, lastly, a phosphatase inhibitor, okadaic acid, potentiates 5-HT. In Fig. 1, the K+ channel is depicted as the phosphorylation substrate; however, it is not clear whether this is the case or whether another protein is phosphorylated and, in turn, depresses K+ channel activity. Depression of K+ current may lead to depolarization, calcium influx, and subsequent enhancement of neurotransmitter release. Such events are known to be important in 5-HT-induced changes in synaptic plasticity in Aplysia. It is intriguing to consider that similar mechanisms may play a role in synaptic plasticity in vertebrates.

5-HT1 receptors are a large family of receptors that are negatively coupled to adenylate cyclase via the Gi family of G proteins (Fig. 2). Additional subtypes in addition to those listed in Table 1 are suggested based on functional assays. The 5-HT1A receptor is the most well-characterized member of this family. In addition to adenylate cyclase coupling, 5-HT1A receptors are linked directly to voltage-sensitive K+ channels via a Gi-like protein (7), with no intervening second messenger signaling (Fig. 2). Dual coupling with both adenylate cyclase and K+ channels is now recognized as a hallmark of Gi-linked receptors and is likely to occur with other members of the 5-HT1 receptor family. In addition, direct coupling to L-type Ca2+ channels has been described as a third signal transduction pathway for the Gi-linked family of receptors (Fig. 2). Activation of Gi-linked receptors leads to enhancement of K+ channel activity and, conversely, blunting of Ca2+ channel activity. One important question is whether these different signal transduction pathways, which are mediated by Gi-like proteins, reflect multiple signaling mechanisms for a single receptor or multiple receptor subtypes that cannot be differentiated with the available drugs. There is good evidence that a single Gi-linked receptor can in fact couple to all three pathways: The cloned a2-adrenergic receptor, expressed in pituitary GH3 cells, couples to inhibition of adenylate cyclase, activation of K+ conductance, and inhibition of a calcium current (67). It is, however, possible that not every receptor couples to all three signals in the native state. Considering the diversity of pertussis-toxin-sensitive Gi-like proteins, it is also possible that different members of the Gi family are involved in the coupling to each of these signaling pathways. Furthermore, evidence suggests that different G proteins may couple different receptors to the same effector. In elegant experiments using antisense strategies, different Go subtypes were found to couple muscarinic and somatostatin receptors to inhibition of calcium currents (43). This diversity at several key points in the signal transduction pathway portends tremendous complexity in Gi-linked responses.

5-HT1A receptors on raphe cells function as somatodendritic autoreceptors, depressing neuronal firing rate when activated (for review see ref. 31). The mechanism involves membrane hyperpolarization elicited by increased K+ conductance; 5-HT1A receptors have also been characterized at postsynaptic sites, such as hippocampus (74) and cortex (9). At these sites, 5-HT1A receptor activation elicits hyperpolarization by enhancing K+ channel activity. Rodent 5-HT1B receptors and the human homologue, 5-HT1Db, function as axon terminal autoreceptors (31), where inactivation of Ca2+ channels may mediate the inhibition of 5-HT release. A significant portion of 5-HT1B receptors are localized on postsynaptic structures where their function is unknown. Interestingly, although 5-HT1B receptor-mediated inhibition of adenylate cyclase is pertussis-toxin-sensitive (52), studies in brain slices suggest that the terminal 5-HT1B receptors in hippocampus regulating neurotransmitter release are not blocked by pertussis toxin (11). These results suggest that axon terminal 5-HT1B autoreceptors may not be coupled to a Gi-like protein.

