REVIEWS

METABOTROPIC GLUTAMATE RECEPTORS AS NOVEL TARGETS FOR ANXIETY AND STRESS DISORDERS Chad J. Swanson, Mark Bures, Michael P. Johnson, Anni-Maija Linden, James A. Monn and Darryle D. Schoepp Abstract | Anxiety and stress disorders are the most commonly occurring of all mental illnesses, and current treatments are less than satisfactory. So, the discovery of novel approaches to treat anxiety disorders remains an important area of neuroscience research. Glutamate is the major excitatory neurotransmitter in the mammalian central nervous system, and G-protein-coupled metabotropic glutamate (mGlu) receptors function to regulate excitability via pre- and postsynaptic mechanisms. Various mGlu receptor subtypes, including group I (mGlu1 and mGlu5), group II (mGlu2 and mGlu3), and group III (mGlu4, mGlu7 and mGlu8) receptors, specifically modulate excitability within crucial brain structures involved in anxiety states. In addition, agonists for group II (mGlu2/3) receptors and antagonists for group I (in particular mGlu5) receptors have shown activity in animal and/or human conditions of fear, anxiety or stress. These studies indicate that metabotropic glutamate receptors are interesting new targets to treat anxiety disorders in humans.

Eli Lilly and Company, Neuroscience Division, Lilly Corporate Center DC 0510, Indianapolis, Indiana 46285, USA. Correspondence to D.D.S: e-mail: [email protected]. doi:10.1038/nrd1630 Published online 24 January 2005

Anxiety disorders represent a range of conditions that include generalized anxiety, panic attacks, post-traumatic stress disorder, obsessive-compulsive syndrome and social phobias1. A number of animal models of anxiety have been devised to evaluate the anxiolytic properties of compounds for clinical use. Much of the current understanding of the neuroanatomical basis of anxiety disorders stems from these models, which are thought to be analogous to certain aspects of the physiological and behavioural constructs of anxiety in humans1,2. Interestingly, although anxiety treatments have been available for decades, the biological basis for anxiety disorders in humans is just beginning to emerge. Since their introduction in the 1960s, benzodiazepine compounds have become the most widely used class of drugs for the treatment of anxiety disorders. Benzodiazepines act by modulating the inhibitory neurotransmitter GABA (γ-aminobutyric acid) through an allosteric site on the GABAA receptor complex, a ligandgated chloride ion channel. However, some forms of anxiety, such as obsessive-compulsive disorder (OCD),

NATURE REVIEWS | DRUG DISCOVERY

are relatively resistant to benzodiazepine treatment; furthermore, the use of benzodiazepines is often associated with serious side effects, such as sedation, memory impairment, ataxia, abuse potential and physical dependence3. It has therefore become increasingly apparent that alternative strategies with limited sideeffect profiles must be devised to treat the varied manifestations of anxiety. For instance, selective serotonin-reuptake inhibitors (SSRIs) have been investigated and recently approved to treat certain anxiety conditions (for example, panic disorder and generalized anxiety disorder (GAD)). Indeed, several recent reviews have focused on novel mechanisms to treat anxiety and stress-related disorders, including corticotropin-releasing factor (CRF) antagonists, neurokinin antagonists, GABAA-selective modulators as well as SSRIs and serotonin noradrenaline-reuptake inhibitors4–9. This article will focus on the emerging preclinical and clinical data that implicate modulation of the metabotropic glutamate (mGlu) receptors as a potential anxiolytic strategy.

VOLUME 4 | FEBRUARY 2005 | 1 3 1

REVIEWS

Glia and astrocytes ? mGlu5

lu 3 mG

mG lu

3

mG

Postsynaptic

lu 7

u2 Gl m

Presynaptic

/5 mGlu 1

EAAT

↓ cAMP NMDA

Glu

↑ Ca2+

AMPA

vGluT

↑ Na+

Kainate

↓ cAMP

lu

4/

↓ Cl–

/8

G

8

mG

lu 1/

5

mGlu 4

m

Modulation of excitation

GA BA

A

EAAT

GAB

AB

mGlu3

GABAB

mGlu7

GABA

Figure 1 | Hypothetical synapse illustrating the general synaptic localization and function of glutamatergic receptors and transporters. The ionotropic glutamate receptors, N-methyl-D-aspartate (NMDA), kainate and α-amino-3-hydroxy-5-methyl4-isoxazolepropionic acid (AMPA) subtypes, largely function to mediate fast receptor transmission, but also mediate the usedependent changes required for neuronal plasticity. The vesicular transporters (vGluT1 and vGluT2) load glutamate into vesicles presynaptically. The glial, astrocyte and postsynaptic glutamate transporters (excitatory amino-acid transporters, EAAT1–5) are thought to mediate the uptake of glutamate and therefore termination of synaptic transmission. The metabotropic glutamate receptors (TABLE 1) have a diverse synaptic localization and function pre- and postsynaptically to modulate neurotransmitter release and postsynaptic excitability, respectively. For instance, the group II mGlu (mGlu2/3) receptor agonists probably serve to stimulate mGlu autoreceptors on glutamatergic terminals to suppress excitatory neurotransmission at selected synapses in the central nervous system. By contrast, mGlu5 receptor antagonists might inhibit mGlu5-mediated potentiation of NMDA receptor ion currents and potentially disrupt the formation of memory processes associated with stressful events. GABA, γ-aminobutyric acid.

Glutamatergic agents for anxiety

IONOTROPIC RECEPTOR

A ligand-gated ion channel receptor that modulates cell excitability.

