Differential Expression of Metabotropic Glutamate Receptors 2 and 3 in Schizophrenia: A Mechanism for Antipsychotic Drug Action?

The GCP II antibody was provided by Drs. Joseph Neale and Tomasz Bzdega (Georgetown University, Washington, DC). Specific antibodies for mGluR2 and mGluR3 were obtained from Abcam (Cambridge, Mass.). The mGluR2 antibody (cat#ab52176) is an affinity-purified antibody produced against an epitope-specific immunogen, and its specificity has been tested by Western blot analysis and pre-adsorption studies. The mGluR3 antibody was generated using a synthetic human peptide predicted not to cross-react with mGluR2. Western blot, pre-adsorption, and coimmunoprecipitation studies using mGluR2 and mGluR3 chimeras have demonstrated specificity of this antibody for mGluR3 (13) . The antibody against beta tubulin (cat#MAB3408) was obtained from Millipore (Billerica, Mass.), and the horseradish peroxidase-coupled secondary antibodies were obtained from Vector Laboratories (Burlingame, Calif.).

To determine potential effects of chronic antipsychotic treatment on GCP II and mGluR3, Sprague-Dawley rats (Charles River, Wilmington, Mass.) were given a first-generation antipsychotic (haloperidol [N=10]), second-generation antipsychotic (risperidone [N=10]), or placebo (N=10) via drinking water continuously for 6 months as previously described (15) . The rats were then sacrificed, trunk blood collected, and plasma frozen at –20°C until analysis. Drug levels were measured in the laboratory of Dr. Thomas Cooper as previously described (15) . The rodent brains were rapidly removed, and the frontal cortex was dissected and frozen by immersion in –40°C isopentane.

The demographic variables of brain pH, postmortem interval, RNA integrity number, and age were compared between the two groups using t tests. Correlations between protein levels and age, RNA integrity number, and postmortem interval were run with a Spearman rank order correlation. The effect of diagnosis on each protein species was analyzed using a mixed-model analysis of variance (ANOVA), with diagnosis as the between-group factor and brain area (dorsolateral prefrontal cortex, temporal cortex, motor cortex) as the within-group factor. Significant findings were further analyzed using post hoc t tests. When a correlation was found between a demographic variable and protein levels, analysis of covariance (ANCOVA) was performed to correct for the confounding variable. All statistical tests were two-tailed, and p values <0.05 were considered statistically significant. Values outside of two standard deviations from the mean were considered outliers and therefore not included in the statistical analyses.

Four normal subjects (age: 32.5 [SD=9.7] years, postmortem interval: 19.5 [SD=5.3] hours, RNA integrity number: 8.35 [SD=0.61]) were used to determine the regional distribution of mGluR2 and mGluR3 in the brain ( Figure 1 ). High levels of mGluR2 expression were observed in the dorsolateral prefrontal cortex, anterior cingulate cortex, orbitofrontal cortex, parietal cortex, and occipital cortex. Lower levels of expression were detected in the caudate nucleus, nucleus accumbens, and cerebellum, with the lowest levels in the thalamus. The hippocampus expressed relatively high levels of mGluR2. The expression of mGluR3, in general, was more evenly distributed within cortical regions, and the hippocampus expressed the highest levels of mGluR3. In contrast to mGluR2, high levels of mGluR3 were seen in the caudate nucleus and nucleus accumbens, with moderate levels in the thalamus and cerebellum. As a result of the relatively high expression of both mGluR2 and mGluR3 in the cortex, we selected two cortical regions consistently implicated in schizophrenia (dorsolateral prefrontal cortex and temporal cortex) to determine differences in protein expression in the illness. The motor cortex was added as a putative control region.

