International Union of Basic and Clinical Pharmacology. CXI. Pharmacology, Signaling, and Physiology of Metabotropic Glutamate Receptors

A. Receptor Subtypes and Splice Variants

The group I mGlu receptors include mGlu₁ and mGlu₅. The mGlu₁ receptor gene (GRM1) and its first three splice variants were cloned in rat in 1992 (Pin et al., 1992; Tanabe et al., 1992). In humans, there are seven mGlu₁ splice variants (a, b, d, f, g, h) that differ in C terminus length (Sugiyama et al., 1987; Tanabe et al., 1992; Laurie et al., 1996; Makoff et al., 1997; Soloviev et al., 1999; DiRaddo et al., 2013) [Ensembl gene identifier: ENSG00000152822]. In addition, 12 single nucleotide polymorphisms within the GRM1 coding region have been identified in patients with schizophrenia (Frank et al., 2011; Ayoub et al., 2012), suggesting mGlu₁ may be a viable therapeutic target for psychosis (Cho et al., 2014b). Spontaneous mutations in GRM1 are also associated with ataxia (Watson et al., 2017). The mGlu₅ receptor is encoded by the GRM5 gene [ENSG00000168959], localized in human chromosome 11, and was first cloned in rat in 1992 (Abe et al., 1992) and in human in 1994 (Minakami et al., 1994). Alternative splicing of GRM5 in humans gives rise to two major isoforms that also differ in C-terminal length; the longer of the two, human mGlu₅a (equivalent to rat mGlu₅b) has a 32–amino acid insertion after residue 876 but is otherwise identical to human mGlu₅b (equivalent to rat mGlu₅a) (Minakami et al., 1993, 1995). Variations in C-terminal length due to alternative splicing of group I receptors influences surface expression, subcellular localization, dimerization, interactions with intracellular proteins, and ultimately cellular responses (Joly et al., 1995; Mion et al., 2001; Francesconi and Duvoisin, 2002; Kumpost et al., 2008; Tateyama and Kubo, 2008; Francesconi et al., 2009a; Techlovská et al., 2014).

B. Localization and Signal Transduction

The group I mGlu receptors are predominantly found in postsynaptic neurons within the CNS (Fig. 2), increasing neuronal excitability and membrane depolarization when activated. In certain circuits, group I mGlu receptors can be found on presynaptic terminals, acting as autoreceptors to modulate neurotransmitter release (reviewed in Pittaluga, 2016). Furthermore, group I mGlu receptors are also expressed in glial cells (reviewed in Spampinato et al., 2018). The cellular responses resulting from group I mGlu receptor activation are highly complex and context dependent.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Synaptic and nonneuronal localization of mGlu receptor subtypes. General overview of metabotropic glutamate receptor neuronal and glial cell localization within glutamatergic and GABAergic synapses. Relationships with other key receptors and transporters that recognize glutamate and GABA as well as ion channels involved in neurotransmission are also shown. Glutamate transporters represent multiple types, namely, excitatory amino acid transporters (EAATs): EAAT1 (also referred to as glutamate aspartate transporter) and EAAT2 (also known as glutamate transporter 1 or solute carrier family 1 member 2). The following abbreviations are used: Ca_V, voltage-gated Ca²⁺ channel; iGluRs, ionotropic glutamate receptors; K2P, two-pore domain K⁺ channel.

Group I mGlu receptors preferentially couple to the G_q/11 family of G proteins, which activate phospholipase C (PLC) β, which hydrolyses phosphatidylinositol 4,5-bisphosphate in the membrane to yield the second messengers: diacylglycerol and inositol 1,4,5-trisphosphate, mobilizing intracellular Ca²⁺ (iCa²⁺) stores (Sugiyama et al., 1987) (Fig. 3). Potentiation of mGlu₅ receptors increases phosphoinositide hydrolysis in vivo in different mouse brain areas, such as prefrontal cortex, cerebellum, hypothalamus, hippocampus, and striatum (Zuena et al., 2018). Interestingly, endogenous activation of mGlu₅ receptors largely accounts for basal phosphoinositide hydrolysis particularly in the prefrontal cortex. Downstream of these second messengers, activation of protein kinase C (PKC) and calmodulin triggers signaling cascades that ultimately phosphorylate and activate extracellular signal–regulated kinases 1 and 2 (ERK1/2), which regulate gene transcription associated with synaptic plasticity (long-term depression or long-term potentiation) (Servitja et al., 1999, 2003; Kanumilli et al., 2002; Page et al., 2006; Jin et al., 2013a; Hong et al., 2016). Diacylglycerol can be further broken down by diacylglycerol lipase to yield endogenous cannabinoid, 2-arachidonoylglycerol (Jung et al., 2005; Gregg et al., 2012). Beyond Gα_q/11-mediated signaling, βγ subunits can enhance or inhibit Ca²⁺ and K⁺ channels, including ionotropic glutamate receptors, to modulate neuronal excitability and membrane potential (reviewed in Valenti et al., 2002), through physical interactions mediated by scaffolding proteins (Tu et al., 1999) or indirect mechanisms driven by intracellular effectors (Fig. 3). The βγ subunits can also activate phosphoinositide-3-kinase (PI3K), which in turn activates RAC-alpha serine/threonine-protein kinase (Akt)–dependent signaling cascades implicated in protein synthesis–dependent long-term depression and cell survival (Hou and Klann, 2004; Page et al., 2006; Hullinger et al., 2015; Zhu et al., 2018). In recombinant systems mGlu₁ and mGlu₅ receptors also couple to Gα_s, stimulating adenylyl cyclases (ACs) and increasing cAMP production (Aramori and Nakanishi, 1992; Joly et al., 1995; Francesconi and Duvoisin, 1998, 2000; Nasrallah et al., 2018). In addition, group I receptors signal via G protein–independent mechanisms (e.g., Homer, src kinases, arrestins, transactivation of tyrosine kinases) to activate different kinase cascades that contribute to synaptic plasticity (Iacovelli et al., 2003; Yang et al., 2006; Emery et al., 2010; Kubota et al., 2014; Eng et al., 2016). Downstream of group I receptor activation diverse transcription factors are activated, including cAMP response element-binding protein (Mao and Wang, 2003b), Elk-1 (Mao and Wang, 2003a; Jong et al., 2005, 2009), c-Jun (Jong et al., 2009), and serum response factor (Kumar et al., 2012).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Signal transduction and regulation of group I mGlu receptors. Overview of group I mGlu receptor scaffolding partners, transducers, downstream effectors, and regulatory proteins; refer to main text for associated primary references. Blue bolded text indicates physiologic consequences linked to specific intracellular responses. The following abbreviations are used: AC, adenylyl cyclase; 2-AG, 2-arachidonoylglycerol; CaMK, Ca²⁺/calmodulin-dependent protein kinase; CREB, cAMP response element-binding protein; DAG, diacylglycerol; DGL, diacylglycerol lipase; EGFR, epidermal growth factor receptor; GKAP, guanylate kinase-associated protein; GRK, G protein–coupled receptor kinase; IP₃, inositol 1,4,5-trisphosphate; JNK, c-Jun N-terminal kinase; MEK1/2, mitogen-activated protein kinase kinases 1 and 2; mTOR, mammalian target of rapamycin; NHERF1/2, Na⁺/H⁺ exchange regulatory cofactors 1 and 2; PKA, protein kinase A; PLD, phospholipase D; PSD-95, postsynaptic density protein 95 (also known as SAP-90 for synapse-associated protein 90); PP2B, protein phosphatase 2B; PP2A/CaN, protein phosphatase 2A/calcineurin; S6K1, ribosomal protein S6 kinase β-1; SHANK, SH3 and multiple ankyrin repeat domains protein; SIAH-1a, E3 ubiquitin-protein ligase SIAH1A; SRF, serum response factor.

The mechanisms that regulate group I mGlu receptor activity are equally complex (Fig. 3). The C termini of group I receptors contain binding sites for diverse scaffolding proteins that regulate receptor localization and recycling in addition to directly linking group I mGlu receptors to other receptors and channels within the postsynaptic density (Roche et al., 1999; Kitano et al., 2002; Lee et al., 2008; Wang et al., 2009; Hu et al., 2012; Wagner et al., 2015; Eng et al., 2016; Gulia et al., 2017). Second messenger-activated kinases also provide negative feedback regulating cellular responses (iCa²⁺ oscillations or receptor desensitization) through phosphorylation of intracellular loops and/or the C terminus (Kawabata et al., 1996; Gereau and Heinemann, 1998; Bhattacharya et al., 2004; Mundell et al., 2004; Kim et al., 2005; Bradley and Challiss, 2011; Ko et al., 2012; Jin et al., 2013a,b, 2018; Raka et al., 2015; Uematsu et al., 2015; Vergouts et al., 2017; Yang et al., 2017; Marks et al., 2018). However, not all functional responses are equally influenced. For example, PKC phosphorylation of mGlu₁ receptors desensitizes signaling to accumulation of inositol phosphate but not cAMP (Francesconi and Duvoisin, 2000). Select kinases bind the C terminus and/or phosphorylate the receptor altering effector coupling, ability to bind other proteins, or receptor endocytosis (Dale et al., 2000; Sallese et al., 2000; Iacovelli et al., 2003; Pula et al., 2004; Nicodemo et al., 2010; Jin et al., 2017). By example, Preso1 enhances cyclin-dependent kinase 5 and ERK1/2 phosphorylation of the Homer binding site within the C terminus (Hu et al., 2012), whereas calmodulin and E3 ubiquitin-protein ligase SIAH-1a recognize overlapping sites in the C tail (Ishikawa et al., 1999), with PKC phosphorylation of this site enhancing SIAH-1a but inhibiting calmodulin binding (Ko et al., 2012). Second messenger-dependent kinases are critical for group I mGlu receptor–dependent long-term depression and potentiation by modulating the activity or promoting endocytosis of ionotropic glutamate receptors including N-methyl-d-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor subtypes (Jia et al., 1998; Snyder et al., 2001; Benquet et al., 2002; Moult et al., 2006; Jin et al., 2013b, 2015; Xu et al., 2013), although PKC-independent mechanisms for NMDA receptor potentiation by group I mGlu receptors have also been reported (Harvey and Collingridge, 1993; Kinney and Slater, 1993; Rahman and Neuman, 1996). In a reciprocal fashion, NMDA receptor-mediated stimulation of protein phosphatase 2A/calcineurin acts to regulate recycling of group I mGlu receptors (Alagarsamy et al., 2005; Pandey et al., 2014), with protein phosphatase 2B playing a similar role (Mahato et al., 2015). It is clear that activation of group I mGlu receptors can trigger a complex network of intracellular effectors that encode the cellular responses that give rise to complex physiologic effects from synaptic plasticity to cell survival.

When coexpressed within the same cell population, activation of mGlu₁ versus mGlu₅ receptors can lead to distinct cellular outcomes (Valenti et al., 2002). Moreover, postsynaptic group I mGlu receptor activation can result in modulation of neurotransmitter release from presynaptic cells via retrograde signaling, for example, endocannabinoids or nitric oxide (Fig. 3) (Maejima et al., 2001; Robbe et al., 2002; Sergeeva et al., 2007; Gregg et al., 2012; Aubrey et al., 2017; Xiang et al., 2019). As a further layer of complexity, both group I mGlu receptors form heteromers and/or larger-order oligomers with other G protein–coupled receptors (GPCRs) (discussed in further detail later), which alters intracellular signaling profiles.

C. Pathophysiology and Therapeutic Potential

A number of reviews provide in depth coverage of the distribution, physiology, and pathophysiological roles of group I mGlu receptors (Ferraguti and Shigemoto, 2006; Niswender and Conn, 2010; Golubeva et al., 2016; Crupi et al., 2019; Pereira and Goudet, 2019). Inhibitors and activators of the individual subtypes are being pursued for a myriad of different psychiatric and neurologic disorders. Briefly, and of relevance to the pharmacological agents reviewed in depth below, distribution of mGlu₁ receptors in regions associated with pain perception as well as mGlu₁ knockout animal phenotypes suggests mGlu₁ inhibitors are potential therapeutic agents for neuropathic pain (Neugebauer, 2002; Schkeryantz et al., 2007). Preclinical studies also indicate mGlu₁ receptor inhibitors may have therapeutic benefit in treating seizures, addiction, anxiety, and certain cancers (Namkoong et al., 2006; Dravolina et al., 2017). mGlu₅^−/− mice have deficits in prepulse inhibition (Brody et al., 2004), impaired learning and memory (Xu et al., 2009; Zeleznikow-Johnston et al., 2018), and reduced propensity for addiction/abuse (Chiamulera et al., 2001) and reverse the majority of fragile X syndrome phenotypes in preclinical models (Dölen and Bear, 2008). As such, selective mGlu₅ receptor inhibitors are desired in the setting of depression (Chaki and Fukumoto, 2018) and anxiety (Ferraguti, 2018) as well as neurodevelopmental disorders such as fragile X syndrome (Yamasue et al., 2019). In this respect multiple mGlu₅ receptor inhibitors have entered phase 2 clinical trials as discussed in detail later; however, to date none have reached market. Potentiation or activation of both group I receptors offers the promise for treating the positive and cognitive symptoms associated with schizophrenia (Walker and Conn, 2015; Nicoletti et al., 2019).

Beyond neurologic and psychiatric disorders, inhibition of mGlu₁ receptors is neuroprotective (in vitro and in vivo) after oxygen-glucose deprivation or ischemic insult (Henrich-Noack et al., 1998; Pellegrini-Giampietro et al., 1999) and may therefore offer a novel intervention for stroke. For multiple preclinical models of neurodegenerative diseases, genetic ablation or pharmacological inhibition of mGlu₅ receptors is neuroprotective and treats associated symptoms, for example, acting procognitively in Alzheimer’s disease or Huntington’s disease (reviewed in Ribeiro et al., 2017) or improving motor deficits in amyotrophic lateral sclerosis (reviewed in Battaglia and Bruno, 2018) or Parkinson’s disease (Mazur, 1995; Battaglia et al., 2004; Armentero et al., 2006; Ambrosi et al., 2010; Black et al., 2010; Masilamoni et al., 2011; Fuzzati-Armentero et al., 2015), although mGlu₅ receptor activators/potentiators may also treat cognitive symptoms associated with Huntington’s disease (Doria et al., 2013, 2015, 2018). Inhibition of mGlu₅ receptors is also indicated for treating neurodegeneration associated with drugs of abuse (Battaglia et al., 2002).

