Modulation of NMDA receptors by GPCRs: role in synaptic transmission, plasticity and beyond

Abstract

NMDA receptors (NMDARs) play a critical role in excitatory synaptic transmission, plasticity and in several forms of learning and memory. In addition, NMDAR dysfunction is believed to underlie a number of neuropsychiatric conditions. Growing evidence has demonstrated that NMDARs are tightly regulated by several G-protein-coupled receptors (GPCRs). Ligands that bind to GPCRs, such as neurotransmitters and neuromodulators, activate intracellular pathways that modulate NMDAR expression, subcellular localization and/or functional properties in a short- or a long-term manner across many synapses throughout the central nervous system. In this review article we summarize current knowledge on the molecular and cellular mechanisms underlying NMDAR modulation by GPCRs, and we discuss the implications of this modulation spanning from synaptic transmission and plasticity to circuit function and brain disease.

Introduction

In the mammalian brain, fast excitatory synaptic transmission is mediated by three glutamate receptors: α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR), N-methylD-aspartate receptor (NMDAR) and kainate receptors. These receptors are molecularly and functionally diverse, and can be regulated by several neuromodulatory systems (Traynelis et al., 2010), making excitatory transmission a highly complex and regulated process. NMDARs stand out because of their dual role as mediators of synaptic transmission and long-term synaptic plasticity, such as long-term potentiation and depression (LTP and LTD) of AMPAR-mediated synaptic transmission. NMDARs are also dynamically regulated and subject to activity-dependent long-term plasticity (Hunt et al., 2012;Rebola et al., 2010). In addition, NMDARs can be found in the presynaptic compartment where they regulate neurotransmitter release and presynaptic plasticity (Bouvier et al., 2018). Furthermore, NMDAR dysfunction has been linked to several neuropsychiatric and neurodegenerative diseases (Lau et al., 2007;Paoletti et al., 2013). Given their critical contribution to brain function, it is not surprising that NMDARs are tightly regulated by a number of mechanisms. A key regulatory mechanism can be found in G-protein-coupled receptors (GPCRs) (Yang et al., 2014). This review article summarizes current evidence in support of GPCRs as key regulators of NMDARs, and the consequences of this process on synaptic transmission and plasticity, circuit function and disease.

NMDARs possess three unique properties (Paoletti, et al., 2013): high calcium permeability, slow gating kinetics, and Mg²⁺ block of the channel pore at resting potential. NMDARs act as coincidence detectors given that both glutamate binding and membrane depolarization, which is required to relieve the Mg²⁺ block of current flow through the channel, must occur in order to allow current flow through the channel. NMDARs are tetramers that can be assembled as di-heteromers, with two obligatory GluN1 and two GluN2 or GluN3 subunits, or exist as tri-heteromers, with a combination of GluN1 and different splicing variants of GluN2 and/or GluN3 (Paoletti, et al., 2013). The subunit composition determines the functional properties of NMDARs (Iacobucci et al., 2018;Paoletti, et al., 2013). Each subunit possesses different intra- and extracellular- motifs that are subject to post-translational modifications that in turn can change NMDAR properties, expression levels, synaptic localization and interaction with other important intracellular components (Kotecha et al., 2003b;Sanz-Clemente et al., 2013;Traynelis, et al., 2010). In addition to their canonical function as ionotropic channels, NMDARs can signal in a metabotropic manner (Dore et al., 2016;Gray et al., 2016). In this case, glutamate binding to NMDARs induces conformational changes of the receptor that activate intracellular messengers, which can trigger synaptic plasticity independently from ion flux through the channel (Dore et al., 2015;Nabavi et al., 2013;Stein et al., 2015). Lastly, NMDAR-mediated transmission and plasticity can be modulated by the release of glutamate and co-agonists (e.g. D-serine) from astrocytes, (Harada et al., 2015;Letellier et al., 2016;Mothet et al., 2006;Navarrete et al., 2019;Panatier et al., 2006).

GPCRs belong to the large family of seven transmembrane domain receptors (7TM) which are activated by neurotransmitters, neuromodulators, peptides and sensory stimuli (Bockaert et al., 1999;Pavlos et al., 2017). GPCRs can be classified in four families, namely, G_s, G_i/o, G_q, and G_12/13. The signaling pathways downstream of these GPCRs have been reviewed elsewhere (Kristiansen, 2004). Briefly, G_s stimulates adenylate cyclase which leads to the activation of the cyclic AMP (cAMP)- protein kinase A (PKA) pathway whereas G_i/o does the opposite (i.e. inhibits adenylate cyclase). G_q leads to protein kinase C (PKC) activation, inositol-3-phosphate (IP3) production and calcium release from internal stores. G_12/13 activates a small Rho-GTPase through guanine nucleotide exchange factor (RhoGEF). In addition, GPCRs can impact neuronal function by regulating several membrane ion channels, such as voltage-gated calcium channels (VGCCs) and potassium channels (Dascal, 2001;McCudden et al., 2005). GPCRs also signal through Gprotein-independent intracellular effectors that rely on direct interactions between the 7TM receptor and intracellular scaffolding proteins or other molecules (Heuss et al., 2000). Thus, GPCRs are well positioned to modulate NMDARs in a number of ways, and by this means strongly regulate synaptic transmission, plasticity, and circuit function (Anwyl, 1999;Hunt, et al., 2012;Rebola, et al., 2010;Tritsch et al., 2012;Yang, et al., 2014).

NMDAR modulation by GPCRs: role of neurotransmitters and neuromodulators

Metabotropic glutamate receptors

Metabotropic glutamatergic receptors (mGluRs) are one of the best-characterized examples of GPCRs that modulate NMDARs. mGluRs are glutamate-activated GPCRs that share no sequence homology with other families of GPCRs (Conn et al., 1997). There are 8 subtypes of mGluRs (1 to 8) that can be subdivided in three groups (I, II, III). Classically, group I mGluRs (mGluR1 and 5) are G_q-coupled, while group II (mGluR2 and 3) and group III (mGluR 4,6,7,8) mGluRs are G_i/o-coupled (Nicoletti et al., 2011). Group I mGluRs have been extensively studied for their interaction with NMDARs.

