Astroglial Serotonin Receptors as the Central Target of Classic Antidepressants

Introduction

Major depressive disorder (MDD) belongs to severe mood disorders. Global prevalence of MDD is about 4.7% with regional differences (Ferrari et al. 2013; Steel et al. 2014). The lifetime prevalence varies from 3.3% to 16% across countries (Gu et al. 2013; Otte et al. 2016). It has been estimated that MDD will be a leading cause of disability by 2030 (Mathers and Loncar 2006). MDD is associated with great individual and family suffering, disability, poor daily functioning, insomnia, and negative prognosis for comorbid medical conditions (Fröjd et al. 2008; Buysse et al. 2008; Lynch and Clarke 2006; Kupfer et al. 2012). MDD is also associated with an elevated risk of suicide, psychotic presentations (Penninx 2017; Kendler et al. 2009), and other psychiatric disorders (Cassano et al. 2004). Treating MDD usually involves pharmacological agents and/or psychotherapy. Treatment coverage of MDD has clear regional differences: the coverage of treatment of psychiatric disorders including MDD varies between 7.0% and 20.3% across countries (Bijl et al. 2003; Demyttenaere et al. 2004). In the United States, less than 30% of adults who are positively screened for depression received any treatment (Olfson et al. 2016). In China, the discordant treatment ranges from 4% to 60% of MDD patients who ever received psychosocial therapy or pharmacotherapy (Yu et al. 2015; Liu et al. 2017; Chin et al. 2015). Overall, variations in epidemiology, clinical presentation, and treatment of MDD are significantly influenced by sociocultural and economic factors (Kleinman 2004; Compton et al. 2006; Zhong et al. 2018).

In the 1950s, imipramine and iproniazid were used, for the first time, for the pharmacotherapy of MDD, thus opening the history of antidepressant treatment (López-Muñoz and Alamo 2009). Consequently, these two agents contributed to the formulation of the first mechanistic hypothesis of depressive disorders, the monoaminergic theory. In the 1960s, the catecholamine hypothesis was further advanced, with the role of noradrenergic imbalance suggested to be the key pathological factor in the occurrence of MDD (Schildkraut 1965). Serotonin was at the focus of these hypotheses, which resulted in the invention of fluoxetine, the first selective serotonin reuptake inhibitor (SSRI). In recent decades, pathophysiological research of depression allowed pharmacological treatments to target specific neurotransmitters as well as the specific brain regions and types of MDD (Kutzer et al. 2020). For example, ketamine and esketamine are proposed to treat resistant depression (Abdallah et al. 2015; Stenovec et al. 2020), while brexanolone, a natural steroid, is the first Food and Drug Administration (FDA)-approved therapeutic medication of postpartum depression (Meltzer-Brody et al. 2018).

Despite considerable advances in studying MDD pathophysiology in recent decades, and an abundance of pharmacological agents approved for MDD treatment, a substantial number of MDD patients remains resistant to available antidepressants. A real-life sequential treatment study demonstrates that nearly 30% of MDD patients fail to achieve remission, even after multiple treatment attempts (Rush et al. 2004, 2006). This may reflect the fact that existing theories cannot fully explain pathological mechanisms of MDD; similarly, pharmacological mechanisms of action of available antidepressants are in need of further research. Overreliance on the monoaminergic hypothesis may stipulate limited success of antidepressants (Ceskova 2016; Ceskova and Silhan 2018).

Growing evidence supports the key role of astrocytes in the pathophysiology and pharmacology of MDD. Astrocytes are homeostatic cells of the central nervous system (CNS). They control multiple aspects of brain physiology at all levels of organization from molecular to organ and systemic (Verkhratsky and Butt 2013; Verkhratsky and Parpura 2015; Verkhratsky and Nedergaard 2018). An individual protoplasmic astrocyte interacts with as many as 100,000 synapses in mice and possibly up to 2,000,000 synapses in humans (Bushong et al. 2002; Oberheim et al. 2009); these astrocytic perisynaptic structures, known as synaptic cradle (Verkhratsky and Nedergaard 2014), are fundamental for maintaining neurotransmission in the CNS. Pathological changes in astrocytes, classified into astrogliosis, astrodegeneration, or atrophy with loss of function, are salient features of all neurological disorders (Pekny et al. 2016; Verkhratsky et al. 2017) including neurodevelopmental (Zeidán-Chuliá et al. 2014) and neuropsychiatric (Verkhratsky et al. 2014; Peng et al. 2015, 2016) disorders. Based on reports of decreased astroglial numbers in postmortem histopathological studies of MDD patients, the dysfunction of astrocytes was postulated to contribute to the pathophysiology of depression (Cotter et al. 2001; Rajkowska et al. 2007; Hercher et al. 2009; Verkhratsky et al. 2014).

Expression of multiple receptors and transporters makes astrocytes potential targets for specific therapy, and indeed, astrocytes respond to many antidepressants (Hertz et al. 2014; Saura et al. 1992; Zhang et al. 2010). Astrocytes possess serotonin receptor subtypes associated with the pharmacology and pathophysiology of depression (Hertz et al. 2010, 2015; Ding et al. 2013; Kirischuk et al. 1996). In this chapter, we focus on astrocytic serotonin receptors and discuss their relation to MDD.

Serotonin Receptors in the Treatments of MDD

A multitude of neurotransmitter receptors is expressed in astrocytes, including ionotropic and metabotropic glutamate receptors, purinoceptors, α/β-adrenoceptors, and serotonin receptors (Hertz et al. 2015), all associated with the intracellular excitability (Verkhratsky and Parpura 2010; Verkhratsky et al. 2020, 2021). Astrocytic serotonin receptors are of particular importance for MDD because of the widespread application of SSRIs in clinical practice. Stimulation of serotonin receptors in astrocytes activates cytosolic and nuclear signaling pathways, altering cellular functions as well as gene expression (Peng et al. 2015).

