Introduction

Schizophrenia is a devastating psychiatric disorder that afflicts approximately 1% of the worldwide population, affects women and men equally, and spans all socioeconomic groups [1]. The disease is characterized by three major symptom domains: positive, negative, and cognitive symptoms [2]. Current antipsychotics are effective at treating the positive symptoms such as auditory and visual hallucinations, delusions, and disorganized thoughts; however, they do not address the negative nor the cognitive symptoms. Negative symptoms (e.g., flattened affect, social withdrawal) and cognitive symptoms (e.g., deficits in working memory, and cognitive flexibility) are believed to be the best predictors of long-term outcome and are estimated to cost the U.S. healthcare system over $60 billion per year [3–5]. Additionally, most patients discontinue current treatments due to adverse effects including extrapyramidal side effects (EPS) (i.e., dystonia, akathisia, parkinsonism, bradykinesia, tremor, and tardive dyskinesia) induced by first-generation typical antipsychotics and metabolic side effects (i.e., weight gain, type II diabetes, and hyperlipidosis) induced by second generation atypical antipsychotics [6–8].

While most current antipsychotics act by antagonizing the hyperdopaminergic and hyperserotonergic states underlying the positive symptoms, there is a growing body of evidence that supports glutamate dysfunction as a contributing factor for the disease [9, 10]. For example, administration of the N-methyl-D-aspartate receptor (NMDAR) antagonist phencyclidine (PCP) [11] induces a schizophrenia-like state that presents clinically with all three symptom clusters in healthy individuals [12, 13]. Similar clinical results have been found with administration of other NMDAR antagonists such as ketamine [14]. NMDAR antagonists also exacerbate or precipitate controlled symptoms when administered to schizophrenia patients [15]. This along with extensive preclinical evidence suggests that NMDAR hypofunction is important in the pathophysiology underlying schizophrenia [10].

Based on the NMDAR hypofunction hypothesis of schizophrenia, pharmacological agents that enhance NMDAR function are not only valuable tools in preclinical animal models but could also provide therapeutic benefits to patients with schizophrenia. Unfortunately, direct activation of NMDARs using traditional orthosteric agonists induces adverse effects such as excitotoxicity and seizures [16–20]. Furthermore, treatments with NMDAR obligate co-agonists such as glycine or serine failed to have consistent efficacy across multiple clinical trials [21]. More recently, selective NMDAR positive allosteric modulators (PAMs) that enhance receptor function in the presence of the endogenous agonists but are devoid of intrinsic activity have been reported [20]. It is possible that NMDAR PAMs could avoid the adverse effects associated with direct activation of NMDARs. The recent development of NMDAR PAMs such as GNE-6901 and GNE-8324 provide proof-of-principle for the development of allosteric modulators of NMDARs, however their poor pharmacokinetic properties and low central nervous system exposures hinder their uses for in vivo studies [20]. Therefore, it will be important to develop more optimized compounds to fully assess the ability of NMDAR PAMs to reverse schizophrenia-like symptoms in animal models without the adverse effects profile of NMDAR agonists.

In addition to NMDARs and other ionotropic glutamate receptors (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) and kainate receptors) that mediate fast excitatory neurotransmission, glutamate also binds to and signals through a family of G-protein coupled metabotropic glutamate (mGlu) receptors [22]. There are eight subtypes of mGlu receptors, mGlu1–8, classified into three groups (group I, mGlu1,5; group II, mGlu2,3; group III, mGlu4,6,7,8). mGlu receptors are class C GPCRs which function primarily as dimers and modulate glutamatergic, GABAergic, and neuromodulatory neurotransmission throughout the central nervous system (CNS) [22]. All three groups of mGlu receptors have been pursued as putative targets for novel antipsychotics due to their ability to directly alter NMDAR function or other aspects of glutamatergic signaling.

The highly conserved orthosteric glutamate binding site among mGlu receptor subtypes has generally precluded the design of subtype-specific receptor agonists or antagonists. This problem has been addressed by designing small molecule modulators that bind to distinct and subtype-unique allosteric sites within the 7 transmembrane domains of mGlu receptors allowing unprecedented selectivity against other glutamate receptors [23]. Positive allosteric modulators (PAMs) generally do not activate the receptor directly but instead potentiate responses to endogenous glutamate. Conversely, negative allosteric modulators (NAMs) act as non-competitive antagonists and may also have inverse agonist activity, reducing constitutive activity of the receptor in the absence of glutamate [24, 25].

Group I mGlu receptors (mGlu1 & mGlu5)

Group I mGlu receptors include mGlu1 [26, 27] and mGlu5 [28], and are primarily coupled to the Gαq subunit of the heterotrimeric G-protein. Canonical Gαq signaling activates phospholipase C beta and causes downstream activation of protein kinase C (PKC) via the generation of the second messengers diacylglycerol (DAG) and inositol triphosphate (IP3). The group I mGlu receptors have emerged as attractive targets, initially based on their close coupling to the NMDAR via intracellular signaling pathways and scaffolding proteins including Homer, SHANK, and GKAP-PSD95 [29–31] and their ability to potentiate NMDA responses in acute brain slices [32]. mGlu1 and mGlu5 are predominantly postsynaptic receptors, but they have also been identified on presynaptic terminals of GABAergic and glutamatergic neurons (Fig. 1) [33]. Group I mGlu receptors are found primarily as homodimers via interactions at the large extracellular N-terminal Venus flytrap domain of each monomer [34]. Emerging evidence points to group I mGlu receptors existing in a monomeric form with distinct neurodevelopmental patterns which may impact their pharmacological profiles at specific ages in rodents [35]. While the potential role of dimeric versus monomeric expression may have interesting implications for schizophrenia, further studies are required to investigate this phenomenon.