The functional correlates of inhibition of adenylate cyclase are not well-defined. Inhibition of protein kinase A with a subsequent reduction in the phosphorylation of its substrates would presumably alter cellular activity because many crucial proteins are regulated by changes in their phosphorylation state. To explore the consequences of 5-HT1A-receptor-mediated inhibition of adenylate cyclase, Andrade (5) recently looked at the functional interaction of 5-HT1A receptors and b-adrenergic receptors linked to activation of adenylate cyclase in single pyramidal cells in the CA1 region of the hippocampus. Because these two receptors have opposing effects on adenylate cyclase activity, 5-HT1A receptor activation was expected to inhibit b-adrenergic-receptor-induced reduction in afterhyperpolarization, which is mediated by activation of adenylate cyclase. Surprisingly, the opposite was found: 5-HT1A receptor activation enhanced the b-adrenergic-receptor-mediated response. An explanation for this paradoxical effect may be related to the recent evidence that G-protein bg subunits can regulate adenylate cyclase activity (68). Interestingly, although Gbg activates type II and type IV adenylate cyclase, it inhibits type I adenylate cyclase. Several other adenylate cyclase isoforms are insensitive to Gbg. This isoform-specific regulation of adenylate cyclase suggests remarkable flexibility. Thus, Gi-linked receptor activation by increasing free Gbg could activate or inhibit adenylate cyclase depending on the isoform of adenylate cyclase present in the particular cell. In hippocampal pyramidal cells, where 5-HT1A receptors potentiate a b - adrenergic -receptor / adenylate - cyclase-mediated response (5), the mechanism may be release of Gbg, which, in turn, activates type II adenylate cyclase. In other cell types, where type I adenylate cyclase predominates, activation of 5-HT1A and other Gi-linked receptors may inhibit adenylate-cyclase-mediated responses. Such heterogeneity may explain the conflicting reports that appeared in the early literature of 5-HT1A-receptor-mediated inhibition and activation of adenylate cyclase in different preparations.

Recombinant 5-HT1A receptors expressed in transfected cells have been found to enhance phosphate uptake (61) and to activate Na+/K+-ATPase (50). Although adenylate cyclase inhibition occurs in these cells, this signal transduction pathway apparently does not mediate these effects. Rather, a previously unrecognized, alternative signal transduction pathway, activation of phospholipase C appears to be involved. The biological significance of this alternative signaling will be discussed later.

One consequence of 5-HT1B-receptor-linked inhibition of adenylate cyclase may be a mitogenic response. In fibroblasts, 5-HT acts via a pertussis-toxin-sensitive G protein to increase DNA synthesis in synergy with fibroblast growth factor (65). Although pertussis-toxin-sensitive inhibition of adenylate cyclase occurred in these cells, evidence suggests that the mitogenic effect of 5-HT in smooth muscle cells is independent of this response (42). One possible mechanism may be pertussis-toxin-sensitive coupling to phospholipase C, which, for the 5-HT1A receptor, predisposes fibroblasts to enhanced proliferation by serum-derived factors (1).

Phospholipase C (PLC), a membrane-bound enzyme, catalyzes the degradation of the inositol lipid, phosphatidylinositol 4,5-bisphosphate (PIP2), with the production of inositol 1,4,5-triphosphate (IP3) and diacylglycerol (DAG) (Fig. 3). IP3 mobilizes Ca2+ from an intracellular storage site by interacting with specific receptors. Ca2+ induces multiple responses in the cell, including activation of calcium/calmodulin-dependent protein kinases, enzymes which phosphorylate/dephosphorylate key protein substrates in the cell. DAG activates another kinase family, protein kinase C (PKC). PKC regulates numerous processes of cell function. DAG is also hydrolyzed by a specific lipase to release arachidonic acid with the subsequent formation of prostaglandins and prostacyclins. Thus, PLC activation induces diverse changes in the cell, leading to the regulation of many cellular processes.

The newly named 5-HT2 receptor family is comprised of previously characterized receptors that are all coupled primarily to PLC (Table 1). The first evidence of a putative coupling of 5-HT receptors to phosphoinositide breakdown came from studies of insect salivary gland. Subsequently, it was shown that 5-HT is able to stimulate PLC activity in slices of rat cerebral cortex. Extensive pharmacological characterization of this response demonstrated the involvement of the 5-HT2A binding sites (64). Interestingly, in newborn rats, 5-HT induced a higher level of phosphoinositide hydrolysis than in adult rats (18). The mechanism of this supersensitivity is still unknown. Also, the exact identity of the G protein involved in 5-HT2A-receptor-mediated activation of PLC remains to be determined. The Gq family of G proteins activates PLC in a pertussis-toxin-insensitive manner. Some investigators have found the stimulation of phosphoinositide hydrolysis by the 5-HT2A receptors to be insensitive to pertussis toxin, supporting the conclusion that a Gq-type protein is involved (8, 38). In contrast, pretreatment with pertussis toxin reduces 5-HT-induced formation of inositol phosphates in cerebellar granule cells (4), suggesting that a pertussis-toxin-sensitive protein of the Gi family couples 5-HT2A receptors to PLC in the immature cerebellum. Thus the 5-HT2A receptor does not agree with the precept that a specific receptor couples with a single G protein. As discussed later, it is becoming increasingly evident that the nature of receptor–G-protein coupling is subject to variation, dependent not only on the structure of the receptor, but also on the particular cell type. In addition, multiple isozymes of PLC exist in the brain. Each one could be activated by different subunits of the G protein: Signals that activate the Gq family could result in the activation of PLC-b1 and PLC-b3 by the corresponding Ga subunits, while the bg subunits of Gi or Go could activate PLC-b and PLC-d1 isozymes (63).

