Electrophysiology of Serotonin Receptor Subtypes and Signal Transduction Pathways

George K. Aghajanian

More than a decade ago a multiplicity of serotonin (5-hydroxytryptamine, 5-HT) receptors in brain were defined pharmacologically by radioligand binding and other methods. Within the past 5 years, molecular cloning techniques have confirmed that 5-HT receptor subtypes (e.g., 5-HT1, 5-HT2, 5-HT3), as predicted from radioligand binding, represent separate and distinct gene products (see Molecular Biology of Serotonin Receptors: A Basis for Understanding and Addressing Brain Function, Serotonin Receptor Subtypes and Ligands, and Serotonin Receptors: Signal Transduction Pathways). This knowledge has revolutionized electrophysiological approaches to the 5-HT system. For example, electrophysiological studies can now be directed toward neurons that express specific 5-HT receptor subtypes based on in situ hybridization maps of receptor mRNA expression. Within each neuron, 5-HT receptor subtypes interact with their own set of G proteins, second messengers, and ion channels, accounting for the wide range of electrophysiological actions produced by 5-HT throughout the brain and spinal cord.

In this rapidly developing field, emphasis must be placed upon more recent contributions to 5-HT electrophysiology, particularly those that have emerged since the publication in 1987 of Psychopharmacology: The Third Generation of Progress (see ref. 2 for a more detailed review of earlier studies).

Physiology

Many electrophysiological studies on the 5-HT1 receptor subtype have been conducted in areas with a dense concentration of 5-HT1A binding sites and a high level of 5-HT1A mRNA expression such as the dorsal raphe nucleus and the hippocampal pyramidal cell layer (18, 42, 51). Studies in these and other regions will be reviewed in the following sections.

Serotonergic neurons of the dorsal raphe nucleus can be inhibited by their own transmitter, 5-HT. Because it is located in the vicinity of the soma, the receptor mediating this effect has been termed a somatodendritic autoreceptor (as opposed to the prejunctional autoreceptor). Early studies showed that lysergic acid diethylamide (LSD) and other indoleamine hallucinogens are powerful agonists at the somatodendritic 5-HT autoreceptor (see ref. 2). Functionally, the somatodendritic 5-HT autoreceptor has been shown to mediate collateral inhibition within the raphe nuclei (see ref. 2). Studies in the brain-slice preparation have revealed that the ionic basis for the autoreceptor-mediated inhibition, by either 5-HT or LSD, is an opening of K+ channels (see ref. 2); these channels are characterized by their inwardly rectifying properties (68).

The somatodendritic autoreceptor of dorsal raphe neurons appears to be predominantly of the 5-HT1A subtype as indicated by the fact that a variety of drugs with 5-HT1A selectivity (e.g., 8-OH-DPAT and the anxiolytic drugs buspirone and ipsapirone) share the ability to potently inhibit raphe cell firing in a dose-dependent manner (see ref. 2). Furthermore, intracellular recordings from dorsal raphe neurons in brain slices show that 5-HT1A agonists fully mimic 5-HT in hyperpolarizing the cell membrane and decreasing input resistance. While the classical 5-HT antagonists have proven ineffective in blocking the electrophysiological effects of 5-HT at the autoreceptor, acute intravenous administration of the drug spiperone, which has moderate affinity for the 5-HT1 receptor binding site, rapidly blocks the effects of the 5-HT1A agonist 8-OH-DPAT in the dorsal raphe (10, 36). Spiperone also rapidly blocks the inhibitory effects of locally applied 5-HT in the dorsal raphe (see ref. 2). In contrast, spiperone is relatively ineffective in blocking the inhibitory effect of 5-HT on postsynaptic CA3 neurons of the hippocampus, suggesting that pre- and postsynaptic 5-HT1A receptors may not be identical (11). However, the basis for this difference is unclear because only one 5-HT1A clone has been reported to date.

Receptor binding and behavioral studies have suggested that the b-adrenoceptor antagonists such as (-)-propranolol may also possess 5-HT1A antagonistic properties (see Serotonin Receptor Subtypes and Ligands). Furthermore, electrophysiological studies have shown that low microiontophoretic currents of the b-blocker (-)-propranolol effectively block the suppressant effects of the 5-HT1A agonists (e.g., ipsapirone and 8-OH-DPAT) on raphe cell firing (see ref. 2). These results fit with cloning data which reveal a remarkable degree of sequence homology between the b2-adrenoceptor and the 5-HT1A receptor (especially in the membrane-spanning segments), providing a molecular basis for the interaction between b-adrenoceptor antagonists and 5-HT1A agonists (see Serotonin Receptors: Signal Transduction Pathways).