mGluRs and LTP of NMDAR

mGluR-mediated modulation of NMDARs was first observed in the CA1 area of the hippocampus, where activation of mGluRs with the broad-spectrum agonist (1S,3R)-1-Aminocyclopentane-1,3-dicarboxylic acid (ACPD) triggered a transient increase in NMDAR currents (Aniksztejn et al., 1991). In addition, tetanic stimulation of synaptic inputs to dentate gyrus granule cells induced LTP of NMDAR-mediated excitatory postsynaptic currents (EPSC), and this LTP and ACPDinduced potentiation occluded each other (O’Connor et al., 1994), suggesting a common expression mechanism. NMDAR LTP induced by synaptic activity was abolished by the application of the non-selective mGluR antagonist (RS)-α-Methyl-4-carboxyphenylglycine (MCPG) or by the NMDAR competitive antagonist D-2-amino-5-phosphonovalerate (D-AP5). These results provided the first evidence that NMDAR LTP required mGluR and NMDAR coactivation by endogenous glutamate. Co-activation of mGluR5 and NMDAR is a common feature across synapses that express NMDAR LTP (Hunt, et al., 2012;Rebola, et al., 2010). In support of this notion NMDAR LTP is absent in mGluR5 knockout mice (Jia et al., 1998). In isolated hippocampal cells, the selective mGluR5 agonist (RS)-2-Chloro-5-hydroxyphenylglycine (CHPG) induced a long-lasting increase of NMDAR EPSCs in an mGluR5, IP3 receptor (IP3R), PKC and non-receptor tyrosine kinase Src (Src)-dependent manner (Kotecha et al., 2003a). The CHPGmediated potentiation required mGluR5 and NMDARs co-activation. At the hippocampal mossy fiber-to-CA3 pyramidal cell (mf-CA3) synapse, brief patterns of presynaptic activity induced mGluR5-dependent LTP of NMDAR-mediated transmission (Kwon et al., 2008;Rebola et al., 2008). This NMDAR LTP not only required mGluR5/NMDAR co-activation, but also postsynaptic calcium influx via NMDARs, G-protein activation, recruitment of PKC and Src kinase, and calcium release from the internal stores (Figure 1). Thus, PKC and Src kinase seem to be the downstream targets of mGluR5 and NMDAR co-activation. In slice cultures, the mGluR-Src pathway potentiates NMDAR-mediated transmission in a G-protein-independent manner (Benquet et al., 2002;Heuss et al., 1999). In hippocampal neurons from both cultures and acute slices, activation of PKC modulates NMDAR gating properties and enhances NMDAR delivery to the plasma membrane following phosphorylation of SNARE-proteins (Kwon, et al., 2008;Lan et al., 2001a;Lan et al., 2001b;Lau et al., 2010;Zheng et al., 1999). Intriguingly, in midbrain dopaminergic neurons, NMDAR LTP requires PKA but not PKC activation (Harnett et al., 2009). Like in the hippocampus, NMDAR LTP in these dopaminergic neurons requires mGluR5 and NMDAR co-activation, which, together with calcium rise generated by postsynaptic action potentials leads to IP3R-mediated calcium-induced calcium release. While the link between mGluR5 and PKA is unclear, recent evidence in organotypic slice cultures showed that activation of G_q-coupled receptors, including group I mGluRs, increases PKA activity (Chen et al., 2017)(Figure 1). Phosphorylation of NMDAR by PKA modifies receptor properties such as calcium permeability, channel gating and currents in both slices and heterologous system (Aman et al., 2014;Murphy et al., 2014;Skeberdis et al., 2006). While group I mGluRs-mediated PKA activation could play a role in NMDAR LTP in midbrain dopaminergic neurons, clear mechanistic evidence for this process is missing.

Figure 1.

Mechanisms of NMDAR modulation by mGluRs

Activation of group I mGluRs leads to calcium release from the endoplasmic reticulum (ER) and activation of PKC via the PLC/IP3 pathway. Recruitment of calcium release from the internal stores requires the interaction of group I mGluRs with Homer 1b/c. PKC can enhance NMDAR-mediated currents and/or increase NMDAR trafficking to the synapse. In addition to PKC, G_q can also modulate NMDAR via the activation of the Src kinase. Group I mGluRs could potentially enhance NMDAR function by non-canonical activation of PKA. Activation of group I mGluRs can also induce depression of NMDAR-mediated transmission. Arrowheads and bars represent positive and negative modulation of downstream targets, respectively. Dashed lines represent a pathway which is largely unknown. Thick and thin lines indicate pathways that are activated or inhibited, respectively.

mGluRs and LTD of NMDARs

mGluRs are also involved in LTD of NMDAR-mediated synaptic transmission. At the mf-CA3 synapse, mGluR5 and mGluR1 induce bidirectional plasticity –i.e. NMDAR LTP/LTD (Hunt et al., 2013). Here, coincident pre- and postsynaptic burst activity induces NMDAR LTP or LTD depending on the timing of the pre and postsynaptic burst. Interestingly, while mGluR5 are required for NMDAR LTP, mGluR1 are required for NMDAR LTD (Hunt, et al., 2013). mGluRdependent NMDAR LTD has also been reported at Schaeffer collateral to CA1 pyramidal cell (Sch-CA1) synapses (Bhouri et al., 2014;Peng et al., 2010;Yi et al., 1995), and at perforant pathway-to-granule cell synapses (Harney et al., 2006).

Molecular mechanisms of mGluR-dependent bidirectional NMDAR plasticity

The reason why two receptors belonging to the same family can engage two opposite forms of plasticity is unclear. A possible explanation is the presence of different splicing variants of mGluR1 and mGluR5 (Conn, et al., 1997). While the two known mGluR5 isoforms possess a long C-terminal tail, some splicing variants of mGluR1, such as mGluR1b, contain a shorter intracellular C-terminal which is crucial for mGluRs coupling to intracellular effectors, such as Homer proteins (Joly et al., 1995;Niswender et al., 2010;Tu et al., 1999;Tu et al., 1998). Homer proteins comprise Homer1b/c and Homer1a. Homer 1a is a dominant negative Homer protein that binds to mGluRs and prevents their binding with Homer1b/c (Kammermeier et al., 2007). Binding of Homer1b/c to group I mGluRs is critical for their perisynaptic localization, functional interaction with NMDAR, and for the coupling to intracellular players such as IP3Rs (Lujan et al., 1996;Ohtani et al., 2014;Prezeau et al., 1996;Tu, et al., 1998;Xiao et al., 1998)(Figure 1). Since plasticity of NMDAR transmission strongly relies on postsynaptic calcium levels (Harnett, et al., 2009;Harney, et al., 2006;Hunt, et al., 2013;Kwon, et al., 2008), impaired coupling of mGluRs to IP3Rs could affect IP3R-mediated calcium rise required for NMDAR plasticity. Consistent with this hypothesis, mfCA3 synapses express the mGluR1b short-tail isoform (Hunt, et al., 2013), suggesting that mGluR1-dependent NMDAR LTD might rely on different postsynaptic calcium dynamics compared to mGluR5-dependent NMDAR LTP. At the Sch-CA1 synapse, expression of mGluR1dependent NMDAR LTD is input specific and correlates with higher levels of Homer1b/c at Schinputs compared to temporoammonic inputs, which lack NMDAR LTD, indicating that levels of Homer proteins can influence the expression of plasticity (Bhouri, et al., 2014). In the Fmr1 knockout (KO) mouse, a model of Fragile X syndrome (FXS), binding of Homer1a to mGluR5 disrupts the mGluR5/NMDAR interaction, thereby impairing NMDAR-mediated transmission, NMDAR-dependent plasticity and cognitive performance (Aloisi et al., 2017). In addition, disrupting the mGluR5/Homer1b/c interaction with a TAT-peptide, a manipulation that recapitulates the effect of Homer1a binding to mGluR5, mimicked the impairment in NMDAR function observed in Fmr1 KO mice, and the functional and behavioral deficit were rescued by selective knockdown of Homer1a (Aloisi, et al., 2017). Together, these observations highlight the role of Homer as a critical molecular hub for the interaction between group I mGluRs and NMDARs and for mGluR-dependent NMDAR plasticity.

