Positive Allosteric Modulators of Metabotropic Glutamate 2 Receptors in Schizophrenia Treatment

Abstract

The last two decades have witnessed a rise in the “NMDA receptor hypofunction” hypothesis for schizophrenia, a devastating disorder that affects around 1% of the population worldwide. A variety of presynaptic, postsynaptic and regulatory proteins involved in glutamatergic signaling have thus been proposed as potential therapeutic targets. This Review focuses on positive allosteric modulation of metabotropic glutamate 2 receptors (mGlu₂Rs) and discusses how recent preclinical epigenetic data may provide a molecular explanation for the discrepant results of clinical studies, further stimulating the field to exploit the promise of mGlu2R as a target for schizophrenia treatment.

Schizophrenia: limitations with currently available drugs

Schizophrenia is a chronic debilitating mental disorder that affects approximately 1% of the general population. Symptoms vary from patient to patient but are generally categorized into positive (e.g., hallucinations, delusions, and disorganized speech and behavior), negative (e.g., social withdrawal, lack of motivation, flat affect), and cognitive (e.g., impairments in memory, attention, and executive function). These symptoms are typically associated with social and/or occupational dysfunction. Although the natural course is heterogeneous, the illness typically strikes in late adolescence or early adulthood and usually continues throughout life [1, 2]. The prognosis of patients is also variable, but is often poor with high rates of unemployment, homelessness, violence, and suicide [3, 4]. Given the combination of onset in early adulthood and the chronic course, schizophrenia presents a staggering drag on the economy and society. It has been estimated that the cost of schizophrenia in the U.S. in 2002 exceeded $62 billion [5] and the World Health Oganization ranks this disorder among the top 10 causes of disability in developed countries [6].

The classical dopamine (DA) hypothesis (see glossary) has dominated the theories of schizophrenia since the mid-20^th century after the observation that first generation, or “typical”, antipsychotic drugs, such as chlorpromazine and haloperidol, are high-affinity antagonists of dopamine D₂-receptors. Hyperactivity in the mesolimbic DA pathway was originally proposed to underlie positive symptoms of schizophrenia [7]. While effective against positive symptoms, typical antipsychotic drugs demonstrate limited efficacy against negative symptoms and cognitive impairments which have been shown to contribute to functional impairment and predict poor prognosis [8]. Moreover, these drugs are associated with a number of side effects including hyperprolactinemia, and extrapyramidal symptoms (EPS). The second generation, or “atypical”, antipsychotic drugs were introduced into the clinical practice in the early 1990s in an attempt to improve clinical efficacy and decrease side effects. Second generation drugs (such as clozapine, olanzapine, risperidone and quetiapine), which have less potential to induce EPS or hyperprolactinemia, differ pharmacologically from the typical ones in that they have less affinity for D₂-receptors and higher affinity/functional interaction with other monoaminergic receptors, including the serotonin 2A receptor (5-HT_2A) [2, 9]. Because of the dual dopamine-serotonin mechanism of action of atypical antipsychotic drugs, a serotonin hypothesis for schizophrenia was also proposed [10]. Although highly effective against a wider range of symptoms than typical agents, atypical antipsychotics still have modest efficacy against negative symptoms. In addition, recent clinical evidence does not support the notion that second generation drugs are superior to first generation in improving neurocognition [11]. Last but not least, atypical agents are associated with side effects, usually metabolic, and in rare cases severe, such as agranulocytosis seen with clozapine [2]. Despite the availability of many typical and atypical antipsychotic drugs, full functional remission is achieved in fewer than 35% of people with schizophrenia [12]. Moreover, most of the responders are “partial responders” with whom negative and cognitive symptoms remain problematic, suggesting that impairment and/or dysregulation in DA and serotonin signaling cannot fully account for the underlying pathology. The limitations of the presently available drugs underscore the need for identification of new antipsychotic compounds aiming new molecular targets. Here we focus on positive allosteric modulation of mGlu₂R as a promising therapeutic strategy to treat schizophrenia.

