131 - 153: Redox Dysregulation and Oxidative Stress in Schizophrenia: Nutrigenetics as a Challenge in Psychiatric Disease Prevention Open Access

Schizophrenia is a major psychiatric brain disease with potentially devastating effects. It strikes in adolescence and young adulthood and can last a lifetime. It affects about 1% of the world's population, is destructive for the individual, family and society, and constitutes a major costly public health problem. It develops progressively, most often undetected during childhood and adolescence in a pre-morbid phase. This usually leads to the onset of psychosis at between 18 and 25 years of age, often evolving toward invalidity. Approximately two-thirds of those who develop schizophrenia require assistance from health care providers (such as government and social security systems) within a few years of onset. The majority of people who develop schizophrenia are unable to return to work or school and may have difficulties in maintaining normal social interactions [1].

The symptoms of schizophrenia are classically divided into categories of positive symptoms (delusions, hallucinations, thought disorder) and negative ones (e.g. deficits in social abilities, poverty of speech, affective flattening). The patients also present other discrete, but more permanent dysfunctions, such as cognitive deficits (problems with attention, specific forms of memory, executive functions) and perceptual instability (basic symptoms) that are now thought to be central to patients’ behavioral disturbances and functional disability. Moreover, patients with schizophrenia also present non-specific symptoms such as anxiety, depression, obsessive behavior, drug and alcohol abuse and suicidal tendency (10% incidence). While present antipsychotic treatments are relatively effective against positive symptoms, they are almost ineffectual for negative and cognitive ones. Indeed, even in patients stabilized with present antipsychotics, these negative and cognitive symptoms are impediments to the social and professional integration of young individuals from the time of disease onset [1–3].

Despite a growing understanding of its neurochemical anomalies, schizophrenia remains an elusive and multifaceted disorder and available evidence regarding its onset and etiology point to a complex interplay of genetic, environmental and developmental factors. Various pathophysiological hypotheses have been put forward, which account for available evidence to varying degrees. Globally, they involve dysfunctions in neurotransmission and impairments of functional connectivity.

It is well established from twin and adoption studies that schizophrenia is highly heritable, but in a complex manner, with a concordance rate of ~50% for monozygotic twins and a heritability of 80% [4]. Numerous studies have focused on identifying genetic vulnerability factors. Results from several genome-wide scans [5–8] have identified chromosomal regions of interest, and cumulative evidence from replication efforts suggest that schizophrenia susceptibility genes may be found on chromosomes 1, 6, 8, 10, 13 and 22 [see reviews in 9–11]. Very recent studies from large genome-wide scans in multiple, large cohorts that have identified both rare high-risk mutations (RR: 2–14) [12–15] and common low-risk variations on chromosome 2 (ZnF804A) and 11 (RR: 1.09–1.19) [16] and in the HLA and histone regions on chromosome 6 [17]. Similarly, studies that have adopted a family-based approach have identified a balanced translocation that disrupts the DISC1 gene [18], as well as the neuregulin gene [19], while hypothesis-driven approaches based on biological findings of deficits in the ability to cope with oxidative stress in patients with schizophrenia have implicated gene variants in the biosynthesis of glutathione as susceptibility factors of the illness [20, 21]. Moreover, understanding how genetic variation at each locus confers susceptibility and/or protection, or what is the contribution of each gene, their relationship with the phenotype and their interaction with environmental risk factors [22, 23] remains a great challenge.

These include exposure to viral infections [24], autoimmune, toxic or traumatic insults and stress during gestation, birth or childhood [25–27] that have been implicated in the pathogenesis of schizophrenia. Recently, models based on epigenetic factors and an interaction between a susceptible genotype and environmental factors have been proposed for this puzzlingly complex disease [28].