mGluR5 can also interact with NMDARs through other intracellular mediators, one being calcium-calmodulin-dependent kinase type 2 (CaMKII) (Jin et al., 2013;Marks et al., 2018). In cultured striatal neurons, mGluR1/5 receptors anchor inactive CaMKII at the synapse. Upon NMDARmGluR1/5 co-activation and calcium influx, CaMKII becomes active and loses its affinity for the C-terminal domain of mGluR1/5 (Jin, et al., 2013). CaMKII then phosphorylates the GluN2B subunit and triggers membrane delivery of GluN2B-containing NMDARs (Jin et al., 2015). It is currently unknown if this mechanism is conserved across other central synapses and in more intact preparations. CaMKII also regulates mGluR5 surface levels and mGluR5-mediated calcium influx in heterologous systems (Marks, et al., 2018). Modulation of mGluR5-mediated calcium dynamics may have an important impact on NMDARs. In the dentate gyrus, mGluR5 mediate a form of NMDAR LTD which can be switched to LTP depending on different intracellular calcium buffer capacity (Harney, et al., 2006). This observation suggests that modulation of calcium dynamics downstream of mGluR5 activation might be important for the bidirectionality of mGluRs modulation of NMDAR –i.e. LTP vs LTD.

GABA_B receptors

GABA_B receptors (GABA_BRs) are GABA-activated G_i/o-coupled receptors that inhibit adenylate cyclase and the cAMP/PKA signaling cascade (Bettler et al., 2004). GABA_BRs also modulate ion channels such as VGCCs and G-protein-coupled inward-rectifying potassium (GIRK) channels (Bettler, et al., 2004) (Figure 2). GABA_BRs are expressed both post- and presynaptically and their activation, respectively, mediates a slow inhibitory postsynaptic potential (slow IPSP) and suppresses neurotransmitter release.

Figure 2.

Mechanisms of NMDAR modulation by GABA_BRs

GABA_BRs negatively regulate the adenylate cyclase (AC)/cAMP/PKA pathway and reduce the function of NMDAR. GABA_BRs can also reduce NMDAR conductance by hyperpolarizing the membrane via the activation of GIRK channels

GABA_BR direct modulation of NMDARs

In layer 2/3 pyramidal neurons of the prefrontal cortex (PFC), activation of postsynaptic GABA_BRs selectively regulates NMDAR calcium permeability but not NMDAR EPSCs (Chalifoux et al., 2010). This effect is the result of G_i/o-mediated inhibition of PKA and is independent of modulation of VGCCs and GIRK channels, consistent with the fact that phosphorylation of NMDAR by PKA at certain residues increases NMDAR calcium permeability but not NMDAR EPSCs (Higley et al., 2010;Lur et al., 2015;Skeberdis, et al., 2006). This mechanism could expand the capability of GABAergic inhibition to change the threshold for plasticity – a phenomenon called metaplasticity– by targeting NMDAR calcium permeability only. In layer 5 pyramidal neurons of the PFC, GABA_BRs are co-expressed with G_i/o-coupled α2 adrenergic receptors (α2ARs) (Lur, et al., 2015). Here, α2ARs activation suppresses AMPAR-EPSCs but not NMDAR-EPSCs, while GABA_BRs selectively inhibit NMDARs despite the common G_i/o coupling. The different effect of α2ARs and GABA_BRs is likely due to Regulator of G-proteins Signaling (RGS) proteins that create synaptic microdomains and prevent crosstalk between GPCRs coupled to the same pathway (Hollinger et al., 2002;Lur, et al., 2015)

GABA_BR indirect modulation of NMDARs

Activation of GIRK channels by GABA_BRs leads to membrane hyperpolarization, which strongly reduces NMDAR-mediated transmission due to enhanced Mg²⁺ block (Morrisett et al., 1991;Otmakhova et al., 2004). Theoretical models predict that the combined conductance of both NMDAR and GABA_BR-activated GIRK channels can generate robust bistability of membrane voltage and persistent firing states, a property thought to be important for working memory (Sanders et al., 2013). Current evidence suggests that GABA_BRs mediate a non-canonical excitatory effect by activating specific subtypes of VGCCs, or by indirect inhibition of largeconductance calcium-activated potassium channels (BK channels) (Garaycochea et al., 2016;Zhang et al., 2016). Such mechanisms could lead to membrane depolarization and facilitate NMDAR-mediated currents and potentials.

Dopamine receptors

Dopamine (DA) receptors can be divided in two families: D1-like (which includes D1R and D5R) and D2-like (which includes D2R, D3R and D4R). D1-like receptors are mainly G_s-coupled while D2-like receptors are coupled to G_i/o (Beaulieu et al., 2011). The two receptor families are widely expressed in the brain, display different affinities for DA, and can be expressed in a cell-specific manner. At the synaptic level, D1- and D2-like receptors can be expressed on the dendritic tree but they also localize at the perisynaptic region of dendritic spines (Ladepeche et al., 2013;Smiley et al., 1994). Due to their localization on dendritic spines, DA receptors are ideally positioned to modulate NMDARs. The interaction between DA receptors and NMDARs has been investigated at several central synapses (Tritsch, et al., 2012). D1- and D2-like receptors seem to play a “pushpull” effect on NMDAR, with D1-like meditating potentiation and D2-like mediating depression of NMDAR-mediated function.

