Update on the neurobiology of schizophrenia: a role for extracellular microdomains

Abstract

The glutamate system includes presynaptic glutamatergic terminals, complex post-synaptic densities found on diverse types of neurons expressing glutamate receptors, as well as glutamate transporters and enzymes that facilitate the glutamate/glutamine cycle. Abnormalities of this system have been implicated in schizophrenia based on an accumulating body of evidence from postmortem, imaging, and preclinical studies. However, recent work has suggested that astrocytes may have more than a bystander role in the synchronization of neuronal responses in the brain. Converging evidence suggests that extrasynaptic glutamate microdomains are formed by astrocytes and may facilitate neuroplasticity via the modulation of extra-synaptic glutamate receptors on neuronal membranes within these domains. In this article the authors propose that the composition and localization of protein complexes in glutamate microdomains is abnormal in schizophrenia, leading to pathological neuroplastic changes in the structure and function of glutamate circuits in this illness.

The study of the pathophysiology of schizophrenia has several complimentary directions, including imaging, behavioral, and genetics studies with living patients, studies of materials obtained from living patients including lymphoblastoid, fibroblast, and nasal epithelium derived cell lines, animal models of schizophreniform endophenotypes, and postmortem studies using brain tissues from afflicted patients. This review will primarily focus on postmortem studies of limbic circuits, and present the hypothesis that extracellular microdomains have a role in the pathophysiology of schizophrenia.

The syndrome of schizophrenia

Schizophrenia is a severe illness that afflicts over 2.4 million people in the USA and over 50 million people worldwide. In many ways, this is the most serious of all psychiatric illnesses: more hospital beds (psychiatric and medical combined) are filled by persons with schizophrenia than due to any other medical condition.¹ Schizophrenia is highly heterogeneous, and most cases are unique in their presentation, response to treatment, and clinical course. Schizophrenia is most properly considered to be a syndrome, and not a distinct disease entity, as its specific causes are unknown, and the diagnosis is based on the presence of a collection of signs and symptoms agreed upon by governing bodies and professional associations. There is no blood test, specific genetic marker, or cognitive battery that can predict who will develop this illness. The schizophrenia phenotype is characterized by the presence of specific clinical findings, often divided into positive, negative, and cognitive symptoms.¹ Positive symptoms include hallucinations, which are often auditory. Patients report that they hear voices that are clearly located outside of their heads, most often engaged in a running commentary on their thoughts and behaviors.^{2, 3} Other common positive symptoms include paranoid delusions and disorders of thought processes.^{2, 3} Much more debilitating are the negative symptoms, which are associated with the diminution of normal social behaviors, and include withdrawal, decreased speech, diminished eye contact, decreased or muted facial expression and vocal inflection, and diminished spontaneous movement.^{1, 4} Cognitive impairments in this illness include, but are not limited to, deficits in verbal fluency, executive function, and working memory.^5–9 Few individuals suffering from schizophrenia have all of these symptoms, but the persistence of several characteristic symptoms, like auditory hallucinations, must be present in order for someone to be diagnosed with this disorder.¹

Prevailing hypotheses of schizophrenia

For decades, schizophrenia research has focused on the dopamine hypothesis of schizophrenia, which postulates that dysregulated dopaminergic neurotransmission is a key feature of the pathophysiology of the illness. The dopamine hypothesis is based on the observation that antipsychotics block D₂ receptors, and antipsychotic affinity for these receptors highly correlates with the ability to ameliorate psychotic symptoms. Although numerous studies point to dopaminergic abnormalities in schizophrenia, dopamine dysfunction cannot completely account for all of the symptoms observed, since neuroleptics typically are effective only for the positive symptoms of the illness while negative symptoms and cognitive deficits are relatively refractory to treatment with typical antipsychotics.^{10, 11} Consequently, alternative hypotheses that may help explain the pathophysiology of schizophrenia have been sought.

