Additional related information may be found at:
Neuropsychopharmacology: The Fifth Generation of Progress

Back to Psychopharmacology - The Fourth Generation of Progress

Schizophrenia and Glutamate: An Update

Blynn Garland Bunney, Ph.D., William E. Bunney, Jr., M.D. and Arvid Carlsson


There is a growing body of evidence suggesting that alterations in neurotransmitter systems, possibly reflecting defects in early neurodevelopmental processes, may play an important role in the etiology of schizophrenia (36, 78). A current focus of research is on two major brain neurotransmitters, dopamine (DA) and glutamate, both of which may be altered in schizophrenia. Briefly, the dopamine hypothesis attributes hyperdopaminergic function as a possible cause of schizophrenia, whereas the glutamate hypothesis proposes a hypofunctional glutamate system. There is substantial evidence for both hypotheses, but some of the most impressive data are from observations that certain classes of drugs can produce schizophrenic-like symptoms in normals. Of all the compounds administered to man, the two classes of drugs that produce the symptoms most similar to those of schizophrenia are the dopamine agonists (e.g., amphetamine) and glutamate antagonists (e.g., PCP). In general, the glutamate antagonists produce somewhat more positive and negative symptoms than the DA agonists, which—when given acutely—fail to produce some of the core symptoms of schizophrenia, such as formal thought disorder and negative symptoms (33). Consequently, a great deal of research emphasis is being placed on the role of glutamate in schizophrenia. Nonetheless, to date, the current antipsychotics do not have significant glutamatergic activity, although promising research at the NMDA receptor site (e.g., the glycine transport inhibitors [5, 39]) may provide new clues for the development of future antipsychotics.

Formulation of the Glutamate Hypothesis

The origins of the glutamate hypothesis of schizophrenia probably can be traced to the observations that the dissociative anesthetic, phencyclidine (PCP, Sernyl) produced psychotomimetic symptoms in a subgroup of surgical patients (40). Anecdotal data compiled by Johnstone et al. (40) revealed that: (a) 7.5% (4/30) of the PCP-treated patients experienced drug-related hallucinations; (b) psychotomimetic effects occurred within 30 minutes of PCP administration (10 mg, intramuscular); and (c) young males were particularly vulnerable to the hallucinatory effects of PCP, especially at higher doses (greater than 20 mg i.v.) [40]. These PCP-related effects appeared to resemble symptoms of schizophrenia and attracted the attention of Luby et al. (54, 67) and others (19), who conducted clinical studies in research settings to determine the relationship between the PCP drug effects and the illness.

Effects of PCP in Normal Volunteers

Studies of PCP in normal volunteers (19,54) showed that subanesthetic doses of PCP (0.1 mg/kg) produced a wide range of schizophrenia-like symptoms including estrangement, loss of body boundaries, formal thought disorder, hallucinations, and psychosis, which were transient (the duration of the acute effects of PCP lasted an average of 1-1.5 hours with general malaise lasting for several hours). These observations led the investigators to propose PCP as a potential drug model for the study of schizophrenia (54).

Effects of PCP in Schizophrenic Patients

Further studies in schizophrenics provided interesting clues to the possible underlying etiology of schizophrenia. Luby et al. () administered subanesthetic doses of PCP (0.1 mg/kg) to four chronic schizophrenics (average length of illness was 9 years) and to one acute, mute, unresponsive catatonic patient. In the chronic schizophrenics, PCP dramatically exacerbated the schizophrenic symptoms (54,24)—"it was as though the acute, agitated phase of the illness had been reinstated" (24). The chronic patients became more assertive, hostile, and unmanageable, and these changes lasted not a few hours (as in normals) but from four to six weeks (55). In contrast, in the acute catatonic patient, PCP initially worsened the symptoms, but four days post-PCP the patient spontaneously emerged from his stuporous state and remembered feeling as though he was "returning to earth from another planet", an improvement that lasted for six weeks (54).

One of the early hypotheses of schizophrenia was that it was related to a sensory deprivation state, similar to the one that is produced by PCP (67). However, not until many years later was the action of PCP on glutamate receptors identified.

