The brain's immune system is controlled by the microglia, a set of cells that when activated can secrete numerous cytokines, chemokines, eicosanoids, proteases, complement and excitotoxins [1]. It is well known that microglial immune cytokines can be activated by, heavy metals (for example, mercury), amyloid, bacteria, and glutamate. During neurodevelopment, the immune cytokines can act as neurotrophic substances, protecting and promoting neurite growth. With excess activation and when chronically activated, these cytokines can be very destructive.

An autotoxic, non-specific immune destruction of neurons, neurites and synaptic connections has been described by McGeer et al. [2]. In this process, either systemic immune factors (cytokines) or local immune factors, such as beta-amyloidcan activate the brain's immune system via activation of astrocytes and microglia. In both instances brain levels of cytokines, reactive oxygen and nitrogen species, cellular immune components, excitotoxins and arachidonic acid are elevated leading to brain dysfunction.

The difference in severity between patients with autism appears to vary with the stage at which the immune/excitotoxic stress arises and its intensity. In humans, it is well known that a considerable amount of postnatal brain development occurs, with the highest synaptogenesis occurring during the last trimester and the first two postnatal years. An excess of extraneuronal glutamate can interfere with neuronal migration patterns, differentiation and synaptic development, resulting in varying degrees of abnormal brain architecture and hence differing severities of autistic features.

Glutamate exerts its effect through metabotropic (mGlu) and ionotropic (iGlu) glutamate receptors localized in the cellular membranes of neurons and glia as neuron-supporting cells. According to their differential affinity for different agonists, ionotropic receptors include N-methyl-D-aspartate (NMDA), kainate (KA), and amino-3-hydroxy-5-methyl-4-isoxazole propionic acid (AMPA) [3]. The metabotropic receptors belong to the G-protein-coupled receptor, and can be divided into group I (mGluR1 and mGluR5), group II (mGluR2 and mGluR3) and group III (mGluR4 and mGluR6-8) according to their primary sequence and pharmacological agonists [4]. The increased probability of epilepsy in patients with autism suggests enhanced glutamatergic signaling with positive correlation between plasma levels of glutamate and the severity of autism and increased expression of mRNAs encoding the AMPA 1 receptor in the cerebellum of autistic patients [5],[6].

While in the adult brain gamma aminobutyric acid (GABA) acts as an inhibitory neurotransmitter, during the perinatal period it depolarizes targeted cells and triggers calcium influx. GABA-mediated calcium signaling regulates a number of important developmental processes which include, cell proliferation, differentiation, synapse maturation, and cell death [7]. A dysfunction of the GABAergic signaling early in development leads to a severe excitatory/inhibitory (E/I) imbalance in neuronal circuits, a condition that may account for some of the behavioral deficits observed in patients with autism [8]. The GABA level, glutamate/GABA and glutamine/glutamate ratios are significantly lower in patients with autism compared to normal controls, thus suggesting a possible abnormality in the regulation between GABA and glutamate that might lead to excitotoxicity [9],[10].

The overall aim of this study is to confirm that GABA and glutamate synapses are abnormal in autistic patients and to determine how these abnormalities may be associated with one another and with neuroinflammatory responses as a pathway involved in the etiology of autism.

GABA and glutamate are the main inhibitory and excitatory neurotransmitters in the human brain. Both have critical roles during early development of the nervous system, a stage when clinical presentation indicates that autism begins. Therefore, it is important to understand the functional status of GABAergic and glutamatergic neurotransmission in the plasma as the body fluid reflects brain physiology and pathology. Clarification of the relationship between abnormality in the absolute and relative GABAergic and glutamatergic synaptic activities and neuroinflammatory biomarkers represented by TNF-α, IL-6, IFN-γ, and IFI16 could help to find a target for pharmacological intervention.