A. Receptor Subtypes and Splice Variants

The group II members, mGlu₂ and mGlu₃ receptors, were first cloned in rat in 1992 (Tanabe et al., 1992) and a few years later in human (Flor et al., 1995a; Emile et al., 1996). Encoded by the GRM2 gene [ENSG00000164082] and localized in human chromosome 3 and rat chromosome 8, no splice variants have been described for mGlu₂ receptor subtype (Sartorius et al., 2006). In human, the GRM3 gene [ENSG00000198822] encodes the mGlu₃ receptor, for which three splice variants are known (Sartorius et al., 2006). The most abundant GRM3 variant lacks exon 4 (GRM3Delta4), encoding a truncated membrane-associated protein that retains the extracellular VFT but lacks the 7TM, which is replaced with a unique 96–amino acid C terminus. mGlu_3delta4 can bind orthosteric ligands and interact with the full-length protein and may thus have a dominant negative effect (García-Bea et al., 2017). Spontaneous mutations in mGlu₃ are associated with melanoma (Prickett et al., 2011; Neto and Ceol, 2018), whereas single nucleotide polymorphisms in GRM3 are linked to cognitive performance in individuals with schizophrenia and are postulated to influence pharmacotherapy (reviewed in Maj et al., 2016 and Saini et al., 2017).

B. Localization and Signal Transduction

The group II mGlu receptors are located both pre- and postsynaptically, with mGlu₃ receptors also found in glial cells throughout the brain (Fotuhi et al., 1994; Testa et al., 1994) (Fig. 2). Using a radiolabeled orthosteric agonist selective for group II receptors, (1R,2S,4R,5R,6R)-2-amino-4-fluorobicyclo[3.1.0]hexane-2,6-dicarboxylic acid, in conjunction with knockout mice for either subtype, mGlu₃ receptor levels were found to be generally higher than mGlu₂ receptors in forebrain regions but similar within the striatum (Wright et al., 2013). In the thalamus and hippocampus, the two subtypes exhibit distinct and, for some nuclei/circuits, segregated expression patterns (Wright et al., 2013). Dissecting the relative contribution of mGlu₂ versus mGlu₃ receptor subtypes has presented a major challenge due to the lack of subtype-selective pharmacological tools (discussed in further detail below).

Group II mGlu receptors preferentially couple to G_i/o proteins, inhibiting adenylyl cyclase and cAMP production as well as inhibiting guanylate cyclase and cGMP production (Wroblewska et al., 2006) (Fig. 4). On presynaptic terminals, βγ subunits modulate ion channel function, inhibiting N-type Ca²⁺ channels (McCool et al., 1996) and activating G protein–coupled inwardly rectifying potassium channel (GIRK) channels (Knoflach and Kemp, 1998; Sharon et al., 1997), thereby decreasing exocytosis of vesicles containing glutamate (Macek et al., 1998; Flavin et al., 2000; Olivero et al., 2017), GABA (Hayashi et al., 1993; Gereau and Conn, 1995; Salt and Eaton, 1995; Schaffhauser et al., 1998) and dopamine (Feenstra et al., 1998; Verma and Moghaddam, 1998; Johnson et al., 2017). The βγ subunits can also activate PI3K-dependent activation of Akt with downstream effectors regulating protein synthesis and gene transcription involved in cytoprotection and synaptic plasticity (Aronica et al., 2003; Ciccarelli et al., 2007; Durand et al., 2011; Li et al., 2015; Ritter-Makinson et al., 2017; Xing et al., 2018). Furthermore, group II receptor activation can trigger transactivation of insulin growth factor-1 receptors via βγ subunit activation of PLC and focal adhesion kinase, leading to ERK1/2 phosphorylation (Hu et al., 2019), a key integrator of multiple convergent pathways that shapes the overall cellular response (Aronica et al., 2003; Ciccarelli et al., 2007). In postsynaptic membranes, group II receptor activation regulates trafficking of NMDA or AMPA receptors with different mechanisms implicated: PKC, soluble N-éthylmaleimide-sensitive-factor attachment protein receptor complexes, Akt/glycogen synthase kinase (GSK) 3β (Tyszkiewicz et al., 2004; Xi et al., 2011; Cheng et al., 2013). Furthermore, mGlu₃ receptor-dependent long-term depression (LTD) in the prefrontal cortex is mediated by functional cross-talk between mGlu₃ and mGlu₅ receptors on postsynaptic neurons (Fig. 2). Activation of mGlu₃ receptors releases βγ subunits to enhance mGlu₅ receptor coupling to G_q (Di Menna et al., 2018). Cross-talk between mGlu₃ and mGlu₅ receptors is postulated to account for the reported dependence on PLC and PKC for mGlu₃ receptor-dependent LTD (Otani et al., 2002; Huang et al., 2007). Recently, however, mGlu₃ receptor-dependent LTD was shown to be modulated by mGlu₅ receptor interactions with homer and signaling via PI3K, Akt, and GSK3β to result in AMPA receptor internalization, a mechanism disrupted by acute stress (Joffe et al., 2019). Indeed, the cellular context is an important contributor to shaping the cellular responses instigated by group II receptor activation. For example, in astrocytes, mGlu₃ receptor activation results in both decreases and increases in cAMP, whereas in neurons the effect on cAMP levels is purely inhibitory (Moldrich et al., 2002). Astrocytic mGlu₃-mediated increases in cAMP levels are dependent on iCa²⁺ levels (mobilization of stores and extracellular influx) and local release of adenosine, which acts at colocated G_s-coupled adenosine A_2A receptors (Moldrich et al., 2002).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Signal transduction and regulation of group II mGlu receptors. Overview of group II mGlu receptor scaffolding partners, transducers, downstream effectors, and regulatory proteins; refer to main text for associated primary references. Blue bolded text indicates physiologic consequences linked to specific intracellular responses. The following abbreviations are used: AC, adenylyl cyclase; CREB, cAMP response element-binding protein; DAG, diacylglycerol; FAK, focal adhesion kinase; GC, guanylate cyclase; GRIP, glutamate-receptor interacting protein; GRK, G protein–coupled receptor kinase; IGF-1R, insulin growth factor-1 receptor; IP₃, inositol 1,4,5-trisphosphate; JNK, c-Jun N-terminal kinase; mTOR, mammalian target of rapamycin; NHERF1/2, Na⁺/H⁺ exchange regulatory cofactors 1 and 2; NMDAR, NMDA receptor; PICK1, protein interacting with C kinase; PKA, protein kinase A; PP2C, protein phosphatase 2C; RanBPM, Ran-binding protein microtubule-organizing center; SNAP-25, synaptosomal-associated protein 25kDa.

With respect to regulation of group II mGlu receptor activity, the cellular context is also a major contributor. Phosphorylation of intracellular loops and/or the C-terminal tail by protein kinase A, PKC, and G protein–coupled receptor kinases have a central role in receptor regulation (Kamiya and Yamamoto, 1997; Macek et al., 1998; Schaffhauser et al., 2000; Cai et al., 2001), uncoupling receptors from G proteins, and promoting interactions with scaffolding proteins such as arrestins, which mediate receptor endocytosis (Iacovelli et al., 2009). The C-terminal tail of group II mGlu receptors binds to multiple scaffolding proteins including protein interacting with C kinase, glutamate receptor-interacting protein, tamalin, Na⁺/H⁺ exchange regulatory cofactors 1 and 2, and Ran-binding protein microtubule-organizing center (Hirbec et al., 2002; Kitano et al., 2002; Seebahn et al., 2008; Ritter-Makinson et al., 2017), as well as protein phosphatase 2C, which dephosphorylates mGlu₃ receptors (Flajolet et al., 2003). Interactions between group II mGlu receptors with diverse intracellular scaffolding proteins regulates receptor localization and functional responses, which likely governs differences observed between cell types or for neurons from different brain regions. Interestingly, mGlu₂, but not mGlu₃, receptors are reportedly resistant to homologous desensitization by G protein–coupled receptor kinases with respect to cAMP signaling (Iacovelli et al., 2009), although heterologous mechanisms, for example, due to PKC activation by colocated adenosine A₃ receptors, affect both subtypes (Macek et al., 1998; Lennon et al., 2010). Functional cross-talk between mGlu₂ receptors and colocated serotonin receptors 2A (5-HT_2A) can also modulate cellular responses to activation of either receptor (Marek et al., 2000; Molinaro et al., 2009; Murat et al., 2019). The interplay of intracellular effectors stimulated by group II mGlu receptors, coupled with regulatory proteins as well as coexpression of other cell surface receptors, gives rise to cell type–specific roles for group II mGlu receptors.

C. Pathophysiology and Therapeutic Potential

The expression patterns of group II mGlu receptors (reviewed in Ferraguti and Shigemoto, 2006) coupled with phenotypes of knockout animals suggest that group II receptors are attractive therapeutic targets for psychosis, cognition, anxiety, pain, and addiction (Cross et al., 2018; Mazzitelli et al., 2018; Pereira and Goudet, 2019), although for many indications establishing whether mGlu₂ or mGlu₃ receptors are the best target in preclinical models of disease has been challenging due to a lack of subtype-selective agents. Despite this shortcoming, both agonists and potentiators of group II receptors have been actively pursued, with multiple agents entering phase 2 or 3 for treating schizophrenia and addiction disorders (Nicoletti et al., 2019). Conversely, group II receptors inhibitors are promising interventions for depression and anxiety and as neuroprotective agents in the setting of ischemia (Celanire et al., 2015; Motolese et al., 2015). However, activation of group II receptors is neuroprotective after excitotoxic insults (Battaglia et al., 2003). In particular, activation of astrocytic mGlu₃ receptors confers neuroprotection to various insults (in vitro and in vivo) via paracrine mechanisms (Bruno et al., 1997; Corti et al., 2007; Cippitelli et al., 2010; Caraci et al., 2011), offering a potential therapeutic target for neurodegenerative diseases and psychiatric conditions associated with neuronal death.

A. Receptor Subtypes and Splice Variants

Before molecular cloning, group III mGlu receptors were termed L-(+)-2-amino-4-phosphonobutyric acid (L-AP4) receptors due to high sensitivity to this ligand, which depresses synaptic transmission in the brain and in retina (Thomsen, 1997). In the 1990s, molecular cloning revealed that four different receptor subtypes mediated the biologic effects of L-AP4: mGlu₄, mGlu₆, mGlu₇, and mGlu₈ receptors. The mGlu₄ receptor is encoded by the GRM4 gene [ENSG00000124493], which is localized in human chromosome 6, rat chromosome 20, and mouse chromosome 17 (Tanabe et al., 1992; Flor et al., 1995b). Two main splice variants were predicted, termed mGlu_4a and _4b (Thomsen et al., 1997), but further studies failed to detect mGlu_4b in different rat brain areas, and no splice site consensus sequences that could support its existence were found in human genomic sequence containing the whole GRM4 gene, suggesting mGlu_4b corresponds to a recombination artifact (Corti et al., 2002). Another variant lacking the first 128 base pairs, termed taste mGlu₄, is found in rat taste buds. The corresponding protein is predicted to lack approximately half the extracellular domain, including a large portion of the glutamate-binding domain (Chaudhari et al., 2000). First cloned in 1993, the mGlu₆ receptor is encoded by GRM6 gene [ENSG00000113262] localized in human chromosome 5 (Nakajima et al., 1993). Two splice variants, termed mGlu_6b and mGlu_6c, are found in both human and rats, which correspond to truncated mGlu₆ receptors lacking the transmembrane domain and intracellular portions of the receptor (Valerio et al., 2001a,b). The mGlu₇ receptor is encoded by the GRM7 gene, localized in rat chromosome 4 and in human chromosome 3 (Okamoto et al., 1994; Saugstad et al., 1994; Makoff et al., 1996). Two main variants, mGlu₇a and mGlu₇b, in rats and humans have different C-terminal tails. The last 16 residues of mGlu_7a are substituted by 23 different residues in mGlu_7b due to the insertion of an out-of-frame 92–base pair exon (Flor et al., 1997; Corti et al., 1998). Three other isoforms have been described, named v3, v4, and v5 (Schulz et al., 2002). Although mGlu_7a and mGlu_7b receptor variants are primarily expressed in the CNS, the expression of v3 and v4 isoforms appears to be restricted in nonneuronal tissues (Schulz et al., 2002). The mGlu₈ receptor, encoded by the GRM8 gene [ENSG00000179603] found in human chromosome 7, was originally cloned in mouse in 1995 (Duvoisin et al., 1995) followed by human (Scherer et al., 1997; Wu et al., 1998) and rat (Saugstad et al., 1997). Three splice variants have been described: mGlu_8a, mGlu_8b, and mGlu_8c (Corti et al., 1998; Malherbe et al., 1999). The last 16 residues of the C-terminal tails of the mGlu_8a and mGlu_8b receptors are different, whereas the mGlu_8c variant is a truncated receptor lacking the transmembrane domains and intracellular C tail. For each of the group III mGlu receptors the different splice variants show distinct tissue distribution and/or changes in intracellular portions of the receptor, which have the potential to alter signal transduction pathways triggered in response to receptor activation.

B. Localization and Signal Transduction

Most group III mGlu receptors are widely expressed throughout the CNS, with the exception of mGlu₆, which is mostly restricted to the retina (reviewed in Ferraguti and Shigemoto, 2006). Group III mGlu receptors are also expressed in glial cell types in the brain, with the exception of mGlu₇ (reviewed in Spampinato et al., 2018). Outside the CNS, group III mGlu receptor expression has been reported in kidney, pancreas, liver, cells from the immune system, and bones, for example (see Julio-Pieper et al., 2011, for review).