Stimulation of phosphoinositide hydrolysis in rat choroid plexus is not mediated by the 5-HT2A receptor. In this structure the 5-HT-induced increase in PLC activity results from the activation of the 5-HT2C receptors (see ref. 64 for review). The only other area of the brain in which 5-HT2C-receptor-mediated phosphoinositide hydrolysis has been definitively characterized is the immature rat hippocampus (19). Recently a new receptor belonging to the 5-HT2 family was cloned: 5-HT2B receptors (29, 44). Present in the stomach fundus, this receptor, when expressed in AV-12 cells, increases the production of IP3. Another 5-HT receptor family, the 5-HT3 receptor, has also been linked to the phosphoinositide hydrolysis cascade as discussed in detail later.

The functional correlates of activation of PLC by 5-HT2 receptors are multiple. An increase in the intracellular concentration of calcium induces a rapid Cl- current through a Ca2+-dependent chloride channel when receptors are expressed in Xenopus oocytes (Fig. 3). This response has been characterized for the three members of the 5-HT2 receptor family and seems to be mediated through the PLC–IP3 pathway (29, 47, 59). 5-HT2C-receptor-mediated activation of chloride current in oocytes involves a pertussis-toxin-sensitive G protein (20) and an upstream calcium/calmodulin-dependent protein kinase (70). 5-HT2A receptor activation also induces the closing of a K+ channel, leading to a depolarization of the cell. This effect (3) is mediated by a G protein. It will be interesting to identify the G protein involved and determine if a direct action inhibits the K+ conductance as suggested by Aghajanian (3) or if a PLC-generated product is involved. More is known about putative mechanisms mediating the closing of K+ channels via activation of the 5-HT2C receptor. In Xenopus oocytes coexpressing 5-HT2C receptors and a K+ channel cloned from the brain, the suppression of K+ conductance by 5-HT involves a calcium/calmodulin-activated phosphatase (34). It was postulated that the activated phosphatase dephosphorylates the K+ channel, leading to its closing as illustrated in Fig. 3. Recovery from suppression seems to be due to the action of a protein kinase, because the protein kinase inhibitor H-7 blocked recovery (34). Cloned 5-HT2C receptors have been expressed in A9 cells as well as Xenopus oocytes and have been found to regulate Cl- and K+ conductance. The exact pathways involved are not completely elucidated and may vary with the model used. For example, both calcium-dependent and calcium-independent mechanisms have been found to regulate K+ currents (34, 55). These results illustrate that a cloned receptor expressed heterologously may couple to multiple signal transduction pathways (see later discussion). Recent patch-clamp studies of choroid plexus (37) demonstrated that 5-HT activates Cl- currents and inhibits K+ currents in choroid plexus. It will be interesting to determine if the intracellular pathways in the choroid plexus epithelium are comparable to those described for oocytes.