Glutamate and schizophrenia

To account for negative and cognitive symptoms of schizophrenia, the DA hypothesis was expanded to postulate hypoactivity in the mesocortical pathway as well. Despite this expansion, the DA hypothesis still likely oversimplifies the neurocircuitry of schizophrenia and does not explain why the mesocortical DA pathway will be hypoactive while the mesolimbic DA pathway is hyperactive. If anything, this dichotomy suggests that either DA is partially involved in the molecular underpinnings of schizophrenia, or alternatively that multiple defective neurotransmitter systems eventually converge on and disrupt the DA system [7]. In an attempt to account for the shortcomings of the DA hypothesis, the “NMDA receptor hypofunction” hypothesis was proposed by Olney and Farber in 1995 [13] based on the observation that non-competitive NMDA antagonists such as phencyclidine (PCP), and ketamine induce a psychotomimetic state that closely resembles schizophrenia in healthy human individuals [14, 15] and exacerbate preexisting symptoms in schizophrenic patients [16, 17]. Although the pharmacology of cocaine and other amphetamine-like psychostimulants is complex [18, 19], they share the ability to bind dopamine transporters and increase synaptic levels of dopamine. Importantly, unlike amphetamine-like psychostimulants, glutamatergic PCP and ketamine dissociative drugs not only induce positive symptoms but also negative symptoms and cognitive dysfunction that better recapitulate the clinical syndrome of schizophrenia. Subsequently, PCP-like drug models have become widely employed in the search for novel treatments. It was proposed that defective NMDA receptors on cortical gamma aminobutyric acid (GABA) interneurons (Figure 1) render these interneurons less effective in inhibiting glutamate projection neurons that project to the ventral tegmental area (VTA). This disinhibition results in an excessive glutamate tone, which now over-stimulates the DA mesolimbic pathway giving rise to the positive symptoms. An alternative pathway that could be fundamental for cognition might be the functional crosstalk between NMDA and 5-HT_2A receptors in pyramidal neurons of prefrontal cortex [20]. Similarly, negative and cognitive symptoms may arise from NMDA receptor hypofunction. Here hyperactive glutamate projection neurons over-activate GABAergic interneurons, located in the VTA, causing them to release excess GABA and overinhibit the mesocortical DA pathway that now becomes unable to adequately supply DA to the prefrontal cortex [7]. Thus, defective glutamate neurocircuitry might actually derive the excess DA in the mesolimbic pathway as well as the deficiency of DA in the mesocortical pathway. Additionally, the resulting glutamate excitotoxicity might drive an ongoing neurodegenerative process responsible for the structural brain changes seen in schizophrenic brains on volumetric MRI [21, 22], although the presence of neurodegeneration in schizophrenia remains controversial [23–27]. Interestingly, group II metabotropic glutamate receptors (mGluRs) are located presynaptically on these hyperactive glutamatergic neurons where they function as autoreceptors to help keep glutamate tone in check [7]. Activation of these receptors can thus attenuate schizophrenia symptoms and might prevent potential neurodegeneration. Recent findings suggest that the mGlu₂Rs are also expressed postsynaptically, where they play a key role in the control of the responses induced by atypical antipsychotics [28–31]

Figure 1. Glutamate neurocircuitry implicated in Schizophrenia.

Normally, cortical GABA interneurons (in purple) exert an inhibitory tone on glutamate neurons that project to the VTA. When optimal, this inhibitory tone controls the amount of glutamate released in the VTA and subsequently the activity of the mesolimbic (1) and mesocortical (2) DA pathways. In schizophrenia, a defective NMDA receptor on the cortical GABA interneurons results in disinhibition of cortical brainstem glutamate projections. Excessive glutamate firing leads to overactivation of the mesolimbic DA pathway (1) and excessive release of DA in the nucleus accumbens. This might be responsible for development of positive symptoms in schizophrenic patients. Similarly, negative and cognitive symptoms may arise from NMDA receptor hypofunction. Hyperactive glutamate tone overactivates GABAergic interneurons in the VTA, overinhibiting the mesocortical DA pathway (2) that now becomes unable to adequately supply DA to the prefrontal cortex resulting in “hypofrontality”. Abbreviations: DLPFC, dorsolateral prefrontal cortex; NAc, nucleus accumbens, VMPFC: ventromedial prefrontal cortex; VTA, ventral tegmental area.

Additional evidence for involvement of glutamate in the pathophysiology of schizophrenia comes from postmortem studies, which revealed altered ionotropic glutamate receptor binding and gene expression mainly in the prefrontal cortex and the hippocampus [32–35]. Furthermore, functional neuroimaging studies have shown reduced NMDA binding in the hippocampus of medication-free schizophrenic patients compared to healthy controls [36]. More recently, proton magnetic resonance spectroscopy has revealed increased cortical glutamate levels in individuals with poor response to antipsychotic treatment in first-episode schizophrenia compared with those who were responders, supporting the hypothesis that poor responders to antidopaminergic therapy may have a glutamatergic basis for their psychosis [37]. Lastly, many genes involved in glutamatergic neurotransmission were found to be associated with schizophrenia in a recent genome wide association study (GWAS) involving a consortium of over 200 institutions worldwide in the largest molecular genetic study conducted to date for a neuropsychiatric disorder [38]. Taken together, the above evidence indicates an important role for dysfunctional glutamatergic neurotransmission in the pathophysiology of schizophrenia.