In the CNS, mGlu₄, mGlu₇, and mGlu₈ receptors are mainly expressed in the active zone of presynaptic glutamatergic and GABAergic neurons (Kinoshita et al., 1996a; Shigemoto et al., 1997; Wada et al., 1998; Corti et al., 2002; Ferraguti et al., 2005; Ferraguti and Shigemoto, 2006) (Fig. 2). Group III receptors can act as autoreceptors reducing glutamate release in the synaptic cleft and as heteroreceptors reducing the release of GABA (Schoepp, 2001). mGlu₄, mGlu₇, and mGlu₈ receptors are preferentially coupled to heterotrimeric G_i/o proteins, leading to the inhibition adenylyl cyclase production of cAMP via the Gα subunit (Fig. 5). On presynaptic terminals, mGlu₄, mGlu₇, or mGlu₈ receptors are well documented to inhibit neurotransmitter release through a direct βγ subunit–mediated inhibition of N or P/Q type of voltage-gated Ca²⁺ channels (Anwyl, 1999; Perroy et al., 2000; Millán et al., 2002a,b; Capogna, 2004; Martín et al., 2007). In addition, mGlu₄ and mGlu₇ receptors also inhibit voltage-gated Ca²⁺ channels via a PKC-dependent mechanism (Perroy et al., 2000; Abitbol et al., 2012), involving interactions between calmodulin and the scaffolding protein, protein interacting with C kinase, for mGlu₇ receptors (Perroy et al., 2002; Suh et al., 2013). Activation of mGlu₄, mGlu₇, or mGlu₈ receptors can also decrease neuronal excitability by the released βγ subunits acting on GIRK channels (Dutar et al., 1999; Saugstad et al., 1996). Other studies suggest that group III mGlu receptors may activate background K⁺ channels such as TREK1 and TREK2 (potassium channel subfamily K member 2 and 10), thereby further decreasing neuronal activity (Lesage et al., 2000; Cain et al., 2008). Group III mGlu receptors are also proposed to inhibit neurotransmitter vesicle exocytosis through direct interactions with the release machinery (Chavis et al., 1998; Erdmann et al., 2012). Coupling of group III receptors to PI3K, probably through Gβγ subunits, and mitogen-activated protein kinase is implicated in neuroprotection mechanisms (Iacovelli et al., 2002).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Signal transduction and regulation of group III mGlu receptors. Overview of group III mGlu receptor scaffolding partners, transducers, downstream effectors, and regulatory proteins; refer to main text for associated primary references. Blue bolded text indicates physiologic consequences linked to specific intracellular responses. The following abbreviations are used: AC, adenylyl cyclase; DAG, diacylglycerol; IP₃, inositol 1,4,5-trisphosphate; PICK1, protein interacting with C kinase; PKA, protein kinase A.

Among group III mGlu receptors, the mGlu₆ receptor is distinct as its expression is mostly restricted to postsynaptic bipolar ON neurons in the retina (Nomura et al., 1994; Vardi et al., 2000), with no strong expression detected in the brain (Nakajima et al., 1993). The mGlu₆ receptor is critical for glutamate-induced signaling in ON-bipolar cells in the retina during darkness (Nomura et al., 1994; Vardi et al., 2000). Glutamate, released from rod photoreceptors in the dark, activates postsynaptic mGlu₆ receptors in bipolar ON cells that lead to the closure of a nonselective ion channel, TRPM1-L [a long form transcript of transient receptor potential cation channel subfamily M member 1 (TRPM1) expressed solely in the dendritic tip of bipolar ON neurons] (Koike et al., 2010). TRPM1 inhibition results in hyperpolarization of bipolar ON neurons, thus inhibiting the ON pathway into darkness. The signaling cascade involves Goα (Koike et al., 2010), Gβγ (Shen et al., 2012), and other proteins such as the orphan GPCR GPR179 or the interacting protein Nyctalopin (Zeitz et al., 2015) (Fig. 5). Collectively, the intracellular signal transduction and regulatory pathways engaged by group III receptor subtypes have not been as well elucidated as group I and group II counterparts.

C. Pathophysiology and Therapeutic Potential

The phenotypes of mice lacking the group III mGlu receptors have revealed physiologic roles and potential as therapeutic targets in several neurologic disorders. Mice lacking the mGlu₄ receptor present deficits in motor performance, spatial memory, and learning of complex motor tasks (Pekhletski et al., 1996; Gerlai et al., 1998), in accordance with its particularly high expression in the cerebellum (Kinoshita et al., 1996b; Corti et al., 2002). The mGlu₄^−/− mice also have enhanced seizure-associated vulnerability (Pitsch et al., 2007) and lack the motor stimulant effect of ethanol (Blednov et al., 2004). The sensitivity to strong noxious stimuli of mGlu₄^−/− mice is altered, and nociceptive behavior in the inflammatory phase of the formalin test is accelerated (Vilar et al., 2013). Knockout phenotypes together with preclinical studies highlight mGlu₄ receptor as a potential therapeutic target in anxiety and depression (Kalinichev et al., 2014), schizophrenia (Wierońska et al., 2012a), epilepsy (Pitsch et al., 2007; Ngomba et al., 2008), neuroinflammation (Fallarino et al., 2010), autism spectrum disorder (Becker et al., 2014), and chronic pain (reviewed in Pereira and Goudet, 2019). In particular, targeting mGlu₄ receptor for the treatment of Parkinson’s disease (PD) has attracted much attention (Célanire and Campo, 2012; Amalric et al., 2013; Charvin, 2018; Volpi et al., 2018). Preclinical studies showed that mGlu₄ receptor activation corrects the imbalance of neurotransmission among the basal ganglia circuitry that is associated with PD (Charvin et al., 2018b), as shown primarily with mGlu₄ potentiation (Marino et al., 2003b) or later with selective agonists (Marino et al., 2003b; Beurrier et al., 2009). Activation or potentiation of mGlu₄ receptors also has neuroprotective effects (Copani et al., 1995; Battaglia et al., 2006). Unfortunately, despite promising preclinical results (Charvin et al., 2017, 2018a), the mGlu₄ receptor potentiator, foliglurax, recently failed to show sufficient efficacy in a phase II clinical trial for PD.

Comparing the phenotypes of mGlu₄, mGlu₇, and mGlu₈ knockout mice indicates the mGlu₄ receptor is most clearly involved in startle and motivational processes, whereas mGlu₇ receptor is involved in hippocampus-dependent spatial learning and fear-related behaviors, and mGlu₈ receptor deletion yields more subtle behavioral changes and influences body weight (Goddyn et al., 2015). The role of mGlu₇ receptors in learning and memory is confirmed by behavioral pharmacology studies (Hikichi et al., 2010a; Klakotskaia et al., 2013). Also, mGlu₇^−/− mice and mice lacking functional mGlu₇ receptors present an increased susceptibility to seizures (Sansig et al., 2001; Bertaso et al., 2008). Accordingly, absence seizures can be induced by the pharmacological blockade of the mGlu₇ receptor (Tassin et al., 2016), whereas mGlu₇ receptor activation protects against epileptogenesis and epileptic seizures (Girard et al., 2019). Widely expressed in the CNS (Kinzie et al., 1997; Shigemoto et al., 1997; Corti et al., 1998; Kinoshita et al., 1998), the mGlu₇ receptor is also considered as a potential therapeutic target for anxiety and depression (Cryan et al., 2003) and neurodevelopmental disorders (O’Connor et al., 2010; Palazzo et al., 2016; Fisher et al., 2018). Genetic disruptions in GRM7 are evident in patients with autism spectrum disorders (Liu et al., 2012; Sanders et al., 2012). Furthermore, the potentiation of mGlu₇ receptor activity improves cognitive and social deficits, as well as respiratory impairments, in a mouse model of Rett syndrome (Gogliotti et al., 2017).

Interestingly, depending on the brain structure, mGlu₇ and mGlu₈ receptors play opposing roles in pain (Boccella et al., 2020). For example, mGlu₇ receptor activation in periaqueductal gray and amygdala is pronociceptive, whereas mGlu₈ receptor activation is antinociceptive (Marabese et al., 2007; Palazzo et al., 2008). Indeed, for mGlu₇ receptors, proalgic or analgesic activity depends on the brain structure and circuits in which the receptor is expressed (see Pereira and Goudet, 2018, for review). For example, activation of mGlu₇ receptors in the nucleus accumbens has an antinociceptive effect (Kahl and Fendt, 2016), opposite to the pronociceptive effect when periaqueductal gray mGlu₇ receptors are activated (Palazzo et al., 2016). Interestingly, mGlu₇ receptor activation also prevents the development of morphine tolerance (Gawel et al., 2018). Additional studies confirmed mGlu₈ receptor modulates sensory symptoms associated to neuropathic pain (Rossi et al., 2014). Furthermore, mGlu₈ receptor activation in the nucleus tractus solitarius enhances cardiac nociception (Liu et al., 2012). Distinguishing the different physiologic and pathophysiological roles for mGlu₇ and mGlu₈ receptors has been hampered by a lack of selective pharmacological tools; however, discovery of new pharmacological tools (discussed in further detail later) have aided dissection of different roles and therapeutic indications.

The mGlu₈ receptor is one of the least studied mGlu receptor family members, due notably to the lack of selective pharmacological tools. Investigations on the phenotype of genetically modified mice lacking the mGlu₈ receptor are thus particularly informative. In various studies, mGlu₈^−/− mice exhibit anxiety-related phenotypes. However, some studies report an anxiogenic-like phenotype (Linden et al., 2003; Duvoisin et al., 2005, 2011), whereas others find an anxiolytic-like one (Gerlai et al., 2002; Fendt et al., 2010, 2013). Also, mGlu₈^−/− mice present robust deficits in contextual fear conditioning, novel object recognition, extinction of operant conditioning, and acoustic startle response (Fendt et al., 2010, 2013). mGlu₈^−/− mice also show enhanced social interaction; however, enhancing mGlu₈ receptor activity does not affect social interaction in wild-type mice (Duvoisin et al., 2011). Further studies are required to better understand the role of mGlu₈ and to clarify its therapeutic potential.

Outside of the CNS, the mGlu₆ receptor plays an important role in visual discrimination in low light conditions (Nomura et al., 1994; Vardi et al., 2000), supported by the mGlu₆^−/− phenotype (Masu et al., 1995). Mutations in proteins involved in the transmission of the signal between rod photoreceptors and bipolar ON cells have been found in patients suffering from congenital stationary night blindness; these include more than 20 loss-of-function mutations in GRM6 (Dryja et al., 2005; O’Connor et al., 2006; Zeitz et al., 2007, 2015). These mutations affect the normal mGlu₆ receptor response to the glutamate released from the photoreceptors, thus impairing signal transmission. The use of an optogenetic tool consisting of a chimera between mGlu₆ receptor and melanopsin receptor has been suggested as a potential approach to restore light sensitivity (van Wyk et al., 2015).

A. Definitions and Mode of Action

By definition, orthosteric ligands act in the same binding pocket as the endogenous ligand, competing to either activate or inhibit mGlu receptor activity. Glutamate and surrogate orthosteric agonists bind in the cleft between the two VFT lobes (Fig. 1). Upon binding, orthosteric agonists stabilize the closed state of the VFT, leading to a change in the relative orientation such that the extracellular domain dimer changes from a “resting” (R) to an active (A) state (Bessis et al., 2002). On the contrary, orthosteric antagonists prevent the full closure of the VFT (Bessis et al., 2000, 2002; Kunishima et al., 2000; Tsuchiya et al., 2002). Based on crystal structures, the main conformations that define the inactive and active states of mGlu receptors are the resting state Roo where both VFTs are open, and the active states Aco or Acc where one or both VFTs are closed, respectively. The two lobes are distant in the resting state and become closer in the active state (Kunishima et al., 2000; Tsuchiya et al., 2002). The closure of one VFT (Aco) is sufficient to induce a functional response from the receptor, but the closure of both VFTs (Acc) is necessary to achieve full activation (Kniazeff et al., 2004).

B. Selectivity

The l-glutamate binding site is highly conserved among the mGlu receptor family, resulting in difficulties in identifying compounds with subtype selectivity. Indeed, glutamate binds to all mGlu receptors under a similar conformation, where the residues participating in direct interactions with its amino acid moiety are fully conserved as well as two residues that interact with the carboxylate moiety (Bertrand et al., 2002; Acher and Bertrand, 2005; Wellendorph and Brauner-Osborne, 2009). Within the orthosteric pocket, several residues that do not directly interact with glutamate are different between the three groups of mGlu receptors, enabling identification of group I, group II, and group III selective ligands (Table 1).

The most commonly used agonists of group I mGlu receptors are 3,5-dihydroxyphenylglycine (DHPG) and quisqualic acid, which are somewhat selective for group I over group II and III receptors (Table 1). Concerning group II mGlu receptors, the classic agonists are (2S,2’R,3′R)-2-(2’,3′-dicarboxycyclopropyl)glycine (DCG-IV) (Brabet et al., 1998) and (1S,2S,5R,6S)-2-amino-bicyclo[3.1.0]hexane-2,6-dicarboxylic acid (LY354740) (Monn et al., 1997), whereas L-AP4 and ACPT-I (Acher et al., 1997) are most commonly used agonists for group III mGlu receptors (Fig. 6). LY341495 is the most used competitive antagonist, which antagonizes all the subtypes, with higher potency at mGlu_2/3 receptors over other subtypes (Kingston et al., 1998) (Fig. 6).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Structures of select orthosteric ligands of mGlu receptors. Representative orthosteric ligands for mGlu receptors. Antagonists are labeled with black text, and agonists are in blue. Detailed in vitro and in vivo pharmacological profiles are listed in Table 1.

For subtype-selective orthosteric ligands, drug designers have to circumvent the highly conserved binding pocket. One way is to design compounds able to interact with proximal residues to the glutamate binding pocket that differ between subtypes. By example, newly discovered orthosteric ligands can discriminate between mGlu₂ and mGlu₃ receptors, for example, (1S,2S,4R,5R,6S)-2-amino-4-methylbicyclo[3.1.0]hexane2,6-dicarboxylic acid and (1R,2S,4R,5R,6R)-2-amino-4-(1H-1,2,4-triazol-3-ylsulfanyl)bicyclo[3.1.0]hexane-2,6-dicarboxylic acid, are two mGlu₂ receptor-selective agonists (Monn et al., 2015a,b), and (1S,2S,4S,5R,6S)-2-amino-4-[(3-methoxybenzoyl)amino]bicyclo[3.1.0]hexane-2,6-dicarboxylic acid is an mGlu₃ receptor-selective agonist (Monn et al., 2018). Cocrystallization of the VFT with each of these ligands, coupled with mutagenesis and molecular modeling, revealed that selectivity is due to interactions with amino acids residing at the periphery of the glutamate binding site (Monn et al., 2015a,b, 2018). In a similar fashion, (2S)-2-amino-4-({[4-(carboxymethoxy)phenyl](hydroxy)methyl}(hydroxy)phosphoryl)butanoic acid (LSP4-2022), an mGlu₄ receptor-selective orthosteric agonist, binds both to the glutamate binding site and to an adjacent pocket (Goudet et al., 2012) (Fig. 6). This adjacent pocket is thought to be one of the sites of action of Cl⁻ ions, which are potent positive allosteric modulators of mGlu receptors (Acher et al., 2011; Tora et al., 2015). Therefore, LSP4-2022 and related derivatives constitute bitopic ligands that simultaneously target an orthosteric and an allosteric binding site (Selvam et al., 2018). Targeting these two sites in tandem provides the means to overcome the difficulty of designing selective orthosteric drugs.