While the dopaminergic neurotransmitter system was implicated due to the effects of antipsychotic drugs, this system does of course not act in isolation. Dopamine receptors are found throughout the brain where they modulate excitatory and inhibitory neurotransmission via G-protein signaling pathways. Blockade of dopamine receptors in corticolimbic circuits directly alters release of other neurotransmitters including glutamate and GABA. Not surprisingly, extensive postmortem studies have found changes in glutamatergic and GABA-ergic systems in this illness.^12–15 However, evidence for involvement of glutamate receptor dysfunction, in particular the NMDA-subtype glutamate receptor, suggests a prominent role for glutamatergic abnormalities. NMDA receptor antagonists (but not GABA receptor antagonists) can induce both the positive and negative symptoms of schizophrenia, including cognitive deficits.^{16, 17} Moreover, these compounds can exacerbate both positive and negative symptoms in schizophrenia.¹⁸ Chronic administration of PCP-like compounds induces a persistent psychotic symptomatology,¹⁴⁸ and reduces frontal lobe blood flow and glucose utilization, which is remarkably similar to the “hypofrontality” described in schizophrenia.¹⁹

Despite these observations, the complexity of schizophrenia is not readily explained by a static neurochemical model. The onset of schizophrenia is typically in late adolescence or early adulthood.¹³ The onset of positive and negative symptoms in a previously normally functioning person, coupled with a lifetime of waxing and waning symptoms, accompanied by the possibility of a steady decline in social, occupational, and cognitive functioning, has led to longitudinal models that take into account genetic and environmental factors. The neurodevelopmental hypothesis of schizophrenia suggests this disease is “a subtle disease process affecting critical circuits in the brain during early development and reaching full-blown consequences during adolescence or early adulthood”.²⁰ Data supporting this hypothesis include studies suggesting that schizophrenia is associated with late winter births in urban environments, as well as a number of other prenatal, perinatal, and postnatal events.^{20, 21} Finally, schizophrenia may also be considered a disorder of neuroplasticity. Plasticity refers to the ability of a system to effect reversible, long-term changes in response to stimuli. Molecular correlates of learning and memory, including long-term potentiation (LTP) and long-term depression (LTD), likely facilitate neuroplasticity in the brain; these processes are significantly impaired in severe mental illnesses, including schizophrenia.^22–24 Considering schizophrenia as a disorder of neuroplasticity allows for integration of neurochemical, developmental, and neurodegenerative hypotheses (and supporting data) under one flag. While neuroplasticity is mediated by a complex web of carefully balanced neurotransmitter systems and circuits, this review will focus on glutamate in schizophrenia for the following reasons: 1) glutamate transmission is a central component of LTP and LTD, and hence has a central role in plasticity; and 2) the effects of NMDA receptor antagonists strongly implicate abnormalities of glutamate in schizophrenia.

Biology of glutamate neurotransmission

Glutamate release and reuptake

The process of release, activity as a ligand, and reuptake of glutamate involves three distinct cell types: the astrocyte, the presynaptic neuron and the postsynaptic neuron ²⁵ (Figure 1). In the presynaptic neuron, glutamine can be converted to glutamate by the enzyme glutaminase, and packaged into vesicles by a family of vesicular glutamate transporters (VGLUT1-3) for release into the synapse.^{26, 27} Glutamate is released into the synapse and may occupy and activate ionotropic (NMDA, AMPA, and kainate) or metabotropic (mGluR1-8) glutamate receptors on both neurons and astrocytes.^{25, 28, 29} Glutamate is rapidly removed from the synapse by a family of plasma membrane excitatory amino acid transporters, localized to postsynaptic neurons and astrocytes.³⁰ Recovered glutamate may enter the TCA cycle via conversion to alpha-ketoglutarate by glutamate dehydrogenase, be converted to glutamine by glutamine synthetase and transported back into the synapse, or be released into the extracellular space by a cystine/glutamate antiporter. Glutamate may also be added to this cycle via synthesis of glutamate in a pathway involving pyruvate carboxylase and transaminases.^31–33 Finally, several families of novel glutamate receptor and transporter associated molecules regulate glutamate release and reuptake through intracellular signaling mechanisms.^34–37

Figure 1.