PKA-mediated modulation of NMDARs by DA receptors

PKA is a key player in the dichotomic effect of D1-like and D2-like receptors on NMDAR function. D1-like receptors are G_s-coupled and their activation increases cAMP levels and PKA activity (Tritsch, et al., 2012) (Figure 3). Several studies in both cultured neurons and acute slices from PFC, striatum and nucleus accumbens have demonstrated that activation of D1-like receptors by DA or selective agonists modulates NMDAR in a PKA-dependent manner (Dudman et al., 2003;Li et al., 2010;Snyder et al., 1998;Wang et al., 2001). Conversely, D2-like receptor activation depresses NMDAR function via PKA inhibition in cultured neurons and acute slices from PFC and hippocampus (Banks et al., 2015;Higley, et al., 2010;Kotecha et al., 2002;Wang et al., 2003). In D2-positive striatal medium-sized spiny neurons (MSNs), activation of D2Rs by an agonist reduces NMDAR-calcium permeability without affecting NMDAR EPSCs (Higley, et al., 2010). The PKA target involved in this process is still a matter of debate. In the CA1 area of the hippocampus PKA selectively regulates NMDAR calcium permeability (Skeberdis, et al., 2006), presumably by phosphorylating the GluN2B subunit at residue Ser¹¹⁶⁶ (Murphy, et al., 2014). However, in D2-positive MSNs, Ser¹¹⁶⁶ is unlikely to be the target of PKA given that GluN2Bcontaining NMDAR in these cells do not seem to contribute to D2R modulation of NMDAR-calcium permeability. In heterologous systems, phosphorylation of GluN1 and not GluN2B regulates NMDAR calcium permeability, suggesting that the GluN1 subunit might be a potential target of PKA in native brain (Aman, et al., 2014). Alternatively, NMDARs could be indirectly modulated by PKA signaling. In the nucleus accumbens, activation of PKA by D1R agonists increases the phosphorylation levels of the GluN1 subunit (Snyder, et al., 1998). This enhancement requires activation of Dopamine- and cAMP-Regulated Phosphoprotein (Mr 32 kDa) (DARPP-32), a phosphoprotein that can inhibit protein phosphatase 1 (PP1) (Figure3). GluN1 phosphorylation levels are also increased by direct blockade of PP1/2A, suggesting a tonic dephosphorylation of GluN1 subunit by PP1. The increase of GluN1 phosphorylation by a D1R agonist was prevented by co-application of D2R agonist quinpirole which dephosphorylates DARPP-32 via calcineurin A. These results indicate that PKA controls GluN1 phosphorylation levels by indirect inhibition of PP1 by DARPP-32. A similar mechanism was described in dissociated striatal MSNs, where D1R agonists increased NMDAR-mediated currents in a DARPP-32-dependent manner (Flores-Hernandez et al., 2002). The same study showed that activation of D2R strongly reduced D1Rmediated potentiation of NMDAR currents, suggesting that DARPP-32 might be a key mediator of the push-pull modulation of NMDARs by DA receptors. It is worth noting that most of these studies employed glutamate uncaging or NMDA application (iontophoresis, puff and or bath application) to test the modulation of NMDAR-mediated currents. While these approaches minimize potential presynaptic effects by potential activation of DA receptors, they do not fully mimic the dynamics of NMDAR activation during glutamate release. This stands true especially during repetitive presynaptic activity where postsynaptic depolarization, by relieving the Mg²⁺ block, increases NMDAR EPSCs (Hunt, et al., 2012). Activation of DA receptors can also change membrane ion channels and thus influence the membrane potential (Cepeda et al., 1998). Since NMDAR are voltage-dependent and voltage-clamp is prone to errors (Beaulieu-Laroche et al., 2018;Williams et al., 2008), changes in membrane potential could lead to misleading results.

Figure 3.

Mechanisms of NMDAR modulation by DA receptors.

D1-like and D2-like-coupled receptors positively and negatively regulate the AC/cAMP/PKA pathway, respectively. PKA modulates NMDAR function directly by regulating NMDAR phosphorylation levels or indirectly by controlling other downstream targets (i.e. inhibition of PP1 via DARPP-32). Direct proteinprotein interaction between NMDARs and DA receptors receptors (red arrows) can also affect the synaptic localization of NMDAR.

PKA-independent modulation of NMDARs by DA receptors

DA receptors can also modulate NMDARs in a PKA-independent manner. In dissociated PFC neurons, D1R agonists increase NMDAR currents via PKC (Chen et al., 2004). The link between D1Rs and PKC is unclear but does not require PLC activation, suggesting that D1R are not acting via G_q -PLC pathway (Chen, et al., 2004). In cultured CA1 pyramidal neurons, activation of D4Rs triggers depression of NMDAR currents, and this effect is mediated by D4R-mediated transactivation of platelet derived growth factor receptor β (PDGFR) (Figure 3), which leads to calcium/calmodulin-dependent desensitization of NMDARs (Kotecha, et al., 2002). In the PFC, activation of D2/3Rs (but not D4Rs) triggers a PDGFR-dependent depression of NMDAR currents (Beazely et al., 2006). These studies suggest that D1- and D2-like receptors can modulate NMDAR currents in a PKA-independent manner, but do not eliminate the possibility that PKA could also modulate NMDAR calcium permeability.

Modulation of NMDARs via direct DA receptor/NMDAR interaction

In addition to modulating NMDARs via intracellular pathways, both D1- and D2-like receptors can modulate NMDARs via direct protein-protein interaction (Cepeda et al., 2006;Tritsch, et al., 2012) (Figure 3). In hippocampal lysates the C-terminal domain of D1R interacts with the GluN1 and GluN2 subunits (Lee et al., 2002). The interaction with the GluN1 subunit seems to be required for D1R-dependent activation of PI3 kinase, while the D1R-GluN2 interaction determines D1Rdependent reduction of NMDAR surface levels. Super-resolution microscopy demonstrated that D1R-NMDAR interaction occurs at perisynaptic areas of excitatory spines of hippocampal neuronal cultures, and that D1R activation decreased the size of D1R and NMDAR clusters, which increased the motility and synaptic localization of NMDARs, thereby facilitating the induction of NMDAR-dependent plasticity in acute hippocampal slices (Ladepeche, et al., 2013). In contrast, acute cocaine administration, which increases synaptic levels of dopamine, led to D2R activation and the formation of D2R/GluN2B complexes that resulted in less phosphorylation of GluN2B by CaMKII (Liu et al., 2006b). This mechanism seems to play a significant role in the behavioral changes induced by cocaine (Liu, et al., 2006b). The D2R/GluN2B complex is also required for the inhibitory effect of D2R-agonists on NMDAR-currents (Liu, et al., 2006b), suggesting that the D2R-NMDAR direct interaction might play a role in the physiological modulation of NMDAR by DA.

Serotonin Receptors

The large family of serotonin (5HT) receptors comprises seven groups (5HT₁ to 5HT₇) and a few subtypes within some groups (Hoyer et al., 2002). Except for the 5HT₃ receptor, which is an ionotropic receptor, all other 5HT receptors are GPCRs. Each group couples to different G proteins: 5HT₁ and 5HT₅ are coupled to G_i/o; 5HT₂ are coupled to G_q; 5HT₄, 5HT₆ and 5HT₇ are coupled to G_s. The presence of multiple splice variants, RNA editing, and other posttranslational modifications add several layers of complexity to GPCR-coupled 5HT receptor signaling. 5HT receptors can be expressed in neuronal subcellular compartments including the soma, dendritic spines and axon terminals (Cornea-Hebert et al., 1999;Peddie et al., 2008;Riad et al., 2000), suggesting that their activation can regulate neuronal excitability, as well as synaptic transmission and plasticity at different levels.