mGluRs and allosteric modulation

mGluRs are class C G protein-coupled receptors (GPCRs) that are characterized by a conserved heptahelical transmembrane domain (TMD), a large N-terminal extracellular domain (ECD), and a C-terminal intracellular domain. The large ECD, characteristic of family C GPCRs, consists of a Venus flytrap domain (VFD), which contains the orthosteric binding site for glutamate, and a cysteine-rich domain (CRD). Another distinguishing feature of class C GPCRs is constitutive homo- or heterodimerization at the cell-surface [39]. It has been demonstrated that class C mGluRs function as homodimers at the plasma membrane in living cells, whereas the class C GABA_B receptor needs to form a heterodimeric complex composed of GABA_B-R1 and GABA_b-R2 to reach the plasma membrane as a functional receptor complex (for review, see [40]). The eight mGluR subtypes identified so far, are classified into three groups based on sequence homology, G protein–coupling, and pharmacology [41]. Group I mGluRs (mGlu₁R and mGlu₅R) are predominantly coupled to Gq/11 and activate the phospholipase C enzyme. Group II mGluRs (mGlu₂R and mGlu₃R) and group III (mGlu₄R, mGlu₆R, mGlu₇R and mGlu₈R) are coupled to G_i/o proteins and typically inhibit adenylyl cyclase activity [42]. Like the majority GPCRs, mGluRs have been classically targeted through their “orthosteric” site (i.e., the binding site recognized by the endogenous ligand). However, since all mGluR orthosteric ligands bind to the VFD, which is highly conserved among all eight mGluRs, subtype selectivity has been difficult to achieve especially for mGluR members from the same group. Moreover, CNS penetration for many of those orthosteric compounds has been limited by their poor physicochemical properties owing to their glutamate-like structures [43]. The difficulty to identify suitable glutamate analogs with adequate subtype selectivity and pharmacokinetic properties triggered the search for an alternative means to target mGluRs. A promising approach has been the use of allosteric modulators; ligands that bind to sites non-overlapping and spatially distinct from, but conformationally linked to, the orthosteric site (Figure 2) [44]. Mutagenesis studies have revealed that the majority of allosteric modulators identified for mGluRs bind to the TMD, which is relatively less conserved among mGluR subtypes than the VFD [45]. This has been recently confirmed by crystal structures for TMDs of both mGlu₁R and mGlu₅R bound by allosteric modulators [46, 47].

Figure 2. Allosteric vs Orthosteric.

(A) A schematic of the dimeric structure of a class C GPCR. mGluRs, as well as other receptors in this class, are homodimers that use the VFD exclusively to bind orthosteric ligands. The TMD may have one or more potential pockets for allosteric ligands. (B) A synthetic orthosteric agonist often produces a bigger effect than the endogenous agonist however its effects may decline by time due to receptor desensitization and/or downregulation. On the other hand, PAM enhances the action of the endogenous agonist on its receptor in a more physiologic temporal pattern and thus is less likely to cause receptor desensitization and/or downregulation.

Allosteric modulators can modify the action of the orthosteric ligand by modulating its affinity and/or efficacy (Figure 3), either in a positive (in case of positive allosteric modulators or PAMs) or a negative direction (in case of negative allosteric modulators or NAMs). This phenomenon is referred to as “cooperativity” [48]. PAMs potentiate the response to an agonist and cause a leftward (and often an upward) shift in the concentration-response curve for the orthosteric agonist. Depending on the degree of stimulus-response coupling in an assay, a potentiator can behave as a pure PAM; an allosteric ligand that elicits no detectable response in the absence of the orthosteric agonist, or an ago-PAM that can directly elicit a response like an agonist [49]. NAMs, noncompetitively antagonize agonists and cause a rightward (and often downward) shift in the agonist concentration-response curves (Figure 3). Neutral allosteric ligands (previously referred to as silent allosteric modulators or SAMs) occupy the allosteric site without affecting the agonist responses on their own, however they can block the allosteric effects of both PAMs and NAMs [44].

Figure 3. Modes of action of allosteric modulators.

(A) The allosteric ligand binds to a site topographically distinct from the orthosteric binding site and modulates affinity (red) and/or efficacy (green) of the orthosteric ligand. Some allosteric ligands are capable of directly eliciting a response on their own (blue). Figure adapted from Conn et al. [35]. (B) Simple schematic representation of effects mediated by different allosteric ligands on the functional response of an orthosteric agonist. The first (solid red) is a PAM that purely enhances orthosteric agonist affinity as evident by an increase in potency (lower EC50) and leftward shift of concentration-response curve compared to the orthosteric agonist alone (solid black). The second (solid green) is a PAM that purely enhances orthosteric agonist efficacy as evident by an increase in Emax and upward shift in its concentration-response curve. The third (blue) demonstrates allosteric agonism that modulates both the affinity and efficacy. The fourth (dashed red) is a NAM that lowers the orthosteric agonist affinity and thus shifts its concentration-response curve to the right. The fifth (dashed green) is a NAM that lowers the orthosteric agonist efficacy and shifts its concentration-response curve downward. Figure adapted from Wootten et al. [101].