Glutamatergic Activity in Schizophrenia

CSF Glutamate

Kim et al. (45) were among the first investigators to propose a glutamate hypothesis of schizophrenia, based on observations that glutamate levels in cerebrospinal fluid (CSF) were decreased by as much as 50% in schizophrenic patients, compared with normal individuals. The hypothesis had two components: (a) schizophrenic patients have a deficiency in glutamatergic function and/or (b) an increase in DA function. Attempts at replicating these findings (45), however, were unsuccessful (27,48,64). Even with the advanced techniques available almost 15 years later (23), CSF levels were not found to be different in the schizophrenic patients. However, significant reductions in gamma-glutamylglutamine (a substance thought to be involved with glutamate uptake and/or glutamate release) were observed.



The cloning, sequencing, and gene expression studies of the glutamate receptor have resulted in the identification of multiple receptor subunits. Currently, there are two major classes of EAA receptors: inotropic (effectors are sodium, potassium and calcium) and metabotropic (effector is G-protein [Gq/11 and Gi/o]) receptors. The inotropic receptors are further subdivided into NMDA and non-NMDA receptors and three classes of respective subunits: (a) the NMDA receptors are divided into NMDAR1 (which has eight splice variants) and NMDAR2 (A-D) subunits, (b) the AMPA/quisqualate receptor subunits are designated as GluR1-GluR4, and (c) kainate receptor subunits consist of low-affinity receptors (GluR5-GluR7) and high-affinity subunits (KA1 and KA2). The metabotropic receptors have been identified as mGlu1-4.

Data from studies of the EAA receptor subunits are consistent with an alteration in schizophrenic postmortem tissue, compared with brain tissue from normal individuals. These and other studies evaluating glutamate, aspartate, and kainate are reviewed below.

Postmortem Studies

In general, an increase in glutamate receptors labeled by the neuroexcitatory agonists 3H-aspartate and 3H-kainate is found in frontal cortical areas of schizophrenics, compared with levels in controls. Specifically, increased 3H-aspartate (21) and 3H-kainate (21,73) binding in the orbital frontal cortex (21) and increased 3H-kainate binding are found in the medial frontal cortical regions of schizophrenics (63,77). In contrast, EAA receptors (3H-aspartate and 3H-kainate) in temporal lobe regions, including the hippocampus, generally did not differ from controls (21,46), although two studies by Kerwin et al. (42,43) reported decreased 3H-kainate binding in these regions.

Additional indicators of brain glutamatergic activity are reflected by alterations in neuropeptides and related enzymes. For example, NAAG (N-acetylaspartylglutamate) is a neuropeptide found in high concentrations in glutamatergic neurons. NAAG has several properties; it serves as a precursor or storage form for glutamate and is thought to act as a glutamate antagonist at the NMDA receptor site (78). High levels of NAAG are associated with hypoglutamatergic function. Results from 12 patients with undifferentiated chronic schizophrenia (11 normal controls and 6 neuroleptic-treated, non-schizophrenic controls) showed that NAAG levels were increased and glutamate levels were decreased in prefrontal and hippocampal regions in schizophrenics, compared with controls. Aspartate levels were also decreased in hippocampal regions of schizophrenics. Based on these and other data (e.g., NAALADase), the investigators concluded that schizophrenia may be associated with a hypofunction of glutamate activity (78).

Non-NMDA Receptors

The non-NMDA receptors may control brain development in the temporal regions (44) and therefore may play a role in neurodevelopmentally-related deficits associated with schizophrenia (1,4,36). As detailed below, the results from these studies (8,14,24,45) are mixed. Nonetheless, they may be relevant, especially in view of data from three (14,24,44) of the four studies suggesting that non-NMDA receptor subunits are decreased in schizophrenics.