In addition to an opening of K+ channels, whole-cell recordings from acutely dissociated raphe neurons have shown that 5-HT decreases high-threshold calcium currents, probably via 5-HT1A receptors because this effect is mimicked by 8-OH-DPAT (49). This calcium current is virtually insensitive to L-type calcium channel blockers but is partially sensitive to w-conotoxin, an N-type channel blocker (50).

Inhibitory or hyperpolarizing responses to 5-HT have been reported in a wide variety of neurons in the spinal cord, brainstem, and diencephalon. In general, such responses have been attributed to an action on 5-HT1 receptors. In sensory neurons of dorsal root ganglia a 5-HT1-like receptor has been reported to reduce the calcium component of action potentials and to produce hyperpolarizations which can be mimicked by 5-HT1A agonists such as 8-OH-DPAT (57, 66). In cerebellar Purkinje cells, 5-HT-induced inhibition but not excitation is mediated through 5-HT1A receptors (23). In brain slices of the nucleus prepositus hypoglossi, focal electrical stimulation evokes inhibitory postsynaptic potentials which have been shown to be mediated by endogenous 5-HT acting on 5-HT1A receptors to activate a K+ conductance (15). In the ventromedial hypothalamus (45) and lateral septum (30), 5-HT produces inhibitory effects through 5-HT1A receptors and also by activating a potassium conductance. In the locus coeruleus, 5-HT suppresses depolarizing synaptic potentials, apparently through both 5-HT1A and 5-HT1B receptors (13). In addition, 5-HT appears to selectively suppress the excitatory response of locus coeruleus neurons to locally applied glutamate through a 5-HT1A receptor (20). In the rat laterodorsal tegmental nucleus (LDT), bursting cholinergic neurons are hyperpolarized by 5-HT via 5-HT1 receptors (35). It has been suggested that the removal of a tonic inhibitory 5-HT influence from these cholinergic neurons may be responsible for the emergence of PGO spikes during rapid eye movement (REM) sleep.

Hippocampus

In CA1 pyramidal cells 5-HT produces a membrane hyperpolarization and reduction in input resistance due to an opening of potassium channels (see ref. 2). The receptors mediating these events appear to be of the 5-HT1A subtype because 8-OH-DPAT and other 5-HT1A (but not 5-HT1B) agonists elicit similar, although weaker, changes in membrane potential and input resistance (5, 55, 59). However, two other postsynaptic responses to 5-HT—a reduction in the amplitude of the slow after hyperpolarization and a late depolarization—do not appear to be mediated by any of the 5-HT1 receptor subtypes (5, 9). In addition to direct postsynaptic effects on pyramidal cells, 5-HT has also been reported to depress spontaneous synaptic potentials (both IPSPs and EPSPs) in CA1 cells; a 5-HT1A receptor may be involved because 8-OH-DPAT produces similar effects (56). Furthermore, there is a subpopulation of interneurons in the hippocampus even more sensitive than pyramidal cells to the hyperpolarizing effect mediated by 5-HT1A receptors (58). Thus, 5-HT may attenuate slow inhibitory postsynaptic potentials in CA1 pyramidal cells by inhibiting a population of feedforward interneurons that are highly sensitive to 5-HT1A receptor stimulation.

Hyperpolarizing 5-HT1A responses in the cerebral cortex have been described in a number of studies (6, 24, 60). However, cortical neurons typically show mixed inhibitory and excitatory responses to 5-HT which involve dual actions at 5-HT2/1C and at 5-HT1A receptors expressed by the same neuron. These mixed responses are described below (see 5-HT2 Receptors: Physiology).

There is evidence that the opening of K+ channels via 5-HT1A receptors in dorsal raphe neurons is mediated by a pertussis-toxin-sensitive G protein. Pertussis toxin catalyzes the ADP ribosylation of the alpha subunit of certain G proteins (e.g., Gi and Go), causing an irreversible uncoupling of the G protein from its receptor. Extracellular and intracellular experiments in the dorsal raphe nucleus have shown that a 48-hr preinjection with pertussis toxin (local or intracerebroventricular) causes an almost total blockade of the inhibitory and hyperpolarizing effect of 5-HT (29, 68). Consistent with their 5-HT1A binding properties, the inhibitory effects of ipsapirone and LSD in the dorsal raphe are also blocked by pertussis toxin (29). G-protein coupling to 5-HT1 receptors in the dorsal raphe is also shown by the fact that intracellular injection of GTPgS, a nonhydrolyzable analogue of GTP which induces an irreversible activation of G proteins, mimics and is nonadditive with the hyperpolarizing action of 5-HT (29). Intracellular GTPgS also renders irreversible the suppression of high-threshold calcium currents in dorsal raphe neurons (50).