There are seven subtypes of serotonin receptors designated as 5-HT_1A-1F, 5-HT_2A-2C, 5-HT₃, 5-HT₄, 5-HT_5A, 5-HT₆, and 5-HT₇. Among these subtypes, 5-HT_1A-1F, 5-HT_2A-2C, 5-HT₆, and 5-HT₇ receptors were discovered in cerebral or spinal astrocytes in humans and rodents (Hirst et al. 1997, 1998; Meuser et al. 2002; Merzak et al. 1996; Li et al. 2012). These serotonin receptors can be involved in the pathophysiology of MDD through numerous cellular signaling pathways. All astrocytic serotonin receptors are G-coupled protein receptors being linked to phospholipase C and inositol-1,4,5-trisphosphate (InsP₃)-dependent Ca²⁺ signaling (Verkhratsky et al. 2012).

5-HT₁ Receptors

The 5-HT_1A and 5-HT_1B receptors attract particular attention because of their involvement in the pathophysiology of MDD; these receptors are also pharmacological targets in the clinical therapeutics of depression (Moret and Briley 2000; Murrough et al. 2011; Murrough and Neumeister 2011; Ruf and Bhagwagar 2009; Sari 2004; Savitz et al. 2009; Tiger et al. 2014). The 5-HT_1A and 5-HT_1B receptors display 43% amino acid sequence homology, and both are the G_i/o protein-coupled receptors (Hoyer et al. 2002). These receptor subtypes have different cellular localizations, with 5-HT_1A receptors mainly expressed in somata and dendrites (Sotelo et al. 1990) and 5-HT_1B receptors being primarily localized in axon terminals (Boschert et al. 1994). Furthermore, 5-HT_1A receptor is one of the most abundant in cortical regions implicated in mood and emotion (Albert and François 2010; Albert and Lemonde 2004). The 5-HT_1B receptors are widely distributed throughout the brain regions, with particularly dense expression in the substantia nigra and globus pallidus (Bonaventure et al. 1997; Varnäs et al. 2011).

The 5-HT_1A receptors act as autoreceptors and heteroreceptors. The presynaptic 5-HT_1A receptors are mainly the autoreceptors localized in the cell bodies and dendrites of serotonergic neurones in dorsal raphe nuclei. These 5-HT_1A receptors modulate the release of serotonin and cell discharge rates. Stimulation of 5-HT_1A autoreceptors decreases the firing activity of central serotonergic neurones and synthesis and release of serotonin by a negative feedback mechanism while promoting the depressive-like behaviors and the resistance responses to antidepressants, thus inhibiting their activity and having opposite results (Richardson-Jones et al. 2010, 2011; Milak et al. 2018). The postsynaptic 5-HT_1A receptors are heteroreceptors, localized in glutamatergic and γ-aminobutyric acid (GABA)-ergic neurones; these receptors show substantial expression in the hippocampus, medial septum, prefrontal cortex, and midbrain ventral tegmental regions where they regulate the release of acetylcholine, glutamate, and dopamine (Comley et al. 2015; Borroto-Escuela et al. 2018; López-Gil et al. 2008; Jeltsch-David et al. 2010; Di Matteo et al. 2008). The 5-HT_1A heteroreceptors play key roles in desensitization of postsynaptic 5-HT_1A receptors, especially in prefrontal cortex regions (Taciak et al. 2018; Naumenko et al. 2014). Inhibition of 5-HT_1A presynaptic autoreceptors or stimulation of 5-HT_1A postsynaptic heteroreceptors may produce antidepressant effects, and hence 5-HT_1A receptor antagonists or agonists may act as antidepressants or antidepressants potentiators, respectively (Richardson-Jones et al. 2011; Philippe et al. 2018).

There are indeed novel antidepressants based on 5-HT_1A receptor agonism; these include YL-0919, brexpiprazole, vilazodone, and vortioxetine hydrobromide. Some of these 5-HT_1A receptor agonists also act as SSRIs and as agonists or antagonists of other serotonin subunits. For example, vortioxetine hydrobromide is the agonist of 5-HT_1A receptor and SSRI; it is a partial agonist of 5-HT_1B receptors and antagonist of 5-HT₃ and 5-HT₇ receptors (Wang et al. 2019). Compound MIN-117, an antagonist of 5-HT_1A autoreceptors, is considered for adjunctive therapy to improve acute antidepressant effects of SSRIs (Artigas et al. 2001; Sahli et al. 2016).

Similar to 5-HT_1A receptors, 5-HT_1B receptors are classified into autoreceptors and heteroreceptors. In postmortem tissue from the suicide MDD subjects, the level of 5-HT_1B receptors is decreased in the hippocampus; the mRNA of 5-HT_1B receptors is also lower in the frontal cortex. Conversely, mRNA of 5-HT_1B receptors is higher in the hypothalamus, compared with a healthy group (Anisman et al. 2008). Serotonin binding to the 5-HT_1B receptors in serotonergic neurones decreases the formation of cAMP and the downstream signaling responses, which results in the reduction of transmitter release (Barnes and Sharp 1999; Leenders and Sheng 2005; Middlemiss and Hutson 1990). Furthermore, 5-HT_1B receptors could modulate serotonin transporters (SERT), and thus 5-HT_1B autoreceptors can regulate the serotonin release from serotonergic neurones (Hagan et al. 2012; Montañez et al. 2013). An agonist of 5-HT_1B receptor CP-93129 decreases 5-HT release in the hippocampus in rats (Hjorth and Tao 1991). In the hippocampus and cortex of wild-type mice, an agonist of 5-HT_1B receptor decreases the level of serotonin, while an antagonist of this receptor has the opposite effect on the serotonin level. Both effects are diminished in the transgenic mice with genetically deleted 5-HT_1B receptors (Rutz et al. 2006). In addition, 5-HT_1B receptors localized in non-serotonergic neurones can modulate the release of glutamate, GABA, acetylcholine, and dopamine (Ruf and Bhagwagar 2009).