Fig. 1
figure 1

Localization of metabotropic glutamate receptor subtypes. Schematic representation of the predominant locations of mGlu receptors at the synapse. mGlu1 (light blue) is found on postsynaptic glutamatergic neurons as well as on GABAergic neurons. mGlu5 (yellow) can be located on the same neurons as mGlu1 as well as on glia. mGlu2 (purple) is found primarily presynaptically as both a homodimer as well as a heterodimer with mGlu4 (red). mGlu3 (dark blue) is found on both presynaptic and postsynaptic glutamatergic, GABAergic, and neuromodulatory neurons as well on glia. mGlu4 is localized to both modulatory neurons as well as on presynaptic glutamatergic neurons as either a homodimer or heterodimer. mGlu7 (green) is localized to presynaptically neurons as well as GABAergic neurons. Lastly, mGlu8 (orange) is primarily localized presynaptically and mGlu6 is not shown since it is restricted to the retina

Full size image

mGlu1

Two recent independent studies have identified 12 rare deleterious nonsynonymous single nucleotide polymorphisms in the GRM1 gene encoding for mGlu1 in schizophrenia [36, 37]. Further support for mGlu1 dysregulation in schizophrenia is evidenced by postmortem findings in which mGlu1 mRNA expression is altered compared to controls [38]. Preclinically, Grm1 knockout mice display deficits in prepulse inhibition (PPI) [39], a behavioral assessment of sensory gating which is the process of filtering unnecessary stimuli from total sensory stimuli and which is impaired in schizophrenia patients [40]. Interestingly, recent studies reveal that GRM1 mutations associated with schizophrenia reduce mGlu1 signaling in cell lines and that selective mGlu1 PAMs can partially rescue the reduction in glutamate-mediated calcium signaling in vitro [41]. Therefore, enhancing mGlu1 signaling through selective agents has the potential to rescue deficits in schizophrenia patients with deleterious GRM1 mutations.

In addition to rescuing mGlu1 signaling deficits, activators or positive modulators of mGlu1 may also act to counteract the hyperdopaminergic signaling in the striatum in schizophrenia patients [42–45]. Multiple studies have demonstrated that the pan-mGlu receptor agonist trans-ACPD is able to attenuate stimulation-induced dopamine release in the dorsal striatum [46], the substantia nigra [47], and the nucleus accumbens [48]. In a follow up study, mGlu1 was identified as the subtype responsible for this effect in the dorsal striatum [49]. Therefore, mGlu1 activation may have the potential to produce similar antipsychotic effects as D2 dopamine receptor antagonist antipsychotics.

Potent first generation mGlu1 PAMs were developed in the early 2000s, but poor drug metabolism and pharmacokinetic (DMPK) profiles limited their use in preclinical studies [50, 51]. More recent efforts yielded VU6000799 and VU6000790 as potent, highly selective mGlu1 PAMs with improved DMPK properties and brain penetrance, and are therefore better suited for in vivo studies [52–55]. In the future, it will be important to evaluate these compounds in animal models that are relevant to the three symptom domains of schizophrenia.

Important to the potential utility of mGlu1 PAMs to treat schizophrenia, it has been reported that the mGlu1 NAMs FTIDC and CFMTI are efficacious in animal models predictive of antipsychotic activity (Table 1), including reducing psychostimulant and NMDAR antagonist-induced hyperlocomotion and deficits in PPI as well as reversing deficits in social interaction induced by the NMDAR antagonist MK-801 in rats [56–58]. The contrasting findings of mGlu1 PAMs, NAMs, and GRM1 knockout animals illustrate the potential complexity of mGlu1 ligands, and suggests that mGlu1 PAMs may only be effective in patients carrying GRM1 mutations. These studies further highlight the heterogeneity of schizophrenia and the critical role of patient selection strategies in psychiatric clinical trials to match genotype with the therapy.

Table 1 Summary of Preclinical Efficacy of Group I mGlu Receptor Ligands
Full size table

mGlu5

In recent years, mGlu5 has emerged as an attractive target for the treatment of schizophrenia [59]. Similar to mGlu1, mGlu5 is primarily postsynaptic but is also located presynaptically and can be expressed on GABAergic neurons and glia (Fig. 1). In the hippocampus, prefrontal cortex (PFC), and other brain regions, mGlu5 plays important roles in synaptic plasticity - the strengthening or weakening of synapses in response to specific activity patterns termed long term potentiation (LTP) and long term depression (LTD), respectively [60, 61]. Early pharmacological and genetic deletion studies in mice have shown that mGlu5 is important in the regulation of specific domains of cognitive function [60, 61] and in behaviors relevant for the positive and negative symptoms of schizophrenia [39, 62, 63]. Interestingly, unlike mGlu1, early studies did not provide evidence that mGlu5 activation reduces dopamine release in the striatum [64, 65] thus any antipsychotic effects of mGlu5 activators may be independent of dopamine modulation.