C. Biased Agonism

Across the GPCR superfamily, it is becoming increasingly appreciated that the cellular response to receptor activation can differ depending on the ligand used, a phenomenon referred to as biased agonism. Biased agonism is thought to originate from the stabilization of different active conformations by distinct ligands, the balance of which is sampled by measuring different downstream measures of receptor activity. For any definition of biased GPCR agonism, it is critical that biased agonism is quantified relative to both a reference agonist and a reference pathway; this is because the relative efficacy of GPCR agonists is influenced by the stimulus-response coupling of the system. The most commonly applied method to quantify GPCR biased agonism is based on the operational model of agonism (Black and Leff, 1983) and subsequent derivation of transduction ratios (Kenakin et al., 2012). For metabotropic glutamate receptor orthosteric agonists, to date observations of biased agonism are limited to group I receptors. For example, relative to glutamate activation of mGlu₁-mediated cytoprotective signaling, quisqualate is biased toward inositol monophosphate (IP₁) accumulation in recombinant and native cells (Emery et al., 2012; Hathaway et al., 2015). At mGlu₅ receptors, biased agonism for DHPG relative to glutamate arises due to “location bias” as DHPG is impermeable and not actively transported across cell membranes. DHPG is therefore unable to stimulate mGlu₅ receptors located on intracellular membranes (Jong et al., 2005). Within different subcellular compartments, mGlu₅ receptors interact with a different complement of transducers, giving rise to different forms of synaptic plasticity (Kumar et al., 2012). It remains to be determined whether other mGlu receptors and associated selective ligands also exhibit location bias that contributes to pharmacological differences.

D. Tolerance

Another layer of complexity with regard to orthosteric agonist drug development is the potential for tolerance development. Under normal conditions glutamate is released transiently into the synapse, briefly activating mGlu receptors before active uptake mechanisms, for example, into astrocytes, reduce synaptic glutamate concentrations. However, these clearance mechanisms are not operative for surrogate orthosteric agonists, resulting in sustained receptor activation, which can lead to tolerance development. The potential for tolerance can be exacerbated for neurologic targets given the need for repeated chronic dosing to achieve a therapeutic effect. Tolerance has been noted for group II orthosteric agonists, where LY354740 efficacy for modulating rapid eye movement sleep wanes with repeated dosing (Ahnaou et al., 2015). Similarly, repeated daily dosing of (1R,4R,5S,6R)-4-amino-2-oxabicyclo[3.1.0]hexane-4,6-dicarboxylic acid (LY379268) results in loss of efficacy as an analgesic (Jones et al., 2005) and antipsychotic-like activity to inhibit phencyclidine- or amphetamine-induced hyperlocomotion (Galici et al., 2005). However, tolerance is not consistently observed with chronic LY379268 dosing and can differ between behavioral paradigms (Cartmell et al., 2000; Anderson et al., 2014; Battaglia et al., 2015; Halberstadt et al., 2019). Whether tolerance development will prove to limit therapeutic efficacy of mGlu receptor orthosteric agonists remains to be seen.

E. Orthosteric Ligands in the Clinic: Success and Failure

The most successful mGlu receptor discovery campaigns focused on orthosteric ligands targeting group II receptors. Multiple group II orthosteric agonists have reached phase II or phase III trials for psychiatric indications. LY354740/eglumegad was well tolerated and showed anxiolytic efficacy in humans (Grillon et al., 2003; Schoepp et al., 2003), with further development focused on a prodrug formulation (LY544344) to improve bioavailability (Rorick-Kehn et al., 2006). However, trials for generalized anxiety disorder were discontinued due to concerns regarding convulsions in animals (Dunayevich et al., 2008). Another group II mGlu receptor-selective agonist, LY2140023/pomaglumetad [prodrug for 4-amino-2-thiabicyclo[3.1.0]hexane-4,6-dicarboxylic acid 2,2-dioxide], improved both negative and positive symptoms in patients with schizophrenia in a randomized phase II clinical trial (Patil et al., 2007). Subsequent phase III trials failed to report significant antipsychotic efficacy for all the patients involved, terminating further development (Kinon et al., 2011; Adams et al., 2013, 2014; Downing et al., 2014). Interestingly, exploratory analyses of multiple phase II and III revealed that some subgroups of patients exhibited improvement after treatment with pomaglumetad, particularly early-in-disease patients or individuals previously treated with D₂ dopamine receptor-targeting drugs but not 5-HT₂ receptor antagonists (Kinon et al., 2015). Findings consistent with preclinical studies showing that mGlu₂ and 5-HT_2A receptors form functional complexes in cortex and that atypical antipsychotic treatment downregulates mGlu₂ receptor expression (González-Maeso et al., 2008; Kurita et al., 2012) (reviewed in Shah and Gonzalez-Maeso, 2019). Trials are ongoing for pomaglumetad for methamphetamine abuse [ClinicalTrials.gov identifier NCT03106571] and psychosis [NCT03321617]. Fasoracetam (also known as NS-105 or NFC-1), which is structurally unrelated to eglumegad or pomaglumegad, has agonist activity at group II and III receptors and showed efficacy for attention deficit hyperactivity disorder in a small cohort of adolescents (Elia et al., 2018), but failed to distinguish from placebo in a subsequent trial [NCT02777931]. In addition to orthosteric agonists, the group II selective orthosteric antagonist (1R,2R,3R,5R,6R)-2-amino-3-((3,4-dichlorobenzyl)oxy)-6-fluorobicyclo[3.1.0]hexane-2,6-dicarboxylic acid (BCI-838; also known as MGS0210), a prodrug of (1R,2R,3R,5R,6R)-2-amino-3-((3,4-dichlorobenzyl)oxy)-6-fluorobicyclo[3.1.0]hexane-2,6-dicarboxylic acid (BCI-632; also known as MGS0039) (Nakamura et al., 2006), completed phase I in healthy volunteers [NCT01546051], although plans for subsequent phase II trial in treatment resistant depression have not eventuated. This same agent has shown preclinical efficacy in models for Alzheimer’s disease (Kim et al., 2014) and posttraumatic stress disorder related to traumatic brain injury (Perez-Garcia et al., 2018). Despite successful discovery efforts for subtype-selective and drug-like orthosteric ligands for the group II mGlu receptors in particular, the majority of mGlu receptor discovery programs are pursuing allosteric modulators as reviewed below.

A. Definitions, Quantification, and Identification

Since glutamate is recognized by two different receptor families (ionotropic and metabotropic receptors) as well as transporters, there remains a concern that orthosteric compounds will suffer from lack of selectivity due to high conservation of glutamate binding sites across different proteins. As such, many discovery programs have focused efforts on identification and development of allosteric modulators. Allosteric modulators interact with sites that are topographically distinct from the orthosteric site, such that a receptor may be simultaneously bound by both an orthosteric and an allosteric ligand (Fig. 1A). For the most part, allosteric sites are located in region of receptors that show greater sequence divergence across subtypes and therefore offer greater selectivity. An allosteric modulator may enhance or inhibit the binding and/or efficacy of an orthosteric ligand, with the magnitude and direction described as “cooperativity.” An allosteric modulator that enhances orthosteric ligand affinity or efficacy is referred to as a positive allosteric modulator (PAM), whereas an inhibitor is a negative allosteric modulator (NAM). In addition, allosteric ligands may also bind to a receptor but have no net effect on either affinity or efficacy of an orthosteric ligand; referred to as neutral allosteric ligands (NALs). Furthermore, allosteric ligands may also possess intrinsic efficacy as either positive or inverse agonists in addition to, or exclusive of, cooperativity with an orthosteric ligand. By example, a PAM with intrinsic agonist activity is referred to as a PAM agonist or ago-PAM.

In addition to potential for increased subtype selectivity, allosteric modulators offer a number of advantages over their orthosteric counterparts. Cooperativity between two ligands is saturable, offering the potential for greater safety in an overdose. Allosteric modulators that have no intrinsic efficacy and are quiescent in the absence of endogenous ligand also provide scope to fine-tune receptor activity in a spatiotemporal fashion, exerting potentiation or inhibition only where, and when, the endogenous ligand is present. For these reasons, discovery programs in industry and academia alike have sought allosteric modulators of mGlu receptors as potential novel therapeutics for a wide array of CNS disorders. However, discovery of allosteric modulators can be associated with considerable challenges with respect to quantification and validation.

Allosteric modulator binding is defined by the law of mass action, where the equilibrium dissociation constant, commonly defined as K_B, describes the affinity of an allosteric modulator for its site. However, the simultaneous binding of an allosteric modulator and orthosteric ligand gives rise to different receptor conformations than can be achieved by the binding of each ligand individually, altering ligand affinity as defined by cooperativity. To quantify cooperativity, the simplest scheme is the allosteric ternary complex model (ATCM) (Fig. 1B; Gregory et al., 2010b), which describes the reciprocal change in ligand affinity when a receptor is simultaneously bound by both an allosteric and orthosteric ligand, defined by the cooperativity factor α. The ATCM is limited to describing allosteric interactions at the level of receptor binding, and for many mGlu allosteric modulators it is apparent these ligands have effects on receptor activity in addition to, or independent of, affinity.

To quantify the full scope of effects an allosteric ligand may have on receptor activity, multiple alternative pharmacological models have been proposed that can accommodate allosteric ligand intrinsic efficacy and efficacy modulation (Slack and Hall, 2012; Hall, 2013; Roche et al., 2014; Hall and Giraldo, 2018; Gregory et al., 2020). A challenge in applying these models is the inclusion of many parameters, which can prohibit fitting to experimental data. In this respect, the most widely adopted framework for quantification of pharmacological activity is an operational model of allosterism (Fig. 1C), which combines the Black and Leff operational model of agonism with the ATCM (Leach et al., 2007; Gregory et al., 2012). Within this framework, the influence of an allosteric modulator on orthosteric agonist efficacy is accounted for by β, an experimentally derived scaling factor. Application of this model therefore allows for delineation of the influence of an allosteric modulator on affinity independently of efficacy. This provides an important distinction given that allosteric modulators can have differential effects, which may be in opposing directions, on affinity versus efficacy. The operational model of allosterism also allows for intrinsic allosteric agonism, defined by τ, but cannot account for inverse agonism.

Accurate quantification of allosteric ligand pharmacology requires appropriately designed experimental paradigms. The definitive experiment to unambiguously demonstrate an allosteric mechanism of action is a kinetic binding paradigm. The simultaneous binding of an allosteric ligand may enhance or slow the dissociation rate (K_off) of the orthosteric radioligand from the receptor, or vice versa. Interaction studies using radiolabeled orthosteric ligands can be used to quantify modulation of affinity (α) as well as ligand affinity for the free receptor (Gregory et al., 2010b). However, it is important to note that the magnitude and direction of cooperativity between two ligands depends on the chemotypes present, a phenomenon known as “probe dependence.” This is an important consideration when extrapolating pharmacological profiles of allosteric ligands based on interactions with a radiolabeled orthosteric antagonist or from a surrogate orthosteric agonist, which is often required in native cells/tissues. Radiolabeled allosteric ligands have been described for multiple subtypes, which can also be used to quantify affinity for the receptor for unlabeled ligands at a common allosteric site (Cosford et al., 2003; Lavreysen et al., 2003; O’Brien et al., 2018) or provide evidence for additional allosteric sites that are conformationally linked such that there is cooperativity between the two allosteric sites.

The vast majority of allosteric ligands for mGlu receptors have been identified and validated using functional assays. The most commonly used approach involves generating modulator titration curves in the presence of either an ∼EC₂₀ agonist concentration for PAM identification, or an EC₈₀ to identify NAMs. The potencies and relative maximum response (for PAMs) or inhibitory effect (for NAMs) from these titration curves are routinely used to drive discovery programs (Lindsley et al., 2016). However, these parameters represent composite values encompassing α, β, K_B, and τ and are also influenced by the concentrations of orthosteric agonist used, orthosteric agonist intrinsic efficacy, and the stimulus-response coupling of the system under investigation (Lindsley et al., 2016). Modulator potencies curves can be analyzed in parallel with an agonist concentration-response curve to estimate K_B and a composite αβ value where the maximum degree of potentiation or inhibition does not reach the limit of the system (Gregory et al., 2012, 2019). However, to quantify the interaction between a modulator and orthosteric agonist, the most robust approach is to perform full agonist concentration-response curves in the absence and presence of increasing concentrations of modulator. Despite the limitations in the most commonly applied screening approaches, drug discovery programs for small molecule synthetic allosteric ligands of mGlu receptors have been largely successful. In addition, there is increasing evidence for endogenous allosteric modulators for mGlu receptor family members.

B. Endogenous Allosteric Modulators

The greater class C GPCR family also includes the calcium-sensing receptor and GPRC6A, two receptors that are known to respond to multiple endogenous ligands including amino acids and cations (Leach and Gregory, 2017). It is perhaps therefore not surprising that divalent and trivalent cations, including Ca²⁺, can directly activate mGlu₁, mGlu₃, and mGlu₅ receptors (Kubo et al., 1998; Miyashita and Kubo, 2000a,b; Jiang et al., 2014). Furthermore, extracellular Ca²⁺ potentiates binding/function of orthosteric ligands at mGlu₁ (Saunders et al., 1998; Jiang et al., 2014). In addition, negatively charged chloride ions activate mGlu₃, mGlu₄, mGlu₆, and mGlu₈ receptors and potentiate glutamate efficacy at mGlu₁, mGlu₂, mGlu₄, mGlu₅, and mGlu₆ receptors (DiRaddo et al., 2015; Tora et al., 2015, 2018). Both cations and anions are thought to mediate activation and/or modulation via interactions with the VFT domain. The extracellular membrane associated cellular prion protein interacts with the mGlu₅ receptor acting as a coreceptor for amyloid β oligomers, although the precise binding interactions within mGlu₅ receptors are unknown (Um et al., 2013). Beyond the extracellular domains, molecular dynamics studies have proposed that lipids can interact with mGlu₅ 7TM (Dalton et al., 2017). Furthermore, cholesterol membrane content enhances mGlu₁ signaling to ERK1/2 phosphorylation mediated via a cholesterol recognition/interaction amino acid consensus motif in transmembrane domain 5 (Kumari et al., 2013). The existence of endogenous allosteric modulators for the mGlu receptors is often overlooked during discovery and validation of synthetic small molecule allosteric modulators.

