The 5-HT₃ receptors are the only ionotropic or ligand gated ion channel of the 5-HT receptor family [14]. They are members of the Cys-loop superfamily of ligand-gated ion channels [15] as are nicotinic acetylcholine (nAch) receptors and gamma-aminobutyric acid-type-A (GABA)_A receptors. Thus, they are composed of five subunits forming a cylinder that can be crossed by cations [16]. In rats and mice, two subunits have been cloned: 5-HT_3A [15] and 5-HT_3B [17] receptor subunits which can be arranged in homomeric 5-HT_3A receptors and heteromeric 5-HT_3A/5-HT_3B receptors [17]. Three others subunits have been also described in multiple mammalian species but not in rodents: 5-HT_3C, 5-HT_3D and 5-HT_3E [18-20] which probably form only heteromeric receptors with 5-HT_3A receptor subunits.

Using different methods including autoradiography, immunohistochemistry and in situ hybridization, the 5-HT₃ receptor distribution has been largely described with some differences between species. 5-HT₃ receptors are expressed both in peripheral and central nervous system. In the periphery, 5-HT₃ receptors are located on pre- and postganglionic neurons from autonomic nervous system and on neurons of the sensory and enteric nervous system [21]. In the central nervous system, 5-HT₃ receptor density appeared low compared to others 5-HT receptors [14,22]. In all species, including humans, the most important densities of 5-HT₃ receptors seems to be found in the hindbrain in particular in the dorsal motor nucleus of the vagus nerve, in the nucleus of the tractus solitaries and in the area postrema [22-32]. These brain areas are involved in the vomiting reflex explaining the relevance of 5-HT₃ receptor antagonist in chemotherapy-induced emesis [14]. Significant densities of 5-HT₃ receptors have also been described in the spinal cord [33,34].

Compared to the hindbrain, the density of 5-HT₃ receptors in the forebrain is lower. Nevertheless, significant levels of these receptors have been found in brain areas involved in the pathophysiology of depression with densities varying across species.

In non primate mammals, 5-HT₃ receptors are found in the limbic areas including the amygdala, hippocampus, nucleus accumbens and in the superficial layers of the cerebral cortex, i.e., the frontal parts (in particular cingulate, prelimbic and infralimbic areas and primary and secondary motor areas), entorhinal and temporal cortex [25,27-29,31,33,35-38]. In addition, some studies have demonstrated that, they are also present at low densities in the dorsal raphe nucleus, striatum, substantia nigria and nucleus accumbens [25,28-30,38].

In the marmoset forebrain, 5-HT₃ receptors are found in medial habenula nucleus and the hippocampus [32]. In the human forebrain, 5-HT₃ receptors are found essentially in limbic structures such as amygdala, hippocampus, nucleus accumbens and striatum whereas only low levels of 5-HT₃ receptors have been found in cortex [23,28,39-42].

A significant amount of 5-HT₃ receptors are localized on presynaptic nerve fibers and terminals [30,43-46]. Indeed, in cortical areas, amygdala [47] and in striatum [48,49], 5-HT₃ receptors are essentially located presynaptically whereas in hippocampus, the postsynaptic receptors are predominant [47].

5-HT₃ receptors are permeable to Na⁺, K⁺ and Ca²⁺ [50,51]. Stimulation leads to opening of the ion channel, inducing a rapid membrane depolarization mediated by cation flow [52,53]. The function of 5-HT₃ receptors depend on their localization: nerve-terminal 5-HT₃ receptors activation leads to release of various neurotransmitters such as 5-HT, DA or GABA [54], whereas the activation of postsynaptic 5-HT₃ receptors is involved in fast synaptic transmission [55,56]. Pre- and postsynaptic receptors are associated with specific characteristics including a different hill coefficient, different single channel conductances, different kinetics, a different re-sensitization time-course [17,57,58]. In particular, presynaptic 5-HT₃ receptor displays a high permeability to Ca²⁺ [45,46,48,59], whereas postsynaptic receptors display a lower permeability to Ca²⁺ compared to Na⁺ and K⁺ [50,60]. Similarly, homomeric 5-HT_3A receptors are equally permeable to monovalent and divalent cations whereas heteromeric 5-HT₃ receptors have a lower permeability to Ca²⁺ [17,61,62]. Moreover, heteromeric receptors display faster activation and deactivation kinetics than homomeric receptors [62]. In vitro studies are generally performed in cultured cells expressing only homomeric 5-HT₃ receptors [61] which can explain several differences obtained by in vitro and in vivo studies.