Serotonin release is suppressed by 5-HT_1B receptor antagonists SB-616234-A (Dawson et al. 2006), AZD3783 (Zhang et al. 2011), and AR-A000002 (Hudzik et al. 2003), which makes them candidates for antidepressive therapeutics. Stimulation of 5-HT₁ receptors may neutralize the positive effects of SSRIs on serotonin levels to produce the therapeutic latency (Blier and de Montigny 1994; Nutt 2002). Therefore, the antagonists of 5-HT_1B receptors may act as rapid antidepressants. Available 5-HT_1B receptor antagonists are not sufficiently effective as antidepressants but can be used as potential adjuvants (Ruf and Bhagwagar 2009). As a result, the combination of SSRIs with 5-HT_1B receptor antagonists represents a new therapeutic scheme for MDD (Matzen et al. 2000).

5-HT₂ Receptors

The 5-HT_2A, 5-HT_2B, and 5-HT_2C receptors are G_q protein-coupled receptors, which activate phospholipase C (PLC) or protein kinase C and trigger rapid InsP₃-dependent Ca²⁺ signaling or stimulate phospholipase A2 (PLA2) thus increasing arachidonic acid release (Masson et al. 2012; Li et al. 2008). The link between 5-HT_2A receptors and astrocytes emerged in 1999 when 5-HT_2A receptors were found to be significantly upregulated in reactive astrocytes (Wu et al. 1999). Trazodone, a triazolopyridine derivative, could play a neuroprotective role by inhibiting 5-HT_2A/2C receptors in human reactive astrocytes, while fluoxetine also has a similar neuroprotective effects by blocking the same receptors (Daniele et al. 2015). Antagonists of 5-HT_2A receptors such as EMD281014, FG5893a, and M100907 suppress depressive-like behavior as reported by forced swim test, the classical rodent behavioral test for depressive phenotype (Patel et al. 2004; Albinsson et al. 1994; Marek et al. 2005). Antagonist of 5-HT_2A receptors promotes noradrenaline release under SSRI treatment (Dremencov et al. 2007), suggesting a feedback loop between 5-HT_1A receptors in glutamatergic neurones and 5-HT_2A receptors in GABAergic neurones. Such interaction may enhance the antidepressant activity of SSRIs and noradrenaline reuptake inhibitor (SNRIs) YM992, which is a potent 5-HT_2A antagonist (Szabo and Blier 2002). Some recent studies show that functional disruption of the 5-HT_2A receptors may be related to postpartum depression disorder or psychotic disorders (Gao et al. 2018).

Even though 5-HT_2A receptors are widely researched with regard to the pharmacological mechanism of antidepressant action, most research focusses on neurones (Szabo and Blier 2001a, b). However, fluoxetine does not affect the mRNA expression of 5-HT_2A receptors in neurones or astrocytes, isolated and sorted from transgenic mice treated with chronic unpredictable mild stress (Li et al. 2012). In contrast, astrocytic 5-HT_2B receptors have shown significant changes.

Fluoxetine increases mRNA expression of 5-HT_2B receptors decreased by the treatment with chronic stress solely in astrocytes, and not in neurones (Dong et al. 2015; Li et al. 2012). Similarly, in the Parkinson’s disease model, 5-HT_2B receptors are downregulated in astrocytes, but not in neurones; this downregulation develops in parallel with depressive-like behaviors (Diaz et al. 2016). The 5-HT_2B receptor is widely distributed in mammalian brains, including the frontal cortex, septal nuclei, dorsal hypothalamus, and medial amygdala (Duxon et al. 1997; Kursar et al. 1994; Li et al. 2012; Dong et al. 2015). In the primary cultured astrocytes or in the fluorescence-activated cell sorted astrocytes from hgfap::EGFP transgenic mice (expressing enhanced green fluorescence protein driven by a fragment of the human glial fibrillary acidic protein promoter), 5-HT_2B receptors are co-expressed with 5-HT_2A and 5-HT_2C receptors (Li et al. 2012; Hertz et al. 2010). However, astrocytic 5-HT_2B receptor has a much higher affinity for serotonin than the other two 5-HT₂ receptors subtypes (Li et al. 2010). Stimulation of 5-HT_2B receptors with 1 nM serotonin triggers phosphorylation of extracellular signal-regulated kinase 1/2 (ERK1/2), while 5-HT_2C receptor-mediated ERK1/2 phosphorylation occurs at 100 nM, and 5-HT_2A receptors are activated by serotonin in concentrations above 10μM (Li et al. 2010). All SSRIs including fluoxetine act as agonists of astrocytic 5-HT_2B receptors suggesting the role of these receptors in antidepressive action (Zhang et al. 2010; Hertz et al. 2012; Li et al. 2009a, 2012, 2017).

Astrocytic 5-HT_2B receptors treated with various concentrations of fluoxetine trigger different signaling pathways to regulate gene expressions. Acute treatment with fluoxetine at concentrations lower than 1μM decreases mRNA expression of c-Fos through phosphoinositide 3 kinase(PI3K)-protein kinase B (also known as AKT) signaling pathway; at higher concentrations (more than 5μM), however, fluoxetine elevates this gene expression by mitogen-activated protein kinase (MAPK)-ERK signaling pathway (Fig. 1; Li et al. 2017). Stimulation of 5-HT_2B receptors by fluoxetine in astrocytes leads to the transactivation of epidermal growth factor receptors (EGFR); this in turn activates downstream MAPK/ERK or PI3K/AKT signaling cascades thus regulating expressions of genes possibly related to mood disorders. These genes include Ca²⁺-dependent phospholipase A2 (cPLA2), subtype 2 of adenosine deaminases acting on RNA’s (ADAR2), or subtype 2 of kainate receptors (GluK2) (Li et al. 2009b, 2011, 2012; Rao et al. 2006). Chronic treatment with fluoxetine also induces RNA editing of ADAR2, resulting in the loss of function of 5-HT_2B receptors, although expression of these receptors is increased (Peng et al. 2014). Rapid increase in [Ca²⁺]_i triggered by an agonist of 5-HT_2B receptors was suppressed by chronic treatment with fluoxetine (Li et al. 2011; Peng et al. 2014). Malfunction of 5-HT_2B receptors has been discovered in chronic stress animal models of depression and in animals with depressive-like behaviors associated with Parkinson’s disease or hyperammonemia (Li et al. 2012; Zhang et al. 2015; Yue et al. 2019). This evidence indicates that 5-HT_2B receptors are a target for SSRIs, and it is indicative of the potential contribution of astrocytic 5-HT_2B receptors to the pathophysiology of MDD (Kennett et al. 1996, 1998). Agonist of 5-HT_2B receptors BW723C86 displays anxiolytic-like effects, whereas 5-HT_2C/2B receptor antagonist SB200646A improves depressive-like behaviors (Kennett et al. 1995).