One of the cellular events controlled by activation of the 5-HT2C receptor may be ion exchange between the central nervous system and the cerebrospinal fluid. 5-HT2C receptors are expressed at high density in the choroid plexus and are concentrated on the apical surface of the epithelial cells (37) in direct contact with the cerebrospinal fluid. Thus, 5-HT2C receptors are located suitably to regulate the composition of the cerebrospinal fluid. In choroid plexus epithelial cells in primary culture, 5-HT regulates the level of transferrin by activation of 5-HT2C receptors (24). Transferrin is an iron carrier protein and also has growth-factor-like actions. The choroid plexus may be an important source of transferrin and iron for brain cells, suggesting a role for the 5-HT2C receptor in brain development and homeostasis. 5-HT2A and 5-HT2C receptors expressed heterologously in fibroblasts activate Na+/K+/2Cl- co-transport (49), but this function could not be reproduced in native choroid plexus epithelial cells. Another important property of the 5-HT2C receptor is its ability to function as a proto-oncogene when expressed in non-neuronal cells such as NIH-3T3 fibroblasts (39). The 5-HT2A receptor similarly acts as a proto-oncogene in NIH-3T3 fibroblasts (40). Introduction of functional 5-HT2A or 5-HT2C receptors into these cells results in the generation of transformed foci. Moreover, generation and maintenance of the transformed foci requires continued activation of the receptor by 5-HT. Nevertheless, when expressed in other types of fibroblast cell lines (41) or in choroid plexus epithelial cells (24), the 5-HT2C receptor does not induce a transformed phenotype, thus distinguishing this protein from strong dominantly acting oncogene products (41).

The 5-HT3 receptor differs from the other 5-HT receptors. The receptor itself forms an ion channel that regulates ion flux in a G-protein-independent manner. The 5-HT3 receptor is a member of the large family of ligand-gated ion channels of which the nicotinic cholinergic receptor is the prototype. 5-HT3 receptors are unique among ligand-gated ion channels because only a single subunit has been found (48), although an alternative splice variant with a six-amino-acid deletion in the cytoplasmic loop has been recently described (35). Many drugs exist that interact specifically with the 5-HT3 receptor, thus facilitating studies of its function. Results obtained in vivo suggest multiple 5-HT3-like receptors, and the search goes on for other subunit proteins that might explain this pharmacological heterogeneity.

5-HT3 receptors were first found on peripheral autonomic, sensory, and enteric neurons, where they mediate excitation. 5-HT3 receptors in brain are primarily localized on nerve terminals (58)), where they function in the regulation of the release of neurotransmitters, including 5-HT (30). The cloned 5-HT3 receptor protein forms a homomeric multisubunit protein that regulates the gating of cations (48), and it presumably mediates the rapid, transient depolarization that occurs on 5-HT3 receptor activation (56).

5-HT3 receptors have been shown to activate phosphoinositide hydrolysis in the medial prefrontal cortex and entorhinal cortices (22), suggesting another family of 5-HT3 receptors linked to G proteins. Thus, 5-HT3 receptors may be analogous to the glutamate receptors, which were first characterized as ionotropic receptors and later shown to also exist as G-protein-linked (metabotropic) receptors. It is difficult, however, to rule out that 5-HT3-receptor-mediated phosphoinositide hydrolysis is secondary to an influx of Ca2+ via the ligand-gated receptor, rather than being an independent signal transduction pathway. Furthermore, there is no convincing evidence of a role for a G protein in the phosphoinositide hydrolysis response. On the other hand, the agonists' and antagonists' profiles for the phosphoinositide-hydrolysis-linked receptor (22) differ somewhat from the profiles for the 5-HT3 ionotropic receptors. Moreover, electrophysiological studies have shown that 5-HT3-like receptors in the medial prefrontal cortex produce a slow depression of cell firing (72), rather than the fast activation of firing that has been described for the ligand-gated 5-HT3 receptor. Thus, both biochemical and electrophysiological results suggest that 5-HT3 receptors are heterogeneous, with respect to subtypes and intracellular signal transduction mechanisms. However, more studies are needed to determine conclusively whether the multiple 5-HT3 receptors include members in both the ionotropic and metabotropic receptor superfamilies. For example, studies in cell-free systems are required in order to demonstrate direct coupling of the 5-HT3 receptor to PLC.

In the early literature, it was found that 5-HT stimulates the activity of phospholipase A2 in membranes of guinea-pig cerebral cortex. Phospholipase A2 releases arachidonic acid, resulting in the production of arachidonic acid metabolites by lipoxygenase and cycloxygenase pathways with the formation of eicosanoids. More recently, it has been found that 5-HT stimulates the production of arachidonic acid in a number of brain tissues—including hippocampal neurons, where its action is mediated by the 5-HT2A receptor subtype (28). Evidence was presented that this effect involves a direct activation of phospholipase A2 and is independent of the release of arachidonic acid induced by hydrolysis of DAG, which occurs when a receptor is coupled to phosphoinositide hydrolysis (28). The mechanism of this apparently direct activation of phospholipase A2 by 5-HT and/or its mediation by a specific G protein is still unknown.