As in the dorsal raphe, the hyperpolarizing effects of 5-HT are blocked by pertussis toxin pretreatment in the hippocampus (see ref. 2). Pertussis toxin also blocks GABAB (see ref. 2) and adenosine (72) responses in the hippocampus. Because the hyperpolarizing effects of 5-HT are nonadditive with GABAB and adenosine agonists, this suggests a common G-protein-mediated transduction mechanism. Although 5-HT is negatively coupled to adenylate cyclase via a pertussis-toxin-sensitive G protein, the hyperpolarizing effect of 5-HT on membrane potential does not appear to involve cAMP because neither bath application of the membrane-soluble analogue of cAMP, 8-Br-cAMP, nor intracellular injection of cAMP reduces the response (see ref. 2!popup(ch43ref2)). On this basis, it appears that the 5-HT1A-induced hyperpolarizing response is mediated by a membrane-delimited coupling of G proteins to K+ channels rather than through a diffusible second messenger system.

Other areas where pertussis toxin blockade of inhibitory responses to 5-HT has been demonstrated include dorsal root ganglion cells (22) and neurons of the ventromedial hypothalamus (46).

Signal transduction pathways for 5-HT1 receptors are depicted schematically in Fig. 1.

Physiology

Quantitative autoradiographic studies show high concentrations of 5-HT2 binding sites and mRNA expression in certain regions of the forebrain such as the neocortex (layers IV/V), piriform cortex, claustrum, and olfactory tubercle (41). With few exceptions (e.g., facial nucleus and the nucleus tractus solitarius), relatively low concentrations of 5-HT2 receptors or mRNA expression are found in the brainstem. Studies aimed at examining the physiological role of 5-HT2 receptors in several of these regions are discussed in the following sections.

Facial motoneurons have a high density of 5-HT2 receptor binding sites, and in situ hybridization shows a high level of 5-HT2 receptor mRNA in these neurons (41). Early studies in vivo showed that 5-HT applied microiontophoretically does not by itself induce firing in the normally quiescent facial motoneurons but does facilitate the subthreshold and threshold excitatory effects of glutamate (see ref. 2). Intracellular recordings from facial motoneurons in vivo or in brain slices in vitro (3, 32) show that 5-HT induces a slow, subthreshold depolarization associated with an increase in input resistance, suggesting a decrease in a resting K+ conductance. Similar effects of 5-HT also have been described in spinal motoneurons (67, 71). Recently, it has been shown that 5-HT also increases the excitability of facial motoneurons by enhancing a hyperpolarization-activated cationic current Ih (26, 31). In contrast, the depolarizing effect of norepinephrine on facial motoneurons appears to involve only a closure of K+ channels (31).

The 5-HT2 antagonist ritanserin is able to selectively block the excitatory effects of 5-HT in facial motoneurons (54). On the other hand, the selective 5-HT1A agonist 8-OH-DPAT, although it increases facial motoneuron excitability when given in vivo by systemic injection, fails to produce excitation when applied locally either by microiontophoresis or by bath application in brain slices. Thus, this selective 5-HT1A ligand does not appear to have any direct excitatory effect on facial motoneurons. However, 5-carboxamidotryptamine (5-CT), a broad-spectrum 5-HT1 agonist, acts directly to enhance facial motoneuron excitability. Surprisingly, ritanserin is able to block the effects of 5-CT as well as those of 5-HT. Because ritanserin has extremely low affinity for all 5-HT1 receptors except 5-HT1C, it is possible that it is the 5-HT1C component of 5-CT's receptor profile that is responsible for its effect on facial motoneurons. The 5-HT1C receptor has a high degree of homology with the 5-HT2 receptor, accounting for the difficulty in distinguishing the two sites pharmacologically. Indeed, it has been proposed that the 5-HT1C now be reclassified as the 5-HT2C receptor and that the original 5-HT2 receptor now be termed 5-HT2A (see Serotonin Receptor Subtypes and Ligands).

