Introduction

As described in recent reviews (Tamminga 1998; Coyle et al. 2002; Harrison and Owen 2003; Konradi and Heckers 2003; Krystal et al. 2003; Moghaddam 2003), genetic linkage studies as well as postmortem and psychopharmacological findings strongly implicate glutamate neurotransmission in the pathophysiology of schizophrenia. In particular, reduced activation of the NMDA receptor subtype may play a major role in cognitive deficits of schizophrenia, because blocking active NMDA channels in healthy volunteers impairs cognitive functioning in a manner that is similar to the cognitive deficits of schizophrenia (Krystal et al. 1994; Adler et al. 1999). These findings suggest that targeting the NMDA receptor may be a feasible strategy for treating the cognitive symptoms of schizophrenia. In addition to these findings, and regardless of the validity of the notion that NMDA dysfunction is involved in schizophrenia, an elementary line of reasoning implicates glutamate-mediated neurotransmission as a primary therapeutic target for cognitive symptoms of schizophrenia. Specifically, abnormal cortical functioning is the most consistent finding in schizophrenia (Weinberger et al. 1986; Benes et al. 1992; Goldman-Rakic 1994; Andreasen et al. 1997; Gur et al. 2000). Although there may be considerable debate about which cortical area is most pertinent to the expression of which symptoms, there is little debate that disruption in cortical processing of information is involved in the expression of the cognitive deficits of schizophrenia. All cortical efferents and cortico-cortical connections are glutamatergic. Therefore, regardless of the nature and the etiology of a cortical dysfunction in schizophrenia, glutamatergic neurons are the pathway by which information abnormally processed in the cortex becomes expressed as aberrant behavior. This suggests that even if the primary pathology in schizophrenia did not involve the glutamate synapse, modulation of glutamate-mediated signal transduction might still provide a mechanism for ameliorating aberrant information transfer within and from cortex.

Options for pharmacotherapeutic approaches that target glutamate receptors

The idea that glutamate functions as a neurotransmitter was introduced in the 1960s (Curtis et al. 1960; Crawford and Curtis 1964). Although the majority of glutamate found in the brain is involved in intermediary metabolism and other non-neuronal functions, the neuronal pool of glutamate is the most prevalent of all neurotransmitter pools. Glutamate receptors mediate nearly half of synaptic transmission throughout the central nervous system (Hollmann and Heinemann 1994). Glutamate neurotransmission is involved in nearly all aspects of brain function, and direct or circumstantial evidence has implicated a role for abnormal glutamate neurotransmission in the etiology or pathophysiology of most neurological and psychiatric disorders. Therefore, there has been a great deal of interest in developing therapeutic strategies that can influence the function of glutamate receptors. However, because of the ubiquitous nature of the glutamate synapse, drugs that interfere with glutamate receptor function were expected to impact glutamatergic function throughout the central nervous system, resulting in an overt disruption in basic brain function. The discovery of a distinct class of glutamate receptors, called metabotropic glutamate (mGlu) receptors, however, has changed the classical view of glutamate receptors and has brought forth the opportunity of developing drugs that modify glutamate neurotransmission in a functionally selective manner (Conn and Pin 1997; Schoepp 2001; Schoepp and Conn 2002). Before the discovery of mGlu receptors, it was thought that glutamate exerts its physiological action through receptors that act directly as ion channels. These “ionotropic” receptors were classified into three broad subtypes according to their preferential agonists as the NMDA, kainate, and AMPA receptors (Hollmann and Heinemann 1994; Huntley et al. 1994). Binding of glutamate to these subclasses of receptors stimulates Ca2+ and Na+ entry into neurons through channels formed either by the receptor itself (as is the case with the NMDA receptor subtype) or by opening voltage-sensitive ion channels that are on the cell membrane. The ionotropic glutamate receptors are expressed by nearly all subtypes of neurons, and mediate fast excitatory neurotransmission throughout the brain. Thus, direct pharmacological manipulation of this group of receptors may produce a global disruption in brain function and produce profound side effects ranging from disruption of movement to impairment of attention and memory. Our understanding of glutamate-mediated neurotransmission was changed profoundly after the discovery of a novel receptor with high affinity for glutamate which, in contrast to fast-acting ionotropic receptors, activate second messenger systems through coupling with G-proteins (Sladeczek et al. 1985; Nicoletti et al. 1986; Sugiyama et al. 1987). The initial reports about these so called metabotropic receptors were followed by an explosion of findings about many other structurally and functionally related receptors (Conn and Pin 1997). Interestingly, the advances in molecular methodologies presented the field with a novel chronology of discoveries: in contrast to nearly all the other neurotransmitter receptors in the brain whereby specific ligands and transduction mechanisms were discovered long before the application of cloning technology allowed for identification of their structure, the amino acid sequence and structure of the metabotropic glutamate receptors were characterized before we knew anything about their functional characteristics or had identified ligands that targeted these receptors.