Over the last two decades, a growing body of evidence suggests that selective mGlu5 PAMs could provide an exciting new approach for the treatment of schizophrenia [66] (Table 1). The first highly selective mGlu5 PAMs DFB [67] and CPPHA [68] demonstrated the viability of developing selective compounds for mGlu5, but lacked properties that would allow their use in vivo. The first major in vivo breakthrough came with the development of CDPPB [69], the first mGlu5 PAM to possess favorable DMPK properties to allow its use in rodent models [69, 70]. Subsequently, it was shown that CDPPB reverses AHL and amphetamine-induced disruption of PPI in rats, providing strong preclinical support for mGlu5 as a potential therapeutic for schizophrenia [70]. In more recent years, there has been tremendous success in developing a large number of structurally distinct, highly selective mGlu5 PAMs that have efficacy in a wide range of animal models relevant to all three symptom domains of schizophrenia [71–76].

Since mGlu5 can potentiate NMDAR responses in select rodent brain regions [77–79], it was initially proposed that mGlu5 PAMs were likely to exert their efficacy through potentiation of mGlu5-induced increases in NMDAR currents in forebrain regions implicated in the pathology of schizophrenia [72, 79]. Unfortunately, some mGlu5 PAMs, such as 5PAM523 which has efficacy in reversing AHL, appear to induce severe adverse effects including seizures and neuronal death which could be related to excessive activation of NMDAR [76, 80]. Until recently, the hypothesis that potentiation of mGlu5 modulation of NMDAR currents was critical for the efficacy of these compounds had not been tested. To systematically test this, a novel biased mGlu5 PAM, VU0409551, was developed that potentiates mGlu5 coupling to Gαq-mediated calcium mobilization and other canonical signaling pathways but does not enhance mGlu5 changes in NMDAR currents (Fig. 2a) [81]. Of interest, VU0409551 produces robust antipsychotic-like effects in pharmacological challenge models of positive psychotic symptoms and cognition-enhancing effects in wild-type animals [81]. VU0409551 also has robust efficacy in reversing deficits in serine racemase knockout (SR−/−) mice, a genetic model of NMDAR hypofunction in which the enzyme that synthesizes the NMDAR co-agonist D-serine is genetically deleted [82]. SR−/− mice display deficits in synaptic plasticity and cognition [83], and recapitulate anhedonic-like symptoms, such as a blunted reward response to cocaine in an intracranial self-stimulation paradigm [84]. Interestingly, VU0409551 rescues signaling, plasticity, and cognitive deficits in this model [82], strengthening the hypothesis that biased mGlu5 PAMs that do not potentiate NMDAR currents still retain efficacy in rodent models relevant for schizophrenia. Furthermore, chronic administration of VU0409551 at doses over 100× those required to achieve in vivo efficacy resulted in no measureable cell death or induction of seizures [81]. In addition, separate studies revealed that eliminating allosteric agonist activity of mGlu5 PAMs is critical for reducing seizure liability [80]. Thus, by developing a detailed understanding of the pharmacodynamic actions of different mGlu5 PAMs, it may be possible to develop mGlu5 PAM clinical candidates that have robust efficacy but are devoid of excitotoxic adverse effects (Fig. 2a).

Fig. 2
figure 2

Emerging concepts in the development of mGlu receptor-targeting antipsychotic therapeutics. a mGlu5 PAMs have recently been developed that bias mGlu5 signaling away from NMDAR modulation but still mobilize intracellular Ca2+ and activate ERK1/2. The mechanism of this bias is still unclear but could involve G-protein dependent versus independent coupling of mGlu5 to NMDAR. The biased mGlu5 PAM VU0409551 retains antipsychotic-like efficacy in vivo but does not cause excitotoxicity or seizures observed with unbiased mGlu5 PAMs that enhance mGlu5-mediated modulation of NMDAR currents. This suggests that NMDAR modulation is not necessary for in vivo efficacy and that this signal bias may provide a means to overcome the NMDAR-mediated excitotoxicity that has stalled mGlu5 PAM development. b Recently, functional mGlu2/4 heterodimers with unique pharmacology have been identified. This suggests that actions at the mGlu2/4 heterodimer rather than at the mGlu4/4 homodimer might underlie the antipsychotic efficacy of mGlu4 PAMs, such as Lu AF21934, consistent with the antipsychotic-like effects of mGlu2-specific ligands. This remains to be tested experimentally but may provide an interesting alternative to failed mGlu2 clinical programs

Full size image

The mechanism by which VU0409551 exerts its antipsychotic-like and procognitive effects in animal models remains unclear. Experiments in wild-type rats suggest that the ability of VU0409551 to enhance certain forms of cognition is independent of NMDAR modulation [81]. It is possible that these effects of the PAM are due to the potentiation of mGlu5-mediated effects on neuronal excitability aside from NMDAR current modulation. In CA1 pyramidal cells mGlu5 activation suppresses the afterhyperpolarization current, thereby increasing the excitability of these neurons [79]. In these same neurons, mGlu5 is critical for a form of long-term plasticity at inhibitory synapses, termed inhibitory long-term depression (iLTD), and an mGlu5 PAM could increase hippocampal transmission via a reduction of inhibitory tone [85]. In layer V pyramidal neurons in the rodent medial PFC, mGlu5 activation increases neuronal excitability and spiking frequency [86, 87] as well as excitatory drive onto these neurons [88]. One hypothesis is that VU0409551 exerts its procognitive effects, especially the augmentation of PFC-dependent recognition memory, working memory, and executive function [81], via increased PFC pyramidal neuron activity but this remains untested.