Fig. 1.

Schematic illustration of biphasic concentration-dependent regulation of Cav-1 gene expression and GSK-3β activity by fluoxetine in astrocytes. Fluoxetine is an agonist to serotonin 2B receptors (5-HT_2BR). Acute treatment with fluoxetine at low concentrations (green arrows) stimulates Src which phosphorylates epidermal growth factor (EGF) receptors (EGFR) at Y845 residue and in turn activates phosphoinositide 3 kinase (PI3K)-AKT signaling pathway. The AKT phosphorylation by fluoxetine at low concentrations inhibits c-Fos gene expression and subsequently decreases caveolin-1 (Cav-1) gene expression (chronic effects) that in turn decreases membrane content of PTEN (phosphatase and tensin homolog), induces phosphorylation and stimulation of PI3K, and elevates glycogen synthase kinase 3 (GSK-3β) phosphorylation thus suppressing its activity. At higher concentrations, fluoxetine (red arrows) stimulates metalloproteinase (MMP) and induces shedding of growth factor which stimulates EGFR and activates mitogen-activated protein kinase (MAPK) extracellular signal-regulated kinases 1/2 (ERK1/2) signal pathway. The ERK1/2 phosphorylation by fluoxetine at high concentrations stimulates cFos gene expression and subsequently increases Cav-1 gene expression (chronic effects), which acts on PTEN/PI3K/AKT/GSK-3β in an inverse fashion. (From Li et al. 2017)

The 5-HT_2C receptors are upregulated by fluoxetine specifically in neurones sorted from Thy1::YFP transgenic mice, expressing yellow fluorescence protein driven by Thy1 promoter (Li et al. 2012). Antagonist of 5-HT_2C receptor S32006 demonstrates antidepressive effects and increases the level of dopamine and noradrenaline in the frontal cortex in rats (Dekeyne et al. 2008). An inverse agonist of 5-HT_2C receptor, S32212, also has antidepressant activity in behavioral tests (Dekeyne et al. 2012). Fluoxetine is reported to act as an antagonist of 5-HT_2C receptors (Ni and Miledi 1997), while chronic treatment with SSRIs may promote receptor desensitization (Yamauchi et al. 2004). Mianserin and several classical tricyclic antidepressants (TCAs), such as amitriptyline, imipramine, and clomipramine, all work as antagonists with moderate to high affinity at 5-HT_2C receptors (Jenck et al. 1993; Millan 2005). SNRIs desipramine and maprotiline as well as atypical antidepressants mirtazapine, trazodone, and nefazodone similarly act as 5-HT_2C antagonists (Ni and Miledi 1997).

Increased density of 5-HT_2C receptors was found in the frontal cortex of subjects who committed suicide (Pandey et al. 2006). This increase may reflect changes in mRNA alternative splicing and/or RNA editing (Martin et al. 2013), which has been proposed as a risk factor for suicidal attempts (Dracheva et al. 2008). Clinical studies similarly indicated increases in mRNA editing of 5-HT_2C receptors in the frontal cortex of depressed suicide individuals (Niswender et al. 2001; Gurevich et al. 2002; Iwamoto and Kato 2003; Schmauss 2003). A recent study demonstrated that astrocytic 5-HT_2C receptors are involved in the depressive- and anxious-like behaviors induced by alcohol addiction. Fluoxetine at low concentration (1μM) can alleviate these anxiety-like behaviors and motor activity induced by alcohol by suppressing the expression of ADAR2 and mRNA editing of 5-HT_2C receptors (Li et al. 2020).

The study in chronic social defeat animal model of depression reported increased sensitivity of 5-HT_2C receptors that can be reversed by chronic antidepressants (Berton et al. 2006; Rygula et al. 2005, 2006). Pretreatment with 5-HT_2C receptor agonist m-chlorophenylpiperazine increases behavioral presentations of defeat in hamsters (Harvey et al. 2012). In rats, however, chronic stress increases depressive-like behavior, this effect being arguably mediated through 5-HT_2C receptors (Kimura et al. 2008; Moreau et al. 1993). Suppression of dopamine release mediated by 5-HT_2C receptors has been observed in rat models of depression (Dremencov et al. 2005; Oba et al. 2013). Antagonists of 5-HT_2C receptors can alleviate stress-induced depressive-like behaviors in mice, while their antidepressive effects may relate to dopaminergic neurones in the ventral tegmental area (Opal et al. 2013). Chronic treatment with 5-HT_2C receptor agonist RO60-0175 alleviates anhedonia in stressed mice by desensitization of the receptor (Moreau et al. 1993). Depressive-like behaviors in the learned helplessness model can be improved by selective antagonists of 5-HT_2C receptors (Strong et al. 2009). Likewise, clomipramine, a potential antagonist of 5-HT_2C receptors, decreases elevated mRNA expression of 5-HT_2C receptors in chronic stress treated rats (Pitychoutis et al. 2012).

All these observations suggest that the dysfunctions of 5-HT₂ receptors may be a consequence of stress and can be linked to anxiety, depression, and suicide. Antidepressive effects of agonists and antagonists of 5-HT₂ receptors remain controversial and poorly understood, thus warranting more research.