A large number of studies, using behavioral, ligand-binding, and electrophysiological techniques, have shown that indoleamine (e.g., LSD and psilocin) and phenethylamine (e.g., mescaline and DOI) hallucinogens share the property of interacting with 5-HT2 receptors (see Glennon). The iontophoretic administration of LSD, mescaline, or psilocin, although having relatively little effect by themselves, produce a prolonged facilitation of facial motoneuron excitability (see ref. 2). Intracellular studies in brain slices show that the enhancement is due in part to a small but persistent depolarizing effect of the hallucinogens (26, 54). In addition, LSD and the phenethylamine hallucinogen DOI enhance the cationic current Ih even to a greater degree than does 5-HT itself, suggesting that this action may be more important quantitatively than the closure of K+ channels in explaining the increase motoneuronal excitability produced by these drugs (26). All of these effects of the hallucinogens are reversed by spiperone and ritanserin, consistent with mediation by 5-HT2 receptors.

Systemically administered mescaline or LSD induces a simultaneous decrease in spontaneous activity and increase in sensory responsivity of noradrenergic cells in the locus coeruleus (LC) (see ref. 2). That the effects of LSD and mescaline (and other phenethylamine hallucinogens) on LC neurons are mediated by 5-HT2 receptors is suggested by the fact that they can be reversed by low doses of 5-HT2 antagonists such as ritanserin and LY-53857. In addition to reversing the effects of hallucinogens, the 5-HT2 antagonists induce a small but significant increase in basal LC firing rates, suggesting the existence of a tonic 5-HT2 inhibitory influence (see ref. 2). The latter electrophysiological findings are paralleled by voltammetric studies which show an increase in the catechol metabolite DOPAC in the LC following systemic injections of the 5-HT2 antagonist ritanserin (21). Antipsychotics with affinity for 5-HT2 binding sites are also able to reverse the actions of hallucinogens in the LC independently of their actions at dopamine and other types of receptors (53). The relative potencies of hallucinogens in their action on LC neurons correlates with their affinity for 5-HT2 receptors (see ref. 2). However, the effects of hallucinogens in the LC are not direct because they are not mimicked by the local, iontophoretic application onto LC cell bodies. Thus, the hallucinogens are likely to be acting indirectly on LC neurons via afferents to this nucleus.

In brain slices of the medial pontine reticular formation, 5-HT induces a hyperpolarization in some cells (34%) and a depolarization in other cells (56%) (63). The hyperpolarizing responses are associated with an increase in membrane conductance and have a 5-HT1 pharmacological profile. The depolarizing responses have a 5-HT2 pharmacology and are associated with a decrease in membrane conductance resulting from a decrease in an outward K+ current. These two actions of 5-HT do not appear to coexist in the same neurons, because none of the cells display dual responses to selective agonists. In the nucleus accumbens the great majority of neurons are depolarized by 5-HT, inducing them to fire (47). The depolarization is associated with an increase in input resistance due to a reduction in an inward rectifier K+ conductance. Pharmacological analysis shows that the depolarization is mediated by a 5-HT2 rather than a 5-HT1 or 5-HT3 receptor.

Marked depolarizing responses to 5-HT, associated with a decrease in a resting or "leak" K+ conductance, have been reported in GABAergic neurons of the nucleus reticularis thalami; these excitatory responses to 5-HT are blocked by the 5-HT2/1C antagonists ketanserin and ritanserin (38). The 5-HT-induced slow depolarization potently inhibits burst firing in these cells and promotes single-spike activity. It has been suggested that this 5-HT-induced switch in firing mode from rhythmic oscillation to single-spike activity, which occurs during states of arousal and attentiveness, contributes to the enhancement of information transfer during these states.

The electrophysiological effects of 5-HT have been studied in several cortical regions. In vivo, 5-HT2 agonists applied by microiontophoresis have been reported to have primarily inhibitory effects on the firing of unidentified neurons in prefrontal cortex (8); these inhibitions are blocked by 5-HT2 antagonists. (Note that the inhibitions produced by 5-HT itself are not blocked by 5-HT2 antagonists but are blocked by 5-HT3 antagonists; see below.) In brain slices, pyramidal cells in various regions of the cerebral cortex have been found to respond to 5-HT by either a small hyperpolarization, depolarization, or no change in potential (6, 24, 39, 60). Based on pharmacological evidence, the depolarizations appear to be mediated by 5-HT2 or 5-HT1C receptors. In rat association cortex it has been shown that there can be a coexistence of 5-HT1A and 5-HT2 receptors on a single neuron (6). Thus, in the presence of the 5-HT2/1C antagonist ketanserin, 5-HT1A-mediated hyperpolarizing responses can be elicited in cells which originally exhibit only a depolarizing response. Excitatory responses to 5-HT in pyramidal cells are due to a reduction in K+ conductances (38, 61). Three different types of conductances appear to be involved: a resting K+ conductance, a depolarization-activated K+ conductance (M current), and a Ca2+-activated K+ conductance. There is pharmacological evidence that the excitatory effects of 5-HT on pyramidal cells in piriform cortex are mediated by 5-HT1C rather than 5-HT2 receptors (61).