132

In general, anxiety- and stress-related illnesses are thought of as a collection of disorders that have in common excessive or inappropriate brain excitability within crucial brain circuits, which leads to the expression of a spectrum of psychic (for example, excessive worry) or somatic (for example, disruptions of sleep, cardiovascular and gastrointestinal functions) symptoms. As glutamate is the major excitatory neurotransmitter in the mammalian brain, it is logical that new approaches for anxiety could include drugs that modulate glutamatergic functions. Within the past decade, it has been realized that glutamate neurotransmission is mediated through, and and regulated by, diverse families of receptors and transporters (FIG. 1; TABLE 1). These families include IONOTROPIC glutamate (iGlu)receptors

| FEBRUARY 2005 | VOLUME 4

(N-methyl-D-aspartate (NMDA), α-amino-3 hydroxy5-methyl-4-isoxazolepropionic acid (AMPA) and kainate types), which directly mediate synaptic excitability and plasticity; metabotropic G-protein-coupled receptors (GPCRs; mGlu1–8), which regulate glutamate release and modify postsynaptic excitability to glutamate; plasma membrane glutamate transporters, which clear the synaptic space of released glutamate (excitatory amino-acid transporters (EAAT) 1–5); and vesicular glutamate transporters, which function to package glutamate for exocytotic release (vGluT1 and vGluT2) (FIG. 1). In general, the involvement of glutamate neurotransmission in psychiatric disorders is, clinically, a relatively unexplored field because of the lack of clearly safe approaches to modulate glutamate neurotransmission.

www.nature.com/reviews/drugdisc

REVIEWS

GLYCINE SITE

The NMDA receptor is unique in that glycine (possibly D-serine) acts as a co-agonist with glutamate at a different site on the receptor complex.

Preclinical work carried out about 20 years ago on MK801, a non-competitive NMDA receptor antagonist, showed that it was anxiolytic in a conditioned emotional response test and a simple conflict test10, which initially supported a role for glutamatergic neurotransmission in anxiety states. Subsequent studies established that competitive NMDA receptor antagonists, GLYCINE-SITE antagonists for the NMDA receptors and AMPA/kainate receptor antagonists also possess anxiolytic properties in animal models11. Overall, the application of these iGlu receptor antagonists to treating anxiety in the clinic has not been successful. So far, compounds that suppress glutamate neurotransmission through a ionotropic receptor mechanism often produce problematic side effects that have relegated their use to more severe and/or acute treatment conditions, such as cerebral ischaemia, head trauma or as amnaesiac/anaesthetics (for example, glycine-site antagonists or ketamine). These clinical shortcomings can generally be attributed to the fairly ubiquitous expression of these targets throughout the brain, the direct influence of these

compounds on fast excitatory neurotransmission, and/or the lack of iGlu-receptor-subtype-selective antagonists. Indeed, these initial glutamatergic agents produce central nervous system (CNS) depression, memory impairment and other undesirable side effects (for example, psychotic-like episodes) in human subjects12,13. Nevertheless, these clinical observations indicate a role of glutamate neurotransmission in mediating fear/anxiety states. Glutamatergic mechanisms that more selectively target and suppress glutamate hyperexcitability within the appropriate fear/anxiety circuits might therefore lead to novel efficacious and well-tolerated anxiolytics. The mGlu receptors represent a novel family of class C GPCRs that comprise at least eight known subtypes (TABLE 1). A growing body of evidence indicates that these receptors might serve as potential therapeutic targets for a variety of pathological states12,14,15. The mGlu receptors are further divided into three groups on the basis of sequence similarity, pharmacology and the preferred signal transduction mechanisms they couple to when expressed in vitro. Group I mGlu

Table 1 | Classification of the metabotropic glutamate (mGlu) receptors Family receptor

Coupling

Key localization and actions

Group/subtype-selective pharmacological agents

mGlu1

Excitatory Gq-coupled

Most often postsynaptic at glutamatergic synapses. Indicated in synaptic plasticity, including long-term. potentiation/depression (LTP/LTD). Cerebellar localization in granular cell and parallel fibre layers.

Agonists: DHPG, 1S,3R-ACPD, quisqualate Antagonist: LY393675 Inverse agonist (or allosteric antagonist): LY367385

mGlu5

Excitatory Gq-coupled

Most often postsynaptic at glutamatergic synapses, also found in glial cells. High expression in several forebrain regions including hippocampus and amygdala. Indicated in synaptic plasticity, especially some forms of cortical and hippocampal LTD.

Agonists: DHPG, 1S,3R-ACPD, quisqualate, CHPG Inverse agonist (or allosteric antagonist): MPEP

mGlu2

Inhibitory Gi / Go-coupled

Localization largely presynaptic on glutamatergic and other neurotransmitter synapses. High expression in forebrain regions including hippocampus and amygdala; also in certain layers with the cortex and cerebellum. Linked to hippocampal LTD and regulation on medial perforant path.

Agonists: DCG-IV, 2R,4R-APDC, 1S,3R-ACPD, LY354740, LY379268 Antagonist: LY341495 Potentiator: 4-MPPTS (LY487379), 4-APPES, CBiPES

mGlu3

Inhibitory Gi / Go-coupled

Widely expressed in glial cells but also discrete localization both pre- and postsynaptic on glutamatergic and other neurotransmitter synapses. Expression within forebrain regions including hippocampus and thalamus. Linked to neurotropin release from glial cells.

Agonists: DCG-IV, 2R,4R-APDC, 1S,3R-ACPD, LY354740, LY379268 Antagonist: LY341495

mGlu4

Inhibitory Gi / Go-coupled

Localization both pre- and postsynaptic on glutamatergic and other neurotransmitter synapses. Presynaptic in cerebellar parallel fibres and linked to cerebellar plasticity and motor learning.