The glutamate synapse. Glutamate is released by presynaptic terminals and removed from the synaptic cleft by a family of excitatory amino acid transporters (EAATs). Glutamate is converted to glutamine, which is shuttled back to the presynaptic terminal, converted to glutamate, and packaged for release. This process is called the glutamate/glutamine cycle.

Glial glutamate transporters

The excitatory amino acid transporters (called EAATs) are expressed in the plasma membranes of neurons and glia throughout the brain in a region and cell specific manner.^{38, 39} EAATs mediate glutamate transport by an electrogenic exchange of 3 Na+, 1 H+, and 1 glutamate molecule into the cell and 1 K+ ion out of the cell, with the net inward movement of one positive charge.^{40, 41} EAATs are likely homomers comprised of 2–3 non-covalently linked subunits that have 6–10 transmembrane domains (Figure 1).⁴² The transporters have specific patterns of cellular localization: EAAT1 and EAAT2 have primarily been localized to astroglia. EAAT3-4 and EAAT5 are primarily localized to neurons and the retina, respectively.^25–27 In the prefrontal cortex, glial transporters (EAAT1 and EAAT2) are predominately expressed in discrete subsets of astrocytes which account for approximately 90% of synaptic glutamate reuptake.⁴³ The glial transporters are localized to perisynaptic processes facing the synaptic cleft.⁴⁴ In the rodent, activation of the promoters and expression of EAAT1 (called GLAST in the rodent) and EAAT2 (GLT-1) is generally non-overlapping.⁴³ In addition, the GLAST, but not GLT-1, promoter was activated and EAAT1 expressed in oligodendrocytes, suggesting that EAAT1 has a role in myelination and CNS connectivity.⁴³ The functional importance, perisynaptic localization, and heterogeneity of expression of the glial glutamate transporters suggests that examination of the expression and function of these molecules may be a high yield target for studies of neuropsychiatric illnesses that involve alterations in glutamate transmission.

EAAT expression, processing, and trafficking

EAATs are synthesized in the ER and have extensive posttranslational modification in the Golgi, including N-linked glycosidation of at least two sites which are important for homomultimer formation.⁴⁵ EAATs are then trafficked to the plasma membrane where localization and clustering are regulated by protein-protein interactions and phosphorylation.^46–50 Ultrastructural studies indicate that most EAAT1-2 expression is in the plasma membrane, suggesting that localization of the glial transporters to the plasma membrane is not regulated via mobilization of an intracellular pool of nascent transporter protein.⁵¹ EAATs may be removed from the plasma membrane via endocytosis and shuttle back to the cell surface via recycling endosomes, or be targeted for degradation in lysosomes.⁵²

Extracellular glutamate

Glutamate released at the terminal diffuses out of the synaptic cleft and may be transported into the astrocyte by excitatory amino acid transporters (EAATs) or spill over to extrasynaptic areas.^{53, 54} Extrasynaptic glutamate may also originate from astrocytes via vesicular release, the cystine/glutamate antiporter (aka system X_c-), or other less prominent mechanisms.^{53, 55, 56} The cystine/glutamate antiporter releases glutamate and transports cystine into the astrocyte for glutathione synthesis. Glutamate levels in the extracellular milieu are postulated to be tightly regulated, as activation of extrasynaptic glutamate receptors has potent effects. For example, activation of extrasynaptic NMDA receptors promotes initiation of NMDA spikes, while LTD and LTP can be readily induced in the adult cortex by activation of extrasynaptic NR2B containing NMDA receptors.^{57, 58}

Removal of glutamate from the synaptic cleft may be conceptualized as a two-step process, involving first the high affinity binding of glutamate by perisynaptic transporters, and second the transport of bound glutamate by the transporter across the plasma membrane.^{44, 59} Once bound, glutamate may be “unbound” or released, or alternatively, transported across the plasma membrane.^{44, 59} The “capture efficiency” of the EAATs is defined as the ratio of the rate of unbinding of glutamate to rate of transport, which is reported to be about 0.5.⁴⁴ The relatively low rate of transport of bound glutamate compared to the capture efficiency suggests that the EAATs first act as buffers for released glutamate.⁴⁴ Thus, glutamate molecules may bounce from one transporter binding site to another, until transported, limiting glutamate spillover from the synaptic cleft.