C. Small Molecule Allosteric Modulators

Concerted discovery efforts from both industrial and academic researchers have yielded a wealth of chemically and pharmacologically diverse small molecule allosteric modulators for the mGlu receptor family (Tables 2⇓⇓–5). The majority of small molecule mGlu receptor allosteric modulators identified to date interact with a common pocket within the 7TM domains. This binding pocket is in a location analogous to the biogenic amine orthosteric site of class A GPCRs, largely lined by residues in transmembrane domains 3, 5, 6, and 7. To date, six X-ray crystal structures of the mGlu₁ or mGlu₅ receptor 7TM domains have been solved with NAMs occupying this common allosteric site (Doré et al., 2014; Wu et al., 2014; Christopher et al., 2015, 2019). A wealth of previous mutagenesis data indicate that this pocket is shared across the mGlu receptor family, and indeed for all class C GPCRs, and can be engaged by both NAMs and PAMs (see Leach and Gregory, 2017, for review). Here we focus on the pharmacological profiles of prototypical and well validated commercially available allosteric modulators for mGlu receptors.

D. Group I PAMs, NAMs, NALs

The first disclosed mGlu receptor allosteric modulator was ethyl (7Z)-7-hydroxyimino-1,7a-dihydrocyclopropa[b]chromene-1a-carboxylate (CPCCOEt) (Annoura et al., 1996; Litschig et al., 1999), a negative allosteric modulator of mGlu₁ receptor. CPCCOEt has low micromolar affinity for mGlu₁ receptors (Lavreysen et al., 2003) and negatively modulates glutamate efficacy but has neutral cooperativity with respect to [³H]glutamate affinity (Litschig et al., 1999). Moreover, CPCCOEt has poor selectivity between group I mGlu receptors (Table 2), negatively modulating mGlu₅ receptor activation with a similar apparent K_B (Hellyer et al., 2018). The discovery of CPCCOEt was followed by EM-TBPC and BAY-36-7620, which showed species differences in mGlu₁ receptor NAM activity with considerably higher affinity for the rat versus human receptor (Malherbe et al., 2003; Cho et al., 2014a). Similar to CPCCOEt, BAY-36-7620 has neutral cooperativity with respect to [³H]quisqualate affinity but inhibits orthosteric agonist efficacy (Carroll et al., 2001; Lavreysen et al., 2003). Since the discovery of these early tool compounds, a wealth of structurally diverse mGlu₁ receptor NAMs have been disclosed that have therapeutic efficacy in preclinical models for analgesia, antipsychotic-like activity, anxiety, addiction, and cancer and as anticonvulsants (Table 2). For diverse scaffolds [Fig. 7, e.g., A-841720 and 1-(3,4-dihydro-2H-pyrano[2,3-b]quinolin-7-yl)-2-phenylethanone], the higher affinity for rat over human (>10-fold) persisted (Cho et al., 2014a). Breakthrough chemotypes represented by FTIDC and JNJ16259685 have similar nanomolar affinities for the rat and human receptors and >100-fold selectivity as NAMs for mGlu₁ over mGlu₅ receptors (Lavreysen et al., 2003, 2004; Suzuki et al., 2007a). Despite ultimate identification of high affinity, in vivo efficacious mGlu₁ receptor NAMs, further development has stalled due to on-target mediated adverse effects such as cognitive impairments from multiple scaffolds (Steckler et al., 2005b; Schröder et al., 2008).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.

Structures of select allosteric modulators of group I mGlu receptors. Representative allosteric modulators for mGlu₁ (A) and mGlu₅ (B) demonstrating the structural diversity for compounds that recognize the common allosteric site within the 7TM. In both panels NAMs are labeled with black text, and PAMs are in blue. Detailed in vitro and in vivo pharmacological profiles for each compound including subtype selectivity are listed in Tables 2 and 3. The following abbreviation is used: RO 01-6128, ethyl N-[2,2-di(phenyl)acetyl]carbamate.

On the other hand, mGlu₁ receptor PAMs have been relatively unexplored, although they may be a promising therapeutic strategy for schizophrenia by restoring function of naturally occurring mutations (Garcia-Barrantes et al., 2015b). The first mGlu₁ receptor PAMs included diverse chemotypes, for example, 2-(4-fluorophenyl)-1-(4-methylphenyl)sulfonylpyrrolidine (RO 67-7476) and ethyl N-[2,2-di(phenyl)acetyl]carbamate, identified from high-throughput screening, which enhanced orthosteric agonist affinity and functional responses at rat mGlu₁ receptor without intrinsic agonist activity (Knoflach et al., 2001) but were not suitable for in vivo studies. Similar to multiple mGlu₁ receptor NAM scaffolds, RO 67-7476 lacks the ability to potentiate glutamate at human mGlu₁ receptors (Knoflach et al., 2001). A subsequent study suggested these mGlu₁ receptor PAMs may have intrinsic efficacy for ERK1/2 and cAMP accumulation; however, this agonist activity could be blocked by both orthosteric and allosteric antagonists, raising the possibility that the apparent intrinsic agonism may be attributable to potentiation of ambient glutamate (Sheffler and Conn, 2008). Of note, both RO 67-7476 and ethyl N-[2,2-di(phenyl)acetyl]carbamate were unable to completely displace binding of the radiolabeled mGlu₁ receptor NAM [³H]1-(3,4-dihydro-2H-pyrano[2,3-b]quinolin-7-yl)-2-phenylethanone (Hemstapat et al., 2006), suggesting these compounds recognize a different site within the 7TM domain. Subsequent discovery efforts identified 3-chloro-N-[3-chloro-4-(4-chloro-1,3-dihydro-1,3-dioxo-2H-isoindol-2-yl)phenyl]-2-pyridinecarboxamide (VU0483605) based on a scaffold hop from an mGlu₄ PAM/mGlu₁ NAM chemotype (Cho et al., 2014b); however, the selectivity of VU0483605 as an mGlu₁ PAM is based on cooperativity, as it has similar affinity for mGlu₅ receptors, albeit with neutral cooperativity with mGlu₅ receptor orthosteric agonist efficacy (Hellyer et al., 2018). Recent medicinal chemistry efforts have yielded the first CNS penetrant mGlu₁ receptor PAMs (Garcia-Barrantes et al., 2015a, 2016a,b; Yohn et al., 2018), paving the way forward to establish therapeutic potential for schizophrenia and beyond.

Although CPCCOEt was the first mGlu receptor NAM disclosed, the mGlu₅ receptor NAM fenobam was discovered earlier (Itil et al., 1978), but its mechanism of action was not elucidated until over 20 years later (Porter et al., 2005). Indeed, of all the subtypes, allosteric ligand discovery against mGlu₅ receptors has proven to be the most fruitful with a wealth of pharmacologically and structurally diverse ligands identified including NAMs, PAMs, and NALs (Table 3). Prototypical mGlu₅ receptor NAMs based on an acetylene core, 2-methyl-6-(phenylethynyl)-pyridine (MPEP) and 3-((2-methyl-4-thiazolyl)ethynyl)pyridine (MTEP), as well as fenobam, have demonstrated the therapeutic potential for mGlu₅ receptor inhibition for addiction, depression, anxiety, neurodegenerative disorders, and autism spectrum disorders (Table 3). Often referred to as “full NAMs,” these ligands have high negative cooperativity with respect to orthosteric agonist efficacy, completely abolishing agonist responses at saturating concentrations, but are neutral with respect to glutamate affinity (Gregory et al., 2012; Sengmany et al., 2019). The relatively high affinity of MPEP and fenobam presented the opportunity to generate radiolabeled versions (Cosford et al., 2003; Porter et al., 2005), which facilitated discovery and validation of novel mGlu₅ receptor NAMs. Of note, many full NAMs have inverse agonist activity (Porter et al., 2005; Sengmany et al., 2019). It has been postulated that the combination of high negative cooperativity and inverse agonism contributes to on-target adverse effect liability of mGlu₅ receptor NAMs, including cognitive impairments and psychotomimetic-like properties (Dekundy et al., 2011; Hughes et al., 2013; Abou Farha et al., 2014; Swedberg et al., 2014; Swedberg and Raboisson, 2014). Repeated exposure to both MTEP and fenobam is associated with tolerance development for reward behaviors (Cleva et al., 2012), but not for fenobam analgesic efficacy (Montana et al., 2011). The limitations associated with mGlu₅ receptor full NAMs stimulated discovery efforts for NAMs with lower negative cooperativity, also referred to as “partial NAMs,” which have limited ability to inhibit glutamate efficacy. Two recent proof-of-concept studies demonstrated that partial NAMs, for example, 2-[2-(3-methoxyphenyl)ethynyl]-5-methylpyridine (M-5MPEP) and N,N-diethyl-5-((3-fluorophenyl)ethynyl)picolinamide, which have limited negative cooperativity with glutamate, elicited anxiolytic, antidepressant activity and reduced cocaine self-administration with comparable efficacy to MTEP (Gould et al., 2016; Nickols et al., 2016). Importantly, unlike MTEP, M-5MPEP did not show psychotomimetic-like effects (Gould et al., 2016); therefore, mGlu₅ receptor NAMs with limited cooperativity may offer improved therapeutic windows.

An inherent challenge for mGlu₅ receptor allosteric ligand discovery has been the prevalence of “molecular switches” where minor substitutions give rise to ligands with reduced or opposing cooperativity (Wood et al., 2011). Although they pose a challenge with respect to structure-activity relationship interpretation, these molecular switches have also offered invaluable tools to dissect mGlu₅ receptor biology, with the MPEP scaffold giving rise to NALs and PAMs. By example, 5MPEP is a neutral mGlu₅ receptor allosteric ligand, which occupies the allosteric site in a competitive manner with MPEP but does not influence orthosteric agonist activity (Rodriguez et al., 2005). Subsequent efforts have identified high affinity mGlu₅ receptor NALs [e.g., 3-azabicyclo[3.1.0]hexan-3-yl-[5-[2-(3-fluorophenyl)ethynyl]pyridin-2-yl]methanone, (4R,5R)-rel-5-(2-chlorophenyl)-4-(5-(phenylethynyl)pyridin-3-yl)oxazolidin-2-one] with suitable properties for in vivo studies (Gregory et al., 2010a; Haas et al., 2017).

Molecular switches within mGlu₅ receptor NAM scaffolds (Fig. 7) have also yielded PAMs and PAM agonists, with advanced compounds from the biaryl acetylene scaffold, such as N-cyclobutyl-6-[2-(3-fluorophenyl)ethynyl]-3-pyridinecarboxamide hydrochloride (VU0360172) and N-(1-methylethyl)-5-(pyridin-4-ylethynyl)pyridine-2-carboxamide, showing high affinity and selectivity for mGlu₅ (Table 3). However, early mGlu₅ receptor PAMs [DFB, ADX47273, 3-cyano-N-(1,3-diphenyl-1H-pyrazol-5-yl)benzamide (CDPPB), N-[4-chloro-2-(phthalimidomethyl)phenyl]salicylamide (CPPHA)] were identified from high-throughput screening using functional assays (O’Brien et al., 2003, 2004; Lindsley et al., 2004; Liu et al., 2008). Structurally diverse mGlu₅ receptor PAM scaffolds compete for the common allosteric site within the 7TM used by MPEP (Gregory et al., 2013b, 2014); however, select PAMs [e.g., CPPHA and 4-butoxy-N-(2,4-difluorophenyl)benzamide] are thought to interact with distinct but as yet unknown site(s) within the 7TM (O’Brien et al., 2004; Chen et al., 2008; Hammond et al., 2010; Noetzel et al., 2013). Mechanistically, mGlu₅ receptor PAMs are largely considered to potentiate mGlu₅ receptor activity in response to glutamate via efficacy modulation (Gregory et al., 2012); however, probe dependence can dictate the nature of these allosteric interactions with multiple PAMs reported to enhance [³H]quisqualate binding (Bradley et al., 2011; Koehl et al., 2019) as well as show different magnitudes of cooperativity depending on the orthosteric agonist used (Sengmany et al., 2017). In this respect the largest magnitude of potentiation observed in mGlu₅ receptor functional assays is ∼20-fold shift in glutamate potency for 1-(4-(2,4-difluorophenyl)piperazin-1-yl)-2-((4-fluorobenzyl)oxy)ethan-1-one (DPFE) and 5-fluoro-2-{3-[(3S,6R)-1-[(4-fluorophenyl)carbonyl]-6-methylpiperidin-3-yl]-1,2,4-oxadiazol-5-yl}pyridine (Gregory et al., 2013a; Parmentier-Batteur et al., 2014). Indeed, the magnitude of cooperativity was the best predictor of in vivo efficacy of an mGlu₅ receptor PAM series using the amphetamine-induced hyperlocomotion assay, when total and free brain and plasma concentrations were determined from the same rats (Gregory et al., 2019). Select mGlu₅ receptor PAMs have intrinsic agonist efficacy; however, in some cases this is linked to high receptor reserve in recombinant systems (Noetzel et al., 2012). However, this is not always the case, with some compounds, for example, DPFE, showing intrinsic efficacy in low expression and native cell systems (Gregory et al., 2013a; Sengmany et al., 2017), and may also be dependent on measure of receptor activation, where mGlu₅ receptor PAMs often activate mGlu₅-ERK1/2 phosphorylation at concentrations that do not elicit iCa²⁺ mobilization responses (Gregory et al., 2012; Rook et al., 2013). Despite the complexity in pharmacology, successful discovery efforts for multiple centrally active mGlu₅ receptor PAMs have established proof-of-concept for procognitive and antipsychotic efficacy of mGlu₅ potentiators (Table 3). However, on-target adverse effect liability has been associated with multiple scaffolds, which has been attributed in part to intrinsic agonist activity and/or magnitude of cooperativity (Rook et al., 2013; Parmentier-Batteur et al., 2014). Recent studies have challenged these conclusions (Rook et al., 2015b; Sengmany et al., 2017), suggesting that biased pharmacology of mGlu₅ receptor PAMs may be linked to adverse versus therapeutic effects (discussed in detail below).