Agonists: L-SOP, ACPT-1, L-AP4 Antagonist: MSOP, MAP4, CPPG

mGlu6

Inhibitory Gi / Go-coupled

Expression confirmed only in retinal bipolar ON cells. Knockout animals reported to have visual acute deficits.

Agonists: L-SOP, L-AP4 Glutamate-site antagonist: MSOP, MAP4

mGlu7

Inhibitory Gi / Go-coupled

Localization both pre- and postsynaptic on glutamatergic and other neurotranmitter synapses in limbic and cortical regions. Has lower affinity for glutamate than other mGlu subtypes and only presynaptic inhibitory mGlu localized to active zone of synapses. Thought to serve a classical autoreceptor function.

Agonists: L-SOP, L-AP4 Antagonist: MSOP, MAP4, LY341495 (100-fold lower affinity than group II)

Inhibitory Gi / Go-coupled

Localization largely presynaptic on glutamatergic and other neurotranmitter synapses. High expression in forebrain regions including hippocampus and amygdala. Linked to regulation of lateral perforant path.

Agonists: L-SOP, L-AP4, 3,4-DCPG Antagonist: MSOP, MAP4

Group I

Group II

Group III

mGlu8

NATURE REVIEWS | DRUG DISCOVERY

VOLUME 4 | FEBRUARY 2005 | 1 3 3

REVIEWS

Box 1 | Discovery of LY354740, an orthosteric-site mGlu receptor agonist Glutamic acid, the H H endogenous agonist for HO 2C HO 2C CO2H Conformational restriction metabotropic glutamate CO2H H NH2 H NH (mGlu) receptors, is a 2 LY354740 L-Glutamate highly flexible molecule Highly potent and selective Non-selective with multiple agonist for mGlu2/3 receptors and low potency energetically accessible conformational states. In an attempt to understand mGlu how glutamate S146 conformation relates to R61 receptor subtype potency R57 and selectivity, several attempts have been made to constrain the glutamate T168 pharmacophore by K377 D295 embedding it into various mono-, bi- and tricyclic ring systems36,118–123. LY354740 is an example of a conformationally constrained glutamate analogue in which the glutamate backbone is locked into a fully extended state by incorporation into a bicyclo[3.1.0]hexane ring system36. This molecule is a highly potent and selective agonist for mGlu2 and mGlu3 receptors124. Mutational analysis of rat mGlu2 identified multiple residues important for LY354740 binding, including R57, Y145, S146, T168, Y216 and D295125. A homology model of the glutamate-recognition site within the amino-terminal domain (ATD) of human mGlu2 was generated on the basis of the X-ray crystal structure of the ATD of mGlu1126. When docked in this model, LY354740 (magenta above) forms an intricate network of hydrogen bonds to key binding-site residues from both upper and lower lobes of the glutamate-binding pocket, including R57, R61, S146, T168, D295 and K377.

receptors (mGlu1 and mGlu5) are positively coupled to phospholipase C, whereas group II (mGlu2 and mGlu3) and group III (mGlu4, mGlu6, mGlu7 and mGlu8) receptors typically inhibit activated adenylate cyclase activity when expressed in vitro16–18. The recent development of potent, systemically active, and subtype-selective mGlu pharmacological agents has allowed for the elucidation of the role that mGlu receptors play in the normal and pathological functioning of the CNS. Relatively recent data indicate that mGlu receptors might have an important role in anxiety-related disorders19,20. In particular, a growing body of evidence implicates the group II mGlu receptor agonists and group I (particularly mGlu5) receptor antagonists as potential anxiolytics. This is supported by both preclinical results and recent clinical data from a study of a group II agonist (TABLE 2). This article will therefore focus on the group I and II mGlu receptors, but will also mention the possible role of group III mGlu receptors in anxiety and stress-related disorders. Group II mGlu receptors in anxiety

Group II mGlu receptors (mGlu2/3) are widely distributed throughout the CNS, where they exhibit moderate to high expression in brain regions that are commonly associated with anxiety disorders, including the hippocampus, prefrontal cortex and amygdala21–26. The mGlu2 receptors are generally expressed at extrasynaptic sites on neuronal terminals, where they have been shown to suppress excitatory amino-acid neurotransmission at a number of synapses27,28. By contrast, mGlu3 receptors 134

| FEBRUARY 2005 | VOLUME 4

display a more diverse localization. Using an mGlu3-preferring antibody and mGlu2 receptor knockout mice, Tamaru and colleagues29 found both pre- and postsynaptic localization on neurons, as well as more widespread localization on certain glial cells. The group II mGlu receptor agonists (acting at mGlu2/3 receptors) seem to be important in reversing anxiety states by maintaining homeostasis in select brain regions related to anxiety. The compound LY354740 has been a pioneering pharmacological agent with group II selectivity, potency and bioavailability as an mGlu2/3 receptor agonist (see BOX 1). LY354740 has progressed into Phase II clinical trials for anxiety disorders, and represents well the state of knowledge for this novel class of putative anxiolytic (see TABLE 2 for a list of supporting data and references). At the synaptic level, LY354740 acts at presynaptic mGlu2/3 receptors to limit the release of glutamate, but might also influence postsynaptic excitability directly via dendritic mGlu2 and/or mGlu3 receptors (FIG. 2). Group II mGlu2/3 receptors are also present at heterosynapses, where they directly limit release of other neurotransmitters, including GABA, monoamines such as dopamine and noradrenaline, and neuropeptides27. LY354740 has now been investigated in several animal models of psychiatric disorders, including models of anxiety and stress in rodents and humans. Fear-potentiated startle is a model of anxiety that uses Pavlovian fear conditioning to assess the ability of a (originally neutral) stimulus to elicit fear in the presence of a noise burst. Similarly to the benzodiazepines22,

www.nature.com/reviews/drugdisc

REVIEWS

Table 2 | Group-II-selective compounds in animal and human anxiety studies Model

mGlu ligand Pharmacology

Route of administration; species

Effect on group II mGlu receptors

References

Fear-potentiated startle

LY354740 LY354740 APDC LY341495 LY354740 LY354740

mGlu2/3 agonist mGlu2/3 agonist mGlu2/3 agonist mGlu2/3 antagonist mGlu2/3 agonist mGlu2/3 agonist