Recently, we have observed a novel effect of 5-HT in piriform cortex, namely, an induction of IPSPs (60) in pyramidal cells. The IPSPs are blocked by the GABA antagonist bicuculline, suggesting that the IPSPs arise from GABAergic interneurons that are excited by 5-HT. Accordingly, a subpopulation of interneurons (at the border of layers II and III) has been found that is excited by 5-HT. Somewhat less frequently, neurons within this same subpopulation of interneurons also tend to be excited by dopamine and norepinephrine (27). Excitation by 5-HT of these interneurons (as well as associated IPSPs in pyramidal cells) is blocked by 5-HT2 antagonists. The hallucinogens LSD and DOM behave as partial agonists in this system, producing a modest activation by themselves but occluding the full effect of 5-HT. We have found a similar partial agonist effect of hallucinogens in the medial septal nucleus, where 5-HT also has an excitatory effect on GABAergic neurons (Alreja and Aghajanian, unpublished data).

In piriform cortex, the 5-HT2 antagonist ritanserin blocks the activation of interneurons more readily than it blocks the depolarization of pyramidal cells (60). Ritanserin has a nearly 10-fold higher affinity for the 5-HT2 receptor than for the 5-HT1C receptor, suggesting that the action of 5-HT on these interneurons might be through 5-HT2 receptors whereas the action of 5-HT on the pyramidal cells might be through 5-HT1C receptors. This hypothesis is consistent with recent findings that mRNA for the 5-HT2 receptor is located in cortical interneurons while mRNA for the 5-HT1C receptor is found in pyramidal cells (40, 41). Recent studies in the piriform cortex employing a new, highly selective antagonist (MDL 100,907) with a 300-fold greater affinity for the 5-HT2 than for 5-HT1C receptors also suggest that 5-HT2 rather than 5-HT1C receptors are responsible for the 5-HT excitation of interneurons (Marek and Aghajanian, unpublished data).

The role of G proteins in mediating the 5-HT2-induced slow inward current has been evaluated in facial motoneurons by using the hydrolysis-resistant guanine nucleotide analogues GTPgS and GTPbS (1). The 5-HT-induced inward current becomes largely irreversible in the presence of intracellular GTPgS. Mediation by G proteins is also suggested by the fact that the inward current is reduced by intracellular GTPbS which prevents G-protein activation. The identity of the G protein(s) mediating the electrophysiological responses remains to be determined.

In addition to the electrophysiological effects mediated by 5-HT2 and 5-HT1C receptors, there is an activation phospholipase C which hydrolyzes phosphatidylinositol (PI) to yield two major intracellular messengers, diacylglycerol (DAG) and inositol trisphosphate (IP3) (see Serotonin Receptors: Signal Transduction Pathways). LSD and DOM act as partial agonists of this effect. An increase in IP3 resulting from the activation of 5-HT2 receptors would be expected to release intracellular stores of Ca2+ from its intracellular stores. Recently, 5-HT-induced increase in cytoplasmic Ca2+ levels has been directly demonstrated in cultured local interneurons of mouse olfactory bulb by Ca2+ imaging techniques using the fluorescent indicator fura-2 (65). This effect of 5-HT was blocked by ritanserin, indicating the involvement of 5-HT2 receptors.