Modulation of NMDAR-mediated transmission and plasticity by 5HT receptors

Early studies showed that 5HT potentiated NMDAR-mediated responses in the neocortex (Nedergaard et al., 1986;Reynolds et al., 1988) a phenomenon presumably mediated by 5HT₂ receptors (Rahman et al., 1993). The first investigations of the role of 5HT on NMDAR-dependent LTP yielded conflicting results. 5HT depletion abolished LTP at PP-GC synapse in the dentate gyrus (Bliss et al., 1983), and others showed a concentration-dependent inhibition of LTP by exogenous 5HT at Sch-CA1 synapse, in part presumably by modulating NMDARs (Staubli et al., 1994). These results highlight that 5HT may have synapse-specific effects that might arise from the diversity of 5HT receptors or from different synapse-specific properties. In the visual cortex, inhibition of NMDAR-dependent LTP relied on inhibition of NMDAR responses via activation of 5HT_1A, 5HT₂ and 5HT₇ receptors (Edagawa et al., 1998, 1999). Another study showed that 5HT application, by co-activation of 5HT₁ and 5HT_2A/C, abolished NMDAR-dependent LTP in the visual cortex of young rats (Kim et al., 2006). This LTP was absent in older animals, which could result from an age-dependent increase in 5HT levels as depletion of endogenous 5HT restored LTP. Thus, the effect of 5HT on NMDAR-dependent plasticity seems to be area-, synapse- and age-specific.

Molecular mechanisms of NMDARs modulation by 5HT₁ and 5HT₂ receptors

5HT modulates NMDAR function through different molecular pathways. In the PFC, 5HT_1A receptor activation decreases NMDAR currents and membrane levels of the GluN2B subunit (Yuen et al., 2005). 5HT_1A activation triggers destabilization of the complex formed by microtubule associated protein 2 (MAP2) and the molecular motor Kif17, which leads to a reduction of NMDAR trafficking to the membrane. Consistent with the G_i/o coupling of 5HT_1A receptors, this effect results from a decrease in PKA activity, which reduces phosphorylation of PKA targets such as CaMKII and extracellular signal-regulated kinase (ERK) that are known to phosphorylate MAP2 on the microtubule binding domain (Figure 4). Blockade and overexpression of RGS4 respectively enhanced and attenuated the modulation of NMDAR by 5HT_1A, suggesting that 5HT_1A-mediated NMDAR inhibition is negatively regulated by RGS4 (Figure 4). The inhibitory effect of 5HT_1A receptors in the PFC is compensated by 5HT_2A/C receptors (Yuen et al., 2008). 5HT_2A/C counteract the inhibitory effect of 5HT_1A by increasing ERK activity in a β-arrestin-dependent manner (Figure 4). A similar G-protein-independent modality for 5HT_2C receptors was also found in motoneurons of the frog spinal cord, where 5HT_2C activation induced a Src-dependent increase in NMDAR currents (Bigford et al., 2012)(Figure 4). In the PFC, 5HT_1A and 5HT_2A/C receptors agonists used at much lower concentration than other studies (Yuen, et al., 2008;Yuen, et al., 2005) regulated NMDAR-induced depolarization and firing in a bidirectional manner (Zhong et al., 2008). At lower doses, 5HT_1A and 5HT_2A/C receptors agonists alone did not change NMDAR currents or membrane properties, but modulated NMDAR-induced depolarization only when co-applied with NMDA. Moreover, low doses of 5HT_1A and 5HT_2A/C agonists alone did not change ERK activity as reported with higher concentrations of the same antagonists, suggesting a synergistic effect of 5HT receptor/NMDAR co-activation (Yuen, et al., 2008;Yuen, et al., 2005). Whether this dosedependent 5HT receptor-activation might somehow mimic physiological (or pathological) changes in 5HT levels is unclear. Of note, both 5HT receptors and NMDARs can be expressed presynaptically (Arvanov et al., 1999;Barre et al., 2016;Bouvier, et al., 2018) (Figure 4). 5HT_2A are expressed at presynaptic terminals of thalamocortical inputs to PFC where they enhance neurotransmitter release via modulation of presynaptic NMDAR in a PLC and PKC-dependent manner (Barre, et al., 2016).

Figure 4.

Mechanisms of NMDAR modulation by 5HT receptors.

5HT₇ and 5HT_1A receptors regulate PKA activity and have a positive and negative effect on NMDAR function respectively. 5HT_1A receptors inhibit ERK phosphorylation by PKA and reduce trafficking of NMDAR to the membrane. RGS4 negatively regulates the effect 5HT_1A receptors effect on PKA activity. 5HT₇ can also negatively regulate NMDAR by inducing transactivation of PDGFRβ which has an inhibitory effect on NMDAR-mediated transmission. 5HT_2A/C receptors enhance NMDAR function by activation of ERK via the β-arrestin pathway or via activation of Src kinase. At the presynaptic level, 5HT_2A increase NMDAR function via the PLC/PKC pathway.

Modulation of NMDARs by 5HT₇ receptors

Activation of 5HT₇ receptors seems to have a differential effect on NMDAR depending on the time of exposure to the agonist. A first report in isolated hippocampal neurons demonstrated that acute application of 5HT₇ agonist induces an increase in NMDAR currents and phosphorylation of GluN1 and GluN2B subunit via PKA activation (Vasefi et al., 2013b) (Figure 4). Instead, a longer exposure to the 5HT₇ agonist induced a PDGFR-dependent decrease in NMDAR surface which endows 5HT₇ receptor with a neuroprotective effect against NMDA-induced excitotoxicity (Vasefi et al., 2013a;Vasefi et al., 2012) (Figure 4). 5HT₇ receptor activation also affects NMDAR-dependent plasticity in the medial vestibular neurons where it blocks NMDAR-dependent LTD in a PKA-dependent manner (Li et al., 2017).

Muscarinic Acetylcholine Receptors

Muscarinic acetylcholine receptors include 5 subtypes of receptors (M1–5) with broad distribution in the CNS. M1, M3 and M5 receptors are G_q -coupled while M2 and M5 receptors are G_i/o-coupled (Thiele, 2013). Muscarinic receptors can be expressed both pre- and postsynaptically. Early evidence suggested that blockade of muscarinic receptors impaired memory performance (Berger et al., 1969;Drachman et al., 1974). In addition, blockade of muscarinic receptors during induction blocked LTP (Hirotsu et al., 1989), suggesting these receptors could facilitate LTP induction presumably by targeting NMDARs.