Oral; rat Intra-amygdala; rat Intra-amygdala; rat I.p.; rat Oral; rat Oral: human

4-MPPTS (LY487379) 4-APPES

mGlu2 potentiator

Oral; rat

Blocked expression Blocked acquisition and expression Blocked expression Reversed LY354740 effects Blocked expression, not acquisition Blocked expression, without reports of sedation Blocked expression

mGlu2 potentiator

Oral; rat

Blocked expression

62

Elevated plus maze

LY354740 LY354740 LY354740

mGlu2/3 agonist mGlu2/3 agonist mGlu2/3 agonist

Oral; mouse Oral; rat Oral; rat

Increased open-arm time Increased open-arm time Increased open-arm time

36 30 37

c-Fos expression

LY354740

mGlu2/3 agonist

S.c.; mouse

Increase open-arm time and reduced plus-mazeinduced c-FOS in hippocampus

43

Conflict tests

LY354740 LY354740 LY354740

mGlu2/3 agonist mGlu2/3 agonist mGlu2/3 agonist

I.p.; rat I.p.; mouse I.p.; rat

LY314582

mGlu2/3 agonist

I.p.; rat,pigeon

LY354740

mGlu2/3 agonist

I.p.; rat

Increased shocks accepted in Vogel test Increased punished crossings in four-plate test Increased punished responding in select operant schedules Increased punished responding in select operant schedules Produced temporal changes in responding dependent on baseline responding

Lactate-induced panic

LY354740

mGlu2/3 agonist

I.p.; rat

Blocked panic response to lactate infusion

42

Stress-induced hyperthermia

LY314582

mGlu2/3 agonist

Oral; rat

80

CBiPES

mGlu2 potentiator

S.c.; rat

Reduced stress-induced increases in core body temperature Reduced stress-induced increases in core body temperature

Drug-withdrawal stress

LY354740

mGlu2/3 agonist

I.p.; rat

Reduced startle response associated with nicotine and diazepam withdrawal

Immobilization stress

LY354740

mGlu2/3 agonist

S.c.: rat

Reduced immobilization-induced increases in noradrenaline and dopamine in prefrontal cortex dialysated

CO2-induced anxiety/panic

LY354740

mGlu2/3 agonist

Oral; human

Prevented CO2-induced anxiety in panic attack patients

Generalized anxiety LY354740 disorder

mGlu2/3 agonist

Oral; human

Reduced anxiety (HAM-A) compared with placebo

30 111 111 111 109 112 62

113 113 114 114 115

62 52 116

20 117

Data were compiled from listed references. 4-MPPTS, 2,2,2-trifluoro-N-[4-(2-methoxyphenoxy)phenyl]-N-(3-pyridinylmethyl)-ethanesulphonamide; 4-APPES, N-[4-(4-carboxamidophenoxy)phenyl]-N-(3-pyridinylmethyl)-ethanesulphonamide hydrochloride monohydrate; APDC, (2R,4R)-4-aminopyrrolidine-2,4-dicarboxylic acid; CBiPES, N-[4′-cyanobiphenyl-3-yl)-N-(3-pyridylmethyl)-ethanesulphonamide hydrochloride; HAM-A, Hamilton Anxiety Scale; i.p., intraperitoneal; LY314582, racemic mixture of LY354740; LY354740, (1S,2S,5R,6S)-(+)-2-aminobicyclo[3.1.0]hexane-2,6-dicarboxylic acid; mGlu, metabotropic glutamate receptor; s.c. subcutaneous.

ELEVATED PLUS MAZE

A commonly used anxiety test for rodents in which the animal can choose to explore an ‘open’ unprotected arm or a ‘closed’ protected arm of a cross-shaped elevated platform. Animals which are less fearful spend more time in the open arm and enter it more often.

administration of the potent and systemically active group II mGlu receptor agonist LY354740 produced a lasting and dose-dependent blockade of the fearpotentiated startle response in rats, with a near complete blockade in fear-enhanced startle with no effects on basal startle30 (FIG. 3). More recent work using this model has indicated that the amygdala might be important in mediating the effects of LY354740 outlined above. It has been demonstrated, for instance, that direct microinfusion of LY354740 into the amygdala prior to testing results in a significant attenuation of the acquisition and expression of fear-potentiated startle31. Interestingly, systemic administration of the selective group II mGlu receptor antagonist LY341495 was able to block the local effects of LY354740 in this study, indicating that the amygdala might be the primary site of LY354740 action in the fear-potentiated startle test following systemic administration30. This idea is supported by the fact that blockade of glutamatergic transmission is