DAG, by activating protein kinase C (PKC), would be expected to affect many long-term cellular responses through protein phosphorylation. The effect of protein kinase inhibitors on the response of facial motoneurons to 5-HT has been tested (1). Two protein kinase inhibitors with different mechanisms of action, 1-(5-isoquinolylsulfonyl)-2-methylpiperazine (H7), a nonselective protein kinase inhibitor, and sphingosine, a selective PKC inhibitor, in concentrations that have no effect of their own (100 and 10 mM, respectively), both markedly enhance and prolong the excitation of facial motoneurons induced by 5-HT. Conversely, phorbol esters that are known to activate PKC reduce the excitatory effect of serotonin. These results suggest that activation of PI turnover, perhaps through receptor phosphorylation, has a negative feedback effect on 5-HT-induced excitations in the facial nucleus. Similar findings have also been obtained in the cerebral cortex using the grease-gap method for assessing the 5-HT enhancement of N-methyl-D-aspartate (NMDA)-induced depolarizations (52). In the latter studies, phorbol esters and DAG rather than mimicking the ability of 5-HT to enhance NMDA depolarizations, promoted 5-HT desensitization. An interesting implication of the negative feedback model is the possibility that the partial agonist properties of hallucinogens with respect to 5-HT-stimulated PI hydrolysis may contribute to, rather than interfere with, their electrophysiological actions (see Serotonin Receptors: Signal Transduction Pathways).

The activation of 5-HT2 receptors can have long-term effects through an alteration in immediate early gene expression. Within 30 min following systemic injections of the 5-HT2 agonist 1-(2,5-dimethoxy-4-iodophenyl)-2-aminopropane (DOI), a dramatic increase in Fos protein (the product of the immediate early gene c-fos) can be detected in neurons in middle layers of cerebral cortex, including piriform cortex (33). It is not known as yet whether this effect on immediate early gene expression is mediated through an activation of PKC.

Signal transduction pathways for 5-HT2 receptors are summarized in Fig. 2.

There are rapidly desensitizing depolarizing responses to 5-HT in the periphery that are mediated by 5-HT3 receptors (formerly known as M receptors). In brain, excitatory responses to 5-HT have been found in cultured mouse hippocampal and striatal neurons which have many of the characteristics of peripheral 5-HT3 responses: rapid onset and rapid desensitization, features that are typical of ligand-gated ion channels rather than G-protein-coupled receptor responses (69, 70). In cultured NG108-15 cells the permeation properties of the 5-HT3 channel are indicative of a cation channel with relatively high permeability to Na+ and K+ and low permeability to Ca2+ (70). Recently, a 5-HT-gated ion channel has been cloned which has physiological and pharmacological properties appropriate for a 5-HT3 receptor (37). In the oocyte expression system, this receptor shows rapid desensitization and is blocked by 5-HT3 antagonists (e.g., ICS 205-930 and MDL 72222). Because of its sequence homology with the nicotinic acetylcholine receptor (27%), the b1 subunit of the GABAA receptor (22%), and the 48K subunit of the glycine receptor (22%), it is likely that this 5-HT3 receptor clone is a member of the ligand-gated ion channel superfamily.

Rapidly desensitizing 5-HT3 responses have also been reported in brain slices. In slices containing the lateral nucleus of the amygdala, 5-HT3-mediated fast excitatory synaptic responses to focal electrical stimulation can be demonstrated when glutamate receptors are blocked; these synaptic responses show rapid cross-desensitization with bath-applied 5-HT (64). In hippocampal slices, 5-HT has been reported to increase GABAergic IPSPs, most likely through a 5-HT3-receptor-mediated excitation of inhibitory interneurons; these responses also show fading with time (55).

While fast, rapidly inactivating excitation has generally become accepted as characteristic of 5-HT3 receptors, nondesensitizing responses have also been reported. In dorsal root ganglion cells a relatively rapid but noninactivating depolarizing response has been described which has a 5-HT3 pharmacological profile (66). In neurons of the nucleus tractus solitarius in a brain-slice preparation, there is a postsynaptic depolarizing response to 5-HT3 agonists which does not appear to be rapidly desensitizing (28). In the latter preparation there are also enhancements of presynaptic responses (both IPSPs and EPSPs). In medial prefrontal cortex a slow inhibitory response to microiontophoretic applied 5-HT has been described which also has a 5-HT3-like pharmacology (7). This effect is mimicked by the 5-HT3 agonist 2-methylserotonin and blocked by the 5-HT3 antagonists BRL 43693 and ICS205930. At this juncture, it is not clear whether there is more than just a superficial pharmacological relationship among these various so-called 5-HT3 responses.

Fig. 3 depicts the 5-HT3 receptor as a ligand-gated cationic channel.