At least eight metabotropic glutamate receptors, termed mGlu1–8, have been cloned. These eight receptors have been classified into three groups (termed groups I–III) primarily based on sequence identity and transduction mechanisms (Table 1; for recent reviews, see De Blasi et al. 2001; Schoepp 2001). Several characteristics of metabotropic glutamate receptors that distinguish them from ionotropic receptors also make them important pharmacotherapeutic targets. First, unlike the ionotropic glutamate receptors that mediate fast synaptic neurotransmission, activation of metabotropic glutamate receptors modulates neuronal activity in a manner similar to neuromodulators such as dopamine and serotonin, which have been effective targets of psychoactive drugs for treatment of most psychiatric disorders (Conn and Pin 1997; Schoepp 2001). Second, the distribution and function of these receptors is highly diverse and heterogeneous. Different subclasses of mGlu receptors are localized differently at both regional and cellular levels. For example, mGlu2 and mGlu3 are found in high density in the cerebral cortex and limbic regions (Phillips et al. 2000; Tamaru et al. 2001), whereas the mGlu6 receptor is found almost exclusively in the retina (Vardi et al. 2000). In addition, different classes of metabotropic glutamate receptors are found on different neuronal and non-neuronal elements such postsynaptic or presynaptic membranes or glial cells. Thus, unlike ligands, that target the ionotropic glutamate receptors and produce non-discriminate excitation or inhibition of fast synaptic neurotransmission throughout the nervous system, heterogeneous distribution of various subtypes of metabotropic glutamate receptors with distinct functional and anatomical properties may allow for modification of glutamate neurotransmission in a functionally selective manner.

Table 1 Classification, function and localization of metabotropic glutamate receptors. AC adenyl cyclase; PLC phospholipase C
Full size table

Targeting metabotropic glutamate receptors to treat the cognitive symptoms of schizophrenia

As mentioned in the Introduction, several lines of evidence make a convincing case that the glutamatergic system may be disrupted in schizophrenia. First, antagonists of the NMDA subtype of glutamate receptors impair cognitive functioning in healthy volunteers in a manner that is very similar to the cognitive deficits observed in patients with schizophrenia (Krystal et al. 1994; Adler et al. 1998, 1999; Newcomer et al. 1999; Krystal et al. 2000). Second, most of the genes that are associated with increased vulnerability to develop schizophrenia express proteins that can, directly or indirectly, affect excitatory neurotransmission (Harrison and Owen 2003; Moghaddam 2003). Third, an accumulating body of postmortem data has found changes in schizophrenic brains that may have resulted from abnormal neurotransmission in the glutamate synapse (Harrison 1999; McCullumsmith and Meador-Woodruff 2002; Clinton et al. 2003). Finally, postmortem, imaging, and psychological lines of investigation consistently point to cortical regions as the primary site of dysfunction in schizophrenia (Goldman-Rakic 1994; Lewis and Anderson 1995; Weinberger et al. 2001), and the only neurons that can transmit abnormally processed information out of the cortex are glutamate-containing neurons. Thus, targeting glutamate neurotransmission is a plausible strategy for treating the cognitive symptoms of schizophrenia.

In general, the cognitive deficit-inducing properties of NMDA antagonists in healthy volunteers suggest that the primary glutamatergic abnormality in schizophrenia may be reduced NMDA function. Therapeutic strategies so far have focused mostly on enhancing NMDA receptor function through mechanisms that produce subtle positive modulation of this receptor. In particular, activation of the glycine/d-serine modulatory site on the NMDA receptor by increasing oral intake of d-serine or exogenous ligands such as d-cycloserine has shown promise in clinical trials (Coyle et al. 2002; Javitt 2002). Modulation of NMDA receptors may also be achieved by targeting metabotropic glutamate receptors (Marino and Conn 2002). Specifically, preclinical studies suggest that two subtypes of mGluRs have the potential of ameliorating cognitive deficits resulting from NMDA receptor dysfunction. These include mGlu5 receptors, which can directly modulate the function of NMDA channel, and the mGlu2/3 receptors, which regulate the release of glutamate (Fig. 1).