Glutamate spillover

The density of perisynaptic glutamate transporter protein, the amount of glutamate released, and the rate of glutamate transport determine, in part, the kinetics of glutamate diffusion away from the synaptic cleft. While several regions have well characterized glutamate spillover between excitatory synapses (cerebellum and hippocampus), there is ongoing debate regarding the extent of glutamate diffusion in other regions, including the frontal cortex, where spillover of glutamate may detrimentally lead to loss of input specificity and activation of cell death pathways.^60–66 Under physiologic conditions, release of glutamate may exceed the capacity of cortical synapses to remove glutamate from the cleft.^{67, 68} Mathematical models suggest that glutamate may diffuse and activate NMDA receptors within a radius of 0.5 μm from the release point.⁶⁹ Thus, the spatial arrangement of glutamate synapses, their glutamate transporter buffering zones, and extrasynaptic glutamate receptors will determine the extent and effect of glutamate spillover.^{67, 70, 71}

Extrasynaptic glutamate receptors

The G-protein linked metabotropic glutamate receptors (mGluR) have a central role in regulating synaptic glutamate. mGluRs are expressed perisynaptically, and activation of mGluR2/3 receptors decreases presynaptic glutamate release.⁷² Thus, activation of mGluRs may serve as a brake on glutamate spillover, preserving input specificity by diminishing synaptic glutamate levels.⁷³ This mechanism has recently been exploited to develop a highly selective mGluR2 agonist that has putative efficacy as an antipsychotic.⁷⁴ This development suggests a role for glutamate spillover in the pathophysiology of schizophrenia and other illnesses where psychosis is a central feature.

Glutamatergic abnormalities in schizophrenia

A number of studies have evaluated glutamate neurotransmission is schizophrenia using different approaches. In this section we first discussed data from magnetic resonance spectroscopy (MRS), a technique which is largely focused on measuring glutamate, glutamine, and associated metabolites in specific brain regions of living patients. A strength of these studies is that the data are collected from afflicted individuals relatively close in time to the onset of the illness (and some studies look at the first break), while postmortem studies are examining brain tissues from more aged individuals who have had a lifetime of psychiatric illness. Next, we will briefly discuss data from postmortem studies, with a particular focus on glutamate receptors and transporters.

Magnetic resonance spectroscopy findings in schizophrenia

The balance of studies using MRS to examine glutamate, glutamine, n-acetylaspartate, and other metabolic intermediates have yielded mixed results.^75–79 While a couple of studies have found changes in glutamate in schizophrenia, one large meta-analysis only found decreases in the glutamate metabolite N-acetylaspartate (NAA) in the basal ganglia and frontal lobe. Changes in NAA levels suggest abnormalities of glutamate synthesis and or cycling in schizophrenia.⁸⁰ A different meta-analysis found decreased glutamate and decreased glutamine in the medial frontal cortex in schizophrenia, suggesting that glutamate neurotransmission is diminished in this illness.⁸¹ One interesting finding from these studies is the loss of correlation between NAA and glutamate levels in subjects with schizophrenia, compared to disease-free control subjects.^{77, 82} Taken together with the meta-analyses, these data suggest a significant abnormality in the glutamate/glutamine cycle in limbic circuits in schizophrenia. One limitation of the MRS approach is that it measures all glutamate, glutamine or NAA, without regard for it being intra or extracellular. For example, there may be a global increase in glutamine in the anterior cingulate cortex, with a large increase in intracellular pools, and a small decrease in extracellular levels.