Advantages of allosteric modulators

Pure allosteric modulators have the advantage over synthetic orthosteric agonists in their ability to preserve the temporal and spatial patterns of receptor signaling since their effects are dependent on the presence of the endogenous ligand. This “state-dependent” mechanism of action explains, for example, why the ability of mGlu₂R PAMs to inhibit striatal excitatory postsynaptic potentials (EPSPs) shows dependence on frequency of presynaptic stimulation of corticostriatal afferents. Unlike agonists, mGlu₂R PAMs inhibit EPSPs only at high frequency stimulation indicating that excessive synaptic glutamate release is required for mGlu₂R PAMS to exert their modulatory effect [50]. Because PAMs will not continuously activate the receptor, receptor desensitization and/or downregulation is less likely to occur, as opposed to synthetic orthosteric agonists. In addition, since no effect is expected at saturating concentrations above that determined by cooperativity, an allosteric drug will have a greater potential to fine-tune physiological responses (Figure 2) with less risk of toxicity [51]. This saturability phenomenon is referred to as the “ceiling level” of the allosteric effect. Lastly, mGluR allosteric modulators generally possess better physicochemical properties and blood brain barrier (BBB) permeability compared to orthosteric drugs, an important feature considering the therapeutic potential of mGluRs in neuropsychiatric disorders [52]

Challenges with allosteric modulators

Despite the above-mentioned advantages, a number of characteristics unique to GPCR allosteric modulators may represent a challenge in the translation of data from such agents. For example, the magnitude and direction of the allosteric effect mediated by the same modulator acting on the same receptor can vary depending on the orthosteric ligand that is used to probe receptor function, a phenomenon referred to as “probe dependence” [51]. Although examples have not been discovered yet for mGluRs [52], the use of glutamate as the orthosteric probe in in vitro assays might be more favorable since results are more likely to be of physiological relevance. Accordingly, any results obtained from assays using non-native agonists as probes should be interpreted cautiously since an allosteric drug may behave differently in vivo in the presence of the endogenous agonist [53]. Moreover, potential species differences in the responses to allosteric drugs may exist since the allosteric sites, presumably having undergone less evolutionary pressure to accommodate an endogenous ligand compared to orthosteric sites, are more likely to show sequence divergence between species [48]. Another challenge is that a number of mGluR allosteric modulator classes are prone to “molecular switches”, by which subtle structural changes within the scaffold can dramatically change the pharmacology of compounds within a class, switching them for example from NAMs to PAMs, PAMs to NAMs, or even change their receptor-subtype selectivity [54–56]. This raises questions over metabolism and pharmacology of metabolites as well [57]. All the above-mentioned complexities, together with the potential implications of mGluR heterodimer formation [58], are issues that need to be addressed before advancing any mGluR allosteric modulator into the clinic.

Potential for biased signaling

Although GPCR activation was first described by a classical two-state model where the receptor exists in an equilibrium between an active and an inactive state, recent evidence supports an alternative multi-state model where GPCRs can adopt multiple conformational states with each state possibly activating a discrete subset of cellular behaviors of a broader spectrum than just G protein coupling or second messenger activation [48, 59]. The ability of ligands to stabilize some unique conformational states activating certain cellular pathways and not others has been termed “biased signaling”, functional selectivity”, and “stimulus trafficking” [48, 60]. Biased allosteric modulation has been demonstrated for a number of mGluR allosteric ligands [61–63]. For instance, the gadolinium ion (Gd³⁺), an allosteric modulator of mGlu_1αR, potentiates Gs-linked cAMP production yet inhibits Gq/11-linked Ca²⁺ mobilization when administered with glutamate [64]. Biased signaling can also involve non-G protein-mediated pathways such as the recruitment of β-arrestin. For instance, TRV130, a G protein-biased agonist of the µ-opioid receptor with minimal β -arrestin recruitment, was shown in mice to produce analgesic effects comparable to morphine with less respiratory depression and gastrointestinal dysfunction [65]. This strongly suggests that biased allosteric agonism and modulation may help selectively target signaling pathways critical for therapeutic efficacy while simultaneously excluding others associated with adverse effects (Figure 4).

Figure 4. Schematic for biased allosteric modulation.

(A) An orthosteric agonist turns the receptor “on” activating signaling pathway 1, associated with therapeutic benefits, and simultaneously pathway 2, associated with adverse effects. (B) Co-binding of an allosteric modulator can bias the signaling of the agonist-bound receptor, selectively engaging pathway 1 while inhibiting pathway 2.

mGlu₂R and schizophrenia

Group II mGluRs are widely expressed in the brain with generally similar distribution patterns in human and rodent brains [66]. mGlu₂R is particularly expressed in regions known to be implicated in schizophrenia such as the prefrontal cortex, hippocampus, striatum, thalamus, and amygdala [67]. Compared to mGlu₂R, mGlu₃R shows a considerably more diffuse distribution pattern in the brain with both showing an overlapping cortical distribution with the 5-HT_2A receptor [54]. At the cellular level, although mGlu₂R is also found postsynaptically [30, 68], both receptors are located presynaptically, where they function as autoreceptors inhibiting glutamate release and modulating synaptic transmission [69]. In contrast to mGlu₂R, mGlu₃R is additionally expressed by astrocytes [70] where it might also be involved in a feedback mechanism to modulate neuronal excitability.