Fig. 1
figure 1

Activation of two types of metabotropic glutamate receptors may attenuate the effects of NMDA receptor function and have utility for treating the cognitive deficits of schizophrenia. These include the mGlu2/3 receptors, which can regulate the release of glutamate and mGlu5 which modulates the activity of NMDA receptors

Full size image

Activation of mGlu5 receptors leads to potentiation of NMDA-evoked responses in cortical and other neural tissues (Doherty et al. 1997; Pisani et al. 1997; Awad et al. 2000; Attucci et al. 2001; Mannaioni et al. 2001). The selective mGlu5 receptor antagonist 2-methyl-6-(phenylethynyl)-pyridine (MPEP) blocks NMDA-induced membrane depolarization in striatal and cortical neurons (Pisani et al. 1997; Attucci et al. 2001). These cellular studies suggest that manipulation of the mGlu5 receptors may influence behavioral effects of NMDA antagonists in vivo. In fact, several recent studies suggest that there is a synergistic interaction between NMDA and mGlu5 receptors at a behavioral level. Specifically, mGlu5 receptor antagonists enhance the effects of NMDA receptor antagonists on hyperlocomotion and disruption of prepulse inhibition (Henry et al. 2002; Kinney et al. 2002, 2003), and working memory (Homayoun et al. 2004). These findings suggest that activation of mGlu5 receptors may be an effective therapeutic strategy for ameliorating NMDA receptor deficiency. Unfortunately, because of the rapid rate of mGlu5 receptor desensitization, mGlu5 receptor agonists are not considered effective therapeutic targets. However, recent studies have identified allosteric sites on the mGlu5 receptor that positively modulate the function of mGlu5 receptors (Knoflach et al. 2001; O’Brien et al. 2003). Preclinical studies so far suggest that these ligands may have therapeutic potential (Marino et al. 2003). Thus, this class of allosteric modulators is expected to reduce the cognitive effects of NMDA antagonists and possibly be of potential of therapeutic value for treating the cognitive symptoms of schizophrenia.

Based on extensive preclinical and recent, albeit limited, clinical data, ligands that activate mGlu2 and/or mGlu3 receptors may also have therapeutic value for treating cognitive deficits of schizophrenia. These receptors are primarily distributed in forebrain regions such as cerebral cortex, caudate putamen, nucleus accumbens, amygdala, and hippocampus (Phillips et al. 2000; Tamaru et al. 2001) that have been implicated in most psychiatric and neurological disorders. They are localized in various combinations of presynaptic, postsynaptic, extrasynaptic, and glial localization. Several studies indicate that activation of this group of receptors “normalizes” glutamate release presumably via presynaptic receptors localized on excitatory terminals (Battaglia et al. 1997; Cartmell and Schoepp 2000). Preclinical studies suggest that agonists of mGlu2/3 receptors diminish the behavioral, including cognitive impairing, effects of NMDA antagonists (Moghaddam and Adams 1998). The mechanism for this normalization of behavior appears to be as follows. One of the secondary effects of systemic exposure to NMDA antagonists, that may be critical to the behavioral effects of these drugs and to the state of glutamate dysfunction in schizophrenia, is an increase in the efflux of glutamate in prefrontal cortex (Moghaddam et al. 1997). The activation of cortical glutamate release by NMDA antagonists probably results from disinhibition of GABAergic input to cortical glutamate neurons (Greene 2001), and would be expected to lead to over-activation of glutamate neurotransmission at non-NMDA receptors and disruption of cognitive functioning (Moghaddam et al. 1997; Lorrain et al. 2003). Blocking this secondary effect by reducing the enhanced release of glutamate in the frontal cortex may provide a feasible strategy to selectively target mechanisms that influence cognitive functioning. The best example of this pharmacological approach, which has been tested in preclinical and clinical models of schizophrenia, involves LY354740 an agonist of mGlu2/3 receptors, which, as alluded to above, autoregulate the release of glutamate. In rodents, agonists of mGlu2/3 receptors normalize the increase in glutamate overflow after NMDA antagonist treatment (Moghaddam and Adams 1998; Lorrain et al. 2003). More importantly, these drugs reduce many of the behavioral abnormalities produced by NMDA receptor antagonist treatment, including stereotypy (Moghaddam and Adams 1998; Cartmell et al. 1999) and working memory deficits (Moghaddam and Adams 1998). Positive allosteric modulators of mGlu2 receptors, which may be more feasible for long-term treatment than direct agonists, have recently been discovered (Johnson et al. 2003). Nonetheless, direct mGlu2/3 agonists have been successfully used in clinical trials for treatment for anxiety (Schoepp et al. 2003), and limited clinical trials with LY354740 in healthy volunteers treated with subanesthetic doses of the NMDA antagonist ketamine indicate that this strategy reverses some of the cognitive impairments induced by ketamine (Krystal et al. 2003). Considering the accumulated evidence suggesting that NMDA receptor dysfunction may be critical to etiology and pathophysiology of schizophrenia (Coyle et al. 2002; Harrison and Owen 2003; Moghaddam 2003), these basic and clinical findings would suggest that potentiating mGlu2/3 receptor function may be an effective treatment strategy for the cognitive symptoms of schizophrenia. Assessment of the effect of this class of drugs on the cognitive functioning in schizophrenia is hopefully not far in the future.