NATURE REVIEWS | DRUG DISCOVERY

associated with anxiolytic effects in the amygdala32 and that agonists of mGlu2/3 receptors have been shown to suppress excitatory neurotransmission in the amygdala33. The ELEVATED PLUS MAZE (EPM) is a widely used anxiety model that exploits the rodent’s innate aversion to heights and open spaces, and anxiolytic benzodiazepine agents such as diazepam are highly effective anxiolytics as measured by an increasing exploration of the open arm34,35. Likewise, LY354740 increases activity on the open arm at oral doses of 1, 3 and 10 mg per kg in mice30,36. Recently, the anxiolytic effects of LY354740 in the EPM were reportedly reversed by the benzodiazepine receptor antagonist flumazenil, whereas the µ-opioid receptor antagonist naloxone was ineffective at blocking the effects of LY35474037. These actions of a benzodiazepine antagonist indicate a mechanism of LY354740 action involving modulation of GABAergic neurotransmission. However, flumazenil could be acting indirectly, because it also

VOLUME 4 | FEBRUARY 2005 | 1 3 5

REVIEWS

Maximal suppression of evoked excitatory EPSPs (%)

100

80

60

40

20

10 –11

–10

–9

–8

–7

–6

–5

LY354740 [–log molar] Amygdala (BLA–CeA) Medial prefrontal cortex

Medial perforant path (hippocampus)

Figure 2 | Brain-region-dependent presynaptic inhibition of excitatory neurotransmission. The selective metabotropic i |D showsDia higher glutamate receptor (mGlu2/3 )NagonistRLY354740 potency for inhibiting excitation in the basolateral (BLA)– centrolateral amygdala (CeA) projections (EC50 of 2 nM) than in the medial perforant pathway between the cortex and hippocampus (EC50 of 334 nM) or medial prefrontal cortex synapses (EC50 of 249 nM). Data adapted from REFS 106,107,108.

MORRIS WATER MAZE

A learning task in which an animal is placed in a pool filled with opaque water and has to use spatial cues to find a hidden platform that is placed at a constant position. DELAYED MATCHING TO POSITION TEST

Rats are trained to press a lever for food rewards, then trained to make either a matching or non-matching response in a task. The task consists of an illuminated lever being presented and retracted (sample stage) followed by a nose poke into the food tray, delivering two levers. For a correct matching response the animal presses the previous (sample stage) lever for a food reward; for a correct nonmatching response the animal presses the other (non-sample stage) lever for a food reward.

136

reverses the anxiolytic effects of serotonin compounds38,39, which do not interact with the GABAA receptor complex. In any case, the mechanism of action of LY354740 is distinct from benzodiazepine, because it can be selectively blocked by the mGlu2/3 receptor antagonist LY341495 (FIG. 4). It therefore seems likely that a number of neurotransmitter systems might converge to influence anxiety states, and that anxiety per se might ultimately arise from a shift towards neuronal hyperexcitability through decreased GABA, or increased glutamate, transmission40, with multiple mechanisms of regulation including monoamines and neuropeptides. In accordance with this hypothesis, the autonomic and behavioural panic responses that are produced by lactate infusion in rats chronically devoid of dorsomedial hypothalamic (DMH) GABA can be reversed by direct infusion of iGlu receptor antagonists in the DMH41 or by systemic administration of LY354740 42. In the latter case, LY354740 probably suppresses glutamate release from excitatory inputs to the DMH through presynaptic group II mGlu receptor activation14,27. A more precise mechanism(s) by which LY354740 acts as an anxiolytic in animal models has emerged from studies using early immediate gene expression as a marker for neuronal excitability. Animals subjected to the EPM show a characteristic induction of c-Fos in several regions of the hippocampus, including the CAl, CA3 and dentate gyrus, and the central nucleus of the amygdala. Pretreatment with LY354740 prevents the c-Fos induction in the hippocampus regions, but does not modify the EPM-induced c-Fos induction seen in the central nucleus of the amygdala. Indeed, LY354740 treatment alone elevated c-Fos in the central nucleus of the amygdala43. Interestingly, anxiolytics

| FEBRUARY 2005 | VOLUME 4

such as benzodiazepines and ethanol similarly induce neuronal activity in the central nucleus of the amygdala44–47, and it has been suggested that inhibitory GABAergic neurons in the central nucleus regulate excitatory output from the medial amygdala48–50. The induction of c-Fos in the central amygdala and inhibition of stress-induced hippocampal c-Fos might therefore be integral to anxiolytic activity in animal models (FIG. 5). It has been noted that LY354740 does not produce ataxia or suppress motor coordination, effects that are typically associated with benzodiazepine treatment. Moreover, LY354740 has also been shown to display a disparate profile with respect to other secondary benzodiazepine pharmacology in some tests. For instance, administration of the CNS depressant hexobarbitol is known to have sedative/hypnotic effects caused by augmenting GABAergic transmission51, and benzodiazepines dose-dependently increase sleep time produced by hexobarbitol when these compounds are coadministered to animals. However, LY354740 did not exacerbate the effects of hexobarbitol30, indicating a mechanism of action that might be devoid of interaction with other CNS depressants or excessive influence on GABAergic transmission. In another report, anxiety states elicited by withdrawal from subchronic nicotine or diazepam administration were evaluated by acoustic startle responses. In this model, the startle amplitude of a subject in response to a series of intermittent acoustic stimuli is monitored, and anxiety associated with withdrawal from nicotine or diazepam exposure is expressed as an enhanced response to loud noise when challenged with the respective treatment compound. LY354740 (administered as the racemic mixture LY314582) attenuated enhanced startle response for up to 4 days in animals that were withdrawn from subchronic nicotine or diazepam administration. Moreover, withdrawal from subchronic exposure to LY354740 itself produced no alterations in subsequent startle responding to challenge with the mGlu2/3 agonist52. These data indicate that agonists of group II mGlu receptors are effective at reducing anxiety states associated with drug withdrawal without producing unwanted side effects, physical dependence or abuse liability in and of themselves. However, there have been reports from preclinical animal models that LY354740 disrupts memory processing. For instance, Higgins and co-workers53 demonstrated a disruption in performance in the MORRIS WATER MAZE and a delay-dependent deficit in matching and non-matching to position operant paradigm. This response was not present in an mGlu2 knockout mouse. Furthermore, in a DELAYED MATCHING TO POSITION TEST the mGlu agonist dis2/3 rupted performance as well. By contrast, when tested at those same doses in a delayed alternation performance task LY354740 did not disrupt performance but rather reversed the disruption induced by phencyclidine54. Significantly, ketamine-induced cognitive impairment in normal human volunteers was prevented by LY354740, which produced results similar to those seen in preclinical models. The role of group II receptors in