While the mGlu₃R gene (GRM3) has been suggested by several independent groups [71–73] as well as by the recent GWAS data [38] to be implicated in schizophrenia, the majority of findings from postmortem human brains have not shown significant changes in mGlu₃R expression in schizophrenia [74–77]. However, when testing either mGlu₂ receptor density or mGlu₂ mRNA expression in postmortem human brain samples, several reports have suggested both down-regulation [74, 78, 79] and up-regulation [75, 80] of mGlu₂R (GRM2) expression in either antipsychotic-free or antipsychotic-treated schizophrenia patients as revealed by absence or presence of antipsychotics in postmortem toxicological analysis (see also [66, 77, 81, 82] for studies suggesting absence of changes in mGlu₂R expression in postmortem human brains of schizophrenic subjects). A potential explanation for these discrepancies is that the expression and function of mGlu₂R in schizophrenic subjects might be related to both the profound effects of age on mGlu₂R density in cortical regions [74, 77] and of antipsychotic drug treatment on the promoter activity of the GRM2 gene (for further discussion, see below).

Group II mGluR agonists have been shown to reverse the effects of PCP on locomotion [83, 84], working memory, stereotypy, and cortical glutamate efflux [83] and to suppress the head-twitch response induced by 2,5-Dimethoxy-4-iodoamphetamine (DOI), a hallucinogenic 5-HT_2A agonist, in rats [68]. Due to the lack of subtype-selective orthosteric ligands, it has been difficult to determine whether the antipsychotic-like effects of mGlu_2/3R agonists are mediated via mGlu₂R, mGlu₃R, or both. Both experiments with knockout (KO) mice and experiments using selective mGlu₂R PAMs suggested that these effects are predominantly mGlu₂R–mediated. Indeed, the inhibitory effects of mGlu_2/3R agonists on PCP- and amphetamine-evoked hyperactivity are absent in mGlu₂R–KO but not in mGlu₃R–KO mice [85, 86].

Recent evidence, however, suggests that mGlu₃R may be a potential therapeutic target for schizophrenia-associated cognitive dysfunction since the mGlu₃R NAMs VU0469942 and VU0477950 reverse the positive effects of the mGlu_2/3R orthosteric agonist LY379268 on synaptic plasticity and learning abilities in mice [87].

Crosstalk between mGlu₂ and 5-HT_2A receptors

Not only does mGlu₂R seem to be the target for many glutamate antipsychotics, but it has also been found to crosstalk with 5-HT_2A receptor, a key target for atypical antipsychotics [30, 74, 88–93]. Interestingly, mGlu₂R has been shown to be necessary for pharmacological and behavioral effects induced by hallucinogenic 5-HT_2A agonists since these effects were abolished in mGlu₂R–KO mice [94]. On the other hand, the locomotor activity induced by the mGlu_2/3R antagonist LY341495 is attenuated in 5-HT_2A-KO mice [74]. Moreover, chronic administration of the hallucinogenic 5-HT_2A agonist 2,5-Dimethoxy-4-bromoamphetamine (DOB) in mice attenuates the behavioral effects of the mGlu_2/3R agonist LY379268 [95] whereas chronic treatment by LY341495 decreases 5-HT_2A binding and the hallucinogenic effects of LSD [96].

Indeed, mGlu₂R and 5-HT_2A were found to co-localize in mouse and human cortical pyramidal neurons and form a heterocomplex with unique signaling properties [30, 74]. Fribourg and colleagues [30] elucidated the functional coupling between mGlu₂R which signals via a Gi heterotrimeric G protein, and 5-HT_2AR which signals via Gq protein. Compared to the homomeric responses of each receptor to its endogenous neurotransmitter, heteromerization of mGlu₂R with 5-HT_2AR potentiates glutamate-induced mGlu₂R–coupled Gi signaling and attenuates the 5-HT-induced 5-HT2AR-coupled Gq signaling. Moreover, drugs that stabilize the active or inactive conformation in one receptor cause the opposite conformation of the partner receptor. This inverse conformational coupling of the two receptors as parts of a heteromer unifies the mechanism of action of antipsychotic drugs targeting these receptors (Figure 5). Using the PCP-like drug MK801 as a model of psychosis, it was demonstrated that the mGlu₂R–dependent antipsychotic-like behavioral effects of LY379268 were absent in 5-HT_2AR-KO mice, while the 5-HT_2AR-dependent antipsychotic-like behavioral effects of clozapine were absent in mGlu₂R–KO mice. In schizophrenia, where the relative expression of these two receptors is dysregulated, a combination of an inverse agonist of 5-HT_2AR and a strong agonist of mGlu₂R may prove beneficial, as suggested by Fribourg and colleagues [30]. Collectively, these data suggest that a common target for psychedelics, atypical and glutamate antipsychotics is the 5-HT_2A–mGlu₂R complex, for which designing a bivalent ligand might be a rational therapeutic strategy.