E. Group II PAMs and NAMs

Discovery and validation of group II mGlu receptor allosteric ligands has benefited from the availability of radiolabeled orthosteric agonists and antagonists owing to the generally higher affinity of orthosteric ligands for group II mGlu receptors (Table 1). Indeed, multiple different mGlu₂ receptor-selective PAMs have been disclosed and established proof-of-concept for mGlu₂ receptor potentiation as a viable therapeutic intervention for anxiety, psychosis, and addiction (Table 4). The first mGlu₂ receptor-selective PAM was 2,2,2-trifluoro-N-[4-(2-methoxyphenoxy)phenyl]-N-(3-pyridinylmethyl)-ethanesulfonamide (LY487379; also referred to as 4-MPPTS) (Johnson et al., 2003), which enhances radiolabeled orthosteric agonist ([³H]LY354740 and [³H]DCG-IV) binding (Schaffhauser et al., 2003; Lundström et al., 2016), orthosteric agonist affinity (Johnson et al., 2005), and functional activity, although LY487379 shows probe dependence as it is neutral with respect to affinity of LY379268 (a high affinity orthosteric agonist) (Johnson et al., 2005). Furthermore, LY487379 has neutral cooperativity with respect to orthosteric antagonist binding (Schaffhauser et al., 2003; Johnson et al., 2005), suggesting LY487379 preferentially interacts with the active receptor state. A similar pharmacological profile has been noted for mGlu₂ receptor-selective PAMs from different chemotypes [including 4-[3-[(2-cyclopentyl-6,7-dimethyl-1-oxo-2,3-dihydroinden-5-yl)oxymethyl]phenyl]benzoic acid (BINA), 5-(7-chloro-2-((S)-1-cyclopropyl-ethyl)-1-oxo-2,3-dihydro-1H-isoindol-5-yl)-isoxazole-3-carboxylic acid dimethylamide, and 8-chloro-3-(cyclopropylmethyl)-7-(4-(3,6-difluoro-2-methoxyphenyl)piperidin-1-yl)-[1,2,4]triazolo[4,3-a]pyridine], which also potentiate glutamate binding and efficacy but are neutral with respect to orthosteric antagonist binding (O’Brien et al., 2018), although the recently disclosed mGlu₂ receptor modulator, (S)-2-(1,1-dimethyl-indan-5-yloxymethyl)-2,3-dihydro-oxazolo[3,2-a]pyrimidin-7-one, has even more pronounced probe dependence, potentiating glutamate affinity but inhibiting [³H]LY341495 binding (Griebel et al., 2016). Further insights into the mechanism of action of mGlu₂ receptor PAMs have been elucidated using radiolabeled mGlu₂ receptor PAMs [2,2,2-trifluoro-N-(3-pentan-2-yloxyphenyl)-N-(pyridin-3-ylmethyl)ethanesulfonamide, JNJ-46281222, and 2-[[(4-bromophenyl)methyl]sulfanyl]-N-[4-(butan-2-yl)phenyl]acetamide]. Compared with orthosteric radioligands, radiolabeled mGlu₂ receptor PAMs recognize fewer binding sites, which is thought to indicate occupation of a single 7TM domain within the dimeric receptor as well as preferential binding to active receptor conformations since orthosteric agonists can increase the number of mGlu₂ receptor PAM binding sites (Lundström et al., 2009, 2011, 2016; Lavreysen et al., 2013; Doornbos et al., 2016; O’Brien et al., 2018). Most mGlu₂ receptor PAMs have intrinsic agonist efficacy (Table 4) with the maximal degree of potentiation observed for glutamate potency of between 10- and 30-fold (Johnson et al., 2005; Galici et al., 2006; Lavreysen et al., 2015; O’Brien et al., 2018). In addition to increased selectivity over mGlu₃ relative to orthosteric agents, mGlu₂ receptor PAMs may also provide improved therapeutic efficacy owing to reduced capacity for induction of tolerance compared with group II receptor orthosteric agonists (Ahnaou et al., 2015). In contrast to successful mGlu₂ receptor PAM discovery campaigns, to date, mGlu₃ receptor-selective PAMs have remained elusive.

Discovery efforts for group II mGlu receptor NAMs have yielded both mGlu₂ and mGlu₃ receptor subtype-selective ligands, although there is less structural diversity available when compared with mGlu₂ receptor PAMs (Fig. 8; Table 4). Negative allosteric modulators of group II mGlu receptors have demonstrated efficacy for improving cognitive deficits and reversing behaviors in preclinical models for depression and anxiety (Woltering et al., 2010; Campo et al., 2011; Goeldner et al., 2013; Engers et al., 2015, 2017) and are neuroprotective under ischemic insult (Motolese et al., 2015). Mechanistically, group II receptor NAMs [including 4-(8-bromo-5-oxo-3,4,5,6-tetrahydro-1,6-benzodiazocin-2-yl)pyridine-2-carbonitrile (MNI-137), 4-[3-(2,6-dimethylpyridin-4-yl)phenyl]-7-methyl-8-(trifluoromethyl)-1,3-dihydro-1,5-benzodiazepin-2-one, 5-[2-[7-trifluoromethyl)-5-[4-(trifluoromethyl)phenyl]pyrazolo[1,5-a]pyrimidin-3-yl]ethynyl]pyridin-2-amine (decoglurant), and related compounds] are neutral with respect to glutamate affinity, primarily acting as negative modulators of glutamate efficacy (Hemstapat et al., 2007; Campo et al., 2011; O’Brien et al., 2018). Akin to observations with group II receptor PAMs, select NAMs have demonstrated probe dependence with respect to modulation of orthosteric agonist affinity, where ligands related to 4-[3-(2,6-dimethylpyridin-4-yl)phenyl]-7-methyl-8-(trifluoromethyl)-1,3-dihydro-1,5-benzodiazepin-2-one or decoglurant are NAMs with respect to [³H]LY354740 binding (Woltering et al., 2008; Lundström et al., 2011). For the majority of pan–group II, mGlu receptor NAMs, including [2-fluoro-4-[2-(4-methoxyphenyl)ethynyl]phenyl][(3R)-3-hydroxy-1-piperidinyl]methanone (ML337), MNI-137, and decoglurant, are full NAMs, completely abolishing the functional response to orthosteric agonists at both mGlu₂ and mGlu₃ receptors (Caraci et al., 2011; Wenthur et al., 2014; O’Brien et al., 2018). However, MNI-137 has differing degrees of negative cooperativity in functional assays of mGlu₂ receptor activity, in some instances showing full blockade but in others limited negative cooperativity (Hemstapat et al., 2007; Yin et al., 2014; O’Brien et al., 2018). Furthermore, both the reported selectivity and inhibitory activity of mGlu_2/3 receptor NAM (3S)-1-(5-bromopyrimidin-2-yl)-N-(2,4-dichlorobenzyl) pyrrolidin-3-amine methanesulfonate hydrate differs depending on the response measured (Caraci et al., 2011; Sheffler et al., 2012). In this respect, it is worth noting that for many ligands and series there has been limited pharmacological profiling to fully discern mechanism of action. Subtype-selective mGlu₂ receptor [e.g., 6-(((2S,6R)-2,6-dimethylmorpholino)methyl)-1-(4-fluorophenyl)-4-oxo-1,4-dihydroquinoline-3-carboxamide, 4-(2-fluoro-4-methoxyphenyl)-7-(2-(2-methylpyrimidin-5-yl)ethyl)quinoline-2-carboxamide, 4-(8-bromo-5-oxo-3,4,5,6-tetrahydro-1,6-benzodiazocin-2-yl)pyridine-2-carbonitrile] or mGlu₃ receptor [[(3R)-3-(hydroxymethyl)-1-pipridinyl][4-[2-(4-methoxyphenyl)ethynyl]phenyl]methanone, (S)-1-(4-fluorophenyl)-4-(2-phenoxypropoxy)pyridin-2(1H)-one] NAMs have been reported (Kolczewski et al., 1999; Sheffler et al., 2012; Felts et al., 2015; Walker et al., 2015; Engers et al., 2017). However, the group II receptor selectivity of ML337 was recently demonstrated to be in part attributable to cooperativity, as this ligand is an NAL at mGlu₅ (Hellyer et al., 2018). Whether other reportedly subtype-selective NAMs are also due to cooperativity rather than affinity remains to be elucidated.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 8.

Structures of select allosteric modulators of group II and III mGlu receptors. Structurally diverse chemotypes allosterically modulate group II (A) and group III (B) mGlu receptors. NAMs are labeled with black text, and PAMs are in blue. Detailed in vitro and in vivo pharmacological profiles for each compound including subtype selectivity are listed in Tables 4 and 5. The following abbreviation is used: LY2389575, (3S)-1-(5-bromopyrimidin-2-yl)-N-(2,4-dichlorobenzyl) pyrrolidin-3-amine methanesulfonate hydrate.

Both selective and pan–group II receptor NAMs from diverse scaffolds interact with a common or overlapping site with that used by mGlu₂ receptor PAMs within the 7TM domain (Schaffhauser et al., 2003; Rowe et al., 2008; Lundström et al., 2011, 2016; O’Brien et al., 2018). Select amino acid residues within this common site can differentially influence group II receptor NAM versus PAM activity (Hemstapat et al., 2007; Lundström et al., 2011; Perez-Benito et al., 2017). These differential effects may be attributable to differential effects on cooperativity or affinity whereby distinct ligand-receptor interactions may contribute to active versus inactive receptor conformations. However, for some scaffolds [e.g., 8-chloro-3-(cyclopropylmethyl)-7-(4-(3,6-difluoro-2-methoxyphenyl)piperidin-1-yl)-[1,2,4]triazolo[4,3-a]pyridine, decoglurant, 6-(((2S,6R)-2,6-dimethylmorpholino)methyl)-1-(4-fluorophenyl)-4-oxo-1,4-dihydroquinoline-3-carboxamide, 4-(2-fluoro-4-methoxyphenyl)-7-(2-(2-methylpyrimidin-5-yl)ethyl)quinoline-2-carboxamide], allosteric interactions have been observed with a mGlu₂ receptor PAM radioligand, indicative of multiple allosteric sites within the 7TM (O’Brien et al., 2018) or possibly more complex interactions due to the dimeric nature of mGlu receptors as has been noted for other class C GPCR allosteric modulators (Gregory et al., 2018).

F. Group III PAMs and NAMs

A list of the some of the commercially available allosteric modulators of group III mGlu receptors is provided in Table 5. The first identified group III mGlu receptor-selective allosteric modulator was N-phenyl-7-(hydroxyimino)cyclopropa[b]chromen-1a-carboxamide (PHCCC) (Maj et al., 2003; Marino et al., 2003b) (Fig. 8). PHCCC acts as an mGlu₄ receptor PAM, increasing potency and efficacy of glutamate or L-AP4 in cell-based assays. PHCCC is closely related to the mGlu₁ receptor-selective NAM CPCCOEt (Annoura et al., 1996) (Fig. 7). Although it has weak potency and poor solubility, PHCCC provided a very useful tool to demonstrate the therapeutic potential of targeting mGlu₄ receptors in Parkinson’s disease and paving the way to drug candidates (Charvin, 2018). Indeed, PHCCC potentiated the inhibitory effect of L-AP4 on transmission at the striatopallidal synapse and reversed akinesia in rats (Marino et al., 2003a,b). PHCCC also reduces hyperalgesia in rat models of chronic pain (Goudet et al., 2008). Subsequently, a new mGlu₄ receptor PAM named (1R,2S)-2-[(3,5-dichlorophenyl)carbamoyl]cyclohexane-1-carboxylic acid (VU0155041) was discovered (Christov et al., 2011), which is more potent and more soluble than PHCCC. Interestingly, VU0155041 is an mGlu₄ receptor allosteric agonist (PAM agonist), contrary to the pure PAM profile of PHCCC. PHCCC and VU0155041 do not compete for the same site (Niswender et al., 2008a). Accordingly, two partially overlapping 7TM binding pockets have been identified in mGlu₄ receptors, a shallow and a deep pocket (Rovira et al., 2015). Analysis of the pharmacological properties and binding modes of several mGlu₄ receptor PAMs, revealed the intrinsic efficacy and cooperativity of mGlu₄ PAMs (both affinity and efficacy modulation of L-AP4 and glutamate) correlate with the binding mode (Rovira et al., 2015). PAMs with intrinsic allosteric agonism bind in the shallow pocket, analogous to the pocket of natural agonists of class A GPCRs, whereas PAMs exhibiting the highest cooperativity with orthosteric agonists bind into a deeper pocket, corresponding to that of mavoglurant in the mGlu₅ receptor 7TM crystal structure (Doré et al., 2014) and pointing toward a site topographically homologous to the Na⁺ binding pocket of class A GPCRs. In preclinical studies, VU0155041 improves symptoms of Parkinson’s disease (Christov et al., 2011), chronic pain (Wang et al., 2011), and autistic-like syndromes (Becker et al., 2014). Foliglurax (PXT002331) is a derivative of PHCCC with good water solubility and high brain exposure after oral administration (Charvin et al., 2017). It is a potent and selective mGlu₄ receptor PAM displaying strong anti-Parkinsonian activity in rodent preclinical models of Parkinson’s disease (Charvin et al., 2017) as well as in primates (Charvin et al., 2018a); however, it recently failed to show efficacy in a phase II clinical trial. Several mGlu₄ receptor PAMs also exhibit PAM activity on mGlu₆ receptors; as yet no selective mGlu₆ receptor allosteric modulators have been described.