Fifteen schizophrenia subjects and 15 normal comparison subjects were used in the analysis of GCP II, mGluR2, and mGluR3 protein expression in the dorsolateral prefrontal cortex, temporal cortex, and motor cortex. There were no significant differences between diagnostic groups with respect to age, postmortem interval, brain pH, or RNA integrity number. No significant correlations were found between mGluR2 or mGluR3 and any demographic variable (all r values between –0.13 and 0.33, all p values >0.18) except between mGluR2 and RNA integrity number in the dorsolateral prefrontal cortex (r=0.56, p=0.01). GCP II levels showed a significant correlation with postmortem interval in the dorsolateral prefrontal cortex (r=0.37, p=0.04) and brain pH (r=0.61, p<0.001) in the temporal cortex.

The question of whether differences in protein expression could be associated with chronic antipsychotic drug treatment was addressed in two different ways. In the first approach, rats were given two different antipsychotics (haloperidol [N=10] and risperidone [N=10]) continuously for 6 months, using doses that produce plasma levels in the human therapeutic range. Tissue from the frontal cortex was taken from the rats that had been receiving an antipsychotic and analyzed for GCP II and mGluR3 protein expression. Haloperidol- and risperidone plus risperidone-9OH levels in plasma from trunk blood collected at the time of decapitation were 6.8 (SD=1.1) ng/ml and 6.2 (SD=3.2) ng/ml, respectively, demonstrating adequate dosing in these animals. There was no effect of antipsychotic treatment on GCP II (F=1.36, df=1, 27, p=0.3) or mGluR3 protein levels (F=1.48, df=1, 25, p=0.20) in the frontal cortex relative to comparison tissue ( Figure 3 ). In addition, differences in protein levels among schizophrenia subjects who were receiving antipsychotic medication at the time of death (N=10) were contrasted with that of schizophrenia subjects who were not receiving antipsychotic medication at the time of death (N=5). Although the number of schizophrenia subjects who had not been receiving an antipsychotic was low, the results support the lack of an antipsychotic effect on these proteins in the prefrontal cortex (GCP II: t=1.62, df=1, 13, p=0.10; mGluR3: t=0.66, df=1, 10, p=0.50 [see the data supplement accompanying the online version of this article]).

There are several prominent findings in the present study. First, the changes in mGluR3 and GCP II protein expression in schizophrenia were localized to the dorsolateral prefrontal cortex (Brodmann’s area 9), a brain region repeatedly implicated in this illness (19, 20) . Many studies have shown that volunteers with schizophrenia perform poorly on working memory tasks and show deficits in dorsolateral prefrontal cortex functional activation (19, 20) . Moreover, molecular studies have identified several replicated alterations in the dorsolateral prefrontal cortex (21) . Our findings are consistent with the localization of molecular alterations to this brain region and extend previous knowledge to the mGluR system. Second, we found a reduction in mGluR3, not mGluR2, leading us to suggest that mGluR3 could be the specific molecular target of the novel mGluR2/mGluR3 agonist LY2140023, mediating its antipsychotic action. Further, both mGluR3 and GCP II are significantly changed in the same brain region, suggesting a potential association between these two proteins, which are already known to be functionally related. NAAG, an endogenous agonist of mGluR3, is the substrate of GCP II. Levels of NAAG are largely determined by GCP II activity, such that GCP II activity may be considered an indirect marker of NAAG levels. Increased GCP II levels indicate low NAAG concentrations (9) . As a result, we reason that NAAG levels are decreased in the dorsolateral prefrontal cortex in schizophrenia. These changes are consistent with phencyclidine studies showing that GCP II inhibition reduces phencyclidine-induced behaviors in laboratory animals (22) . Third, we have provided evidence in the present study that the changes in GCP II and mGluR3 expression are not likely to be secondary to antipsychotic treatment.

GCP II levels and activity in the prefrontal cortex of schizophrenia subjects have been previously examined but with inconsistent results. Some of these studies have reported no change in mRNA expression (23) . Some have reported a decrease in quantitative binding (24), while others have found reduced enzyme activity (25) . Although these studies examined GCP II in Brodmann’s area 46, in the present study, we used tissue from Brodmann’s area 9. In addition, we report increases in protein expression not in mRNA levels or enzyme activity. Of note, we found an influence of postmortem interval on GCP II levels, which is a variable that may influence results across laboratories. Further, disease heterogeneity in schizophrenia must always be considered.