In vitro assays indicate that VU0409551 exerts both PAM and robust agonist activity with respect to mGlu5-mediated extracellular-signal regulated kinase (ERK) activity. This is in agreement with the ability of VU0409551 to enhance LTD at the Schaffer Collateral-CA1 (SC-CA1) synapse induced by the group I mGlu receptor agonist DHPG, a form of plasticity that involves rapid protein synthesis and ERK activation [89–91]. Additionally, augmentation of early-phase LTP (E-LTP; < 3 h) by mGlu5 PAMs may require NMDAR current potentiation, explaining why VU0409551 does not augment E-LTP. While E-LTP is not dependent on ERK activity nor protein synthesis, late-phase LTP (>3 h) is ERK-dependent [92], can be enhanced by mGlu5 PAMs [93] and is closely linked to long term memory consolidation. Therefore, VU0409551 via its positive effects on ERK activation may exert its procognitive effects by potentiating late-phase LTP, although this remains to be tested experimentally.

Interestingly, VU0409551 is able to rescue deficits in hippocampal E-LTP in SR−/− mice without any augmentation in littermate controls [82]. This effect also correlates with the ability of VU0409551 to enhance NMDAR synaptic responses exclusively in the knockout mice. How VU0409551 exerts these effects in animals with marked NMDAR hypofunction [94] but not in wildtype animals remains to be determined. It is still unclear how prototypical mGlu5 PAMs enhance NMDAR function in wild-type animals [68, 81] as studies have implicated both G-protein-dependent [95–98] and G-protein-independent [99–102] pathways in the mGlu5-NMDAR interaction (Fig. 2a). Therefore, the actions of VU0409551 in SR−/− mice could involve a rearrangement of the postsynaptic density to prefer G-protein-independent mGlu5-mediated NMDAR current enhancement or differential spatial and/or temporal coupling of mGlu5 to G-protein-dependent downstream effectors that could augment NMDARs such as PKC and CaMKII. Future work is still needed to determine how mGlu5 PAMs enhance NMDAR function in wild-type animals and schizophrenia-like animal models.

Group II mGlu receptors (mGlu2 & mGlu3)

The group II mGlu receptors, mGlu2 and mGlu3 [103], are expressed presynaptically [22] outside of the active zone on pre-terminal regions of axons where they can be activated by astrocytic glutamate release or excessive synaptic glutamate [104] (Fig. 1). mGlu3 is also found postsynaptically as well as on astrocytes where it mediates neuroprotective effects [104] and participates in astrocytic-neuronal communication [105, 106] (Fig. 1). Group II mGlu receptors traditionally couple to the Gαi/o subunits of G-proteins, leading to inhibition of adenylyl cyclase and other signaling pathways [22].

Due primarily to their inhibition of neurotransmitter release from glutamatergic, GABAergic, and neuromodulatory (dopaminergic, noradrenergic, etc.) presynaptic terminals (Fig. 1), the group II mGlu receptors have attracted considerable attention as potential targets for novel antipsychotics. mGlu2/3 activation has been shown to reduce extracellular dopamine efflux in the nucleus accumbens [48, 64], the substantia nigra [47], and recently afferent-driven dopamine release in the dorsal striatum [107]. Additionally, elevated post-mortem glutamatergic markers are observed in the cortex of schizophrenia patients [108], commonly attributed to NMDAR dysfunction on GABAergic interneurons leading to an overall net disinhibition [10, 109, 110]. Furthermore, in both humans and animal models, NMDAR antagonist psychotomimetics increase glutamatergic transmission in the prefrontal cortex and increase CNS glutamate levels [10, 111–114]. mGlu2/3 activation can inhibit glutamate release, therefore providing rationale for the development of agonists or potentiators of mGlu2/3 to reduce the excessive glutamatergic tone in the brain of patients with schizophrenia [115].

Directly relevant to NMDAR hypofunction, postsynaptic mGlu2/3 activation can also enhance NMDAR currents in CA1 hippocampal pyramidal cells via Src kinase [116] as well as in PFC pyramidal cells via PKC [117] and SNARE-dependent mechanisms [118]. It is currently unclear whether this enhancement of NMDAR currents contributes to the antipsychotic efficacy of mGlu2/3 activators but it is interesting that compounds targeting these receptors could provide a two-pronged approach: normalizing both NMDAR function and excessive glutamate levels in schizophrenia.

Of further importance, a functional heteromeric complex between the 5-HT2A serotonin receptor (5-HT2AR) and mGlu2 has been postulated based on biochemical, behavioral, and pharmacological data [119–121]. 5-HT2AR agonists such as psychedelic hallucinogens typically induce a psychotomimetic state in humans [122] similar to observations with NMDAR antagonists. Activation of 5-HT2AR enhances thalamocortical neurotransmission in rodents [123, 124] and this effect is antagonized by activation of group II mGlu receptors [125]. Orthosteric agonists of mGlu2/3 functionally antagonize 5-HT2A receptor signaling [119], therefore activators of mGlu2 may possess antipsychotic properties similar to atypical antipsychotics that partially rely on 5-HT2AR antagonism for their efficacy. An issue this raises for clinical trials is that antagonism of 5-HT2A with atypical antipsychotics has been shown to downregulate expression of GRM2 (the gene encoding mGlu2) in rodents and decrease positive epigenetic markers of GRM2 expression in both mice and humans [126]. Based on these data, it may be essential to stratify future patient populations based on prior use of atypical serotonergic antipsychotics to ensure adequate target engagement and therapeutic efficacy of mGlu2 ligands.