5-HT₃ receptor agonists and antagonists present different affinity and efficacy depending on the structure of 5-HT₃ receptors, i.e., heteromeric or homomeric [17,58,61] (Table 1). 5-HT₃ receptor agonists seem not have clinical interests. Frequently used preclinical tool agonists are 1-(m-chlorophenyl)-biguanide (mCPBG) and 2-methyl-5-HT (2-Me-5-HT) which do not cross the blood brain barrier. These compounds are not selective for 5-HT₃ receptors since mCPBG has notable affinity for the DA transporter [63] while 2-Me-5-HT has notable affinity for other 5-HT receptor subtypes [64,65]. SR57227A is the mostly used 5-HT₃ receptor agonist that crosses the blood brain barrier [66].

In comparison to 5-HT3 receptor agonists, many 5-HT₃ receptor antagonists have been developed and they are widely used in the clinic. The main therapeutic use of 5-HT₃ receptor antagonist is for chemotherapy-induced emesis [67]. However, other therapeutic uses of 5-HT₃ receptor antagonist have been suggested e.g. pain, addiction and psychiatric disorders [68]. In regard with psychiatric disorders, 5-HT₃ receptor antagonists present anxiolytic and antidepressants effect (see below) but they may also have antipsychotic effect even if data are yet controversial [21,69]. The 5-HT₃ antagonists may be identified by the suffix setron. Different drugs belonging to the “setron class” are used in the clinic: MDL 73,147EF (dolasetron), GR38032F (ondansetron), BRL 43694 (granisetron), ICS 205-939 (tropisetron), palanosetron listed according to ascending receptor binding affinity (7.73 to 10.45 nM) [67]. Other “setrons” have been used in preclinical studies including DAU 6215 (itasetron), BRL-46470A (ricasetron), MDL 72222 and LY277359 (zatosetron).

Curiously, 5-HT₃ receptor agonist or antagonist responses are frequently associated with a bell-shaped dose-response curve and this is the case for both clinical and preclinical studies. Generally, the maximum effect is typically observed at very low dose, in the microgram range, while higher doses are ineffective [21]. For example, such responses were observed in the rat learned helplessness [70], the forced swim test and the tail suspension test [71] as well as in the induction of theta rhythms [72]. The inverse dose-response relation of 5-HT₃ receptor agonists may be explained by receptor desensitization. Receptor desensitization is involved in mediation of short-term plasticity of synapses. Such mechanisms seem to be involved in regulation of 5-HT₃ receptors activity in cultured cells [52]. This desensitization can be explained by receptor internalization. In fact, when cells expressing 5-HT₃ receptors in the plasma membrane were activated by the 5-HT₃ receptor agonist, mCPBG, a decrease of 5-HT₃ receptor density was observed after a few minutes [73]. Internalization can be prevented by the 5-HT₃ receptor antagonist ondansetron [74]. Interestingly, it has been recently demonstrated that the 5-HT₃ receptor antagonist palanosetron can also induce 5-HT₃ receptor internalization and cause prolonged inhibition of receptor function whereas ondansetron and granisetron do not [75]. Moreover, MDL7222 fails to induce internalization of 5-HT₃ receptors since a chronic treatment with this compound does not alter sensitivity to a 5-HT₃ receptor agonist [76]. Thus, internalization of 5-HT₃ receptors is not sufficient to explain the bell-shaped dose-response curve observed under various conditions [21] brought forward hypothesizes of steric hindrance at higher concentrations, different effect on hetero/homoreceptors inducing an effect on some receptors at a low concentration and another effect at higher concentration explaining a heterogeneous response depending on concentration.