Bidirectional modulation of NMDARs by M1 receptors

In the hippocampus, both acetylcholine and muscarinic agonists transiently increased NMDAR-mediated transmission in a calcium- and IP3R-dependent manner (Harvey et al., 1993;Markram et al., 1990, 1992). This effect was mediated by M1 receptors, which colocalize with NMDARs at excitatory synapses in hippocampal pyramidal neurons (Marino et al., 1998). The mechanisms underlying M1-dependent potentiation of NMDAR are diverse. In CA1 pyramidal neurons, M1dependent potentiation requires postsynaptic calcium rise and activation of IP3Rs, but not PKC (Harvey, et al., 1993;Markram, et al., 1992). Other studies reported that M1 activation with acetylcholine was sufficient to trigger LTP of NMDAR-mediated synaptic transmission, and this potentiation relied on CaMKII, PKC and Src activation and was independent of SNAREdependent exocytosis (Fernandez de Sevilla et al., 2010) (Figure 5). As in the hippocampus, activation of M1 receptors by exogenous agonists and endogenous acetylcholine triggered a PKC-dependent increase of NMDAR-evoked depolarizations in striatal neurons (Calabresi et al., 1998). While most studies have reported M1-dependent potentiation of NMDAR-mediated transmission, there is also evidence for an M1-dependent form of NMDAR LTD at Sch-CA1 synapses that is induced by agonists or endogenous acetylcholine (Jo et al., 2010). This form of LTD requires calcium rise, ryanodine receptor activation and the calcium sensor hippocalcin, and is expressed via dynamin-dependent internalization of NMDARs (Figure 5). Thus, activation of M1 receptors can induce potentiation or depression of NMDAR-mediated synaptic transmission. A clear explanation has not emerged but these opposite effects could be due to a different developmental stage – i.e. 14–16 day-old rats (Fernandez de Sevilla, et al., 2010) vs 4–5 week-old rats (Jo, et al., 2010) –, or the duration of M1 activation, where brief activation (< 1s) leads to potentiation (Fernandez de Sevilla, et al., 2010), whereas longer M1 activation (~10 min) leads to depression (Jo, et al., 2010).

Figure 5.

Mechanisms of NMDAR modulation by muscarinic acetylcholine receptors.

Activation of M1 receptors leads to calcium release from the ER and recruitment of PKC, CaMKII and Src which in turn enhance NMDAR-mediated transmission. M1 receptors reduce membrane hyperpolarization by blocking SK-channel thus disinhibiting NMDAR function. M1 receptors can induce depression of NMDAR-mediated transmission via a pathway that involves hippocalcin-mediated dynamin-dependent internalization of NMDAR.

Indirect modulation of NMDARs by M1 receptor and SK channels

Activation of M1 receptors can indirectly affect NMDARs via inhibition of small conductance calcium-activated potassium (SK) channels (Giessel et al., 2010;Tigaret et al., 2018). Upon intracellular calcium rise, SK channels open and hyperpolarize the membrane. In dendritic spines of CA1 pyramidal neurons, activation of M1 receptors decreases the sensitivity of SK channels for calcium (Giessel, et al., 2010) (Figure 5), which boosts both NMDAR- and AMPAR-mediated depolarizations. This M1-mediated modulation of SK channels is likely implicated in the regulation of postsynaptic calcium dynamics during the induction of LTP at Sch-CA1 synapse (Faber et al., 2005;Ngo-Anh et al., 2005;Tigaret, et al., 2018).

Adrenergic receptors

The major source of norepinephrine (NE) in the brain is the locus coeruleus from which thousands of NE-producing neurons send extensive projections throughout the brain (Berridge et al., 2003). NE is released in the extracellular space and by volume transmission targets α(1–2) and β(1–3) adrenergic receptors (ARs). α1ARs and α2ARs couple to G_q and G_i/o proteins, respectively, while βARs are G_s-coupled (Berridge, et al., 2003). At central synapses ARs can be expressed pre- and postsynaptically where they modulate synaptic transmission and neuronal properties.

Modulation of NMDARs by βAR

Given its implication in behavior (Bouret et al., 2005) and in learning and memory (Tully et al., 2010), several groups have tested the role of NE in NMDAR-dependent LTP in the hippocampus. In the dentate gyrus, manipulations that depleted NE content, or blocked βARs impaired LTP (Bliss, et al., 1983;Stanton et al., 1985), while NE or a βARs agonist was sufficient to trigger long-lasting potentiation of EPSPs and population spikes both in the dentate gyrus (Lacaille et al., 1985;Neuman et al., 1983) and in the amygdala (Gean et al., 1992). Moreover, activation of βARs facilitates the induction of AMPAR LTP at Sch-CA1 in a NMDAR-dependent manner (Thomas et al., 1996), consistent with an effect of βARs activation on NMDAR.

PKA-dependent modulation of NMDARs by βARs

In autaptic CA1 neurons, NMDARs are continuously phosphorylated by PKA under basal conditions, and a brief burst of synaptic activity induced calcineurin-dependent dephosphorylation of NMDARs that reduced NMDAR currents (Raman et al., 1996) (Figure 6). Activation of βARs prevented calcineurin-induced desensitization of NMDARs by increasing the activity of PKA. The putative aminoacidic residue responsible for βARs-mediated phosphorylation of NMDAR by PKA was found nearly 20 years after. A serine-to-alanine mutation on Ser¹¹⁶⁶ of the GluN2B prevented βAR-mediated increase in NMDAR EPSCs, dendritic spine calcium rise and phosphorylation of GluN2B subunit by PKA (Murphy, et al., 2014). Remarkably, PKA phosphorylation can also modulate NMDAR calcium permeability without significantly changing NMDAR-mediated EPSCs (Chalifoux, et al., 2010;Higley, et al., 2010;Lur, et al., 2015;Skeberdis, et al., 2006). Since PKA has several targets on the NMDAR (Aman, et al., 2014), different neuromodulators might induce NMDAR phosphorylation by PKA at specific residues/subunits with diverse consequences for NMDAR function.

Figure 6.

Mechanisms of NMDAR modulation by adrenergic receptors.

α2ARs inhibit NMDAR function via negative-regulation of the AC/cAMP/PKA pathway. α2AR inhibitory effect on NMDAR function relies on the interaction with RGS4. βARs enhance NMDAR function by positively modulating the cAMP/PKA pathway. Activation of α1ARs triggers a PLC/IP3/calcium-dependent depression of NMDAR transmission. The inhibitory effect of α1ARs relies on the interaction with RGS2/4 and the scaffold protein spinophilin.