Abnormalities of glutamatergic enzymes in schizophrenia

There are several key enzymes involved in the glutamate/glutamine cycle as well as the synthesis or break-down of glutamate (Figure 1). Changes in enzymes levels may impact the amount of glutamate available for release from neurons and glial cells. Several studies have found decreased expression of carboxypeptidase II (binding and activity), glutaminase (mRNA), and glutamaine synthetase (mRNA and protein) in limbic regions in schizophrenia.^83–86 Other studies have found increases in glutaminase (mRNA and activity).^{83, 87} While these data support the hypothesis that glutamate synthesis and cycling may be impaired in schizophrenia, all of these studies were done at the regional level, and thus fail to capture the complexity of glutamate synapses at the cellular or subcellular level. For example, there may be diminished expression of glutamate enzymes in astrocytes, but increased expression in pyramidal neurons. Finally, one of the most interesting findings is a decrease in the dipeptidase glutamate carboxypeptidase II (GCP II; also known as NAALADase) activity in the frontal cortex and hippocampus in schizophrenia. GCPII catabolizes N-acetylaspartyl glutamate (NAAG) to glutamate and NAA.⁸⁸ These findings are consistent with the MRS data discussed above which found decreased levels of NAA. NAAG antagonizes NMDA receptors, and increased levels (secondary to diminished GCP II activity) might contribute to NMDA receptor hypofunction. One strength of this study is that the authors measured enzyme activity, and not just transcript or protein levels, a technically demanding approach.⁸⁹

Glutamate receptor abnormalities in schizophrenia

The observation that phencyclidine may cause a schizophreniform psychosis in persons without a prior diagnosis of schizophrenia led to investigation of ionotropic glutamate receptor expression in schizophrenia. Initial hypotheses were focused on the idea that a loss or hypofunction of NMDA receptor activity would be reflected by diminished expression of NMDA receptor subunits as well as NMDA receptor binding sites. However, on balance, studies of NMDA receptor expression in the postmortem brain in schizophrenia have no clear or consistent pattern of findings ⁹⁰. For example, there are over 18 studies of NMDA receptor subunit expression in the frontal cortex alone, and other than some changes in binding site expression, the hypothesis that there is deficient NMDA receptor expression stands largely unproven.⁹⁰ Similar to NMDA receptors, AMPA and kainate receptor studies generally do not have a consistent pattern of abnormalities other than perhaps changes in AMPA receptor GluA2 subunit expression in the hippocampus.^{17, 91, 92} Interestingly, administration of PCP, which blocks the NMDA receptor channel, leads to increased glutamate release, which may lead to spillover of glutamate from the synaptic cleft to extrasynaptic areas, activating extrasynaptic (non-NMDA) glutamate receptors.

Metabotropic receptor expression in schizophrenia

While there are fewer postmortem studies of metabotropic glutamate receptors, compared to ionotropic receptors, the data are no less contradictory. For example, mGluR3 protein expression has been reported as increased, decreased and unchanged in the frontal cortex.^93–96 Genetic linkage studies suggest that mGluR5 is involved in schizophrenia, and increased mGluR5 and mGluR1 transcript, and mGluR1 protein expression have been found in prefrontal cortex in this illness.^{94, 97–99}

Abnormalities of glutamate transporters in schizophrenia

Several studies have reported region-level changes in the expression of the glial glutamate transporters EAAT1 and EAAT2 in schizophrenia. EAAT1 protein expression was decreased and EAAT1 glycosylation altered in the DLPFC.^{100, 101} In contrast to these protein studies, increased levels of EAAT1 mRNA were found in the anterior cingulate cortex (ACC) and thalamus,^{100, 102, 103} suggesting a compensatory response to diminished glutamate reuptake capacity. Alterations in EAAT2 mRNA have been reported in the hippocampus (decreased) and neocortex (increased, decreased and unchanged) in schizophrenia.^{98, 100, 102, 104–106} The neuronal transporters have also been studied. We have previously reported increased expression of EAAT3 protein and mRNA in the anterior congulate cortex, while other studies have measured EAAT3 mRNA expression in the frontal cortex (increased), DLPFC (no change) and striatum (decreased).^{100, 102, 106–109} These conflicting data for neuronal glutamate transporters mirror the findings of glutamate receptor subunit expression, and are limited by the likelihood that glutamate transporter expression changes may be cell-specific, and change in different directions in different populations of cells.