www.nature.com/reviews/drugdisc

REVIEWS

Table 3 | Group-I-selective compounds in animal models of anxiety Model

mGlu ligand

Pharmacology

Route of administration; species

Effect

References

Acoustic startle

ACPD MPEP

Group I/II agonist mGlu5 antagonist

Intra-amygdala; rat I.p.; rat

Prevented habituation of acoustic startle Blocked acquisition/expression of fear-potentiated startle

74 84

Fear-conditioned freezing (auditory and contextual)

MPEP

mGlu5 antagonist

Intra-amygdala; rat consolidation or expression

Blocked acquisition but not of fear-conditioned freezing

86

Elevated plus maze

MPEP MPEP

mGlu5 antagonist mGlu5 antagonist

I.p., oral; rat Oral; rat

Increased maze open-arm time Increased open-arm time

78 79

Conflict tests

MPEP MPEP MPEP

mGlu5 antagonist mGlu5 antagonist mGlu5 antagonist

I.p.; rat I.p.; mouse Oral; rat

Increased shocks accepted in Vogel test Increased punished crossings in four-plate test Increased punished responding in Geller-Seifter test

78 78 79

Stress-induced hyperthermia

MPEP

mGlu5 antagonist

Oral; rat

Reduced stress-induced hyperthermia

Other tests

MPEP MPEP DHPG CHPG ACPD LY393675 LY367385 MPEP

mGlu5 antagonist mGlu5 antagonist Group I agonist mGlu5 agonist Group I/II agonist Group I antagonist mGlu1 antagonist mGlu5 antagonist

Oral; mice Oral; rat I.c.v; rat I.c.v; rat I.c.v; rat I.c.v; rat I.c.v; rat I.c.v; rat

Decreased number of buried marbles Increased social interaction Increased plasma corticosterone levels Increased plasma corticosterone levels Increased plasma corticosterone levels Increased plasma corticosterone levels Increased plasma corticosterone levels Increased plasma corticosterone levels at high doses (>30 mg per kg)

79,80 79 79 81 81 81 81 81 81

Data were compiled from listed references. ACPD, (1S,3S)-1-aminocyclopentane-1,3-dicarboxylic acid; CHPG, (RS)-2-chloro-5-hydroxyphenylglycine; DHPG, (S)-3,5dihydroxyphenylglycine; i.c.v., intracerebroventricular; i.p., intraperitoneal; LY393675, 2-(S)-amino-2-(3-cis-carboxycyclobutyl)-3-(9H-thioxanthen-9-yl) propionic acid; LY367385, [(+)-2-methyl-4-carboxyphenylglycine]; mGlu, metabotropic glutamate receptor; MPEP, 2-methyl-6-(phenylethylnyl)-pyridine.

memory function therefore remains unclear, and will require careful examination as the appropriate tools (selective agents and/or receptor knockout animals) become available. Collectively, these findings indicate that group II mGlu compounds (that is, those active at mGlu2 and mGlu3) might function to maintain brain excitability in anxiety states by modulating excitatory neurotransmission at specific synapses within the CNS. Recent advances in the molecular biology of the GABAA receptor complex have indicated that the sedation and memory impairment linked to benzodiazepine treatment are attributable to actions on α1-containing GABA receptor complexes55. By contrast, the anxiolytic properties of benzodiazepines have been attributed to their interaction with receptors containing the α2-receptor subtype56. The α2-GABAA receptors display a distinct and robust distribution in cerebral cortex and hippocampus, where they are found on glutamatergic pyramidal cells57,58. Interestingly, LY354740 binding exhibits a strikingly similar overlap with α2-GABAA receptor binding in the rat brain26,59. These findings are indicative of a potential mechanism through which benzodiazepines and group II mGlu receptors might exert similar effects on CNS excitability. For example, facilitation of GABAergic transmission by stimulation of α2-containing receptors on pyramidal cells would result in hyperpolarization of these cells and a decrease in excitability of the major efferent projections from the cortex and hippocampus. Likewise, stimulation of presynaptic mGlu2/3 receptors in these crucial brain regions would result in suppression of excitatory input to pyramidal cells and therefore a reduction in excitatory outflow from these neurons14,27.