mGlu2/3 agonists

Interest in targeting mGlu2/3 for the treatment of schizophrenia began with the initial findings that the selective group II mGlu receptor orthosteric agonists LY354740 and LY379268 can reverse the ability of NMDAR antagonists to induce hyperlocomotion, stereotypies, deficits in working memory, cortical glutamate efflux, and increased firing of PFC neurons in rats [114, 127, 128] (Table 2). One of the most intriguing results of these studies was that activation of group II mGlu receptors had no effect on the NMDAR antagonist-induced rise in extracellular dopamine in the CNS [127]. Despite this, group II mGlu receptor agonists still retained antipsychotic-like efficacy in a range of animal models, suggesting the possibility of treating psychosis without the extrapyramidal or other adverse effects associated with dopaminergic antipsychotics. On the other hand, when the mGlu2/3 agonist LY404039 was administered to rats, it caused an increase in extracellular dopamine, the dopamine metabolites 3,4-dihydroxyphenylacetic acid (DOPAC) and homovanillic (HVA), and the serotonin metabolite 5-hydroxyindoleacetic acid (5-HIAA) in the PFC [129, 130]. In contrast to the hyperdopaminergic state in the striatum thought to underlie the positive symptoms of schizophrenia, it is hypothesized that the negative symptoms of the disease are contributed to by a dearth of dopamine and serotonin release in cortical regions [131–134]. Therefore, the increased dopamine as well as dopamine and serotonin turnover observed in the rat PFC in response to LY404039 [129] may predict efficacy in treating some aspects of the negative symptoms of schizophrenia.

Table 2 Summary of Preclinical Efficacy of Group II mGlu Receptor Ligands
Full size table

Based on the extensive preclinical evidence in support of mGlu2/3 agonists as novel antipsychotics, Eli Lilly & Co. progressed LY2140023 monohydrate (pomaglumetad methionil; prodrug of the active mGlu2/3 agonist LY404039) into clinical trials and demonstrated safety and tolerability in humans [135]. In a 4-week multicenter proof-of-concept phase II clinical trial of 196 patients randomly assigned to receive LY2140023, olanzapine, or placebo, LY2140023 showed statistically significant improvements in positive and negative symptoms (assessed by the Positive And Negative Symptom Scale, PANSS) relative to placebo and was comparable to the currently approved atypical antipsychotic olanzapine [136]. Most excitingly, this study found that LY2140023 was well-tolerated and did not produce any EPS or elevated prolactin levels [136].

Following these promising initial results, a second 4-week phase II dose-ranging study found that neither LY2140023 nor olanzapine was more efficacious than placebo. Thus, the results were inconclusive due to an abnormally high placebo effect [137]. In a subsequent 24 week phase II study LY2140023 was found to significantly reduce PANSS scores over the 24-week period but from weeks 16 to 24 it was less effective than the current standard of care group (treatment with olanzapine, aripiprazole, or risperidone) [138]. Discouragingly, in a larger phase II trial of 1013 patients, LY2140023 failed to show improvements in PANSS total score compared to placebo, while the atypical antipsychotic risperidone significantly separated from placebo [139]. A separate phase 1b study found that LY2140023 also failed to demonstrate efficacy in alleviating negative symptoms when administered concurrently with atypical antipsychotics although this has yet to be analyzed post-hoc based on prior patient antipsychotic use [140]. In response to these undesirable larger-scale clinical trial results, Eli Lilly and Co. terminated the development of LY2140023.

mGlu2 PAMs

Although preclinical studies with group II mGlu agonists appeared promising, chronic administration of group II mGlu receptor agonists resulted in robust tolerance and loss of their ability to reverse amphetamine- and PCP-induced hyperlocomotion [141]. It is possible that this contributed to the lack of reliable clinical efficacy outlined above. Additionally, group II mGlu receptor agonists can impair working and spatial memory in rodent models [142, 143]. However, studies with mGlu2 and mGlu3 knockout mice suggest that the reversal of amphetamine- and PCP-induced hyperlocomotion by group II mGlu agonists was dependent on activation of mGlu2, not mGlu3 [144, 145], prompting the development of mGlu2 selective PAMs. By potentiating responses to endogenous glutamate, it is possible that mGlu2 PAMs could reverse excessive glutamatergic signaling only at synapses where this pathophysiology is present, potentially avoiding the tolerance and cognition-impairing effects seen with orthosteric agonists and providing an alternative path forward for therapeutics targeting these receptors.

Two prototypical mGlu2 PAMs, LY487379 [146, 147] and biphenyl-indanone A (BINA) [148, 149], showed efficacy in reversing amphetamine- and PCP-induced hyperlocomotion and disruptions in PPI (Table 2). Furthermore, BINA was able to attenuate the serotonin-induced increase in excitatory transmission in the PFC and reduce head twitch behavior induced by the 5-HT2A receptor agonist (−)-DOB [150]. Therefore, mGlu2 PAMs were efficacious in dopaminergic, glutamatergic, and serotonergic pharmacological models of the positive symptoms of schizophrenia. These studies provided foundational research which motivated multiple drug discovery programs to develop selective mGlu2 PAMs [151–155] that have efficacy in animal models of schizophrenia including TASP0443294 [156], JNJ-40411813/ADX71149 [157, 158], AZD8529 [159], and SAR218645 [160] (Table 2).