Taken together, these findings have led to reformulation of the NMDA receptor hypofunction hypothesis, with the idea that changes in glutamate receptor and/or transporter expression in schizophrenia is not a problem of too much or too little protein expression, but a problem with protein trafficking or signaling. For example, NMDA receptor signaling was diminished in postsynaptic densities isolated from subjects with schizophrenia, and NMDA and AMPA receptor subunit expression was altered in specific subcellular compartments, suggesting abnormalities of receptor trafficking to the cell surface.^110–112 These data have led to the most general question of whether or not there is an abnormality of receptor or transporter trafficking machinery in schizophrenia.

Glutamate receptor trafficking protein abnormalities

The glutamate receptors interact with numerous proteins that link the receptors to trafficking machinery, scaffolding proteins, and signaling patways. For example, the NMDA receptors specifically associate with PSD95, PSD93, SAP102, and NF-L, proteins which modulate NMDA receptor function by promoting clustering and anchoring at the synaptic membrane and regulating intracellular signaling.¹¹³ Our laboratory and others have found profound abnormalities of these proteins in thalamocortical regions in this illness.^{98, 114–120} For example, NF-L and SAP97 protein expression was decreased in the DLPFC, while the signaling molecule SynGAP was decreased in the ACC.^{116, 119–121}

Interestingly, while there is some overlap between AMPA and NMDA receptor interacting proteins, AMPA receptors have a rich trafficking biology, which includes de novo, recycling, and constituitive pools of receptors that may laterally translocate to and from the postsynaptic density, facilitating some types of LTP and LTD. Changes in AMPA receptor protein expression typically exceed changes seen for the receptor subunits themselves, and are more consistent in their valence, with a consistent trend for diminished expression of protein levels for trafficking molecules in schizophrenia.^{112, 116, 122–127}

EAAT interacting proteins in schizophrenia

Less well studied than the receptor interacting proteins, several glutamate transporter interacting molecules have been identified, including G-protein suppressor pathway 1 (GPS1), JWA, ARHGEF11, and KIAA0302 (also called beta III spectrin). These molecules can affect glutamate transport function through trafficking, anchoring, phosphorylation, glycosylation, and degradation of transporters in the brain. For example, G-protein suppressor pathway 1 (GPS1) decreases EAAT2 mediated glutamate reuptake through a direct protein-protein interaction, and levels of GPS1 protein were elevated in the frontal cortex in schizophrenia. These data suggest that there may be normal levels, but decreased activity, of a specific transporter due to modulation of transporter function or localization to the plasma membrane.¹⁰⁰

In summary, changes in glutamate neurotransmission in schizophrenia point towards complex abnormalities of protein localization and function, leading to alterations of the composition of postsynaptic densities and other related protein complexes that contribute to the molecular neuropathology that underlies the schizophrenia phenotype.

Glutamate spillover in schizophrenia

Glutamate spillover in schizophrenia

As outlined above, several postmortem studies have found changes in EAAT expression in schizophrenia, as well as changes in the molecules that regulate EAAT localization and activity.^{98, 100, 103, 104, 106, 107} In general, these changes in gene expression are consistent with diminished regional expression of glial glutamate transporter expression and activity. Further, a recent genetic study analyzing copy number variants reported a subject with schizophrenia with a deletion of several EAAT1 exons ¹²⁸. Finally, characterization of the complete GLAST (EAAT1) knockout found changes consistent with behavioral endophenotypes associated with schizophrenia, including locomotor hyperactivity and abnormal social behavior ^{129, 130}. Several of these abnormal behavioral findings were reversed by administration of antipsychotic medication or mGluR2/3 receptor agonist administration, which decreases presynaptic glutamate release. These data suggest that there are region specific deficits in EAAT reuptake capacity in schizophrenia which could lead to glutamate spillover.