Modulation of NMDARs by αARs

In PFC slices and dissociated neurons, activation of α1ARs and α2ARs by agonists triggers a transient and reversible reduction in NMDAR-mediated transmission (Liu et al., 2006a). Activation of α1Rs triggers a PLC/IP3/calcium-dependent reduction of NMDAR EPSCs (Figure 6). α1ARs modulation of NMDAR required both RGS2/4 and spinophilin, a scaffold protein that binds GPCRs and RGS allowing their interaction(Liu, et al., 2006a). On the other hand, α2ARs activation induced microtubule destabilization by inhibiting the cAMP/PKA/ERK pathway. The effect of α2ARs also required the activation of RGS4 but not RGS2, suggesting that different RGS can shape the net effect of neuromodulators (Liu, et al., 2006a) (Figure 6).

Adenosine receptors

Adenosine is a purinergic neuromodulator widely released throughout the brain. Unlike classic neurotransmitters, adenosine is not stored in vesicles, but it is either a product of extracellular ATP hydrolysis or is released via specific membrane transporters (Chen et al., 2014). Adenosine binds to four metabotropic receptors: A1, A2A, A2B and A3. Adenosine has been shown to modulate NMDARs mainly through G_s–coupled A2A receptors (A2ARs) (Figure 7).

Figure 7.

Mechanisms of NMDAR modulation by A2ARs.

Activation of A2ARs can both inhibit or enhance NMDAR function. A2ARs-mediate inhibition of NMDAR relies on the activation of the PLC/IP3 pathway and on the activation of CaMKII. Positive modulation of NMDAR by A2ARs occurs through the activation of the AC/cAMP/PKA pathway.

Inhibition of NMDARs by A2A receptors

Early studies using acute striatal slices have shown that activation of A2ARs in MSNs reduced NMDA-evoked currents in a PLC/IP3/calcium/CaMKII-dependent manner (Norenberg et al., 1998;Norenberg et al., 1997;Wirkner et al., 2000) (Figure 7). Intriguingly, despite the G_s coupling, the inhibitory effect of A2ARs did not rely on the cAMP/PKA pathway, suggesting these receptors in MSNs might signal in a non-canonical manner. It should be highlighted that these studies did not target a specific subset of striatal MSN (i.e. D1-positive or D2-positive MSNs). A subset of MSNs did not display a decrease of NMDA-evoked currents in response to A2AR activation (Wirkner, et al., 2000) suggesting that A2ARs might have a cell-type specific effect (i.e. D1positive vs D2-positive MSNs).

Potentiation of NMDARs by A2ARs

More recently, it was found that activation of A2ARs in D2R-positive MSNs enhanced NMDAR-EPSCs and calcium signals in dendritic spines (Higley, et al., 2010). In the hippocampus, A2AR activation in CA1 pyramidal neurons enhanced NMDAR currents in a transient manner, while at the mf-CA3 synapse activation of A2ARs is necessary and sufficient to trigger LTP of NMDARmediated synaptic transmission (Rebola, et al., 2008). A possible explanation for these diverse effects is the colocalization and interaction of A2ARs with other GPCRs (Lohse, 2010;Prezeau et al., 2010). A2ARs colocalize and display structural/functional interactions with both mGluR5 (Ferre et al., 2002;Nishi et al., 2003;Tebano et al., 2005) and D2Rs (Fuxe et al., 2005;Hettinger et al., 2001;Torvinen et al., 2004).

NMDAR modulation by GPCRs: implications for circuit function and disease

Modulation of NMDARs by GPCRs in synaptic plasticity and circuit function

NMDARs are a critical trigger of most forms LTP and LTD of AMPAR-mediated transmission (Luscher et al., 2012). NMDAR-mediated calcium influx during the induction of plasticity is a key step in the induction of LTP and LTD. By changing the gain of NMDAR-mediated transmission, GPCRs can either decrease or increase the threshold for NMDAR-mediated forms of plasticity in a process known as metaplasticity (Abraham et al., 1996).

NMDARs also play a major role in temporal and spatial integration of excitatory inputs (Branco et al., 2010b;Hunt, et al., 2012). Activation of spatially clustered NMDARs allow supralinear summation of excitatory inputs and the generation of regenerative “all-or-none” potentials called NMDA-spikes/plateau potentials (Branco et al., 2010a;Schiller et al., 2001). NMDA spikes are important computational units that allow for the communication between remote dendritic inputs and the soma (Larkum et al., 2009) and for the encoding and decoding of incoming synaptic activity (Polsky et al., 2009). These functions of NMDA spikes may contribute to the encoding of sensory inputs (Manita et al., 2017). Because of their slow kinetics NMDA-spikes/plateau potentials might participate in the generation of persistent firing states (Wang, 1999). Persistent activity likely underlies short-term retention of stimulus-related information (Constantinidis et al., 2018;Zylberberg et al., 2017), a putative mechanism of working memory (Riley et al., 2015). In the PFC, a brain region involved in working memory (Riley, et al., 2015), D2R-dependent LTD of NMDAR-mediated transmission reduces the temporal integration of incoming hippocampal inputs with possible repercussions on pyramidal cell firing and PFC function (Banks, et al., 2015). This effect could be counterbalanced by D1Rs that induce NMDAR LTP and enhance temporal integration of synaptic inputs (Seamans et al., 2001). 5HT_1A and 5HT_2A receptors also bidirectionally modulate NMDARs (Barre, et al., 2016;Yuen, et al., 2008;Yuen, et al., 2005;Zhong, et al., 2008). 5HT_2A receptors in the PFC have recently been implicated in cognitive processes and associative learning (Barre, et al., 2016), but it is unclear whether this action is due to pre- or postsynaptic modulation of NMDAR. At the mf-CA3 synapse, mGluR5-dependent LTP and mGluR1-dependent MDAR LTD modulate the output firing of CA3 pyramidal cells induced by presynaptic bursts of activity (Hunt, et al., 2013). There is also evidence that NMDAR LTP at this synapse can trigger homo- and heterosynaptic metaplasticity (Hunt, et al., 2013;Rebola et al., 2011). Both, the increase of CA3 output and metaplasticity triggered by NMDAR LTP could have important implications in hippocampal computations and in the creation/dismantling of CA3 neuronal assemblies during memory storage (Hunt, et al., 2013), consistent with a role of CA3 NMDARs in memory retrieval (Nakazawa et al., 2002). Thus, by regulating NMDAR-transmission and plasticity, group I mGluRs, and presumably other GPCRs, could contribute significantly to memory processing in the hippocampus. While GPCR modulation of NMDARs can strongly affect important brain functions and computations, the diverse effects of GPCRs on neuronal function impose significant challenges for understanding the precise impact of GPCR modulation of NMDARs on behavior. Future work will be required to develop novel pharmacological and optical tools that modulate specific signaling pathways linking GPCRs to NMDARs with high spatiotemporal resolution.