Using nonspecific probes for the whole GluR1 family, Kerwin et al. (44) found reductions in mRNA GluR1 levels in CA3 regions in hippocampi of schizophrenics (45). Investigations by the same group with more specific probes (i.e., GluR1, GluR2, GluR6&7 and KA 1&2) showed that all of these were significantly reduced in schizophrenics (45). Similarly, Collinge and Curtis (14) observed a dramatic loss of mRNA AMPA receptors in the hippocampus of schizophrenics, compared with controls. Eastwood et al. (25), using in situ hybridization histochemistry with probes specific for two non-NMDA (AMPA) receptor subtypes (GluR1 and GluR2, showed that mRNAs for GluR1 and GluR2 were regionally decreased in dentate gyrus, CA4, CA3 and subiculum, while GluR2 mRNA was reduced in the parahippocampal gyrus. Breese et al. (8) measured antibodies to functional AMPA/kainate (GluR1-3) sites and to kainate binding sites (GluR5-7) in the hippocampi and cingulate cortex of schizophrenic subjects and non-psychotic controls (normals and subjects with a previous history of alcohol abuse). No significant differences were found for any of the GluR receptor subtypes in schizophrenics, compared with only the normal controls.

NMDA Receptors

Some evidence from animal studies suggests that specific components of the NMDA receptor unit (e.g., NMDAR2D) may be developmentally regulated (60). In one of the first studies of NMDA receptor subunits in schizophrenia, Akbarian et al. (3) presented data that suggested a regional deficit in NMDA subunit density in brain tissue of schizophrenics. The expression patterns of five NMDA receptor subunits (NMDAR1/NMDAR2 [A-D]) were measured in postmortem prefrontal, parieto-temporal and cerebellar cortical tissue of schizophrenics (N=15) and age-, gender- and autolysis time-matched controls (N=15). A significant 53% proportional increase in NMDAR2D subunit mRNA levels was found in prefrontal areas of schizophrenics, but no significant changes were observed in the expression patterns of the other NMDA receptor subunits. Other brain regions were similar to those in controls, and no medication effects were apparent. Furthermore, some data suggest that NMDAR2D may be necessary for the formation of specific neuronal connections that are, at least in part, mediated by NMDA receptors (53). Special kinetic properties of postsynaptic NMDA receptors containing the NR2D polypeptide may effect postsynaptic depolarization when presynaptic activity is reduced (60). These data support the hypothesis by Akbarian et al. (3) that an increase in NMDAR2D subunits could reflect a compensatory response to decreased function (hypofrontality) in prefrontal cortical regions of the schizophrenic brain.

Additional NMDA receptor sites of interest to schizophrenia research are those sites located within the ion channel of the NMDA receptor complex, particularly those relevant to PCP binding. The next section presents a description of the ion channel and the compounds that bind to the PCP site. The PCP site probably plays a significant, if not critical role in the production of psychotomimetic effects.

PCP-Receptor Site

Non-competitive antagonists such as PCP and PCP-like compounds bind to a site within the NMDA ion channel at the PCP receptor site (there is also a Mg+ site within the ion channel). Binding inside the channel is dependent on the state of the ion channel (open or closed). Non-competitive antagonists such as PCP and PCP-like compounds appear to bind only when the channel is open. In the presence of NMDA agonists, binding of PCP to the receptor site is enhanced. Data from MacDonald et al. (56) provide some clues as to why PCP may have more potent effects than other PCP-like compounds (e.g., ketamine). Basically, the hypothesis is that once a compound binds to the PCP site, the ion channel closes and the trapped molecules cannot escape until the channel reopens. The potency of the drug is determined by its relative rate of escape from the open ion channel. The rate of escape for PCP molecules is 10 times slower than for ketamine—which may account for the higher potency of PCP, since it remains in the channel for longer periods of time (56). However, since PCP and PCP-like compounds do not bind exclusively to the PCP receptor, sites such as other inotropic receptors, the metabotropic site, and the sigma site may also be involved in the production of psychotomimetic effects.