TASP0443294 dose-dependently attenuated methamphetamine-induced hyperlocomotion, MK-801-induced deficits in social memory, and ketamine-induced increases in cortical gamma power, as well as reducing the duration of REM sleep in rats [156]. JNJ-40411813/ADX71149 also dose-dependently inhibited PCP- and scopolamine-induced but intriguingly not amphetamine-induced hyperlocomotion. Furthermore, JNJ40411813/ADX71149 reduced brain glucose metabolism induced by the NMDAR antagonist memantine and head twitch response induced by the 5-HT2A agonist DOM [158]. Recently, SAR218645 was shown to reduce DOI-induced cortical glutamate release and head twitch behavior but had no effect in pharmacological or genetic dopaminergic and glutamatergic models of the positive symptoms of schizophrenia [160]. SAR218645 did improve MK-801-induced short-term episodic memory as well as working memory deficits in GluN1 knockdown mice, providing the first evidence of cognition-enhancing effects of mGlu2 PAMs in a genetic model of schizophrenia [160]. Based on these results, the authors suggested that mGlu2 PAMs with profiles like SAR218645 might be efficacious in treating the cognitive deficits in schizophrenia but not the positive symptoms [160].

To date, two mGlu2 PAMs have progressed to clinical trials: JNJ40411813/ADX71149 [161] and AZD8529 [159]. Phase I assessment in healthy volunteers indicated that JNJ40411813 was generally well tolerated in healthy men and women--with adverse events such as ataxia and somnolence emerging only at high doses [161]. However, secondary measures of cognition endpoints suggested that the mGlu2 PAM decreased accuracy in an attention task in healthy men. Although, JNJ40411813 did trend to reduce cognitive deficits in attention and episodic memory precipitated by smoking withdrawal in a subpopulation of healthy volunteers, this was not statistically significant compared to placebo. Promisingly in a proportion of volunteers, 500 mg JNJ40411813 reduced the increase in Brief Psychiatric Rating Scale (BPRS) total score and negative symptom score induced by a low dose of (S)-ketamine [161]. Based on its tolerability and promising initial results in the ketamine challenge, it will be interesting to see if Johnson & Johnson will progress the compound further.

Recently, the phase II trial results of AstraZeneca’s mGlu2 PAM AZD8529 were disclosed [159]. Despite being well tolerated with mild adverse events, AZD8529 did not show any improvements in PANSS total score or PANSS positive and negative subscale scores compared to placebo. While AZD8529 did not produce any extrapyramidal side effects or elevation of prolactin (an effect observed with the comparator risperidone) it failed to demonstrate efficacy in this study of 104 patients with schizophrenia [159]. Possible explanations for this lack of efficacy include lack of sufficient target engagement and the use of a less symptomatic patient population. However, CNS activity suggesting target engagement was subsequently validated using fMRI, and risperidone significantly improved PANSS scores compared to placebo, suggesting that this mGlu2 PAM may lack sufficient efficacy even at doses that provide CNS effects [159].

Together with the disappointing results of the group II agonist LY2140023 trials, there is a significant discrepancy between these preclinical data implicating glutamatergic dysfunction and mGlu2 agonist or PAM efficacy and these clinical data. This could be in part due to improper patient selection, as hyperactivity of cortical regions correlates with psychosis only early on in disease progression [162, 163]. Furthermore, since atypical antipsychotics may decrease mGlu2 levels via the 5HT2A/mGlu2 heteromer [126], lower receptor levels might contribute to the lack of efficacy in the patient populations used in either study. While an intriguing possibility, this remains to be tested.

mGlu3

While pharmacological manipulation of group II mGlu receptors was based on the normalization of aberrant glutamatergic signaling downstream of NMDAR hypofunction, single nucleotide polymorphisms (SNPs) in the GRM3 gene encoding mGlu3 have been associated with schizophrenia in multiple studies [164–167]. No studies to date have found statistically significant associations with GRM2 SNPs [168, 169]. The association between GRM3 and schizophrenia has been extensively reviewed in the past, with certain SNPs associated with deficits in working and episodic memory [166]. More recently, a large-scale genome-wide association study of almost 37,000 patients with schizophrenia identified the GRM3 locus, as well as 108 other loci, associated with schizophrenia [170], supporting the idea that mGlu3 may be a viable target along with mGlu2, despite the antipsychotic-like efficacy of mGlu2 specific potentiators in rodent models.

Supporting this, a recent study using the mGlu2 agonist/mGlu3 antagonist LY395756 [171] showed that mGlu2 agonism was sufficient to enhance NMDAR function but the combination of mGlu2 agonism and mGlu3 antagonism could not reverse MK801-induced deficits in working memory [172]. This is consistent with the finding that mGlu3 is required for a form of LTD in the mouse PFC and that a selective mGlu3 negative allosteric modulator impairs PFC-dependent cognition [173]. Based on these findings and the neuroprotective role of mGlu3 [174–177], agonism or enhancement of mGlu3 signaling may provide pro-cognitive benefits in addition to ameliorating some of the neuroinflammatory pathology seen in schizophrenia [178, 179]. Finally, it has recently been reported that mGlu3 activation can positively modulate mGlu5 signaling [180], providing a potential mechanism to enhance NMDAR function (via mGlu3-mGlu5-NMDAR interactions) and consequently provide both antipsychotic and pro-cognitive efficacy. Though this hypothesis remains to be tested, the biological role and preclinical pharmacology indicate that enhancement of mGlu3 might be a promising strategy for the treatment of schizophrenia, especially with potential for improving cognitive disturbances in patients with schizophrenia.