Other Serotonin Receptors

The 5-HT₃ is the only ionotropic member of the serotonin receptors family. This 5HT-gated channel is assembled from five subunits, 5HT_3A, 5HT_3B, 5HT_3D, 5HT_3E, and 5HT_3F, expression of which was confirmed in humans (Niesler et al. 2007). In rodents, however, only 5-HT_3A and 5HT_3B subunits have been identified (Karnovsky et al. 2003). Ondansetron and zacopride, antagonists of 5HT₃ receptors, have been used in clinical therapeutics of depression, and both reversed the depressive-like behaviors in the rat learned helplessness model (Martin et al. 1992; Thiebot and Martin 1991). Antagonists of 5HT₃ receptors potentiate the pharmacological efficiency of clinical antidepressants (Nakagawa et al. 1998; Redrobe and Bourin 1997). Vortioxetine is a multimodal drug which binds to 5-HT_1A, 5-HT_1B, 5-HT_1D, and 5-HT₇ receptors and to SERT; it also acts as a potent antagonist of 5HT₃ receptors. Vortioxetine demonstrated antidepressive effects in rodent models of depression (Guilloux et al. 2013; Mork et al. 2012; Wallace et al. 2014). Furthermore, preclinical data also show that vortioxetine as an antagonist of 5HT₃ receptors produced antidepressant response faster than classical antidepressants (Bétry et al. 2013). In addition, vortioxetine demonstrates antidepressive performance in patients resistant to SSRIs or SNRIs (Alvarez et al. 2014). 5HT₃ receptor antagonists may be promising therapeutic agents, although certain unresolved issues remain, including those of heterogeneity of the species used and of the inconsistent data on effective doses as antidepressants (Adrien et al. 1992; Costall et al. 1987; Johnson et al. 2002, 2006; Martin et al. 1992; Ramamoorthy et al. 2008).

5-HT₄ receptors are widely distributed in limbic regions, such as the amygdala and hippocampus (Hannon and Hoyer 2008; Tanaka et al. 2012). Being G_s-coupled, 5-HT₄ receptors increase intracellular cAMP by stimulating adenylyl cyclase (Dumuis et al. 1988; Fagni et al. 1992). Decrease in 5-HT₄ receptors was observed in the striatum of depressed patients and in mice models (Amigó et al. 2016; Madsen et al. 2014). In postmortem studies of depressed suicide victims, various binding forms of 5-HT₄ receptors and the different levels of cAMP in distinct brain areas have been reported (Rosel et al. 2004). A selective partial agonist of 5-HT₄ receptors, RS67333, has a rapid antidepressant effect in chronic depression models of rodents (Lucas et al. 2007).

In MDD context, studies of 5-HT₅, 5-HT₆, and 5-HT₇ receptors are scarce. Administration of N-methyl-D-aspartate receptor antagonist ketamine, at a dose that leads to drug abuse, can trigger schizophrenic-like symptoms in animals (de la Salle et al. 2016). In these ketamine pretreated mice, an agonist of 5-HT₆ receptors E-6837 as well as an antagonist of these receptors SB-271046 can both produce depressive-like behaviors (Suárez-Santiago et al. 2017). Antagonists of 5-HT₇ receptors have faster antidepressant effects when compared to fluoxetine in the rat model of depression (Mnie-Filali et al. 2011).

To summarize, there are many uncertainties and controversies concerning MDD-related pharmacology of serotonin receptors. In general, agonists of 5-HT_1A, 5-HT_2B, and 5-HT₄ receptors and antagonists of 5-HT_1B, 5-HT_2A, 5-HT_2C, 5-HT₃, 5-HT₆, and 5-HT₇ act as antidepressants. None of these drugs, however, can substantially affect clinical manifestations of depression.

Astrocytes as a Target for Classic Antidepressants

Classic antidepressants are SSRIs, SNRIs, TCAs, inhibitors of monoamine oxidase (MAO), and lithium. Major theories of MDD pathophysiology focus on neurones, with particular emphasis on monoamines (Marathe et al. 2018), aberrant signaling of brain-derived neurotrophic factor (BDNF), leptin dysregulation (Milaneschi et al. 2017), myo-inositol abnormalities (Yu and Greenberg 2016), and glutamate homeostasis failure (Shirayama et al. 2017). Here, we discuss the MDD from the astrocytic angle and portray astrocytes as targets for therapy.

SSRIs

SSRIs, which include fluoxetine, fluvoxamine, sertraline, paroxetine, and citalopram, are the most popular antidepressants and anxiolytics. An increase in extracellular serotonin availability following SERT inhibition has been regarded as the main and the only pharmacological mechanism of SSRIs, despite the well-known delay in clinical effect contrasting to the immediate inhibition of SERT. Recently, the contribution of astrocytes to MDD has gained increased attention. Postmortem histological analysis of the frontal cortex reported a reduced number of glia in patients suffering from MDD (Cotter et al. 2001, 2002). Decreased density of neuroglia and decrease in glia-to-neurone ratio have also been observed in the amygdala (Bowley et al. 2002; Hamidi et al. 2004). Experimental ablation of astrocytes (but not neurones) in the frontal cortex is sufficient to initiate depressive-like behavior in rats (Banasr and Duman 2008). Similarly, loss of astrocyte-derived ATP induces depressive phenotypes in rodents (Cao et al. 2013). Fluoxetine reverses the loss of hippocampal astrocytes in an animal model of depression (Czéh et al. 2006), while increasing the release of ATP from astrocytes by regulating vesicular nucleotide transporter (Kinoshita et al. 2018).