The existence of the 5-HT4 receptor in the central nervous system was first suspected on the basis of biochemical data showing positive coupling of 5-HT responses to adenylyl cyclase (16). Thus, this receptor differs from the 5-HT1 subfamily, which is negatively coupled to adenylyl cyclase. Very recently, two novel 5-HT receptors positively coupled to adenylyl cyclase have been cloned; however, because their pharmacology differs from that of the previously described 5-HT4 site, they have been tentatively designated as 5-HT6 and 5-HT7 receptors (34, 43). The original 5-HT4 receptor has not as yet been cloned and is characterized primarily by its pharmacological properties. The latter is generally unaffected by 5-HT1, 5-HT2, and 5-HT3 agonists or antagonists, but certain benzamides (e.g., BRL 24923, zacopride, and metoclopramide) can act as agonists while ICS 205-930 (in contrast to other 5-HT3 antagonists) shows antagonism, but with low potency and poor selectivity (16).

Up to this point there have been few electrophysiological studies performed on putative 5-HT4 responses in the brain. In hippocampal CA1 pyramidal cells, 5-HT induces a slow depolarization (unmasked after blockade of 5-HT1A hyperpolarizing responses) and a reduction in the amplitude of the calcium-activated potassium conductance (afterhyperpolarization). These responses are not sensitive to the usual 5-HT1, 5-HT2, and 5-HT3 agents, whereas they are blocked by BRL 24924 and certain other substituted benzamides and ICS 205-930 (4, 19), suggesting a 5-HT4-like pharmacology. However, these responses do not appear to be mediated through the cAMP signal transduction pathway (4). Another electrophysiological action of 5-HT with a 5-HT4 profile has been reported in cultured mouse collicular neurons (25). In these cells, 5-HT reduces slowly inactivating voltageactivated potassium currents; this effect is generally insensitive to 5-HT1, 5-HT2, and 5-HT3 drugs but is responsive to a variety of 5-HT4-active agents including the substituted benzamides. However, in contrast to CA1 pyramidal cells, the effects of 5-HT in collicular neurons do appear to be mediated through the cAMP transduction pathway because they are mimicked by cAMP analogues and are blocked by intracellular application of PKI, a specific inhibitor of cAMP-dependent protein kinase.

In a number of regions, cAMP has been shown to mimic the ability of 5-HT to enhance the hyperpolarization activated cationic current (Ih). For example, in neurons of the nucleus prepositus hypoglossi the augmentation of Ih can be mimicked by cAMP-active agents (e.g., 8-bromo-cAMP and forskolin) (14). However, since this effect is also mimicked by 5-HT1 agonists, the precise identification of the 5-HT receptor subtype is unclear because 5-HT1 receptors are generally coupled negatively to adenylate cyclase. The newly cloned 5-HT7 receptor, which has relatively high affinity for certain 5-HT1A ligands (34), must now be considered as a possible mediator of the increase in Ih in these cells. An augmentation of Ih by 5-HT (and norepinephrine) has also been reported for several thalamic nuclei such as the dorsal lateral and medial geniculate nuclei of the thalamus (48). The increase in Ih results in a reduced ability of thalamic neurons to generate rhythmic burst firing. It has been suggested that this reduction in burst firing (which is associated with sleep spindles) increases the efficacy of transfer of information through the thalamus during periods of increased arousal and attentiveness. The increase in Ih in the thalamus is not blocked by 5-HT1 or 5-HT2 antagonists but is mimicked by membrane-permeable cAMP analogues or the adenylate cyclase activator forskolin. However, it has not been determined as yet whether this response is mediated by 5-HT4 receptors. In brain slices, a non-5-HT1, 5-HT2-mediated increase in Ih has also been reported for dopaminergic neurons of the substantia nigra pars compacta (44). However, as yet there have been no reports on the role of cAMP or 5-HT4 receptors in this effect.

Possible signal transduction pathways for 5-HT4 receptors are depicted in Fig. 4.

The extraordinarily diverse electrophysiological actions of 5-HT in the central nervous system can now be categorized according to receptor subtypes and their respective effector mechanisms. The following generalizations are emerging: (a) Inhibitory effects of 5-HT are mediated by 5-HT1 receptors linked to the opening of K+ channels or the closing of Ca2+ channels, both via pertussis-toxin-sensitive G proteins; (b) there are facilitatory effects of 5-HT that involve the closing of K+ channels which are mediated by 5-HT2 receptors, with the PI second messenger system and PKC acting as a negative feedback loop; (c) other facilitatory effects of 5-HT are mediated by 5-HT4 receptors through a reduction in IK(Ca2+) or voltage-dependent K+ current, apparently in some cases through the cAMP pathway and in other cases not; and (d) fast excitations are mediated by 5-HT3 receptors through a ligand-gated cationic ion channel which does not require coupling with a G protein or a second messenger. Thus, the electrophysiological actions of 5-HT encompass the two major neurotransmitter superfamilies: the G-protein-coupled receptors (i.e., 5-HT1, 5-HT2, 5-HT4) and the ligand-gated channels (5-HT3).