Group III mGlu receptors (mGlu4, mGlu7, & mGlu8)

The group III mGlu receptors are grouped based on high sequence homology and consist of mGlu4 [103], mGlu6 [181], mGlu7 [182], and mGlu8 [22, 183]. mGlu6 is expressed exclusively in the retina, whereas the other group III mGlu receptors are primarily expressed in the CNS [181]. Similar to group II, group III mGlu receptors canonically signal via the Gαi/o subunits of the heterotrimeric G-protein complex, leading to inhibition of adenylyl cyclase and cAMP production [22]. Activation of group III mGlu receptors can also regulate neurotransmitter release via activation/inhibition of different ion channels and GβY-dependent inhibition of vesicular fusion [184]. Akin to the group II mGlu receptors, the therapeutic promise of group III mGlu receptor activators or potentiators arises from their hypothesized ability to ameliorate the hyperglutamatergic state proposed to take place in schizophrenia. Also, group III mGlu receptor activation reduces dopamine release in the nucleus accumbens [64] but more work is needed to ascertain if this reduction of dopamine release would contribute to potential antipsychotic efficacy of group III mGlu receptor agonists and/or PAMs.

mGlu4

mGlu4 is expressed predominantly on presynaptic glutamatergic and GABAergic terminals [22] (Fig. 1). In multiple immunohistochemistry studies, mGlu4 has been shown to localize to the presynaptic active zone, where it is situated to function as an auto- and heteroreceptor upon the release of glutamate into the synaptic cleft [185, 186]. mGlu4 is highly expressed in the cerebellum, moderately expressed in the olfactory bulb and thalamus, and lowly expressed in the hippocampus and the striatum [187]. Likely due to high levels of mGlu4 in the cerebellum, mGlu4 KO mice have deficits in cerebellar synaptic plasticity and impaired ability to learn complicated motor tasks [188]. Mice lacking mGlu4 also display deficits in spatial reversal and long-term memory [189], indicating a role of mGlu4 in cognition and cognitive flexibility, both of which are impaired in schizophrenia.

Multiple studies suggest that activation of mGlu4 may have antipsychotic-like effects in rodent models (Table 3). The pan-group III agonist ACPT-I reduced PCP- and amphetamine-induced hyperlocomotion as well as DOI-induced head twitches [190], and these actions of ACPT-I are also observed with mGlu4-selective agonists, LSP1-2111 [191] (>30-fold selective for mGlu4 vs. mGlu8) and LSP4-2022 [192] (>100-fold selective for mGlu4 vs. mGlu7; >300-fold vs. mGlu8). In addition to efficacy in models of the positive symptoms of schizophrenia, both LSP1-2111 and LSP4-2022 have efficacy in models of negative symptoms and cognitive deficits [191, 192]. Furthermore, the mGlu4-selective PAMs Lu AF21934 [193], Lu AF32615 [194], and ADX88178 [195] displayed similar promise in models of all three symptom clusters of schizophrenia [196, 197], providing further support for potential therapeutic utility of selective mGlu4 activators.

Table 3 Summary of Preclinical Efficacy of Group III mGlu Receptor Ligands
Full size table

Interestingly, recent studies raise the possibility that some of the in vivo actions of mGlu4 agonists or PAMs could be mediated by actions on mGlu2/4 heterodimers (Fig. 2b). While mGlu receptors are thought to function primarily as homodimers [198], in recent years it has become apparent that functional mGlu heterodimers exist and can have unique profiles in terms of altered signaling and pharmacology [199–201]. Recent studies reveal that a heterodimer between mGlu2 and mGlu4 exists and displays unique pharmacology compared to mGlu2 or mGlu4 homodimers [201, 202]. Interestingly, mGlu2/4 heterodimers are activated by orthosteric agonists of either mGlu2/3 or mGlu4 [201]. Furthermore, Lu AF21934, an mGlu4 PAM that has efficacy in rodent models of antipsychotic-like effects, has robust efficacy as an mGlu2/4 heterodimer PAM (Fig. 2b). Thus, while studies have yet to directly test the hypothesis that mGlu2/4 heterodimers are involved in the antipsychotic-like effects of these compounds, it will be important to consider this possibility in future studies.

mGlu7

A polymorphism in the GRM7 gene encoding mGlu7 that reduced transcription in vitro was found to be positively associated with schizophrenia in a large Japanese cohort [203], indicating that hypofunction of mGlu7 may contribute to this disorder. However, at present, few studies have focused on a potential role of mGlu7 in the pathophysiology of schizophrenia. Interestingly, mGlu7 exhibits the widest expression of group III receptors [187, 204], with high expression in cortex, hippocampus, and other forebrain regions [205]. Studies with mGlu7 KO mice demonstrated a role of mGlu7 in hippocampal short-term plasticity [206], amygdala-dependent learning processes [206], short-term working memory [207, 208], and extinction learning [208, 209]. Also, activation of mGlu7 reduces glutamatergic neurotransmission at the SC-CA1 synapse in the hippocampus [210, 211] and acts as a heteroreceptor (Fig. 1) to modulate GABA release and the induction of LTP at SC-CA1 [212]. Thus, selective activators of mGlu7 have the potential to enhance some aspects of hippocampal-dependent cognitive function. In addition, evidence suggests that mGlu7 activation reduces thalamocortical neurotransmission [213], a circuit thought to be overactive in schizophrenia [214]. However, the mGlu7 allosteric agonist AMN082 [215] exacerbates MK-801-induced hyperlocomotion and DOI-induced head twitches [191] (Table 3). While this may be due to off-target effects of AMN082 or its metabolites in vivo [216], these pro-psychotic effects were absent in mGlu7 KO mice [191] suggesting that they are mediated by mGlu7. It remains to be seen if the same pro-psychotic effects are observed using selective PAMs and future studies are needed to fully evaluate the potential utility of mGlu7 agonists or PAMs in schizophrenia-related models.