Dorsal raphe nucleus is the brain structure with the highest density of 5-HT cell bodies. Electrophysiological studies demonstrated that the 5-HT₃ receptor agonist, phenylbiguanide, has no effect on the dorsal raphe nucleus 5-HT cell firing whereas the other 5-HT₃ receptor agonist 2-Me-5-HT has an inhibitory action [65,77]. However, the interpretation of these findings are difficult as there are reports showing that a 5-HT_1A receptor antagonist can prevent the suppressant effect of a 5-HT₃ agonist [77], and others reporting a lack of effect of a 5-HT₃ receptor antagonist and 5-HT_1A antagonist to reverse the effect of the agonist [65]. Furthermore, both in vivo and in vitro, studies of various 5-HT₃ receptor antagonists including ondansetron and zacopride report no significant effect on the dorsal raphe nucleus 5-HT cell firing [65,77]. Interestingly, it has been shown in vitro that 2-Me-5-HT induces a release of 5-HT in raphe nuclei slices both under basal conditions and also after stimulations [78]. This release of 5-HT explains the decrease of firing observed with 5-HT₃ receptor agonists since dorsal raphe nucleus neurons mediate inhibitory effects at 5-HT_1A autoreceptors. The 5-HT release induced by 5-HT₃ receptor agonists is also not specific in basal conditions: it can be antagonized by fluoxetine and not by 5-HT₃ receptor antagonists whereas it is reversed in stimulated conditions by ondansetron [78]. Thus, it seems that in dorsal raphe nucleus, there is no detectable 5-HT tone at 5-HT₃ receptors and that contradictory results obtained in basal conditions can be explained by the poor selectivity of 5-HT₃ receptors agonists.

The role of GABAergic neurotransmission in depression is a relative new area of research [79]. Changes in GABAergic function has been observed in animal model of depression [80] and GABA receptor agonists seems to present antidepressant-like properties [81]. Moreover, serotonergic neurons connect mainly GABAergic interneurons suggesting a strong interaction between the two systems [82,83]. GABAergic interneurons expressing 5-HT₃ receptors have been essentially detected in the hippocampus and prefrontal cortex [84].

The dopamine system can be divided into three pathways. The nigrostriatal dopaminergic pathway consists of the substantia nigra pars compacta (SNc, A9) and their associated efferent targets in the dorsal striatum. The mesolimbic dopaminergic pathway contains A10 dopaminergic neurons located within the ventral tegmental area (VTA) and their associated efferent targets in the ventral striatum including nucleus accumbens and limbic structures (e.g., amygdala and hippocampus). The mesocortical dopaminergic pathway consists of the A10 dopaminergic neurons and their associated efferent targets in the prefrontal cortex. It has been suggested that abnormalities of dopaminergic neurotransmission have been implicated in the pathogenesis of affective disorders [113,114]. In particular, several studies have suggested that the mesolimbic pathway can be altered in mood disorders. For example, stress, in animal models of depression, activates VTA DA neurons and their limbic efferent targets [115]. Moreover, the DA system is now one of the targets of depression treatment, for example drugs initially used in schizophrenia like aripiprazole or quetiapine are now used in depressed patients [8].

There is less data concerning other neurotransmitters release and this relevance of this data for depression is less clear data. Nevertheless, there is some evidence that 5-HT₃ receptors may also induce a release of glutamate from nerve terminals [148-150]. Also, there might be a role for acetylcholine (Ach), since 5-HT₃ receptor activation inhibits cortical release of Ach. It is not clear, however, whether this is a direct effect. Moreover 5-HT₃ receptor antagonist induced an increase of cortical Ach release, an effect that was potentiated by GABA receptor antagonists. This has led to the suggestion that the augmented ACh release by 5-HT₃ antagonists causes a blockade of GABA-mediated inhibition on cholinergic neurons [14,151]. Finally, some studies evoked an increase of noradrenaline induced by activation of 5-HT₃ stimulation but some others data are controversial and this effect seems not specific [85].