The first allosteric modulator acting at the mGlu₇ receptor to be described was N,N′-dibenzhydrylethane-1,2-diamine dihydrochloride (AMN082) (Mitsukawa et al., 2005). Since there was a lack of pharmacological tools to study mGlu₇ receptor, this mGlu₇ PAM agonist attracted much interest. However, AMN082 presents off-target effects, as it retains activity in mGlu₇ receptor knockout mice (Ahnaou et al., 2016b). Indeed, AMN082 is rapidly metabolized in vivo, with the major metabolite being a potent monoamine transporter inhibitor (Sukoff Rizzo et al., 2011). Thus, preclinical results obtained with AMN082 have to be carefully interpreted, since its actions may not be driven solely by mGlu₇ receptors. Selective mGlu₇ receptor NAMs have been described and may be more adequate for investigating the role of mGlu₇ receptor in vivo: 6-(4-methoxyphenyl)-5-methyl-3-(4-pyridinyl)-isoxazolo[4,5-c]pyridin-4(5H)-one (MMPIP) (Suzuki et al., 2007b), 6-(2,4-dimethylphenyl)-2-ethyl-4,5,6,7-tetrahydro-1,3-benzoxazol-4-one (ADX71743) (Kalinichev et al., 2013), and 7-hydroxy-3-(4-iodophenoxy)-4H-chromen-4-one (XAP044) (Gee et al., 2014). Interestingly, the inhibitory activity of MMPIP is context dependent, where MMPIP may not antagonize mGlu₇ receptor activity in all cellular contexts. Indeed, MMPIP is unable to block agonist-mediated responses at the Schaffer collateral-CA1 synapse, where mGlu₇ receptor is known to modulate neurotransmission (Niswender et al., 2010). ADX71743 is a bioavailable and brain penetrant mGlu₇ NAM that induces a robust anxiolytic effect in rodents (Kalinichev et al., 2013). Most small allosteric modulators described so far act via a binding pocket located within the transmembrane domain; however, XAP044 mediates its action through an interaction with the extracellular domain of mGlu₇ receptor (Gee et al., 2014). The exact binding pocket of XAP044 is not known at the moment, with chimeric mGlu₇/mGlu₆ receptors used to map its action to the extracellular domains (Gee et al., 2014). Recent efforts aiming to identify novel mGlu₇ receptor PAM scaffolds have turned to cheminformatics-based approaches; however, to date these have yielded low potency potentiators (Tresadern et al., 2017).

Only few mGlu₈ receptor allosteric modulators have been identified. 2-[[(4-Bromophenyl)methyl]sulfanyl]-N-[4-(butan-2-yl)phenyl]acetamide (AZ12216052) is an mGlu₈ receptor PAM agonist of glutamate at mGlu₈ receptors (Duvoisin et al., 2010). In vivo, AZ12216052 is anxiolytic in apolipoprotein E–deficient mice, which show increased levels of anxiety-like behaviors (Duvoisin et al., 2010). AZ12216052 also displays analgesic activity after injection into the dorsal striatum of neuropathic rats (Rossi et al., 2014). However, AZ12216052 possesses some off-target effects since it retains anxiolytic activity in mGlu₈^−/− mice (Duvoisin et al., 2011). Another useful pharmacological tool for mGlu₈ receptor is VU6005649, a brain penetrant PAM of mGlu₇ and mGlu₈ receptors that displays in vivo efficacy in a mouse contextual fear conditioning model (Abe et al., 2017). The pan–group III receptor PAM, N-[3-chloro-4-[(5-chloro-2-pyridinyl)oxy]phenyl]-2-pyridinecarboxamide (VU0422288), which has similar affinity for mGlu₄, mGlu₇, and mGlu₈ receptors (Jalan-Sakrikar et al., 2014), rescues deficits (synaptic plasticity and behavioral phenotypes) in a mouse model of Rett syndrome (Gogliotti et al., 2017). VU0422288 also shows probe dependence with respect to both its apparent affinity and magnitude of positive cooperativity (Jalan-Sakrikar et al., 2014). Whether targeting one or multiple of the group III receptor subtypes will best treat this neurologic disorder remains to be explored with subtype-selective pharmacological agents.

G. Allosteric Modulators Progressing to Clinical Trials

With respect to clinical translation of promising preclinical efficacy for mGlu receptor allosteric modulators, mGlu₅ NAMs have demonstrated the most progress with multiple agents reaching phase II trials for a variety of indications. Prior to elucidation of its mechanism of action, fenobam was assessed in a small double-blind placebo-controlled study as a single agent for treatment of anxiety compared with diazepam and was reported to have fewer adverse effects (Pecknold et al., 1982). Subsequently, fenobam was trialed in an open-label pilot study for treatment of fragile X syndrome behavioral deficits, where it was well tolerated but lacked efficacy (Berry-Kravis et al., 2009). Multiple phase II trials of mGlu₅ receptor NAMs in patients with fragile X syndrome (both adolescents and adults) have now been completed, with basimglurant (RO4917523) and mavoglurant (AFQ056) being well tolerated, improving behavioral symptoms but failing to meet primary outcomes (Jacquemont et al., 2011; Bailey et al., 2016; Berry-Kravis et al., 2016; Youssef et al., 2018). These failures may in part be attributable to the difficulties associated with study design for indications lacking rigorous criteria for assessment of behavioral symptoms, or a need to stratify patient populations. Basimglurant also failed to show efficacy in primary clinician assessed measures for major depressive disorder, although patient-reported outcomes suggested an antidepressive effect (Quiroz et al., 2016). Mavoglurant entered phase II trials for obsessive-compulsive disorder but was terminated early due to lack of efficacy and a higher incidence of adverse effects (Rutrick et al., 2017). Beyond psychiatric indications, mavoglurant lacked efficacy in treating patients with levodopa-induced dyskinesia (LID) in PD or chorea in patients with Huntington’s disease (Reilmann et al., 2015; Trenkwalder et al., 2016). In contrast, dipraglurant (ADX48621) was also assessed in a phase IIa trial for PD-LID, showing promising indications of antidyskinetic efficacy (Tison et al., 2016). Raseglurant (ADX10059) showed antireflux efficacy for gastroesophageal reflux disease (Keywood et al., 2009; Zerbib et al., 2010, 2011), but further development was ultimately discontinued due to liver toxicity concerns.

With respect to modulators of other mGlu receptor subtypes, building on preclinical efficacy in addiction models for nicotine and methamphetamine (Caprioli et al., 2015; Justinova et al., 2015; Li et al., 2016), AZD8529, an mGlu₂ receptor PAM, is currently in phase II trials for smoking cessation but failed to demonstrate efficacy as an antipsychotic or for negative symptoms in patients with schizophrenia (Litman et al., 2016). Another mGlu₂ receptor PAM, ADX71149/JNJ-40411813, was also trialed for smoking cessation and found to improve attention and memory as well as reverse effects of ketamine (Salih et al., 2015). In a phase II trial of major depressive disorder patients with significant anxiety, JNJ-40411813 as an adjunct therapy to standard of care failed to relieve anxiety (Kent et al., 2016). A group II receptor NAM, decoglurant, also commenced trials for major depressive disorder; results are yet to be posted, but development was discontinued. The efficacy and safety of an mGlu₄ PAM, foliglurax, was recently evaluated in a phase II clinical trials in patients with PD treated with levodopa, experiencing end-of-dose wearing off and levodopa-induced dyskinesia (Charvin et al., 2017, 2018a), but the program was discontinued due to insufficient efficacy. The varying degrees of success and failures with clinical translation for mGlu receptor allosteric modulators are in keeping with high attrition rates for neuroscience targets. Moreover, these results speak to a need to better understand the pharmacological properties of allosteric modulators and harness novel modes of action and activity.

H. Secondary Allosteric Sites within the VFT and 7TM Domains

The vast majority of allosteric modulators for mGlu receptors are believed to interact with a common allosteric site within the 7TM domain, analogous to the biogenic amine orthosteric binding pocket of class A GPCRs. However, allosteric modulators interacting at alternate allosteric sites may offer the means to engender unique pharmacological profiles and increased subtype selectivity. Within the context of a full-length dimeric mGlu receptor, there are multiple possible sites to exploit for allosteric ligands. Multiple subtype-selective single domain antibodies, also referred to as nanobodies, have now been described that recognize epitopes within the VFT. For mGlu₂, three nanobodies that recognize overlapping epitopes but have different pharmacological properties have been described (Scholler et al., 2017). DN1 recognizes both active and inactive mGlu₂ receptors, whereas DN10 and DN13 require active homodimeric mGlu₂ receptor states to bind and potentiate orthosteric agonist activity (Scholler et al., 2017). Similarly, a nanobody that recognizes a loop region with lobe 1 of the mGlu₅ receptor VFT potentiates agonist binding and function but can also recognize both active and inactive receptor states (Koehl et al., 2019). To date, nanobodies have been used to facilitate structural studies or as biosensors for active receptor conformations. For mGlu₇ receptors, a monoclonal antibody, MB1/28, binds to the dimeric VFT, inhibiting receptor activation but is able to induce receptor internalization (Ullmer et al., 2012). Beyond antibodies, the naturally sourced sweet protein monellin was recently revealed as an mGlu₅ receptor allosteric agonist that is also thought to interact with the N terminus, and interacts allosterically with small molecule allosteric modulators (Chen et al., 2020).

Multiple allosteric sites have been postulated for the mGlu₅ receptor 7TM domain; however, the precise location of these secondary allosteric sites has proven elusive (Chen et al., 2008; Hammond et al., 2010; Noetzel et al., 2013). With the recent publication of a full-length cryo-electron microscopy structure of mGlu₅ receptors, we now appreciate that the cysteine-rich domain is a stalk that holds the VFT above the 7TM domains and interacts with the second extracellular loop to transmit conformational changes (Koehl et al., 2019). It is tempting to speculate that the inability to identify these secondary pockets may have been due to a monomeric view of the 7TM domain. However, biophysical studies as well as the new structures demonstrate that the 7TM domains themselves dimerize when activated (El Moustaine et al., 2012; Doumazane et al., 2013; Xue et al., 2015; Koehl et al., 2019). Appreciation of the full-length dimeric structure offers the possibility of identifying new allosteric sites to exploit through targeting these newly appreciated interfaces.

A. Biased Modulators

Allosteric modulators elicit potentiation or inhibition through stabilizing different receptor conformations than can be achieved with an orthosteric ligand alone. Therefore, there is the potential that these conformations can give rise to biased pharmacology. Where intrinsic efficacy differs between pathways relative to a reference agonist, this is referred to as biased agonism and can be quantified as discussed previously for biased orthosteric agonists. For mGlu₅ receptors, PAM agonists from diverse scaffolds are biased agonists relative to DHPG in both recombinant and native cells; however, the bias profile differs between scaffolds, with 5-[2-(2-(3-fluorophenyl)ethynyl]-N-[(1R)-2-hydroxy-1,2-dimethylpropyl]-2-pyridinecarboxamide, DPFE, and [6,7-dihydro-2-(phenoxymethyl)oxazolo[5,4-c]pyridin-5(4H)-yl](fluorophenyl)methanone (VU0409551) each exhibiting different bias profiles for mGlu₅ receptor signaling and receptor desensitization (Sengmany et al., 2017; Hellyer et al., 2019). However, biased modulation is also possible, where the direction or magnitude of modulation of the same agonist differs between pathways and may manifest either as differential apparent affinity or cooperativity (Sengmany et al., 2019). For structurally diverse mGlu₅ receptor PAMs (VU0360172, DPFE, VU0409551), the magnitude of cooperativity with DHPG was lower when measured in IP₁ accumulation compared with iCa²⁺ mobilization (Sengmany et al., 2017). Moreover, DPFE and CDPPB inhibit mGlu₅ receptor orthosteric agonist stimulated ERK1/2 phosphorylation in primary cultures (Zhang et al., 2005; Sengmany et al., 2017). Importantly, biased allosteric agonism and potentiation of mGlu₅ receptors, as well as probe dependence, are mediated via dynamic interactions within the common allosteric pocket (Hellyer et al., 2020; Sengmany et al., 2020). Allosteric interactions with distinct binding sites would be expected to offer further diversity in these biased pharmacological fingerprints. Furthermore, differential cooperativity has also been noted for mGlu₅ receptor NAMs, where N-(3-chloro-2-fluorophenyl)-3-cyano-5-fluoro-benzamide inhibits iCa²⁺ mobilization but is an NAL with respect to IP₁ accumulation (Sengmany et al., 2019). Recent studies indicate biased agonism and modulation of mGlu₅ receptor NAMs and PAMs extends to receptor regulatory processes such as internalization and desensitization (Hellyer et al., 2019; Arsova et al., 2020). Another contributing factor to biased modulator pharmacology at mGlu₅ receptors may be ligand binding kinetics as suggested by two recent studies on structurally diverse NAMs (Sengmany et al., 2019; Arsova et al., 2020).

Biased pharmacology of mGlu₅ receptor PAMs extends beyond second messenger signaling in cultures to intact circuitry. VU0409551 potentiates mGlu₅-dependent long-term depression in the hippocampus (Rook et al., 2015b), prefrontal cortex (Ghoshal et al., 2017), and nucleus accumbens (Turner et al., 2018) but unlike other PAMs is unable to potentiate DHPG stimulation of mGlu₅ receptor-mediated modulation of NMDA receptor currents (Rook et al., 2015b). Differential potentiation of mGlu₅ receptor-dependent synaptic plasticity and modulation of NMDA receptor activity in the hippocampus has also been noted for mGlu₅ receptor PAMs structurally unrelated to VU0409551. N-(1,3-Diphenyl-1H-pyrazolo-5-yl)-4-nitrobenzamide (VU29) potentiates hippocampal long-term potentiation but not NMDA receptor currents (Xiang et al., 2019). Within these brain slice electrophysiology experiments, mGlu₅ receptor PAM effects are mediated via enhancement of endogenous glutamate tone, or exogenously applied DHPG, suggesting that similar to VU0409551, VU29 has biased cooperativity between these two measures of mGlu₅ activity. Another structurally unrelated mGlu₅ receptor PAM, N-[4-chloro-2-[(4-fluoro-1,3-dioxoisoindol-2-yl)methyl]phenyl]pyridine-2-carboxamide, failed to potentiate DHPG-mGlu₅ receptor long-term potentiation and long-term depression in brain slice electrophysiology experiments (Noetzel et al., 2013). For mGlu₅ receptor allosteric ligands, biased pharmacology may offer the means to selectively modulate therapeutically beneficial effects while avoiding those linked to adverse effects. To realize this potential, there remains a need to better understand how different mGlu₅ receptor signaling and cellular responses are linked to behavioral effects in the whole animal and the translation of these effects to the clinic. By example, it was recently demonstrated for a series of mGlu₅ receptor PAMs closely related to VU0409551 that cooperativity with glutamate (for iCa²⁺ mobilization) rather than ligand affinity was predictive of relative efficacy in rats for reversing amphetamine-induced hyperlocomotion (Gregory et al., 2019). However, whether mGlu₅ receptor PAM affinity, cooperativity, bias, or agonism proves to be the best predictor for therapeutic efficacy and safety across multiple in vivo measures for different scaffolds remains to be rigorously tested.