Binding studies in brain tissue from schizophrenics on PCP sites with PCP ligands (e.g., 3H-MK-801, 3H-TCP ligands) show significant differences between schizophrenics and to normals. The regional distribution of NMDA receptors (labeled with 3H-MK-801) in normals is the highest in frontal cortex, followed, in descending order, by entorhinal cortex, hippocampus and amygdala. Lower concentrations of 3H-MK-801 receptor sites are found in the putamen and thalamus, while the fewest number of 3H-MK-801 receptors are in the substantia nigra and the nucleus dentatus (47). PCP sites (labeled with 3H-MK-801) are more abundant in schizophrenics in most regions, including frontal cortex, amygdala, entorhinal area, hippocampus and the putamen, in which the difference is statistically significant (44% increase in binding). In another study using 3H-TCP-labeled receptors, an increase in receptor density was reported in the orbital frontal cortex of schizophrenics, compared with controls (73). In contrast, Weissman et al. (83) failed to find a difference in 3H-TCP binding in frontal cortex, but did observe a reduction in 3H-TCP binding in occipital cortical regions of schizophrenics.

PCP-like Compounds and Psychoses

One of the major clues to the possible underlying deficits is the observation that most, if not all, of the compounds that bind to the PCP site and have been administered to humans are capable of producing psychosis. Drug-associated psychoses have been reported with PCP-like compounds including ketamine (6,70,15,49), dexoxadrol (52,72, 85) and MK-801 (Zukin, personal communication), in addition to PCP.

One of the more interesting aspects of the clinical effects of PCP is the wide variability in the duration of clinical symptoms elicited by the drug, which suggests that some individuals, particularly schizophrenics (54,55), may be more sensitive to PCP. Weissman et al. (84) described three stages of PCP effects (i.e, acute, prolonged and recurrent). The acute stage is most common. In the prolonged state, PCP effects last beyond detectable PCP plasma levels—perhaps for up to four weeks. The recurrent state may consist of PCP-induced changes which may not appear until up to two weeks post-PCP exposure, and symptoms may take an additional four weeks to remit. PCP urine and plasma levels in the recurrent state are usually undetectable. Schizophrenics in remission can experience recurrence of symptoms after single doses of PCP (84), or as described earlier, some patients may experience symptoms lasting from four to six weeks (54).

Ketamine-induced Psychoses


The risk of potentially serious adverse psychological events associated with PCP make it a less than ideal compound for study and precludes careful research investigation. The PCP-like compound, ketamine, is an alternative compound, and although it is less potent than PCP, it offers some advantages, including a shorter half-life than PCP and transient and reversible psychotomimetic effects that can be observed at acute, subanesthetic doses. Earlier, uncontrolled studies documented the psychotomimetic effects of ketamine (6, 15, 70). However, in the past two years, with the use of standardized rating instruments, three groups have used double-blind methodology in normal volunteers to gain a better understanding of ketamine (32, 49, 57). In a randomized, double-blind placebo-controlled study, Krystal et al. (50) administered ketamine hydrochloride (0.1 or 0.5 mg/kg) to a group of 19 healthy volunteers. Ketamine produced positive and negative symptoms (BPRS), in a dose-dependent manner, that have been associated with schizophrenia. The higher dose of ketamine (0.5 mg/kg) elicited significant perceptual effects, including alterations of body, environment, and time perceptions, although illusions rather than hallucinations were reported. In another double-blind, placebo-controlled study, Malhotra et al. (57) found that in healthy volunteers ketamine produced a brief psychosis marked by thought disorder and withdrawal-retardation. Hartvig et al. (32) in a double-blind, crossover design study, reported similar mood altering effects of ketamine in five volunteers.


Lahti et al. (51) studied the effects of ketamine in schizophrenic patients in a double-blind, placebo-controlled investigation. Using a within-patient design, patients were administered varying doses (0.1, 0.3, or 0.5 mg/kg) of ketamine. Results indicated that, for patients maintained on haloperidol (N=9), ketamine induced a brief (less than 30 minutes), dose-related worsening of BPRS positive (but not negative) symptoms. In contrast, in six patients retested off-medications (four week washout from haloperidol), there was only a slight exacerbation of BPRS symptoms (lasting less than 20 minutes), which did not reach significance. These results suggest that haloperidol-treated patients may actually be more susceptible to the effects of ketamine.

Lahti et al. (51) also provided other clues relevant to the schizophrenic process. Significantly, the ketamine-related symptoms were not novel and had been previously experienced by the patients independent of medication. Second, almost half of the patients (4/9) experienced a delayed or prolonged (8-24 hours) psychotic reaction, which suggests that schizophrenics may be particularly vulnerable to the effects of non-competitive NMDA antagonists.