Brain GABA and glutamate metabolism include a series of integrated and interconverted pathways. Glutamate, in addition to its importance in the clearance of brain ammonia through glutamine synthesis, serves as an important energy source through the tricarboxylic acid cycle (TCA) cycle after its deamination by glutamate dehydrogenase and oxidative decarboxylation by α-ketoglutarate dehydrogenase. Glutamate is converted to GABA by glutamic acid decarboxylase. Finally, GABA is metabolized to succinate by the combined reactions of GABA transaminase and succinic semialdehyde dehydrogenase. The significant increase of the absolute concentration of both neurotransmitters reported in the present study, together with the remarkable decrease in the glutamate/GABA ratio in autistic patients compared to healthy controls (Table 1), can be easily related to autistic characteristics. Since GABA derives from glutamate and glutamate derives from GABA, alterations in both neurotransmitters can affect each other. The unexpected elevation of GABA as an inhibitory neurotransmitter in autistic patients with a high rate of onset of epilepsy could be explained on the basis that high plasma GABA could be concomitant with lower brain GABA due to a smaller number or dysfunctional neuronal GABA receptors. GABA exerts its functions by binding to chloride-permeable ionotropic GABA_A receptors and metabotropic GABA_B receptors. This explanation can be acceptable because GABAergic transmission through receptors is achieved by different mechanisms such as modulation of GABA_A receptors and variation of intracellular chloride concentration together with alteration in GABA concentration [14]. Elevation of GABA level, which was recorded in the present study, can be easily related to the significant elevation of glutamate as substrate of glutamate decarboxylase. The suggested role of GABA receptors in the excitotoxicity or inhibitory/excitatory imbalance in autistic patient could be supported through the recent work of Han et al. [15] in which low doses of benzodiazepines increase inhibitory neurotransmission through positive allosteric modulation of post-synaptic GABA_A receptors, improved impaired social interaction, repetitive behavior, and cognitive ability in an animal model of autism. In contrast, negative allosteric modulation of GABA_A receptors impaired social behavior in wild-type mice, supporting the notion that reduced inhibitory neurotransmission may contribute to social and cognitive deficits as autistic characteristics.

The recorded decrease of IL-6 and TNF-α in plasma of autistic patients compared to control healthy participants (Table 1) could be a consequence of an early elevation of these cytokines in plasma followed by influx of both to the brain across the blood-brain barrier (BBB). This explanation can be supported by many studies that proved the elevation of IL-6 in autistic brains [16]. Although elevation of IL-6 is a repeated finding in autism, the exact mechanism by which an IL-6 increase may contribute to the pathogenesis of this disorder remains undefined. The association between IL-6 and low glutamate/GABA ratio (Table 5) observed in the present study, could help in clarifying this mechanism. Firstly, a key role for inhibitory/excitatory imbalance in autism is supported by the fact that 10 to 30% of autistic patients suffer from epilepsy [17]. This synaptic abnormality hypothesis was further supported by the identification of mutations affecting synaptic cell adhesion molecules, as well as synaptic proteins in autistic subjects [18]-[20]. Secondly, IL-6 elevation was found to stimulate excitatory synapse formation and impair the development of inhibitory synapses. This role of IL-6 was supported by the detection of a reduced post-excitatory inhibition in IL-6 overexpressed mice. Post-excitatory inhibition, which is usually measured by paired-pulse inhibition (PPI) has been suggested as being caused by a reduction in the release of glutamate from terminals of afferent axons [21],[22]. This effect seems to be the result of an inhibition of calcium influx through presynaptic receptors, which play a critical role in the release of glutamate from synaptic vesicles on afferent stimulation [23],[24]. Thirdly, mice with elevated IL-6 in the brain display many autistic features, including impaired cognitive abilities, deficits in learning, abnormal anxiety traits and habituations, as well as decreased social interactions [13].