Chronic glutamate spillover may lead to remodeling of synapses

In the PFC in schizophrenia, there are changes in the structure, composition, and numbers of excitatory synapses.^131–133 Increased packing density, decreased numbers of dendritic spines and diminished expression of structural proteins suggest significant alterations of synapses in this region.^134–136 Several reports have found specific alterations in layers III and IV of the PFC, including abnormalities of pyramidal cells and interneurons.^{131, 137–140} One well replicated finding is a decrease in parvalbumin positive inter-neurons in the middle cortical layers.^{137, 141} A lamina specific de cit in inhibitory tone could lead to increased release of glutamate, which combined with diminished reuptake capacity could lead to increase glutamate spillover.

Accumulating evidence from postmortem gene expression studies suggest neurochemical alterations consistent with spillover. We have found increased mGluR2/3 protein in the PFC,⁹⁴ which may be interpreted as an attempt by synapses to decrease spillover by decreasing presynaptic release. Expression of the cystine/glutamate antiporter (xCT) was also increased in the DLPFC.¹⁴² This molecule is expressed on glia and releases glutamate into the extrasynaptic space in exchange for uptake of cystine, which is required for glutathione synthesis. The effect of increased xCT protein expression on glutamate release is not known, because it is the activity, and not expression level, of this molecule that determines the rate of glutamate release. However, changes in the expression of xCT minimally suggest abnormalities of the regulation of extracellular glutamate levels.¹⁴² Finally, a number of studies have also described changes in ionotropic glutamate receptor binding site suggesting a change in NMDA and AMPA receptor stoichiometry in the frontal cortex in schizophrenia.^{124, 143–148} Interestingly, pre-clinical studies have shown that glutamate spillover is associated with alterations in ionotropic receptor subunit composition and function.^{145, 149}

According to us there is remodeling of glutamate synapses in schizophrenia secondary to glutamate spillover. Glutamate spillover may be secondary to increased release (in a misguided attempt to activate sick NMDA receptors), as well as deficits in glutamate reuptake capacity. In this setting, we would predict that perisynaptic localization of glutamate transporters is diminished, either as a primary de cit in transporter localization, as a compensation for increased extrasynaptic glutamate release, or both. Redistribution of glutamate transporters would lead or contribute to increased spillover, causing excitotoxicity and loss of input specificity. Further, we postulate that these deficits are initially relatively subtle, but chronic in nature, leading to inappropriate remodeling of excitatory synapses which do not function normally. This idea is supported by the phenotype of the GLAST/EAAT1 knockout mice, which have moderate cognitive and behavioral impairment, but no morbidity or mortality associated with seizures.^{129, 130, 150}

A role for extracellular glutamate microdomains

Glutamate microdomains

Critical to cellular function of proteins is subcellular locality or microenvironment, in which proteins cluster and interact with numerous others. These biologically and morphologically discrete microdomains, such as the PSD, require tightly regulated trafficking of component proteins and thus organize the intermolecular environment for proteins and their interactions.^{44, 151–155} Evidence for cortical glutamate microdomains is based on several discrete observations: 1) extracellular glutamate levels may vary between 0.2–7 mm in the extrasynaptic space in a region- and milieu- dependent manner; 2) a large body of work has described localization of functional glutamate receptors outside of synapses; 3) the localization and buffering/transport properties of glutamate transporters strongly suggests partitioning of non-synaptic extracellular spaces; 4) several recent studies have characterized the functional coupling of protein complexes, structural proteins, organelles, and signaling pathways that converge on cellular processes involving glutamate, and finally; 5) glial cells may form electronically-independent morphological structures that ensheath neuronal structures of unknown function.^{151, 156–159} Taken together, these data suggest that glutamate microdomains are formed by specialized protein clusters on the membranes of astrocytic processes apposed to extrasynaptic glutamate receptors expressed on specialized regions of neuronal membranes (Figure 2).^{151, 157} Diffusion of glutamate between domains or domains and synapses would be limited by the dense expression of glutamate transporters between these specialized structures (Figure 2).⁷¹

Figure 2.