Fluoxetine acts as an agonist of astrocytic 5-HT_2B receptors. However, acute and chronic treatments of astrocytes with fluoxetine produce different outcomes. Acute treatment with 10μM fluoxetine triggers Ca²⁺ signals, transactivates the EGFR, and phosphorylates downstream ERK1/2 signaling cascade; ERK1/2 enters nuclei and regulates the mRNA and protein expression of c-Fos and FosB, consequently modulating the expression of several key proteins (Li et al. 2008). For instance, fluoxetine, by stimulating 5-HT_2B receptors, increases protein expression of cPLA2, GluK2, ADAR2, transient receptor potential canonical 1 channels, L-type Ca_v1.2 Ca²⁺ channels, caveolin-1, and BDNF (Li et al. 2009a, 2011, 2017, 2019; Hertz et al. 2015; Peng et al. 2018). Increased expression of ADAR2 in astrocytes following chronic treatment with fluoxetine catalyzes RNA editing, promoting loss of function of 5-HT_2B receptors and GluK2, which affects Ca²⁺ signaling and phosphorylation of ERK1/2 (Li et al. 2011; Hertz et al. 2012). Decreased expression of astrocytic 5-HT_2B receptors was also observed in chronic stress models and Parkinson’s disease (Dong et al. 2015; Zhang et al. 2015). Fluoxetine can selectively upregulate the expression of 5-HT_2B receptors in astrocytes (Li et al. 2012). However, PI3K/AKT and MAPK/ERK signaling pathways induced by low and high concentrations of fluoxetine result in the opposite regulation of caveolin-1 and glycogen synthase kinase-3β (GSK-3β) (Fig. 1) (Li et al. 2017). Some effects of fluoxetine mediated by 5-HT_2B receptors may contribute to both therapeutic and adverse effects of this drug; hence, selective stimulation or inhibition of various signaling pathways triggered by fluoxetine may promote the efficacy or lessen the side effects of antidepressants.

Depressive disorders are associated with many neurological conditions including sleep disturbances, Alzheimer’s, and Parkinson’s diseases (Xia et al. 2018; Orgeta et al. 2017; Huang et al. 2019). Aging of astrocytes (Verkhratsky et al. 2020, 2021) may contribute to age-dependent diseases as well as to depressive phenotypes. Sleep disturbance in particular is linked to both depression and neurodegeneration. Insomnia and depression can easily combine into a vicious cycle, as decreased sleep time increases the risk for MDD, which further exacerbates sleep disorders (Roberts and Duong 2014). The glymphatic system (Iliff et al. 2012) is responsible for the clearance of brain metabolite waste via paravascular pathways (Xie et al. 2013). Astrocytic endfeet forming glia limitans vascularis and endfeet-polarized expression of aquaporin 4 (AQP4) support the glymphatic system. In aging, astroglial atrophy decreases glymphatic clearance facilitating extracellular accumulation of disease-related proteins, such as β-amyloid (Kress et al. 2014). Increased accumulation of β-amyloid has been reported by positron emission tomography in the right hippocampus and thalamus of human subjects having their moods affected by sleep deprivation (Shokri-Kojori et al. 2018). In animals, chronic stress increases β-amyloid₄₂ in the frontal cortex and hippocampus. Fluoxetine upregulates the expression of AQP4 and promotes clearance of β-amyloid₄₂ by improving glymphatic clearance; this also rescues depressive-like behaviors (Xia et al. 2017). Toxic load with iron also impairs the glymphatic system, thus exacerbating depressive-like phenotypes and increasing neuronal apoptosis (Fig. 2; Liang et al. 2020).

Fig. 2.

Treatment with chronic stress enhances neuronal apoptosis triggered by overload with iron. Mice were pretreated without or with chronic unpredictable mild stress (CUMS) for 6 weeks; in week 6, animals were randomly separated to be injected with dextran (CUMS group) or iron dextran (CUMS+Iron group) for 6 days. Apoptosis was detected by TUNEL assay in the cortex (a) and hippocampus (c), where neuronal nuclei were stained with NeuN (red), while all the cell nuclei were labeled with DAPI (blue in merge). Percentage of cell death was determined by the ratio of TUNEL+ and NeuN+ cells, with the cortical and hippocampal regions analyzed in (b, d), respectively. Data are presented as mean ± SEM, n = 6. *p < 0.05, statistically significant difference compared with control, untreated group; **p < 0.05, statistically significant difference compared with any other group. (From Liang et al. 2020)

Growing evidence indicates a strong link between sleep deprivation and activation of nucleotide-binding domain and leucine-rich repeat protein-3 (NLRP3) inflammasome; this activation is mediated by astrocytic P2X₇ receptors in astrocytes, and it facilitates neuronal apoptosis (Xia et al. 2018; Zielinski et al. 2017). Chronic sleep deprivation selectively downregulates expression of astrocytic 5-HT_2B receptors, the effect of which is similarly mediated by P2X₇ receptors (Fig. 3; Xia et al. 2020). These changes result in depressive-like behaviors associated with the gradual elevation of extracellular ATP (Xia et al. 2020). Fluoxetine increases the phosphorylation of signal transducer and activator of transcription 3, thus suppressing astrocytic NLRP3 activation and neuronal apoptosis (Xia et al. 2018; Zielinski et al. 2017). Inhibition of astrocytic NLRP3 inflammasome by fluoxetine is also mediated by 5-HT_2B receptors (Li et al. 2019). In addition, fluoxetine, again acting through 5-HT_2B receptors, recovers BDNF levels suppressed by sleep deprivation (Li et al. 2019). Pre-treatment with the 16 kDa hormone leptin increases expression of astrocytic 5-HT_2B receptors, thus enhancing positive effects of fluoxetine on BDNF and depressive-like behaviors (Li et al. 2019). Of note, long-term, chronic stress may induce sleep disorders and depression associated with decreased levels of BDNF (Schmitt et al. 2016).

Fig. 3.