Figure 5. Gi-Gq Balance Model of the Mechanism of Action of Antipsychotic and Psychedelic Drugs through the mGluR2/2AR Complex.

(A) Heteromerization of mGlu₂R and 5-HT_2AR establishes an optimal Gi-Gq signaling balance in response to the endogenous neurotransmitters, glutamate and serotonin (an increase in Gi and a decrease in Gq when compared to the homomeric receptor signaling; dashed lines). Psychedelics (mGlu₂R inverse agonists and 5-HT_2AR agonists) invert the signaling balance through the complex tipping the balance from being in favor of Gi signaling (normal complex) to becoming in favor of Gq signaling (propsychotic). (B) Disruption of the optimal balance in psychotic states can be compensated for by antipsychotics (mGlu₂R agonists and 5-HT_2AR inverse agonists) that tip the balance in favor of Gi signaling thus restoring the physiological signaling. Figure reprinted from Fribourg et al. [21].

Clinical effects of mGlu_2/3R agonists

In 2007, a phase II proof-of-concept clinical trial compared pomaglumetad methionil (LY2140023 monohydrate; a prodrug of the mGlu_2/3R agonist LY404039) to the atypical antipsychotic olanzapine or placebo. The study showed a positive effect for pomaglumetad against positive and negative symptoms of schizophrenia compared to placebo. Although the efficacy was not as significant as for olanzapine, pomaglumetad was safe and well-tolerated and, importantly, was neither associated with extrapyramidal symptoms nor with metabolic abnormalities [97]. Interestingly, a pharmacogenetic analysis identified 23 single nucleotide polymorphisms (SNPs) significantly associated with pomaglumetad response, sixteen of which were located in the 5-HT_2A receptor (HTR_2A) gene [98]. However, subsequent clinical trials with pomaglumetad showed either inconclusive results [99], or results that are no different from placebo [100]. Importantly, results form a recent post-hoc analysis suggest that the therapeutic effects of pomaglumetad depend on previous exposure to either typical or atypical antipsychotic drugs [101]. These findings and their basic mechanism based on previous preclinical findings [102] are discussed in Box 1

Box 1. Can epigenetics explain discrepant clinical data?

Interestingly, recent preclinical and postmortem human brain findings may provide an epigenetic explanation of the apparently discordant results with both mGlu_2/3R agonists and mGlu₂R PAMs. It was demonstrated that chronic atypical antipsychotic therapy induces a highly selective down-regulation of mGlu₂R expression in mouse frontal cortex (Figure I). This effect was mediated via a signaling mechanism that involved 5-HT_2A receptor-dependent modulation of the promoter activity of the Histone deacetylase 2 (Hdac2) gene. Thus, the presence of serotonin resulted in inhibition of HDAC2 promoter activity, an effect that was reversed in vitro and in mouse frontal cortex by clozapine. Importantly, chronic treatment with atypical antipsychotics induced a 5-HT_2A receptor-dependent up-regulation and increased binding of HDAC2 to the mGlu₂R promoter. These epigenetic changes occurred in association with repressive histone modifications at the promoter region of the mGlu₂R gene in mouse and human frontal cortex [102]. Because chronic antipsychotic drug treatment induced a 5-HT_2A receptor-dependent decrease in mGlu₂R transcription and its electrophysiological properties, together these findings suggest that previous medication with atypical antipsychotic drugs may prevent the therapeutic responses to mGlu₂R in schizophrenia patients. This is supported by results from a recent post-hoc analysis disclosed by Eli Lilly investigators where the subgroup of patients who were earlier in disease course (<3 years of illness) responded to pomaglumetad, as opposed to patients late in disease (>10 years of illness). More interestingly, in patients with previous exposure to predominantly D2-blocking drugs, the efficacy of pomaglumetad was comparable to risperidone, whereas it did not differ from placebo in patients previously treated predominantly with 5-HT_2A antagonists [101]. These findings also suggest that inhibition of HDAC2 may reverse the repressive epigenetic changes induced by chronic atypical antipsychotic drugs and hence improve the antipsychotic properties of mGlu_2/3R agonists and mGlu₂R PAMs.