Glutamate microdomains may be formed by specialized protein complexes found on plasma membranes in extrasynaptic regions where astrocytic and neuronal membranes are apposed to one another. Specific protein complexes that may help form these domains have been identified. One complex (1) comprised primarily of glutamate transporters and supporting molecules may limit diffusion of glutamate from the synaptic cleft, while another protein complex (2) may regulate extracellular glutamate levels in glutamate microenvironments (3) which in turn would modulate activation of extrasynaptic glutamate receptors on neurons and glia (4).

The composition and function of specialized protein clusters in astrocytes has recently been investigated in rodent brain tissues. Using immunoisolation, one study found a complex comprised of GLT-1 (the rodent isoform of EAAT2), Na+/K+ATPase, hexokinase, and mitochondria,¹⁵⁷ while another found Na+/K+ATPase, the water channel aquaporin 4, and mGluR5 (Figure 2).¹⁶⁰ Other studies have found functional coupling of glutamate reuptake, cytosolic and mitochondrial sodium exchange, and glucose utilization in astrocytes.^{161, 162} Interacting partners and ultrastructural localization have not been determined for the cystine/glutamate antiporter.

Evidence for abnormalities of glutamate microdomains in schizophrenia: structural abnormalities

In the frontal cortex in schizophrenia there are changes in volume and cell density suggesting significant alterations in the spatial arrangement of synapses and microdomains in this illness.^{132–136, 163} Specifically, there is thinning of cortical gray matter accompanied by decreased density of astrocytes as well as a loss of neuropil, while the balance of studies has typically found no changes in the number of neurons.^{132–136, 163} These findings suggest a marked abnormality in the ultrastructural elements that account for the large volume of gray matter not occupied by cell bodies or synapses.

Abnormalities of mitochondria

A few studies have assessed the density of mitochondria in schizophrenia. One study found a decrease in the number of mitochondria per synapse in the stratum in treatment responsive subjects with schizophrenia, while another found decreased volume fraction and area density of mitochondria in subjects with duration of disease more than 21 years.^164–166 In addition, decreased expression of transcripts for a mitochondrial proton transporter were found in the frontal cortex in schizophrenia, and association of the glycolytic enzyme hexokinase 1 with mitochondria was decreased in the parietal cortex in this illness.^{167, 168} These data suggest an abnormality of mitochondrial coupling in schizophrenia.

Changes in glutamate receptor and transporter expression

The changes in glutamate receptor and transporter expression detailed above are consistent with diminished glutamate reuptake capacity, indicating increased diffusion of glutamate between synapses and microdomains. In addition, changes in extrasynaptic expression of the cystine/glutamate antiporter (which releases glutamate from astrocytes) indicate an increased capacity for extrasynaptic release of glutamate.^{111, 142} Both of these mechanisms could lead to increased activation of extrasynaptic glutamate receptors in extracellular glutamate microdomains.

Conclusions

We propose that there is increased activity of extrasynaptic glutamate receptors in schizophrenia secondary to increased levels of extrasynaptic glutamate (Figure 2). Increased extrasynaptic glutamate may be due to increased presynaptic release and spillover (in a misguided attempt to activate sick NMDA receptors in the PSD), increased diffusion of glutamate out of the cleft secondary to deficits in glutamate buffering and reuptake capacity, and/or increased release of glutamate from astrocytes. In this setting, we would predict that integrity of glutamate microdomains is disrupted, either as a primary de cit in the assembly and localization of these domains, as a compensation for increased synaptic glutamate release and spillover, or both. Regardless of the mechanism, we hypothesize that the composition and localization of protein complexes in glutamate microdomains is abnormal in schizophrenia, leading to pathological neuroplastic changes in the structure and function of glutamate circuits in schizophrenia.