P2X₇ receptors mediate the selective decrease of 5-HT_2B receptors in astrocytes treated with chronic sleep deprivation. (a) Immunolabelled 5-HT_2B receptors (green) were co-stained with glial fibrillary acidic protein (GFAP, red) and NeuN (blue) in the mice frontal cortex treated with sham (Control) or exposed to sleep deprivation for 3 weeks of wild-type (WT) and P2X₇ receptor knockout (P2X₇R-KO) mice. Scale bar, 20μm. (b) 5-HT_2B receptors immunolabeling intensity of neurones and astrocytes, respectively, relative to the cell-free parenchyma in the cortex, normalized to the intensity of a matching sham group. (c) RT-PCR analysis of 5-HT_2B receptors mRNA expression in WT or P2X₇R-KO mice treated with sleep deprivation for 3 weeks, expressed as the relative expression ratio of 5-HT_2B receptor/GAPDH, the latter used as loading control. Data represent mean ± SEM. *Indicates statistically significant (p < 0.05) difference from matching sham groups, n = 6. (From Xia et al. 2020)

Pathological features and pathogenesis of sleep disturbances, depressive disorders, and neurodegenerative pathology, such as Alzheimer’s disease, usually overlap, and all these may be linked to astrocytes. Morphological and functional abnormalities of astrocytes can impair BDNF signaling and the glymphatic system, with failed glymphatic clearance being a key contributor to diseases of cognition (Nedergaard and Goldman 2020), which could further worsen the deposition of brain waste and impair neuronal functions. Fluoxetine and other SSRIs may provide neuroprotection by arresting astrogliopathology through direct stimulation of 5-HT_2B receptors.

SNRIs

Serotonin/noradrenaline reuptake inhibitors exert their effects through increasing availability of serotonin and noradrenaline in the nervous tissue, and they are commonly used to treat patients with refractory depression (Wang et al. 2014). Venlafaxine is one of the most commonly prescribed SNRI with a broad range of antidepressant effects. Accumulating evidence shows that venlafaxine may target astrocytes by affecting the metabolism of lysine, tyrosine, glutamate, methionine, ethanolamine, fructose-6-phosphate, and phosphorylethanolamine (Sun et al. 2017). Duloxetine, another SNRI widely used in the treatment of MDD (Nemeroff et al. 2002), was reported to suppress ischemia-induced astrocytic reactivity and oxidative stress (Lee et al. 2016).

TCAs and MAOIs

Astrocytic-derived neurotrophic growth factors receive attention because of their links to the mechanism of action of antidepressants (Tsybko et al. 2017). In primary cultured astrocytes, amitriptyline increases expression of BDNF, vascular endothelial growth factor (VEGF), fibroblast growth factor 2 (FGF2), and glial cell line-derived neurotrophic factor (GDNF) (Kajitani et al. 2012). Furthermore, clomipramine (TCA), fluvoxamine (SSRI), and duloxetine (SNRI) all elevate the expression of FGF2 and BDNF in astrocytes (Hisaoka-Nakashima et al. 2016; Kajitani et al. 2012). Fluoxetine and paroxetine both enhance the mRNA expression of VEGF in primary astrocytes (Allaman et al. 2011).

Mirtazapine, a tetracyclic piperazinoazepine antidepressant, has a distinct pharmacological mechanism associated with inhibition of α₂-adrenoceptors and 5-HT₂ and 5-HT₃ receptors (Benjamin and Doraiswamy 2011; Croom et al. 2009). Unlike amitriptyline and other antidepressants, mirtazapine does not affect noradrenaline or serotonin reuptake (Al-Majed et al. 2018). Moreover, mirtazapine has a faster onset time as compared with SNRIs in the treatment of acute-phase MDD (Nagao et al. 2013; Watanabe et al. 2011). In primary culture of astrocytes, mirtazapine increases the expression of GDNF and BDNF by stimulating phosphorylation of ERK^1/2 (Hisaoka-Nakashima et al. 2016).

Astrocytes in Treatments of Bipolar Disorders

Therapeutic strategy for bipolar disorders (BD) is distinct from treating MDD, due to reflecting different pathophysiology. Lithium ion (Li⁺) in various salts, valproic acid (VPA), and carbamazepine (CBZ) are three classical anti-bipolar drugs. Lithium salt was first to be used for treating mania and depressive episodes of bipolar depression; importantly, lithium may also prevent relapses (Brown and Tracy 2013). Both CBZ and VPA are used for treating manic episodes and epileptic seizures (Nevitt et al. 2019; Xu et al. 2019; Post et al. 2007). These three classical anti-BD drugs have nothing in common in their chemical structures, and only a few metabolic parameters in the brain are affected by them in a similar manner (Li et al. 2007). Revealing shared effects of these drugs is important because common mechanisms of action may provide information about the pathophysiology of BP and facilitate drug development.

Astroglial cell numbers in the cortex from BD patients are reduced (Ongür et al. 1998, 2008; Rajkowska 2000). This coincides with exaggerated activity of the glutamatergic system, as demonstrated by MRI, in the brains of BD patients (Ongür et al. 2008; Michael et al. 2003; Chen et al. 2010; Eastwood and Harrison 2010), which could be due to suppressed capacity of astrocytes to maintain glutamate homeostasis. In astrocytes, riluzole can increase the expression and activity of plasmalemmal glutamate transporters in order to increase the glutamate clearance from the extracellular space (Carbone et al. 2012). Riluzole can also improve the glutamine-glutamate cycle and the glutamatergic transmission in brains of BD patients (Brennan et al. 2010) and even alleviate related disease symptoms (Zarate and Manji 2008), which is also indicative of potential astrocytic glutamatergic dysfunction.

We outline some common targets of these three classical anti-BD drugs on astrocytes. Firstly, the decreased expression of GluK2 was reported in the entorhinal cortex and the hippocampus of BD patients, most of whom had received drug therapy (Beneyto et al. 2007; Benes et al. 2001). In primary cultures of astrocytes treated with Li⁺, VPA, or CBZ at clinically relevant doses the mRNA expression of GluK2 receptors decreased, whereas the expression of other subunits of GluK receptors remained unchanged (Li et al. 2009b). In vivo, after 2 weeks of intraperitoneal injection, CBZ reduced the gene expression of GluK2 which is encoded by GRIK2 (Li et al. 2009b); of note, this gene in a genetic linkage region (6q21) is associated with BD (Schulze et al. 2004). In other studies, CBZ, VPA, or Li⁺ suppresses GluK2-mediated phosphorylation of ERK1/2 in primary cultured astrocytes, while anticonvulsant topiramate is ineffective (Li et al. 2009b). Release of glutamate from astrocytes induced by ATP is inhibited by all three anti-BD drugs; this may suppress hyperactivity of the glutamatergic transmission in brains of patients with BD (Liu et al. 2015).