Modulation of NMDARs by GPCRs in brain disease

Modulation of NMDAR by GPCRs has been proposed to contribute to neuropsychiatric diseases like autism and schizophrenia. NMDAR dysfunction has been strongly linked to autism spectrum disorder, highlighting NMDAR modulation as a potential therapeutic target for autistic patients (Lee et al., 2015). Modulation of NMDAR using allosteric modulators of mGluR5 seems to ameliorate hallmark symptoms of autism such as impaired social behavior and repetitive behavior (Silverman et al., 2010;Won et al., 2012). In Fmr1 KO mice, excessive mGluR5 surface mobility impairs NMDAR-mediated transmission and mGluR-dependent LTD and leads to cognitive dysfunction (Aloisi, et al., 2017). Both pharmacological blockade and genetic deletion of mGluR5 abolish NMDAR LTP (Harnett, et al., 2009;Hunt, et al., 2013;Jia, et al., 1998;Kotecha, et al., 2003a;Kwon, et al., 2008;Rebola, et al., 2008), suggesting that impaired modulation of NMDARs by mGluR5 could be involved in fragile X syndrome.

Modulation of NMDAR by DA receptors might play an important role in schizophrenia (Javitt, 2010). Both D2Rs and NMDARs have been strongly implicated in schizophrenia (Howes et al., 2015). Antipsychotic drugs that target D2Rs are still one of the most effective treatments for schizophrenia, and NMDAR antagonists such as ketamine and phencyclidine (PCP) can mimic both the positive and negative symptoms of schizophrenia (Javitt, 2010;Javitt et al., 1991). This evidence strongly supports the dopamine-glutamate crosstalk hypothesis of schizophrenia (Howes, et al., 2015). It also suggests that alterations of the DA receptor-mediated regulation of NMDARs could contribute to some symptoms of schizophrenia and that this interaction could be a therapeutic target. Targeting GPCRs with higher selectivity has been one of the main challenges in the development of new therapies for neuropsychiatric diseases. Allosteric modulators of GPCRs represent a novel approach that is gaining attention in the treatment of diseases such as schizophrenia and autism disorder (Conn et al., 2014;Foster et al., 2017). Allosteric modulators target GPCRs on a different binding site from the endogenous ligand (the orthosteric site) and can enhance (positive allosteric modulator, PAM) or inhibit (negative allosteric modulator, NAM) the effects of endogenous ligands on their target GPCRs. PAMs and NAMs can also have intrinsic activity independent of agonist binding to the orthosteric site (Conn, et al., 2014). In addition, GPCRs possess multiple conformational states that are linked to the activation of distinct signaling pathways (e.g. G-protein-dependent vs G-protein-independent). Allosteric modulators can stabilize GPCRs in a specific conformational state and thus bias the signaling pathway of a receptor in a specific manner. This approach might be employed in the future to treat NMDAR-linked disorders without directly targeting the NMDAR itself but by binding to GPCRs that modulate NMDAR properties.

Conclusions and future directions

By changing the number and functional properties of NMDARs, several GPCRs strongly modulate NMDAR-mediated processes, including synaptic transmission and plasticity, and by this means significantly modify brain function. Depending on the context –e.g. brain area, synapse type, developmental stage– and extent of GPCR activation, the same GPCR may engage different intracellular pathways and as a result have diverse effects on NMDARs. In addition, different GPCRs and their signaling cascades can converge in a synaptic compartment. Thus, coincident activation of multiple GPCRs may have synergistic or opposite effects on NMDARs. Because many GPCRs are coupled to the same G-protein type (i.e. G_s, G_i/o, G_q), the pathways involved in the modulation of NMDARs are highly redundant. However, GPCRs with common G-protein coupling can modulate different downstream targets (i.e. AMPAR vs NMDAR) and thus be more selective than previously thought. NMDARs and GPCRs have been implicated in neuropsychiatric diseases and might be key targets for the development of new therapeutic strategies. Biased allosteric modulators could be employed to target specific GPCR pathways that are selectively involved in the modulation of NMDARs, thus avoiding harmful off-target effects.

While the molecular mechanisms that link GPCRs to NMDAR have been extensively investigated, several questions remain unanswered. It is unclear whether release of neuromodulators with physiological patterns of activity in vivo would yield similar results to those observed with pharmacological application of exogenous agonists in vitro. Classic pharmacological tools do not allow manipulation of GPCRs with high spatiotemporal resolution. In addition, studying the effect of GPCR activation on NMDARs in subcellular compartments in vivo (i.e. dendrites, dendritic spines, presynaptic boutons) is challenging in part due to the poor resolution of most imaging techniques (e.g. two-photon calcium imaging). New tools could be combined in order to achieve the spatiotemporal resolution required to measure and manipulate GPCR activity. Optogenetic activation of specific neuromodulatory inputs can mimic the release of endogenous neuromodulators, and light-activated molecules such as caged compounds and photoswitchable ligands (Paoletti et al., 2019;Spangler et al., 2017) could be combined with the measurement of subthreshold and/or population activity with genetically encoded voltage-indicators (Abdelfattah et al., 2019;Kwon et al., 2017).

Several outstanding questions about the modulation of NMDAR by GPCRs remain unanswered, including the role of heterodimerization and/or crosstalk of GPCRs targeting NMDARs. Multiple GPCRs can be expressed in the same neuron and even in the same subcellular compartment (e.g. dendritic spines, presynaptic terminals) (Hunt, et al., 2013;Lur, et al., 2015;Tebano, et al., 2005). Although the functional relevance of GPCRs heterodimers is still matter of debate (Gurevich et al., 2018), heterodimerization and/or simultaneous activation of different GPCRs might lead to the activation of unexpected signaling cascades that normally would not be engaged by single protomers (Lohse, 2010;Prezeau, et al., 2010). It is currently unknown whether gliotransmitters released by astrocytes could influence NMDAR via GPCRs. Astrocytes can release both glutamate and ATP (Harada, et al., 2015), which can target mGluRs and adenosine receptors (upon ATP conversion to adenosine), respectively. However it is unclear whether modulation of NMDARs by astrocytes occurs through activation of GPCRs by gliotransmitters. Lastly, it would be important to know whether GPCRs could also modify the metabotropic function of NMDARs.

GPCRs are powerful modulators of NMDAR function. While several mechanisms involved in the GPCR-mediated modulation of NMDAR properties have been elucidated, we are still far from understanding the exact role of this modulation in brain function and disease. The development of new optical and pharmacological tools that allow controlling GPCRs with high spatiotemporal resolution will not only advance our understanding on the role of NMDARs in circuit function and behavior, but may also reveal novel targets to control NMDAR dysregulations implicated in brain disorders.

HIGHLIGHTS.

• GPCRs are strong modulators of NMDAR function throughout the brain

• Modulation of NMDARs by GPCRs affects neuronal integrative properties and can induce metaplasticity

• Multiple GPCRs can coexist at the same synapse and shape NMDAR function in an complex manner

• Novel pharmacological and optical tools that target GPCRs will help to achieve precise spatiotemporal control of NMDARs