NATURE REVIEWS | DRUG DISCOVERY

One of several remaining questions concerns the relative significance of mGlu2 and mGlu3 receptors in the anxiolytic actions of the group II agonists. Addressing this question has been hampered by a relative lack of pharmacological tools to differentiate between the two receptors. Recently, however, reports describing allosteric modulators that act as potentiators for mGlu2 have emerged60–63. These potentiators are distinct from the mGlu2/3 agonists in several ways, including their selectivity (>1,000-fold selectivity for mGlu2 compared with the other seven mGlu receptors), their site of action (acting through modulatory sites on receptors that are distinct from the glutamate site), and their mode of action (enhancing the normal activation of mGlu2 versus continuous stimulation of the receptor)60–63. Three of these novel mGlu2-subtype-selective potentiators have shown efficacy in one of two separate preclinical models of anxiety, the fear-potentiated startle and stress-induced hyperthermia models (TABLE 2)61,62. In these particular preclinical anxiety models, modulation of the mGlu2 receptor alone seems to be sufficient to achieve efficacy. However, caution should be used in interpreting these results because there is evidence that mGlu3 receptors may have a role in the actions of the mGlu2/3 agonists in these or other models of anxiety/ stress. Specifically, the anxiolytic actions of LY354740 in the EPM test are absent in both mGlu2 and mGlu3 knockout mice131. Group I mGlu receptors in anxiety

In general, group I mGlu receptors (mGlu1/5) display reciprocal distributions within a given brain region; that is, whereas mGlu5 is expressed at high levels in ventral striatum, cortex and hippocampus, little or no mGlu1 is VOLUME 4 | FEBRUARY 2005 | 1 3 7

REVIEWS

a

b

Diazepam

Post-conditioning

Post-conditioning

500 400 300 200 100

*

*

0 –100 –200

LY354740 600

Difference in startle amplitude (Vmax)

Difference in startle amplitude (Vmax)

600

Control

0.3

0.6

500 400 300 200

0

1.0

Diazepam (mg per kg intraperitoneally)

c

*

*

100 Control

0.1

0.3

0.6

1.0

LY354740 (mg per kg per orally)

d 600

600

Pre-conditioning

500 400 300 200

* 100 0

Control

0.1

0.3

0.6

* 1.0

Diazepam (mg per kg intraperitoneally)

Difference in startle amplitude (Vmax)

Difference in startle amplitude (Vmax)

Pre-conditioning

500 400 300 200 100 0

Control

0.003

0.03

0.3

3.0

LY354740 (mg per kg per orally)

Figure 3 | Effects of diazepam and LY354740 in an animal model of anxiety: fear-induced potentiated startle. Male rats were dosed with 0.1 to 1.0 mg per kg diazepam intraperitoneally (a and c), or 0.003 to 3.0 mg per kg LY354740 per orally (b and d) and compared with the appropriate vehicle control group. Dosing occurred 30 minutes before the post-conditioning phase when animals had learned to associate a cue to impending foot shock, and therefore tested the effect on a fear response (a and b). Alternatively, drugs were tested in the pre-conditioning model by injecting diazepam or LY354740 before animal training/initial exposure to the cue-associated foot shock and tested the next day without treatment as a measurement of memory/learning deficits. Diazepam showed equal potency in both the post-conditioning (anxiolytic) and pre-conditioning (learning impairment) models. In contrast, LY354740 was as effective as diazepam for anxiolytic action, but at all doses tested did not impair animals in the pre-conditioning paradigm (that is, no learning impairment). Adapted from REF. 109.

VOGEL TEST

Also known as the conflict drinking test, this test uses waterdeprived rats that are punished with an electric shock when drinking. Anxiolytic drugs increase drinking behaviour that has been suppressed under threat of shock in this test.

138

found in cells of the rodent cortex or hippocampus64,65, although it has recently been suggested that primate prefrontal cortex pyrimidal cell dendrites do express mGlu166. Group I mGlu receptors are predominantly localized on postsynaptic neurons, just lateral to the postsynaptic density. However, there is some anatomical and biochemical evidence for presynaptic localization of mGlu1 receptors67–70 and, in non-human primates, mGlu5 receptors66. In rodents, mGlu5 receptors are generally located postsynaptically, and their activation has been shown to influence the excitability of postsynaptic neurons through modulation of a variety of ligandgated ion channel currents and intracellular signalling molecules28. A preponderance of evidence points to a role for mGlu5 receptor blockade in producing anxiolytic effects in rodents (TABLE 3). A role for group I mGlu receptors in anxiety states was first indicated by the finding that (S)-4-carboxy-3hydroxyphenylglycine (S-4C3HPG), an antagonist of group I mGlu receptors that also displays partial agonist properties at group II mGlu receptors71, produces anxiolytic effects in rats. Direct microinfusion of 4C3HPG into the CA1 region of the rat hippocampus was reported to produce an anticonflict effect in the 72 VOGEL TEST . The CA1 region of hippocampus contains relatively higher levels of mRNA for group I mGlu receptors73 than group II mGlu receptors73, and so it is

| FEBRUARY 2005 | VOLUME 4

tempting to speculate that 4C3HPG exerts its anxiolytic effects by blocking group I mGlu receptors, although involvement of activation of the group II receptors cannot be excluded by these studies. Another study evaluated the effects of a nonselective group I/II mGlu agonist on the acoustic startle response. Local microinjection of 1-aminocyclopentane-trans-1,3-dicarboxylic acid (transACPD) into the central nucleus of the amygdala prevented the normal habituation of acoustic startle that is typically observed in this model at a time point 4 hours after infusion74. This finding indicates that activation of mGlu receptors within the central nucleus of the amygdala might act to enhance the startle response and, therefore, mediate anxiogenic effects in this brain region. Given the results noted above, and subsequent studies using subtype-selective mGlu compounds such as LY354740 and MPEP (TABLES 2 and 3), the facilitation of acoustic startle response witnessed in this study is most likely due to the activation of mGlu5 receptors (see below and section on group II mGlu receptors). 2-Methyl-6-(phenylethynyl)pyradine (MPEP) (BOX 2) is a potent, systemically active antagonist of the mGlu5 receptor subtype75 that has recently been shown to possess anxiolytic properties in a number of animal models. For instance, intraperitoneal (i.p.) or oral doses of MPEP significantly increased the number of accepted shocks in the CONFLICT DRINKING TEST (1, 10 mg per kg