Recently, it has been shown that the 5-HT2C receptor is linked not only to activation of PLC, but also to an elevation of cyclic-3¢,5¢-guanosine monophosphate (cGMP) in the choroid plexus (33). The mechanism presumably involves the activation of phospholipase A2 and the release of arachidonic acid, a potent activator of guanylate cyclase. Other pathways might also contribute to elevation of arachidonic acid, such as activation of PLC by the 5-HT2C receptor and subsequent liberation of arachidonate by hydrolysis of DAG. 5-HT also has been shown to stimulate the formation of cGMP in a clonal cell line of glial origin (54) by a mechanism which could involve the 5-HT2A receptor and an increase of intracellular calcium mediated by PLC. A rise in the level of cGMP apparently mediated by the 5-HT3 receptor was described in neuronal cell lines (62), the mechanism of which may involve the production of arachidonic acid and/or formation of nitric oxide. Because the elevation of cGMP by 5-HT does not seem to involve a direct activation of guanylate cyclase but rather is the consequence of interacting pathways, this may be an example of cross-talk between the various signal transduction mechanisms after activation of 5-HT receptors (see later section for a further discussion of cross-talk).

Cloned receptors expressed in heterologous cell lines (cells that have been genetically engineered to express receptors that are not naturally present) have many advantages, including a pure population of receptors that can be studied without problems associated with receptor interactions or drug nonspecificity. These scientific factors combined with practical considerations, such as improved sensitivity and convenience, are responsible for the surging popularity of transfected cell lines. However, artifacts such as aberrant G-protein coupling or receptor cross-talk may be a consequence of overexpression of receptors in cells that do not normally synthesize those receptors (see next section). The pharmacological properties of the receptor are generally thought to be independent of other cellular constituents, so studies of receptor pharmacology in transfected cells should be valid. For example, the intrinsic activity of a drug (full agonist versus partial agonist versus pure antagonist) can be accurately accessed in transfected cells that give a strong biochemical signal. However, intrinsic activity varies with receptor density (71), and transfected cells with high receptor levels may not reproduce the endogenous system. For example, the partial agonist properties of pindolol at the 5-HT1B receptor were masked in high-expressing transfected cells, where this partial agonist behaved as a full agonist (2). The opposite may also occur—a partial agonist converts to an antagonist—as has been illustrated for 5-HT1A receptors expressed in HeLa cells (13). In these instances, receptor density appears to be the key variable, although it is also conceivable that the properties of a drug may change depending on the G protein involved or the cellular response. All of these considerations suggest that caution should be used when interpreting pharmacological studies in transfected cell lines. In vivo studies of 5-HT1 receptors illustrate that drug properties at a given receptor may also be site-specific. For example, the degree of 5-HT1A receptor reserve varies markedly in different brain sites. 5-HT1A somatodendritic autoreceptors in raphe possess a large receptor reserve, whereas postsynaptic 5-HT1A receptors in hippocampus lack receptor reserve (73). Partial agonists at postsynaptic 5-HT1A receptors became full agonists at raphe 5-HT1A receptors.

An example of the utility of heterologous expression systems is the resolution of a recent controversy concerning the existence of multiple, distinct 5-HT2A receptors versus multiple affinity states. Pierce and Peroutka (57) proposed the existence of two different 5-HT2A receptors, based on agonist radioligand binding studies in membranes from rat and human cerebral cortex. One subtype was characterized as having high affinity for agonists; the other, low affinity for agonists. Antagonists possessed equally high affinities for both of the putative receptor subtypes. In contrast, other investigators proposed the existence of two different affinity states for agonists at a single 5-HT2A receptor. Using cloned rat 5-HT2A receptors expressed in heterologous cell lines, Teitler et al. (69) and Branchek et al. (14) showed that a single 5-HT2A receptor protein binds agonists with two different affinity states, ruling out the postulated two-receptor theory. Thus, in this example, unambiguous data were obtained using a pure population of receptors expressed in cells that do not normally express the protein.