Some of the currently used antidepressants show affinity for 5-HT₃ receptors. Thus, the tricyclic antidepressant (TCA) Imipramine, the SSRI fluoxetine, the non-selective α2-adrenoceptor antagonist mirtazapine and the MAOI phenelzine dose-dependently block the inward current mediated by 5-HT₃ receptors expressed in cultured cells [152-156]. Moreover, several antidepressants, including fluoxetine, can inhibit binding of a 5-HT₃ receptor antagonist [157]. Interestingly, fluoxetine induced an increase of Polysialic Acid Neural Cell Adhesion Molecule (PSA-NCAM), a molecule implicated in neuroplasticity, an effect that was blocked by administration of 5-HT₃ receptor antagonist, thus suggesting a direct effect of fluoxetine on 5-HT₃ receptors [158]. Finally, fluoxetine inhibits 5-HT release induced by 5-HT₃ receptor agonists in the dorsal raphe nucleus [78]. Except for mirtazapine, the latter effect of antidepressants on 5-HT₃ receptors seems non-competitive[155,156,159]. Similarly, typical and atypical antipsychotics antagonize 5-HT₃ receptor in a non competitive manner [160]. This has led to the suggestion of an allosteric recognition site different from the 5-HT₃ binding site [155]. Interestingly, the non-competitive 5-HT3 antagonism of antidepressants seem not associated with an increase of internalization [161]. It has also been shown that 5-HT₃ receptors and antidepressants are colocalized in specific domain of membrane cells, the raft-like domains. Nevertheless antidepressants may exert effect on 5-HT₃ receptor despite disruption of lipid rafts [161,162].

Electrophysiological studies exploring recovery of firing after chronic (from 2 days to 3 weeks) antidepressant treatments in anesthetized rats suggested that mirtazapine may be more rapid than SSRIs in reversing the decrease of 5-HT dorsal raphe nucleus firing rate. Also, co-administration of paroxetine and mirtazapine may accelerate this index of antidepressant response compared to either drug alone [163]. In preclinical studies, cyamemazine, an atypical antipsychotic with D2-like, 5-HT_2A, 5-HT_2C and 5-HT₃ receptor antagonism properties [164] presents anxiolytic-like activity. In clinical studies, cyamemazine improved the anxious syndrome [164]. Finally, it has been suggested that electroconvulsive therapy may potentiate 5-HT₃ receptor function in hippocampal CA1 pyramidal cells [165].

Lu AA21004 is a 5-HT₃ receptor antagonist, 5-HT_1A receptor agonist (h5-HT3 and h5-HT_1A receptors: Ki = 4.5 and 15 nM, respectively) and an inhibitor of the 5-HT transporter (5HTT) (h5HTT: IC₅₀ = 5.4 nM) [166]. This in vitro profile translates into enhanced levels of 5-HT as well as other neurotransmitters (i.e., noradrenaline, dopamine, acetylcholine) in hippocampus and prefrontal cortex after in vivo administration [91,166]. It is different from others antidepressants since it has an effect on 5-HT levels even at a low transporter occupancy [91,166]. Moreover, it displays anxiolytic-like and antidepressant-like activity in various validated rodents models with a better efficacy than current antidepressant drugs [91,166]. Interestingly, in electrophysiological studies, Lu AA21004 has a preclinical profile of a fast-acting antidepressant, since it induced early 5-HT_1A receptor desensitization compared to fluoxetine, an effect probably mediated by 5-HT₃ receptors blockade [167]. Currently in phase III trials, LuAA21004 displays very good efficacy and well tolerance [168].

Poncelet et al. [169] demonstrate that the selective 5-HT₃ receptor agonist SR57227A produces antidepressant-like effect in different behavioral tests (forced swimming test, learned helplessness) in rodents. In contrast, other investigators report 5-HT₃ receptor agonists alone or in combination with antidepressants to be ineffective in the forced swimming test [170] and others report that 5-HT₃ receptor agonists attenuate the effects of antidepressants in this animal model [171]. These variable results may be explained by the different doses tested and may also ascribed to SR57227A being a partial 5HT₃ agonist.

In the learned helplessness test, zacopride, ondansetron and tropisetron reverse the escape failure with a biphasic dose effect relationship [70]. In the forced swim test and tail suspension test, acute and chronic treatments with 5-HT₃ receptor antagonists decrease the immobility time [71,172-175]. Interestingly, in the forced swim test, ondansetron increases the efficacy of fluoxetine, venlafaxine and citalopram [71,170]. In bulbectomized rats (another animal model of depression) 5-HT₃ receptor antagonists reverse their depression-like phenotype [71,175]. The novel putative antidepressant (4-benzylpiperazin-1-yl) (quinoxalin-2-yl) methanone (QCF-3) which is a 5-HT₃ receptor antagonist displayed also antidepressant properties both alone and in association with fluoxetine [176]. Finally, antidepressants induce a decrease in rapid eyes movements (REM) sleep. Microinjection of the 5-HT₃ receptor agonist mCPBG in the dorsal raphe nucleus causes a reduction of rapid eye movement sleep (REMS) whereas ondansetron prevented this reduction. [150]. Thus, administration of a 5-HT₃ receptor antagonist may have a beneficial effect on sleep disturbances induced by antidepressants.