mGlu8

mGlu8 is widely expressed throughout the brain, although at relatively low levels compared to other group III mGlu receptors [22]. Like mGlu4 and mGlu7, mGlu8 is expressed in the presynaptic active zone of mainly glutamatergic synapses [185, 217] (Fig. 1) where it functions to modulate neurotransmitter release. It has also been identified in the postsynaptic compartment in the retina, medulla, and periphery [205]. mGlu8 has been shown to function as an autoreceptor at the lateral perforant path synapse in the dentate gyrus [218], thus gating glutamatergic transmission into the hippocampus. Consistent with this, mGlu8 KO mice display deficits in hippocampal-dependent learning [219]. Additionally, mGlu8 suppresses glutamatergic input into the bed nucleus of the stria terminalis (BNST) implicating a role for this receptor in anxiety and stress [220], consistent with results observed in the mGlu8 KO mice [221]. Similar to both mGlu4 and mGlu7, the neuromodulatory role of mGlu8 in brain regions implicated in learning and memory suggests that mGlu8 ligands could be beneficial in treating the cognitive deficits in patients with schizophrenia.

In studies investigating the potential antipsychotic efficacy of targeting mGlu8, researchers from GlaxoSmithKline found that the relatively selective orthosteric mGlu8 agonist (S)-3,4-DCPG [222] was unable to reverse PCP-induced or amphetamine-induced hyperactivity in Sprague-Dawley rats [221] (Table 3). Furthermore, mGlu8 KO mice had no significant deficits in PPI and thus it was concluded that mGlu8 does not appear to be involved in the etiology of schizophrenia nor does it appear to be a potential target for a novel antipsychotic [221]. This may be true with respect to positive symptoms but, based on the role mGlu8 plays in hippocampal neurotransmission [218, 221] it is still possible that agonists or potentiators of mGlu8 can have cognitive-enhancing properties. While exciting, this remains to be tested.

Conclusion

Extensive preclinical evidence has implicated all three groups of mGlu receptors as viable targets for the development of novel therapeutics for the treatment of schizophrenia. Agonists and subtype-selective PAMs for these receptors have efficacy in dopaminergic, serotonergic, and glutamatergic models of the positive and negative symptoms of the disease, and also demonstrate nootropic, or cognition-enhancing effects (Tables 1, 2, and 3). Based on preclinical findings, mGlu receptor modulators have the potential to be major improvements over currently approved dopaminergic and serotonergic antipsychotics. Negative results in clinical trials for the group II agonist pomaglumetad (LY2140023) and the mGlu2 PAM AZD8529 are disappointing but could be explained by patient selection issues. Potential future trials with patients selected based on stage of the disease and prior antipsychotic usage may yield different results in light of our advanced knowledge of the pathophysiology of schizophrenia. These negative clinical results suggest that there are still important gaps in our knowledge of how to translate preclinical results into clinical efficacy in schizophrenia. Taken together, these studies point to potential challenges in selecting the most appropriate patient populations for evaluating different mechanisms for improving different symptoms observed in schizophrenia patients.

In addition to potential antipsychotic efficacy, selective mGlu receptor ligands could provide cognition-enhancing effects targeting a major unmet need of this disorder. To this end, the key role of mGlu5 in hippocampal and cortical plasticity suggests that mGlu5 PAMs might have their greatest effects on cognition. Furthermore, the interaction between mGlu3 and mGlu5 and the involvement of mGlu3 in cortical plasticity suggests that potentiators of mGlu3 may also exert cognition-enhancing effects. Lastly, the notion that mGlu receptor ligands would provide a powerful cognition-enhancing approach to mitigate the deficits observed in schizophrenia is further evidenced by the mGlu2 PAM SAR218645 improving learning and memory in rodent models of schizophrenia as well as the preclinical cognition-enhancement observed with mGlu4 agonists and PAMs.

Of particular importance to future development efforts, both the discovery of biased PAMs (Fig. 2a) and the existence of mGlu heterodimers with distinct pharmacology (Fig. 2b) could provide novel approaches to optimize efficacy while avoiding toxic or other adverse effects. Although relatively unexplored, these emerging concepts in the pharmacology and biology of mGlu receptors provide a new path forward in the face of negative clinical results.

In conclusion, the metabotropic glutamate receptors represent a large group of promising targets for novel therapeutics to treat all three symptom domains of schizophrenia. While many discovery efforts are still in preclinical phases of development, they have yielded several subtype-selective tool compounds with minimal adverse effect profiles and promising preclinical efficacy. These compounds provide the unprecedented opportunity to further our fundamental understanding of the therapeutic role of mGlu receptor modulation in schizophrenia and represent a potential breakthrough in treating patients suffering from this disorder.