Figure I (in box 1). Chronic atypical antipsychotic therapy downregulates mGlu₂R. Schematic model of the effect of chronic atypical antipsychotic treatment to the epigenetic status of the mGlu₂R (Grm2) gene. (A) Activation of the 5-HT_2A receptor by the endogenous neurotransmitter serotonin represses the promoter activity of the HDAC2 gene in mouse and in human frontal cortex. (B) Atypical antipsychotic drugs, such as clozapine and risperidone, reverse the 5-HT_2A receptor-dependent repression of HDAC2, an effect that is associated with increased HDAC2 promoter activity and repressive histone modifications at the mGlu₂R promoter. This epigenetic effect of chronic atypical antipsychotic treatment may limit the therapeutic effects of mGlu_2/3R agonists such as pomaglumetad.

Promise of mGluR positive allosteric modulators

LY487379, reported by researchers at Eli Lilly in 2003, was the first identified mGlu₂R–selective PAM [103]. LY487379 showed comparable efficacy to mGlu_2/3R agonists in preclinical psychosis models [104, 105]. However, the poor bioavailability, modest potency, and short duration of action limited the further in vivo characterization of compounds within the LY487379 class. Soon after, the biphenyl-indanone class was reported [106, 107]. Biphenyl-indanone A (BINA) became the prototype mGlu₂R PAM for in vivo studies after it was shown to be a potent mGlu₂R–selective PAM with robust long-lasting in vivo activity [108]. BINA blocks the PCP-induced hyperlocomotor activity [108, 109] as well as the DOB-induced head twitch response [89]. Pharmacologic magnetic resonance imaging (phMRI) revealed that BINA blocks PCP-induced blood oxygenation level-dependent (BOLD) activation in the rat brain. This allowed defining BINA’s mechanism of action at the brain circuitry level where its effect was apparent in the prefrontal cortex, caudaute–putamen, nucleus accumbens, and mediodorsal thalamus, structures linked to schizophrenia [109]. After the introduction of LY487379 and BINA, a variety of structurally distinct classes of mGlu₂R PAMs were discovered, many of which proved to have in vivo activity as well [110]. Early mutagenesis studies identified three amino acid residues in transmembrane (TM) segments IV and V of mGlu₂R to be critical for the activity of LY487379 and several other mGlu₂R PAMs [111, 112]. A more recent extensive study identified additional residues in TM III, V, and VI to play an important role in the activity of mGlu₂R PAMs [113]. Interestingly, 3 residues in TM IV of mGlu₂R have been found to be essential for the heteromerization of 5-HT_2A-mGlu₂R [68] indicating that TMD region of mGlu₂R is implicated in different types of allosteric interactions. However, the implication of mGlu₂R involvement in heteromeric receptor complexes on PAM pharmacology remains unknown. Although glutamate-induced signaling through mGlu₂R can be potentiated by a PAM, mGlu₂R in its heteromeric state may not necessarily respond to a PAM in the same manner. Moreover, it is not known yet whether mGlu₂R PAMs can crosstalk and alter the signaling of 5-HT_2A through the 5-HT_2A–mGlu₂ receptor heteromer.

So far, two mGlu₂R PAMs have made it into clinical trials. The first compound is ADX71149, from Addex and Janssen, which showed the first successful clinical proof-of-concept in a Phase IIa study. Data reported in late 2012, demonstrated safety and tolerability and identified patients with residual negative symptoms as the population most likely to benefit from adjunctive treatment with ADX71149 [114]. The second compound, AZD8529, from Astrazeneca, failed to separate from placebo in a phase IIa study, unlike the active control risperidone [115]. However, it is worth mentioning that this study tested a single dose of AZD8529 as a monotherapy, thus further investigation is warranted.

One more ongoing effort by Janssen is to identify potential positron emission tomography (PET) tracers for imaging mGlu₂R [116]. Preliminary evaluation of [11C]JNJ-42491293 in rat brain demonstrated specific and reversible binding to an mGlu₂R allosteric site [117]. Subsequent evaluation in 20 healthy male subjects confirmed radioactivity uptake consistent with reported distribution of the mGlu₂R in human brain [118]. These encouraging results would likely allow quantifying mGlu₂R in human brain as well as for assessment of occupancy by drug candidates targeting this allosteric site.

mGlu₅R is another receptor within the mGluR superfamily for which positive allosteric modulation is a potential treatment strategy for schizophrenia. The efficacy of mGlu₅R PAMs in animal models of schizophrenia and cognitive dysfunction were thought to be mediated by mGlu₅R–induced potentiation of NMDA receptors [119, 120]. However, a recent study has challenged this hypothesis by showing that VU0409551, a novel mGlu₅R PAM with robust antipsychotic and cognition-enhancing effects in animal models, exhibits biased allosteric modulation by selectively potentiating mGlu₅R coupling to Gq-mediated signaling but not mGlu₅R modulation of NMDA receptors [61]. This raises an interesting question regarding the signaling mechanisms underlying the antipsychotic effects of mGlu₂R PAMs and whether they could exert some of their beneficial effects through NMDA-independent signaling as well. It would also be interesting to compare mGlu₂R vs. mGlu₅R PAMs, particularly which classes of drugs are superior for treatment of cognitive dysfunction associated with schizophrenia.