If schizophrenia is a hypoglutamatergic illness, the challenge to be addressed is whether the illness can be pharmacologically altered by potentiating glutamatergic activity at the NMDA receptor without producing toxic effects. Attempts to treat NMDA-related deficits in schizophrenia involve the administration of compounds acting as agonists at the NMDA receptor site. Some major potential candidates for antipsychotic compounds are glycine and glycine-related compounds.


Glycine binds to a strychnine-insensitive site outside of the NMDA ion channel. As reviewed by Strous and Javitt (74), the opening of the unblocked NMDA channel is complex, and the channel is only able to function efficiently in the presence of glycine. More recent work has shown that glycine is an obligatory co-agonist at NMDA-R.

Measurements of glycine binding (3H-glycine) in postmortem tissue of schizophrenics indicate that glycine receptors are increased in brain (e.g., in sensory cortex and premotor cortex) [35]. Other data show that glycine levels are increased in mesial but not lateral regions of temporal lobe in schizophrenics, (81). These data are compatible with a compensatory increase in 3H-glycine receptors, possibly in response to reduced glutamate activity. There is, however, some evidence that neuroleptics may increase glycine levels (78).

Clinical Studies

Glycine and Glycine-like Compounds

Increasing NMDA function with glycine agonists may be a potential new strategy for the management of schizophrenia (22), particularly for the treatment of negative symptomatology (34,38). Since the conventional, DA-blocking antipsychotics are generally not effective for negative symptoms, augmentation of therapeutic effects with glycine and glycine-like compounds may be an effective treatment alternative.

An open study by Waziri et al. (80) was one of the first to report that glycine has antipsychotic properties. Eleven schizophrenic patients were given high doses (5–25 g/day) of glycine in addition to treatment with typical neuroleptics. Clinical improvement was observed in 4/11 patients and was maintained after the neuroleptics were withdrawn. An open study by Costa et al. (16) reported improvement in two of six chronic schizophrenic patients treated with glycine (15 g/day) and typical neuroleptics. In another open investigation, Rosse et al. (69) administered glycine (10.8 g/day) to six chronic schizophrenics and observed improvement in BPRS and SANS ratings in three of the patients. Potkin et al. (66) reported improvement in two of 11 patients maintained on neuroleptics and treated with glycine (15 g/day) in a double-blind, placebo-controlled study. Heresco-Levy et al. (34) administered high-dose glycine (0.8 g/kg) as adjuvant therapy in a double-blind, placebo-controlled crossover trial. Improvement in negative symptoms (PANSS) was observed only after six weeks of glycine treatment in 8/11 patients; improvement was maintained after glycine was withdrawn. Measurements of glycine levels revealed that those patients with the lowest pretreatment levels showed the most improvement in negative symptoms. One of the most convincing pieces of data for the antipsychotic effects of glycine is from a double-blind, placebo-controlled study by Javitt et al. (38) which showed that high doses of glycine (30 g/day) improved negative symptoms in all seven of the neuroleptic-treated schizophrenics in the study.

Other glycine-related compounds administered to schizophrenics include milacemide, an acylated "prodrug" of glycine that is converted into glycine in the brain (68), and D-cycloserine, a partial agonist at the glycine site (13,28). Open studies of "low" and "high" dose milacemide by Rosse et al. (69) showed no significant therapeutic effects of the drug. Data from an increasing number of studies suggest that D-cycloserine may be effective in a narrow dose range (e.g., 50–100 mg) in the treatment of negative symptoms (28,29,18,79). However, one open study by Cascella et al. (13) used high doses of milacemide (250 mg), which produced improvement in 1/7 patients and clinical worsening in 4/7 patients. Dramatic improvement in negative symptoms was seen in five of nine patients in a dose-finding trial (28). A more recent study (29) of 50 schizophrenic outpatients in an eight week, placebo-controlled trial showed that SANS scores were reduced by 20% in the D-cycloserine-treated group, specifically on the subscales measuring flat affect and anhedonia. DeSouza et al. (18) similarly observed an improvement in negative symptoms after treatment with D-cycloserine. Finally, a single-blind study in drug-free schizophrenics showed that D-cycloserine-treated patients had significant reductions in negative symptoms (79).