The diversity of receptors and transduction pathways that underlie the varied electrophysiological actions of 5-HT, together with the differential expression of these receptors in different neuronal populations, helps to explain how it is possible for one transmitter to be linked to such a large array of behaviors, clinical conditions, and drug actions. For example, alterations in 5-HT function have been implicated in affective disorders, anxiety states, schizophrenia, obsessive compulsive disorder, eating disorders, migraine, and sleep disorders (see Indoleamines: The Role of Serotonin in Clinical Disorders). There is an equally wide range of drugs that interact with 5-HT neurotransmission, including antidepressants (e.g., selective 5-HT uptake blockers), atypical antipsychotics (e.g., clozapine), anxiolytics (e.g., buspirone), antiemetics (e.g., ondansetron), hallucinogens (e.g., LSD), antimigraine drugs (e.g., sumatriptan), and appetite suppressants (e.g., fenfluramine).

The direct actions of 5-HT at a cellular level, as detailed in this review, do not have a simple relationship to the overall input–output relations. For example, while 5-HT1A agonists are directly inhibitory upon serotonergic neurons in the raphe nuclei (possibly at doses insufficient to affect postsynaptic 5-HT1A receptors), the net effect of this inhibition may be disinhibitory for postsynaptic neurons that express 5-HT1A receptors and simultaneously disfacilitatory for postsynaptic neurons that express 5-HT2, 5-HT3, and 5-HT4 receptors. Similarly, while 5-HT2 agonists may be directly excitatory at a subpopulation of GABAergic interneurons in the cerebral cortex that express 5-HT2 receptors, the net effect of this excitation may be inhibitory for those pyramidal cells that receive inputs from these interneurons. In general, to understand the functional consequences of the discrete cellular actions of 5-HT, these must be viewed within the context of the neuronal networks where they occur.

This review has described the individual cellular actions of 5-HT in many different regions of the brain and spinal cord. Do these discrete and disparate effects of 5-HT have an integrated function greater than the sum of the parts? Suggestive of this possibility is the fact that (a) serotonergic neurons are clustered in a relatively small number nuclei within the brainstem (i.e., the raphe nuclei), projecting diffusely to almost every other region of the neuraxis (see Anatomy, Cell Biology, and Maturation of the Serotonergic System: Neurotrophic Implications for the Actions of Psychotropic Drugs), and (b) in unanesthetized animals, the tonic firing of serotonergic neurons as a group varies according to behavioral state, such that activity is greatest during behavioral arousal, diminished during slow-wave sleep, and virtually absent during REM sleep (see Serotonin and Behavior: A General Hypothesis).

Do the individual cellular actions of 5-HT in different regions of the central nervous system serve to coordinate overall behavioral states? The possibility of an integrative action of 5-HT can be illustrated for three disparate sets of neurons as follows. During the waking state, when serotonergic neurons are in a tonic firing mode, the following conditions would prevail: (a) Motoneurons would be in a relatively depolarized, excitable state (via 5-HT2 receptors) and thus would be receptive to the initiation of movement; (b) neurons of the nucleus reticularis thalami would be in a depolarized, single-spike mode (via 5-HT2 receptors) and thus would be conducive to thalamocortical sensory information transfer (38, 48); and (c) neurons of the laterodorsal tegmental nucleus would be hyperpolarized (via 5-HT1 receptors) and therefore not able to generate the bursting activity of REM sleep (35). With a reduction in serotonergic activity during various stages of sleep, the above conditions would reverse such that motoneurons would become less excitable, thalamocortical sensory information transfer would be diminished, and sleep spindles and PGO waves would emerge. Thus, serotonergic neurons can be seen as functioning as part of a complex, coordinated modulation of motor, sensory, and other systems to promote a given behavioral state. It remains to be seen whether all the diverse cellular actions of 5-HT can be incorporated into this kind of holistic scheme or whether there are subsets of serotonergic neurons that operate different groups of postsynaptic neurons in an independent fashion.

This work was supported by USPHS grants MH 17871 and MH 15642 and by the State of Connecticut.

published 2000