Recent findings suggest that ligands that interact with putative receptor heteromers, such as µ-opioid-mGlu₅R and µ-δ opioid receptor heteromer, modulate unique phenotypes in rodent models [121, 122]. Since the heteromer between mGlu₂R and 5-HT_2AR has been suggested to function as a primary molecular target responsible for the therapeutic effects of both atypical and glutamate antipsychotic drugs (Figure 5), further medicinal chemistry work with bivalent compounds that specifically target the 5-HT_2AR-mGlu₂R heteromer inducing an antipsychotic-like pattern of G protein signaling in frontal cortex pyramidal neurons may provide a route for the development of new and more effective agents for the treatment of schizophrenia.

Concluding remarks

This review has summarized recent findings suggesting that mGlu₂R activation may provide a novel therapeutic strategy for treatment of schizophrenia and help avoid the adverse effects associated with currently available antipsychotic drugs. As pharmacologic tools, PAMs might be superior to orthosteric agonists since they have a unique ability to modulate glutamate release in a “state-dependent” manner that helps fine-tune physiological responses. An important challenge in the field is related to the potential adverse signaling effects induced by mGluR PAMs in the absence of orthosteric agonists [123]. This is particularly important considering the recent findings that mGluRs with VFTs deleted, when reconstituted into nanodiscs, are able to couple to and activate heterotrimeric G proteins in the presence of PAMs [124]. Whether the expression of alternatively spliced variants that generate truncated isoforms of mGluRs [125] might lead to unfavorable PAM-dependent signaling events in the absence of endogenous agonists, needs to be explored. Despite the advances in our knowledge of medicinal chemistry and basic mechanism of action of mGluRs, many questions are yet to be addressed with both biophysical approaches and whole animal experimental systems (Box 2). Only time and additional research will provide us with answers.

Box 2. Outstanding questions.

• How do mGlu₂R PAMs modulate mGlu₂R in its heteromeric states?

• Do mGlu₂R PAMs have the ability to crosstalk to 5-HT_2A receptors?

• Can an mGlu₂R PAM be a part of a bivalent ligand targeting the 5-HT_2A–mGlu₂ receptor complex implicated in schizophrenia? And if yes, can a bivalent PET tracer be used diagnostically for quantifying such receptor complex?

• Would an HDAC2 inhibitor improve the therapeutic efficacy of mGlu₂R antipsychotics in patients with a history of chronic atypical antipsychotic therapy?

• What is the right patient population that is likely to benefit from mGlu₂R allosteric modulation?

Highlights.

• mGlu₂ activation is a potential therapeutic strategy for schizophrenia.

• Unlike orthosteric agonists, PAMs have a unique ability to fine-tune physiological responses.

• Chronic atypical antipsychotic therapy may reduce the efficacy of mGlu₂ antipsychotics.

Glossary Box

Allosteric site

a binding site on a receptor macromolecule that is non-overlapping and spatially distinct from, but conformationally linked to, the orthosteric binding site

Dopamine (DA) hypothesis

the most classical schizophrenia hypothesis that attributes psychotic symptoms to hyperactive DA signaling in the brain

Epigenetics

literally means "above" or "on top of" genetics. It refers to external modifications to DNA that turn genes "on" or "of" without changes in DNA sequence

Excitatory postsynaptic potential (EPSP)

a temporary depolarization of postsynaptic neuron membrane potential that makes it more likely for this neuron to trigger an action potential

Glutamate

the main excitatory neurotransmitter in the brain

G protein-coupled receptor (GPCR)

an integral plasma membrane protein that senses molecules outside the cell and transduce those extracellular signals to intracellular relay proteins, the heterotrimeric GTP-binding proteins (G proteins)

Heteromer

structural assembly composed of two or more different components

Histone

Evolutionary conserved proteins found in eukaryotic cell nuclei that package DNA into structural units called nucleosomes. Covalent modifications at the N-terminal tail of histones correlate with open or closed states of chromatin, depending on the type of modification, and thus lead to changes in gene expression

Histone deacetylase (HDAC)

a class of enzymes that remove acetyl groups from histones allowing the histones to wrap the DNA more tightly and repress gene transcription

Metabotropic glutamate receptors (mGluRs)

GPCRs that bind glutamate and function to modulate, rather than mediate, synaptic transmission. This is to differentiate them from “ionotropic” glutamate receptors like NMDA, which are ion channels mediating excitatory neurotransmission

N-methyl-D-aspartate (NMDA) receptor

is a glutamate receptor and ion channel protein found in nerve cells. It is involved in synaptic transmission and plays an important role in learning and memory

NMDA hypofunction hypothesis

a glutamate-based hypothesis which postulates that reduced activation of NMDA glutamate receptor underlies development of different schizophrenia symptoms

Orthosteric site

the binding site on a receptor macromolecule that is recognized by the endogenous agonist for that receptor