Beyond mGlu₅ receptors, the mGlu₇ receptor-targeting monoclonal antibody, MB1/28, is an NAM for orthosteric agonist inhibition of cAMP accumulation but has intrinsic efficacy for inducing receptor internalization (Ullmer et al., 2012). At mGlu₂ receptors, the PAM agonist BINA was found to be a biased agonist relative to LY354740, favoring ERK1/2 phosphorylation over coupling to Gα₁₅ (Hellyer et al., 2020). Furthermore, biased pharmacology can contribute to apparent selectivity of allosteric ligands. Reportedly, selective allosteric ligands across the class C GPCR family were recently shown to either have neutral cooperativity for mGlu₅ receptors or have biased pharmacology (Hellyer et al., 2018). Collectively, the potential for biased allosteric ligand pharmacology highlights the importance of considering the assay/system/approach used to define mechanism of action, classify pharmacological effects, and selectivity.

B. Location- and Context-Dependent Pharmacology

Related to the phenomenon of biased allosteric agonism and modulation is the concept of context-dependent pharmacology, including the contribution of “location bias.” Quite simply put, the observed pharmacological effect is influenced by the cellular context within which it is studied. The first report for context-dependent pharmacology was for the mGlu₇ receptor NAM, MMPIP, which shows different magnitudes of inhibition of mGlu₇ receptor activity for the same agonist between different recombinant cell lines (Niswender et al., 2010). For mGlu₅ receptor modulators, both PAMs and NAMs have context-dependent pharmacology, manifested as distinct biased agonism profiles (Sengmany et al., 2017; Hellyer et al., 2019) or differential apparent affinities (Sengmany et al., 2019) or potencies (Jong et al., 2019) of NAMs between recombinant and native cells from different brain regions. Quantitative pharmacological differences between cell types may be a consequence of different stimulus-response coupling efficiencies, the presence or absence of receptor interacting proteins (other GPCRs, transducers, or scaffolding partners), or differences in receptor subcellular compartmentalization and relative accessibility by different ligands.

Group I mGlu receptors associate with different lipid microdomains (Burgueño et al., 2003; Francesconi et al., 2009b), the balance of which can be altered by receptor activation or membrane cholesterol content, which in turn can modulate signaling to ERK1/2 phosphorylation (Kumar et al., 2008, 2013). Furthermore, mGlu₅ is also found on intracellular membranes (e.g., nucleus; Jong et al., 2005), with signaling arising from these intracellular sites differing from that elicited by plasma membrane receptors (Jong et al., 2009; Kumar et al., 2012; Purgert et al., 2014). Subcellular compartmentalization of mGlu receptors can shape the physiologic responses to orthosteric agonists, particularly for surrogate agonists that cannot access intracellular receptors. Such effects may contribute to observations of probe dependence by allosteric ligands. To date, the influence of allosteric ligands on GPCR subcellular localization (and vice versa) has been relatively unexplored. However, Ca-sensing receptor allosteric ligands can act as “pharmacochaperones” to increase Ca-sensing receptor cell surface expression (White et al., 2009), suggesting that mGlu receptor allosteric ligands may also have the potential to alter receptor location.

In addition to physiologic context differences, the disease state can also impact mGlu receptor signaling and subsequently ligand pharmacology. In the setting of melanoma, mGlu₁ loses the capacity to signal via classic G protein pathways; however, glutamate retains the ability to stimulate mGlu₁ internalization, which promotes melanoma cell survival (Gelb et al., 2015b). In a preclinical model of Huntington’s disease, the balance of mGlu₅ receptor signaling pathways are perturbed, where IP₁ accumulation is reduced, but iCa²⁺ mobilization, Akt, and ERK1/2 phosphorylation are increased (Ribeiro et al., 2010). Brain region–specific changes in group I mGlu receptor signaling have also been noted after chronic cocaine administration, such that the mGlu₅ receptor NAM MPEP becomes an agonist for inducing cAMP response element-binding protein phosphorylation in the nucleus accumbens but not striatum (Hoffmann et al., 2017). Changes in the balance of intracellular signaling responses specific to the disease setting could be exploited by the development of biased ligands; however, they also offer another layer of complexity with respect to designing appropriate discovery pipelines for the translation of biased ligands. Indeed, VU0409551, which does not potentiate mGlu₅ receptor modulation of NMDA receptor currents in wild-type animals, does potentiate these responses in a genetic model of schizophrenia (Balu et al., 2016). These data highlight the paucity in our understanding of how different diseases and pathologic processes reshape the intracellular responses to mGlu receptors. Parsing out these differences will offer the prospect for rational discovery efforts to tailor therapeutic efficacy to the pathway level to restore neurotransmission to the healthy setting.

C. Heteromerization of mGlu Receptors

Historically, mGlu receptors were believed to form strict homodimers, unlike other class C GPCRs such as the obligatory heterodimeric GABA_B or sweet or umami taste receptors. The first evidence of mGlu receptor heteromers came from the demonstration that group I receptors can assemble and function together when cotransfected in HEK293 cells but cannot assemble with either group II or group III receptors (Doumazane et al., 2011). This same study demonstrated that heteromers are also formed among group II and group III receptors, prompting investigations into the existence and function of native mGlu receptors heteromers in vivo.

One anticipates that the formation of heterocomplexes by mGlu receptors, where natively expressed in the same cells, should result in specific pharmacological signatures that differ from homomeric receptors. Intriguing pharmacological responses aroused suspicion on the existence of mGlu₂-mGlu₄ receptor heteromers in rat dorsal striatum where the well established mGlu₄ receptor PAM PHCCC failed to potentiate mGlu₄ receptor activity at corticostriatal synapses in rat dorsal striatum, whereas VU0155041 retained its expected mGlu₄ receptor PAM activity (Yin et al., 2014). In vitro studies confirmed that mGlu₂-mGlu₄ receptor heteromers are differentially potentiated by mGlu₄ receptor PAMs from different scaffolds (Kammermeier, 2012; Yin et al., 2014; Niswender et al., 2016). At the mechanistic level, these pharmacological differences arise from complex asymmetric functioning of mGlu₂-mGlu₄ receptor heteromers. Indeed, after orthosteric agonist activation, the signaling of mGlu₂-mGlu₄ receptor heteromer only occurs through the transmembrane domain of mGlu₄ receptor (Liu et al., 2017); however, the mGlu₂ receptor subunit can signal if potentiated by an mGlu₂ receptor PAM. This is reminiscent of previous studies showing that only one subunit is active at a time in an mGlu receptor homodimer (Goudet et al., 2005; Hlavackova et al., 2005). Evidence of the presence of mGlu₂-mGlu₄ receptor heteromers at this corticostriatal synapses were reinforced by immunoprecipitation studies showing the presence of protein complexes containing mGlu₂ and mGlu₄ receptors in striatum (Yin et al., 2014). Pharmacological evidence of mGlu₂-mGlu₄ receptor heteromers has also been detected in lateral perforant path terminals in rat hippocampus (Moreno Delgado et al., 2017).

Recently, mGlu₂-mGlu₇ receptor heteromers were reported in the hippocampus and mGlu₃-mGlu₇ receptor heteromers in the cortex (Habrian et al., 2019). Interestingly, further in vitro investigations using a single molecular Förster resonance energy transfer approach revealed that the glutamate affinity and efficacy at mGlu₇ receptors are greatly enhanced when associated to an mGlu₂ receptor subunit, as compared with the mGlu₇ receptor homodimer. Also, association with mGlu₂ receptors confers to the mGlu₇ receptor subunit the ability to be fully activated by the selective group III agonist LSP4-2022. Of note, previous neuroanatomical study revealed that mGlu₇ and mGlu₈ receptors may be expressed in the same boutons in the hippocampus (Ferraguti et al., 2005), raising the possibility the mGlu₇-mGlu₈ receptor heteromers may also be of relevance in the hippocampus. Heterodimerization of mGlu₇ receptors with other mGlu receptor subtypes may provide a means to enhance the range of glutamate concentrations sensed by the mGlu₇ receptor, which is otherwise insensitive to low glutamate levels.

There is also evidence of mGlu₁ and mGlu₅ receptors forming complexes in mouse hippocampus and cortex as shown by a knockout-controlled interaction proteomics strategy and further confirmed by immunoprecipitation and superresolution microscopy imaging of hippocampal primary neurons revealing mGlu₁-mGlu₅ receptor coexpression at the synaptic level (Pandya et al., 2016). Indeed, these data are in keeping with evidence that blockade of both group I receptor subtypes is required to ablate DHPG-induced long-term depression in the hippocampus (Volk et al., 2006). More recently, a single-cell RNA sequencing study revealed the coexpression of different mGlu subtypes within the same cell in the adult mouse cortex (Lee et al., 2020). Notably, most pyramidal cells contained at least four receptor subtypes. Probing the propensity of different mGlu receptors to coassemble by fluorescent-based complementation assays, the authors concluded that mGlu₂ and mGlu₃ receptors are particularly prone to form heteromers when coexpressed in heterologous cells (Lee et al., 2020). Other prominent mGlu receptor pairs included mGlu_2/4, mGlu_1/5, mGlu_3/4, and mGlu_3/7. The coexpression of native mGlu₂ and mGlu₃ receptors in mouse frontal cortex was confirmed by in situ hybridization and coimmunoprecipitation (Lee et al., 2020). The prevalence of heteromerization between mGlu receptors adds considerable complexity to understanding and interpreting molecular pharmacological properties of ligands and particularly the notion of selectivity.

Beyond heteromerization with other mGlu receptor subtypes, increasing evidence suggests mGlu receptors form heteromers and larger-order complexes with class A GPCRs, including mGlu₂-5-HT_2A receptors (González-Maeso et al., 2008; Fribourg et al., 2011; Delille et al., 2013; Moreno et al., 2013; Moreno et al., 2016; Felsing et al., 2018), group I receptors with multiple adenosine receptor subtypes (Ciruela et al., 2001; Ferré et al., 2002; Nishi et al., 2003; Domenici et al., 2004; Rodrigues et al., 2005), mGlu₅-dopamine D₁ receptors (Sebastianutto et al., 2020), mGlu₅-dopamine D₂ receptors (Ferré et al., 1999; Popoli et al., 2001), and mGlu₅-dopamine D₂-adenosine A_2A receptors (Díaz-Cabiale et al., 2002; Cabello et al., 2009). For each pairing with a class A GPCR, the functional responses arising when receptors are coactivated or coincidentally inhibited changes the pharmacological profile to when mGlu receptor is activated in isolation. By example, heteromerization with dopamine D₁ receptors enhances the proportion of mGlu₅ receptors in active states, elevating basal G_q coupling and signaling toward iCa²⁺ mobilization over cAMP pathways (Sebastianutto et al., 2020). Heteromerization is often observed in a cell type– or brain region–specific fashion. In this respect, selectively targeting mGlu receptor heteromers offers the intriguing prospect of achieving tissue-level selectivity of drug action. The study of mGlu receptors heteromers is still in its infancy, and it is clear that further investigations will be needed to better understand its functional consequences in brain function and therapeutic potential.

D. Optical Tools to Probe and Control mGlu Receptors

Irreversible ligands or photoaffinity probes have been widely used to study ligand-receptor interactions and aid structural determinations across diverse protein targets. Such tools have not been available for mGlu receptors; however, recent efforts exploiting selective allosteric chemotypes have proven successful. The first in class were bifunctional mGlu₅ receptor NAMs that included a photoactivatable moiety to irreversibly bind receptors and a click chemistry handle to allow secondary attachment of clickable reporter (e.g., fluorophore) for identification (Gregory et al., 2016). Installation of a covalent or photoreactive moiety has been successfully achieved within three distinct mGlu₂ receptor PAM scaffolds (Doornbos et al., 2019; Hellyer et al., 2020). The development of covalent or photoactivatable ligands is not without its challenges. Covalent ligands require proximity to an appropriate amino acid for reactivity. Furthermore, the bifunctional clickable photoprobes for mGlu₂ and mGlu₅ receptors revealed substantial nonspecific interactions, which may limit how these tools can be applied. There has been considerable interest in alternative approaches to optically control mGlu receptor function. Two main strategies exist to control mGlu receptors by light: an optogenetic pharmacology approach based on attached photoswitchable ligands (Fig. 9) and a photopharmacology approach based on freely diffusible light-operated ligands (Fig. 10) (Goudet et al., 2018). The aim of both strategies is to use light to achieve precise spatiotemporal control over receptor activity.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 9.

Optogenetic pharmacology of mGlu receptors. Optogenetic pharmacology consists in covalently attaching a photoswitchable tethered ligand to a genetically modified protein, enabling the photoactivation or photoantagonism of the receptor. Several strategies have been applied to mGlu receptors. (A) Photoswitchable tethered ligands. (B) PORTL. (C) Membrane-anchored PORTL. (D) Antibody-based photocontrol. ST, SNAP-tag; GFP, green fluorescent protein.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 10.

Photopharmacology of mGlu receptors. Photopharmacology is based on the use of small, diffusible, drug-like, photoregulated ligands, enabling the photoactivation or photoantagonism of the receptor. (A and B) Photocaged ligands (also named photoactivable ligands) possess a protecting group that can be removed after illumination, enabling the onset of drug activity on the receptor. Example: An inactive photocaged derivative of the mGlu₅ NAM raseglurant. Uncaging is provoked by violet illumination, allowing raseglurant to effectively block mGlu₅ activity in cells or in living mice. (C and D) Photoswitchable ligands are rapidly and reversibly photoisomerizing at specific wavelengths, modifying their overall structure and thus their ability to interact with their target. Azobenzene is the most common photoisomerizable core used to design photoswitchable ligands. In the dark or under white light, the azobenzene moiety is in a trans configuration converting to a cis configuration upon illumination with an appropriate wavelength (usually in the UV range). Example: Optogluram, a photoswitchable mGlu₄ PAM. DEACM, 7-(diethylamino)coumarin-4-yl]methyl.