www.nature.com/reviews/drugdisc

REVIEWS

Box 2 | Discovery of MPEP, a non-competitive allosteric-site mGlu receptor antagonist Advances in automated OH high-throughput N Chemical optimization N screening technologies N N have enabled the evaluation of large MPEP SIB-1757 Highly potent and selective mGlu5-selective antagonist compound libraries with modest potency against recombinant metabotropic glutamate (mGlu) receptors mGlu expressed in nonS809 neuronal cells. This leadgeneration strategy led to the identification of SIBP655 1757, a relatively potent and highly selective A810 antagonist of mGlu5 K638 receptors127,128. Chemical Y659 optimization of SIB-1757 ultimately led to the identification of the phenylethynyl pyridine derivative MPEP, a highly potent and selective mGlu5 receptor antagonist75. Both SIB-1757 and MPEP display non-competitive kinetics in inhibiting mGlu5 receptor function through interaction with a binding site located within the transmembrane (TM)-spanning region. Mutational analysis of this region identified residues P655, S658 and A810 as crucial for high-affinity MPEP binding to human mGlu5 129. Additional residues identified by mutational analysis of the rat mGlu5 receptor include Y568, L743, T780, W784, F787 and Y791130. A homology model of the 7-TM domain of hmGlu5 was generated on the basis of the X-ray crystal structure of bovine rhodopsin. When docked in this model, MPEP (magenta above) is predicted to fit into a pocket formed primarily by residues from TM III (P655, S658 (not shown) and Y659) and TM VII (S809 and A810). K638 originates from TM II and might also contribute to the binding of ligands interacting at this site.

FOUR-PLATE TEST

A test in which the test animal is placed on a plate and allowed to explore, after which each time the animal passes from one plate to another a shock is delivered. Anxiolytic drugs reduce the fear of shock and increase the number of punished crossings. GELLER–SEIFTER TEST

An anxiety test in which animals learn to press a lever for food, after which food is only delivered under threat of shock. Anxiolytic drugs increase lever pressing that is normally suppressed by the threat of shock punishment.

i.p.76) and enhanced the time spent in the open arm on the EPM77,78 in rats, without impeding exploratory behaviour or motor coordination at these doses. MPEP also increased punished crossings in mice, similarly to diazepam, using the FOUR-PLATE TEST76 and increased punished responding in the GELLER–SEIFTER TEST in rats79. In another study, MPEP was shown to block the autonomic nervous system response to stress-induced hyperthermia. MPEP (1, 7.5, 15, 30 mg per kg, p.o.) significantly reversed the hyperthermic response associated with stress experienced during core temperature evaluation using a rectal probe in mice79,80. To this end, intracerebroventricular administration of selective ((RS)3,5-dihydroxyphenylglycine (DHPG): mGlu1/5; (RS)-2chloro-5-hydroxyphenylglycine (CHPG): mGlu5) and nonselective (1S,3R-ACPD, group I/II) group I mGlu receptor agonists have been shown to dose-dependently increase serum corticosterone levels in the rat, whereas selective group II mGlu receptor compounds did not produce this effect (although group I antagonists also elevated corticosterone levels81). These latter results indicate that group I mGlu receptors in the paraventricular nucleus of the hypothalamus might be directly linked to the autonomic response associated with stressful stimuli, possibly by acting on cells that contain corticotropin-releasing factor in this region82. Recent evidence has indicated that the central anxiolytic effects of group I mGlu receptor antagonists (and, in particular, MPEP) following systemic administration

NATURE REVIEWS | DRUG DISCOVERY

might be the product of mGlu5 receptor blockade within the hippocampus and/or amygdala. Riedel and colleagues83 have shown that fear conditioning in rats results in a transient, time-dependent shift in mGlu5 protein expression from the CA3 hippocampal region (1 day after last acquisition trial) to CA1 and the dentate gyrus (10 days after last acquisition trial). Interestingly, systemically administered MPEP blocked both the acquisition and expression of conditioned fear in the fearpotentiated startle model84, indicating that mGlu5 might have a crucial role in the acquisition and/or expression of fear-related short- and long-term memory. In support of this idea, mGlu5 receptor expression in CA3 returned to baseline levels 10 days post-acquisition, and CA3 activity has been implicated in early, but not late, phases of memory consolidation85. In contrast, Rodrigues and colleagues86 found that intra-amygdala injection of MPEP disrupted the acquisition but not the consolidation or expression of auditory or contextual fear-conditioned freezing in rats. In addition, MPEP was found to disrupt the thalamic–lateral amygdala long-term potentiation in rat brain slices86. Taken together, these results indicate that MPEP acts in a time- and sub-region-specific manner — disrupting acquisition of fear-conditioning through actions within the amygdala and/or hippocampus, but also blocking expression of fear-conditioned responses through blockade of hippocampal (and/or other brain regions than the amygdala) mGlu5 receptors. VOLUME 4 | FEBRUARY 2005 | 1 3 9

REVIEWS

Vehicle (n = 10)

LY341495 (0.6 mg per kg, n = 10)

***

40

*** *

Time (%)

30

20

10

0

Vehicle

LY354740 (15 mg per kg intraperitoneally)

Chlordiazepoxide (10 mg per kg intraperitoneally)

* p