Glycine Transport Inhibitors

The identification of glycine transport inhibitors provided a new and exciting approach to the development of alternative antipsychotic compounds (5). Two classes of glycine transporters have been cloned and identified. These are GlyT1 and related subunits (GlyT1a-c), which are present in forebrain and colocalize with NMDA-R, and Gly T2, found in spinal cord and brain stem, and which colocalizes with strychnine-sensitive glycine receptors. The glycine transporters are high-affinity and regulate extracellular glycine. Inhibitors to the Gly transporters have been identified. Of these, the most potent is TxRx 5311 which is active at the GlyT1c receptors. In vivo experiments in rats showed that the GlyT1 transport inhibitors decreased PCP-induced increases in locomotor activity. Research data suggests that these compounds are selective for NMDA-R channel activity, the site where PCP is active (5). These substrates may be more effective than glycine in the treatment of schizophrenia (39). Since schizophrenic patients with low pretreatment glycine levels are the most likely to respond to glycine agonists (34), this subgroup of patients may be particularly good responders to treatment with glycine transport inhibitors.

Brain Imaging Studies of NMDA Receptors

The use of brain imaging technology to investigate the effects of PCP provides clues to the acute and chronic regional effects of the drug on brain metabolism. Data from receptor binding studies in animals show that limbic regions (especially the hippocampus) contain high concentrations of NMDA receptors (47,59). Although it may be supposed that PCP would reduce brain metabolism (especially in frontal brain regions to mimic the hypofrontality seen in schizophrenics) [9,76], studies in animals administered PCP and PCP-like compounds show that PCP increased regional glucose metabolism, specifically in hippocampal regions (61,65,75,82). Ketamine, a PCP-like compound, was similarly shown in five of six studies to increase glucose metabolism in hippocampal regions in animals (17,20,31,62,71). However, schizophrenia is a chronic illness associated with hypothesized reductions in glutamate activity and hypofrontality (10). A chronic PCP study by Gao et al. (26) in rats showed that glucose metabolism was decreased in limbic regions. Lahti et al. (50) administered ketamine (0.3 mg/kg, i.v.) to a group of schizophrenics stabilized on haloperidol and measured metabolic activity with H215O and PET. In these patients, ketamine increased cerebral blood flow in the anterior cingulate cortex and in an area which extended inferiorly to the medial wall of the prefrontal cortex. In contrast, blood flow to the hippocampus and the primary visual cortex was decreased (50).

Glutamate and Possible Filtering Deficits in Schizophrenia

One of the major deficits in schizophrenia, documented by a number of investigators over the years, is the general inability of patients to filter incoming external sensory information efficiently (11). Historically, sensory gating deficits are reported to affect primary auditory and visual sensory modalities (54,67). The acoustic startle response is one of the models for the study of filtering deficits in schizophrenics. In this paradigm, a small auditory stimulus is presented at 80–120 ms prior to a larger stimulus pulse, and eyeblink responses are recorded. In normal individuals, an inhibition of eyeblink responses follows the second intensified pulse of sound. In schizophrenics, however, the inhibitory response to the second auditory stimulus is significantly diminished, indicating a lack of inhibition (7!popup(ch114). Studies using electrophysiological recordings of evoked potentials in animals demonstrated that drugs which antagonize glutamate (PCP, MK-801, and ketamine) reliably produce startle deficits in animals (58), similar to the changes observed in gating studies of schizophrenic patients (e.g., ref. 41).



A review of the glutamate data in schizophrenia provides evidence consistent with the observation that alterations in glutamate function may play a role in this disease. Data from postmortem brain tissue, in particular, suggests that receptor systems may be altered. The characterization, cloning and expression of the NMDA receptor subunits provide an opportunity for future investigations in the study of potential defects in schizophrenia. Finally, the identification of novel drug target sites may open up new alternative strategies for the next generation of antipsychotics.


published 2000