Expression profiles of mGluR2 and mGluR3 in the dorsolateral prefrontal cortex of schizophrenia subjects have also been previously examined (11, 12, 18, 26, 27) but with inconsistent results. To the best of our knowledge, at the mRNA level, no changes in mGluR3 in the dorsolateral prefrontal cortex have been reported (18, 26), although alterations in expression of a splice variant may exist (27) . Crook et al. (11) did not find any change in the combined expression of mGluR2/mGluR3 protein in Brodmann’s area 46. However, a later study reported an increase in mGluR2/mGluR3 immunoreactivity in Brodmann’s area 46 but not in Brodmann’s area 9. Corti et al. (28) developed a mGluR3-specific antibody and tested it on Brodmann’s area 10 schizophrenia tissue but found no change in the total or monomeric form of mGluR3. However, they did find a decrease in the dimeric form of mGluR3. In the present study, we found a reduced expression of monomeric mGluR3 protein. The apparent discrepancy between the Corti et al. report and our data is not clear. Differences in brain regions analyzed (Brodmann’s area 9 versus Brodmann’s area 10), differences in tissue quality characteristics (such as age and postmortem interval), or differences in assays between laboratories may be contributory factors.

The lack of effect of diagnosis on mGluR2 expression reflects normal levels of mGluR2 protein and is consistent with our previous study, which showed unchanged levels of mGluR2 mRNA in the gray matter of the dorsolateral prefrontal cortex in schizophrenia (18), but differs from a recent report of a small cohort of seven pairs of medication-free schizophrenia subjects showing reduced mGluR2 mRNA (13) . In the present study, no differences exist between schizophrenia subjects who received drug treatment and those who did not receive drug treatment. However, as a result of the small sample size, further examination may be warranted. We did not find an effect of antipsychotic treatment on mGluR3 or GCP II protein expression, an outcome consistent with previous studies (27 – 29) .

The mGluR3 receptors are located at the presynaptic terminal, postsynaptic terminal, and glial cells (30) and have distinct functions at each site. Overall, activation of mGluR3 receptors has been linked to the modulation of a variety of physiological functions, including release of neurotransmitters, regulation of synaptic plasticity, and enhancing neuroprotection (9) . Several mGluR3-mediated effects may be of particular significance in schizophrenia.

Neuronal mGluR3 receptors modulate the glutamate and gamma-aminobutyric acid (GABA) systems, both of which are implicated in the pathophysiology of schizophrenia (9) . Presynaptic mGluR3 receptors inhibit the release of glutamate and GABA into the synapse, while postsynaptic mGluR3 receptors have been shown to modulate NMDA and GABA type A (GABA _A ) receptor function (9) . In the prefrontal cortex, the enhancement of NMDA receptor function by mGluR2/mGluR3 receptors (31), most likely mGluR3 (10), has been suggested as the physiological basis for reversal of phencyclidine-induced behaviors (10, 31) . Activation of mGluR3 receptors also regulates GABA _A receptor subunit expression, specifically α ₁, α ₅, α ₆, β ₂, γ ₂, and δ subunits ( 32 [unpublished data available upon request from Ghose S.]). In the schizophrenia prefrontal cortex, alterations in several GABA _A subunits, including α ₁, γ ₂, and δ, have been reported (33) . It is plausible that the mGluR3 decreases in the dorsolateral prefrontal cortex reported in the present study are related to the GABA _A abnormalities in schizophrenia.

Astrocytic mGluR3 receptor activation generates the formation of transforming growth factor β ₁ and transforming growth factor β ₂ (34), factors that are neuroprotective. The transforming growth factor β pathway proteins modulate synaptic plasticity (35) and glutamatergic function (36) and enhance dendritic growth and spine formation (37) . Moreover, activation of astrocytic mGluR3 receptors plays a role in modulating brain microcirculation (38) . Baslow et al. (38) suggested that NAAG activates astrocytic mGluR3 and triggers Ca ²⁺ currents that spread to astrocytic end-feet in contact with the vascular system, where a secondary release of vasoactive agents increases blood flow. As a result, we speculate that low NAAG levels are associated with the low levels of regional blood flow often reported in the prefrontal cortex in schizophrenia subjects (39) .