An intriguing result, obtained in a fibroblast cell line expressing the human 5-HT1A receptor, suggests that some 5-HT antagonists distinguish between G-protein-coupled and -uncoupled receptors (66). This study showed that the binding of 5-HT1A receptor agonists decreases in the presence of guanine nucleotides, a common property of G-protein receptor systems. Unexpectedly, the binding of the antagonist spiperone actually increases. Because guanine nucleotides destabilize the receptor–G-protein coupled form, these data suggest that spiperone has a higher affinity for the uncoupled receptor. Other antagonists were equipotent at binding to both the coupled and uncoupled state. If confirmed, studies such as these may lead to a reclassification of drugs that act at the 5-HT1A receptor. We have found evidence for a novel classification of 5-HT2C receptor antagonists using fibroblast cell lines expressing the cloned 5-HT2C receptor (9a). One cell line had a high basal level of phosphoinositide hydrolysis, indicating that the receptor was constitutively active in these cells. Common 5-HT antagonists reduced basal activity in the absence of agonist (Fig. 4), suggesting that these drugs have negative intrinsic activity at the receptor. This represents a novel property because drugs such as these are generally thought to function as silent antagonists, occupying the receptor and blocking an agonist, but having no effect in the absence of agonist. We are currently exploring the significance of negative intrinsic activity in the mechanism of the atypical down-regulation of 5-HT2C receptors by these antagonists and also in behavioral paradigms that may be sensitive to negative intrinsic activity.

A given receptor may couple to more than one signal transduction pathway. The example of 5-HT1A receptor coupling to inhibition of adenylate cyclase as well as directly activating a K+ channel has already been discussed. When a receptor couples to a previously unrecognized signal cascade, especially if the cloned receptor is overexpressed in cell lines, this multiple signaling is referred to as promiscuous coupling. An example is Gi-linked receptors that couple to activation of adenylate cyclase in transfected cell lines. It is entirely possible that many, if not all, of the incidences of promiscuous coupling can be explained by cross-talk between the various signal transduction pathways. The numerous and diverse possibilities of cross-talk are just beginning to be worked out. Recent evidence suggest that G-protein bg subunits may play a central role in cross-talk between signaling pathways (10). Both adenylate cyclase and phospholipase Cb are activated by Gbg in an isoform-specific manner. Gbg stimulation of PLC and adenylate cyclase requires high receptor occupancy and high expression levels. Thus the functional consequences of receptor–G-protein activation may vary from cell to cell, depending on both (a) the receptor and its level of expression and (b) the component of effector molecules within a given cell. Such complexities make results obtained in transfected cell lines difficult to relate back to the in vivo situation.

Emerging evidence suggests that, in addition to coupling to inhibition of adenylate cyclase and to the regulation of K+ and Ca2+ channels, the 5-HT1A receptor also couples to PLC. These studies were done in transfected cell lines expressing a high density of the cloned human 5-HT1A receptor. In COS and HeLa cells, 5-HT is 100 times more potent at inhibiting adenylate cyclase than activating PLC (26). In other cell lines, the difference in potency for activating the two signals is smaller or nonexistent (46). Still another signal pathway is activated in transfected CHO cells, a 5-HT1A-receptor-mediated augmentation of the release of arachidonic acid by substances such as ATP and thrombin (60). All three responses—inhibition of adenylate cyclase, activation of PLC, and augmentation of arachidonic acid release—are mediated by a Gi protein(s) with equal sensitivity to inactivation by pertussis toxin. Arachidonic acid release has only been demonstrated in intact cells. This, combined with the finding that down-regulation of PKC abolishes 5-HT augmentation of arachidonic acid release (60), is consistent with an indirect mechanism for this response. On the other hand, an indirect mechanism for the activation of PLC in cell membranes is less likely. As discussed earlier, some isoforms of PLC are activated by bg subunits of G proteins. Coupling of the 5-HT1A receptor to Gi protein and inhibition of adenylate cyclase would lead to release of Gbg, which in turn could regulate PLC (Fig. 5). Based on the antibody blocking profile, Gi3 appears to mediate both inhibition adenylate cyclase and stimulation of PLC (25), which is consistent with interacting mechanisms. Alternatively, it is possible that the same receptor, coupled to the same G protein, can independently regulate distinct transmembrane effector pathways (Fig. 5). Definitive studies of coupling to multiple G proteins require antisense strategies or blocking antibodies directed against specific G proteins. Regardless of the mechanism, it is not at all clear that multiple signaling occurs with the native 5-HT1A receptor. It is possible that expression of high levels of 5-HT1A receptors in cells that do not normally express the protein creates artifacts, such as aberrant coupling to alternative G proteins or effector pathways. Consistent with this interpretation is the finding of cell-specific signaling of the 5-HT1A receptor—that is, different patterns of coupling of the cloned transfected receptor in different cell types (26, 46). Also, the physiological consequences of alternative coupling of the 5-HT1A receptor to PLC varies as a function of cell type. In HeLa cells, an epithelial cell line, activation of 5-HT1A receptors enhances phosphate uptake, but this effect is not evident in transfected CHO cells, a fibroblast line (61). These examples of cell-type-specific coupling to G proteins, effector enzymes, and downstream regulatory events highlight the limitations of studying receptor function in transfected cell lines.