Anxiolytic activity of 5-HT₃ receptor antagonists have been extensively studied in animal models [177,178]. Griebel et al. [178] analyzed more than 75 5-HT₃ receptor related experiments in which two-third of the studies demonstrated anxiolytic-like effect of 5-HT₃ receptor antagonists and one-third of the studies failed to reveal an effect. For example, in some studies, 5-HT₃ receptor antagonists disinhibited punished behavior in the Vogel test [145,179,180] and in other studies 5-HT₃ receptor antagonists were inactive [181]. Interestingly, after chronic treatment with diazepam, 5-HT₃ receptor agonists had no more anxiogenic effect, suggesting that diazepam may have induced a 5-HT₃ receptor desensitization [182].

Another index used to evaluate anxiolytic effect of drugs is by examining its effect on cholecystokinin [183]. Interestingly, GABAergic interneurons expressing 5-HT₃ receptors co-express cholecystokinin [88,104]. It has been shown that 5-HT or 5-HT₃ receptor agonists induce a release of cholecystokinin in cortical or accumbal synaptosomes while 5-HT₃ receptor antagonists decrease spontaneous- or induced-release of CCK [184,185]. However, in another study, 5-HT₃ receptor antagonists fail to prevent the increase of CCK induced by stress in rat prefrontal cortex [186]. Finally, it is well known that amygdala is involved in physiology of anxiety [187]. As described previously, 5-HT₃ receptors are present at significant levels in the amygdala. In the mouse amygdala, local administration of 5-HT₃ receptor antagonists attenuates aversive response whereas 5-HT₃ receptor agonists increase this [146]. Under the same conditions, in the social interaction test, 5-HT₃ receptor antagonists have also an anxiolytic-like activity [146]. It has been suggested that the amygdala is involved in disinhibitory effects induced by various 5-HT₃ receptor antagonists [188].

5-HT_3A receptor knockout male mice present anxiety-like behavior in elevated plus maze, novelty interaction animal models and light/dark box models of anxiety [189,190]. In the forced swim test, male 5-HT_3A knock-out (KO) did not differ from the wild type rats, whereas female 5-HT₃ knockout mice showed increased immobility [191]. Mice overexpressing 5-HT₃ receptors show decreased anxiety in the elevated plus-maze and in the exploration paradigm [192].

5-HT₃ receptor antagonists have been tested on anxiety and depressive syndromes associated to other diseases. For example, ondansetron reduced depressive symptoms in patients with chronic hepatitis C [193], with alcoholism [194] and in bulimic patients [195]. Similarly in fibromyalgic patients, tropisetron improved anxiety and depressive scores [196]. Moreover, in pathology with common symptoms of depression like chronic fatigue or fibromyalgia, 5-HT₃ receptor antagonist have been reported efficacious [193,197-199], although negative results in chronic fatigue syndrome have been also reported [200]. Suspicion of 5-HT syndrome after use of setron alone or in combination with mirtazapine has reported in one study [201].

In healthy volunteers, ondansetron has been explored in emotional processing tasks. It abolishes the emotion potentiated startle effect, revealing an anxiolytic-like activity [202]. In anxious patients, zatosetron tends to reduce anxiety however the results from this study was not significant [203]. Similarly, tropisteron improves anxiety scores in patients suffering from generalized anxiety disorders [204]. In clinical studies, results concerning CCK release were contradictory as in preclinical studies. Indeed, in patients with panic disorder and social phobia, ondansetron fails to prevent the anxiety induced by pentagastrin, a CCK agonist [205]. Whereas, in healthy volunteers, ondansetron prevents panic symptoms induced by CCK tetrapeptide [206].