In addition to classical targets, Li⁺ can play a key role in the metabolism of inositol. The intracellular myoinositol needs continuous re-supply from the diet (Spector and Lorenzo 1975) and de novo synthesis, which occurs in the vasculature (Wong et al. 1987). In primary culture of mouse astrocytes, chronic treatment with Li⁺ at 1 mM inhibits the uptake of inositol, although this effect is absent during acute Li⁺ treatment (Wolfson et al. 1998). Changes in myoinositol levels may have opposite effects depending on its concentration. For example, myoinositol at low concentration promotes the uptake of the extracellular inositol, whereas at high concentrations the uptake is inhibited. This may be related to the existence of two inositol transporters in astrocytes, the high-affinity Na⁺-dependent myoinositol transporter (SMIT/SLC5A3), which accounts for most of the uptake at low (25μM) inositol concentration, and the lower-affinity H⁺-dependent myoinositol transporter (HMIT/SLC2A13), which dominates the uptake at higher (50μM) inositol concentrations (Wolfson et al. 2000). Because the elevated intracellular pH (pH_i) stimulates SMIT but inhibits HMIT, the effects of anti-BD drugs on the inhibition of inositol may be caused by the intracellular alkalinization. This was further confirmed by studying the uptake of [³H]myoinositol in astrocytes at various myoinositol doses or at various pH_i and corroborated by analysis of different myoinositol transporter operation (Fu et al. 2012). Chronic treatment by any of the three anti-BD drugs increases astrocytic pH_i (Song et al. 2008, 2013). The Na⁺–H⁺ exchanger (NHE) exchanger (NHE) is one of the acid-controlling transporters, widely distributed in almost all cell types, including neurones and astrocytes (McAlear and Bevensee 2006). The NHE/SLC9 family has seven subtypes, NHE1-NHE7. The NHE1/SLC9A1 is expressed in cultured astrocytes from the rat hippocampus (Pizzonia et al. 1996) and from the mouse cerebral cortex (Song et al. 2008). Chronic treatment with Li⁺ at 1 mM alkalinize pH_i by about 0.10, but 0.5 mM is ineffective (Song et al. 2008).

The anti-bipolar effects of lithium are often attributed to the classical hypothesis of the depletion of myoinositol induced by Li⁺ (Berridge et al. 1989). According to this hypothesis, Li⁺ causes the non-competitive inhibition of inositol-phosphate hydrolysis; InsP₃ is produced by PLC and degraded to myoinositol by inositol phosphatases, whereas diacylglycerol is converted to phosphatidate and condensed with cytidine triphosphate to form cytidine mono-phosphoryl-phosphatidate, which together with myoinositol regenerates phosphatidylinositol 4,5-bisphosphate (PIP₂) (Hallcher and Sherman 1980; Inhorn and Majerus 1987). Because Li⁺ inhibits the formation of inositol from InsP₃, it decreases the regeneration of PIP₂ and blocks PLC pathway. The [Ca²⁺]_i response to noradrenaline in cultured astrocytes is suppressed by chronic treatments with Li⁺ (Chen and Hertz 1996).

As astrocytes are the direct targets of Li⁺, identification of the unique astroglial transcriptional networks is of importance. Recently, it was found that Li⁺ can induce a specific astrocytic phenotype (Rivera and Butt 2019). By pharmacogenomic analyses, the roles of the extracellular matrix (ECM) regulatory enzyme lysyl oxidase (LOX) and peroxisome proliferator-activated receptor γ (PPAR-γ) were shown to modulate astrocyte morphogenesis. According to genomic analyses, LOX is the most highly regulated Li⁺-responsive astroglial gene. Irreversible LOX inhibitor mimics effects of Li⁺ suggesting that LOX is a major regulator of astrocyte morphology and proliferation. Treatment with Li⁺ also actives PPAR-γ (Liu et al. 2011), while PPAR-γ has been reported to inhibit LOX (Segond et al. 2013). These effects may involve GSK3β-Wnt signaling, which is implicated in neuropsychiatric disorders and interacts with ECM remodeling. In addition, PPAR-γ agonists decrease TGF-β1 signaling leading to decreased fibrosis (Vallee et al. 2017), providing a further link with LOX and the actions of Li⁺.

Neuroprotective effects of anti-bipolar drugs have been discovered in other neuronal diseases. Pretreatment with VPA increases expression of plasmalemmal glutamate transporter 1 (GLT-1) in astrocytes and promotes the recovery of locomotor activities in a rodent model (Johnson Jr et al. 2018). Lithium can mediate neuroprotection by inhibition of GSK3β after spinal cord injury (Li et al. 2018). Chronic treatment with Li⁺ activates α1 subunit of neuronal Na⁺-K⁺-ATPase (Li et al. 2018). In contrast, chronic treatment with CBZ increases astrocyte specific α2 subunit containing Na⁺/K⁺ ATPase (Li et al. 2013).

Conclusion

In this chapter, we introduced the contribution of astrocytic serotonin receptors to the pathophysiology of MDD and to pharmacological mechanisms of classic antidepressants. Direct stimulation of astroglial 5-HT_2B receptors may be the common target of SSRIs. Activation of astrocytic serotonin receptors associates with several signaling pathways and regulates expression of numerous genes that are related to depressive-like behaviors. Recent discoveries overviewed in this chapter highlight the key roles of astrocytes in the pathogenesis and treatment of mood disorders.