Surprisingly, functional coupling of the 5-HT1A receptor to activation of adenylate cyclase via GS has not been demonstrated in transfected cells even though 5-HT1A-receptor-mediated activation of adenylate cyclase has been described in vivo. The a2-adrenergic receptor, initially described in vivo as a Gi-linked receptor coupled to inhibition of adenylate cyclase, has been found to activate adenylate cyclase via GS in transfected cells. The coupling to GS requires high receptor expression and high agonist concentration and may be mediated by G protein bg subunits liberated by activation of the Gi-mediated pathway (27).

Examples of cross-talk between receptor/signal transduction pathways have been described in native expression systems. The enhancement of b-adrenergic-receptor-mediated hyperpolarization by 5-HT1A receptor activation may be an in vivo physiological correlate of Gbg activation of type II adenylate cyclase (5). In astroglial cells in primary culture, 5-HT augments b-adrenergic receptor stimulation of cAMP formation (32). This effect is mediated by the 5-HT2A receptor, which activates phosphoinositide hydrolysis in these cells. This may be a version of the mechanism of cross-talk described many years ago (23) in which PKC activation potentiates cAMP formation by b-adrenergic agonists, perhaps by PKC-mediated phosphorylation of the catalytic subunit of adenylate cyclase. Also, in cultured astrocytes, the alpha2 agonist clonidine potentiates 5-HT2A-receptor-mediated phosphoinositide hydrolysis (32), perhaps reflecting Gbg activation of phospholipase Cb. Astrocyte cultures appear to be especially suitable for studying cross-talk between intracellular signaling cascades, although subpopulations of astrocytes that respond differently to 5-HT may compromise interpretation of results in these cells (53). An example of cross-talk between Gi-linked 5-HT1A receptors and muscarinic receptors linked to phosphoinositide hydrolysis was described, where 5-HT attenuates carbachol-stimulated phosphoinositide hydrolysis (16). This was initially interpreted as evidence for negative coupling of the 5-HT1A receptor to PLC. More recently, a mechanism involving activation of phospholipase A2 has been proposed in which phospholipase-A2-dependent release of arachidonic acid activates PKC, which then down-regulates either the muscarinic receptor or PLC (17). Interested readers are referred to a recent review in which many other possible mechanisms of cross-talk among the various signal-activated phospholipases are discussed (45).

This chapter has attempted to bring the reader up-to-date about the signal transduction pathways for 5-HT receptors. Additional 5-HT receptors will undoubtedly be discovered, and it is possible that the signaling of these new receptors will follow one of the pathways described in this review. However, the recent description of novel mediators of signal transduction, such as nitric oxide and carbon monoxide, suggests that we should keep an open mind about the future possibilities.

Dr. Sanders-Bush's research is supported by USPHS grants MH 34007 and DA 05181 and by an unrestricted research grant from Bristol Myers Squibb Corporation. Hervé Canton is partially supported by a fellowship from Groupe De Recherche Servier. The authors thank Ms. Edna Kunkel for preparing the illustrations.

published 2000