We propose a formulation of glutamatergic function in the prefrontal cortex of individuals with schizophrenia that recognizes the complex roles of mGluR3 in brain function and is consistent with a putative glutamate reduction in schizophrenia ( Figure 4 ). At the postsynaptic mGluR3 receptor, the reduction in protein would diminish glutamate signal through a loss of positive modulation at the NMDA receptor ( Figure 4 B). Moreover, at the glial mGluR3, reduced receptor activation would decrease transforming growth factor β release. The reduced level of this growth factor may be associated with the smaller size of prefrontal neurons and reduction in dendritic branching reported in schizophrenia. Further, reduced astrocytic mGluR3 activation can lead to the decrease in Ca ²⁺ signaling and subsequent reduction in local blood flow in the dorsolateral prefrontal cortex of individuals with schizophrenia demonstrated by numerous previous studies in schizophrenia (39) . In addition, the increase in GCP II on glial cells decreases NAAG levels through active degradation, further reducing mGluR3-mediated signaling.

We also propose that activation of mGluR3 by agonists such as LY2140023 would reverse the changes brought about by mGluR3 reduction in the prefrontal cortex of individuals with schizophrenia ( Figure 4 C). Although mGluR3 activation would further reduce NAAG release via presynaptic inhibition, it would generate an increased response at the postsynaptic mGluR3 receptor to potentiate NMDA function. The net effect would be an increase in NMDA-mediated signaling at the glutamate synapse. In addition, disease-associated GABA deficits mediated by low mGluR3 would also be reversed by LY2140023. In the glial compartment, mGluR3 activation would augment transforming growth factor β levels, reversing the potentially deleterious effects on neuronal morphology and function. Further, glial mGluR3 activation could restore local blood flow in the prefrontal cortex.

In summary, LY2140023 could 1) reverse mGluR3-mediated deficits in schizophrenia to increase function at the NMDA receptor, 2) “normalize” GABA function, 3) increase transforming growth factor β, and 4) restore dorsolateral prefrontal cortex perfusion. These interpretations provide a model of prefrontal cortical dysfunction in schizophrenia in which to conceptualize our data and postulate mechanisms for the action of the selective mGluR2/mGluR3 agonist.

The extent to which human postmortem tissue reflects in vivo conditions is always a question but is buttressed in the present study by the selection of high-quality tissue characteristics. Several parameters have been identified to mark tissue quality, such as RNA integrity number and postmortem interval (14) . The tissue used in the present study was of high quality as evaluated by these parameters. Moreover, the potential effects of antemortem antipsychotic treatment on gene expression products can be an important potential confound. Although the present study attempted to address the latter issue using two approaches, and both suggested no chronic medication effect, the possibility of a drug effect must always be considered. Additionally, one cannot exclude the possibility that these drugs have distinct effects in the human brain relative to the rodent brain. We examined protein levels in the gray matter of the cortical regions. The possibility of changes in the dorsolateral prefrontal cortex white matter (18) will need further evaluation. In addition, we are unable to comment on whether the mGluR3 change we found localizes to any particular receptor population (i.e., presynaptic, postsynaptic, or glial), nor can we draw conclusions about the dynamic regulation between GCP II and mGluR3.

In closing, we have provided evidence that NAAG-mediated neurotransmission at the mGluR3 receptor is disrupted in the dorsolateral prefrontal cortex in schizophrenia, based on human postmortem tissue measures of the proteins involved. The defects we have reported could be attenuated by mGluR3 agonists reversing the consequences of the protein changes, putatively ameliorating the symptoms of the illness. This leads us to speculate that this molecular target could mediate the therapeutic response to LY2140023, the first mGluR2/mGluR3 agonist with an antipsychotic action in schizophrenia (5) .