The suggested increase of brain TNF-α and the association between this cytokine and the impaired glutamate/GABA ratio reported in the present study (Table 5) could be explained on the basis of the interesting relationship between neurons, glial cells and TNF-α leading to glutamate excitotoxicity [25]. In general, excitotoxicity is related to the activation of NMDA receptors by excessive glutamate. Brain cell death usually occurs after over stimulation of glutamate receptors due to impaired re-uptake of glutamate by the EAAT2/GLT-1 transporters on glial cells [26]. The expression of this transporter has been shown to be regulated by NF-κB [27]. Pickering et al. [28] showed that increased levels of TNF-α might be easily related to glutamate excitotoxicity through the decrease of EAAT2/GLT-1 expression. TNF-α induces the classical I kappa B (IαB) degradation pathway to trigger NF-αB nuclear translocation and DNA binding to repress EAAT2 expression. In this situation, the presence of elevated TNF-α concentrations leads to elevated extracellular glutamate concentration, thereby increasing the risk of glutamate excitotoxicity. Elevated TNF-α concentrations could be also contributed to the decrease of the inhibitory transmission. Recently, an in vitro culture study in mature rat and mouse hippocampal neurons demonstrated that acute (45 minute) application of TNF-α induced a rapid and persistent decrease of inhibitory synaptic strength as well as a downregulation of cell-surface levels of GABA_A receptors [29].

Table 1 demonstrates the large significant increase of IFN-γ in plasma of autistic patients compared to healthy control participants. This increase was associated with the low glutamate/GABA ratio representing inhibitory/excitatory synaptic transmission (Table 5). This association could be reasonably accepted and explained on the basis of the key role of IFN-γ in the induction of neurotoxicity. Recently, it has been proposed that IFN-γ directly induces neuronal dysfunction through the formation of dendritic beads in mouse cortical neurons and the enhancement of glutamate neurotoxicity via AMPA but not NMDA receptors. IFN-γ forms a unique, neuron-specific, calcium-permeable receptor complex with AMPA receptor subunit GluR1. Through this receptor complex, IFN-γ phosphorylates GluR1 at serine 845 position by the JAK1-2/STAT1 pathway, increases Ca⁺² influx and nitric oxide production, and subsequently decreases ATP production, leading to the dendritic bead formation. This explanation could provide a mechanisms through which inflammation could be related to glutamate excitotoxicity as an etiological signaling involved in autism pathology [30].

ROC analyses of glutamate, GABA, glutamate/GABA, TNF-α, IL-6, IFN-γ and IFI16 are presented in Table 2 and Figures 4 (a-d). All measured parameters demonstrated satisfactory sensitivity and very high specificity, which confirmed the hypothesis that autistic patients are characterized by excitotoxicity and suffer from neuroinflammation.

Multiple regression analysis, as one of the most powerful methods of data analysis, was used between the measured parameters to determine the relative contribution of impaired GABAergic and glutamatergic neurotransmission in the recorded neuroinflammation and to predict the role of the measured cytokines in the glutamatergic/GABAergic imbalance in autistic patients compared to healthy control subjects (Tables 3, 4 and 5).

The finding that the high plasma GABA levels are positively correlated to glutamate and glutamate/GABA levels (Table 3) led us to propose that the unexpected increase in GABA concentration in autistic patients is induced partly by glutamate bioavailability as substrate to glutamate decarboxylase (GAD) enzyme. This can be supported by the value of R², which shows that 0.973 or 97% of the variance in GABA was related to the elevation of glutamate. Despite the elevated GABA in plasma of autistic patients, a dysfunction in inhibitory GABAergic transmission is suggested due to a reduced density of GABA_A receptors in the hippocampus [31],[32], reduced GABA_A and GABA_B receptor protein subunits in the cerebellum, prefrontal Brodmann area 9 and parietal Brodmann area 40 [33],[34], and a reduced expression of the GABA-synthesizing iso-enzymes GAD65 and GAD67 in the parietal and cerebellar cortices in autistic brains [35].

Table 4 demonstrates the stepwise multiple regression analysis using glutamate as dependent variable and TNF-α, IL-6, IFN-γ and IFI16 as independent variables. The value of R² shows that 0.869 or almost 87% of the variance in glutamate was explained by, or related to, neuroinflammation as discussed above. While, IFN-γ cytokine was the most significantly related to glutamate excitotoxicity, IFI16 did not contribute.

Based on the present study, it could be concluded that glutamate/GABA balance or excitatory/inhibitory balance is crucial for the functioning of the synapse. Multiple mechanisms may compromise this balance; neuroinflammation highly contributes to the imbalance and consequently to the etiology of autism.