In attempting to produce a unifying concept of the etiology of schizophrenia, researchers have posited that these biological mechanisms have their origins in developmental processes that emerge prior to the onset of clinical symptoms. Indeed, evidence for pre- and perinatal epidemiological risk factors of schizophrenia, and for premorbid dysfunction during infancy and childhood have led to the formulation of the so-called neurodevelopmental hypothesis: schizophrenia is viewed as resulting from etiological events acting between conception and birth, and interfering with normal maturational processes of the central nervous system [29–31]. Moreover, it is also hypothesized that the interaction between a hereditary predisposition and early neurodevelopmental insults results in defective connectivity between a number of brain regions, including the midbrain, nucleus accumbens, thalamus, temporo-limbic (including hippocampus) and prefrontal cortices [2, 32–34]. This defective neural circuitry is then vulnerable to dysfunction when unmasked by developmental processes and events of adolescence (myelination, synaptic pruning and hormonal effects of puberty on the central nervous system) and exposure to stressors as the individual enters high-risk ages [3, 31, 35].

A number of theories implicate aberrant neurotransmission systems in schizophrenia, in particular, aberrant dopaminergic [36–38], glutamatergic [39–41] and γ-aminobutyric acid (GABA)-ergic systems [42–46] involving dysfunctions in presynaptic storage, vesicular transport, release, re-uptake and metabolic mechanisms [3, 47]. It is unclear, however, to what extent such neurochemical findings reflect primary causes rather than secondary effects of the pathology, including compensatory mechanisms or environmental interactions.

Multiple lines of evidence suggest that schizophrenia is associated with abnormalities in neural circuitry and impaired structural connectivity. Post-mortem histological studies have shown anomalies at the level of dendritic spines [48–51] and decreases in numbers of inhibitory GABA-parvalbumin interneurons in the prefrontal cortex [46, 52]. Moreover, recent advances in diffusion tensor imaging have allowed in vivo explorations of anatomical connectivity in the human brain. These have pointed to connectivity abnormalities in fronto-parietal and fronto-temporal circuitry in schizophrenia [for reviews see 53, 54]. Further evidence for anomalies in information integration across brain networks is accumulating.

This is based on the study of dynamic, context-dependent processes, which require the preferential recruitment of context-relevant networks over others [55–57]. Evidence is emerging in schizophrenia for an impairment in both local and long-range synchronization in a range of cognitive and perceptual tasks [58–61]. Such perturbation of brain connectivity might be associated with functional anomalies of dopaminergic, glutamatergic and GABA-ergic systems [62–64]. The connectivity argument is reinforced by the fact that the age of onset of full-blown psychosis corresponds to the maturation of myelinated pathways, in particular those involving the prefrontal cortex.

In summary, existing neuroanatomical, neurochemical, neurophysiological and psychopathological arguments converge to suggest that schizophrenia may be considered as a developmental syndrome involving faulty connectivity and neurotransmission and it is likely to have complex origins deriving from multiple genetic and environmental factors.

In the present review, we will emphasize the need to identify a ‘hub’ or ‘final common pathway’ leading to schizophrenia, a hub on which various known causal factors converge and from which established patho-physiological impairments originate. Through a reverse translational approach [65], we have identified a candidate hub related to redox dysregulation. The hub of redox dysregulation/oxidative stress resulting from a genetic impairment of glutathione (GSH) synthesis fulfills such requirements: it represents a complex interplay between genetic and environmental factors during brain development, which leads to impaired neuronal integrity and connectivity and sets off a cascade of events that extend into adult life (fig. 1).

The tripeptide GSH (γ-glutamyl-cysteine-glycine), known as the major intracellular non-protein antioxidant, is required (1) for protection against cellular damage due to reactive oxygen and nitrogen species (ROS and RNS) and detoxification of environmental toxins and reactive metabolites [66], and (2) for the maintenance of the thiol redox status which is critical for redox-sensitive processes [67] such as cell cycle regulation and cell differentiation [68], receptor activation (e.g. N-methyl-daspartate, NMDA, receptor [69]), signal transduction (e.g. H-Ras, PTP-1B) and transcription factor binding to DNA (e.g. Nrf-2, NF-κB) [67]. GSH deficiency will induce oxidative stress, leading to deleterious peroxidations of lipids, proteins and DNAs, altering lipid metabolism and affecting mitochondrial function [70].

Substantial evidence of oxidative damage has been observed in peripheral tissues and post-mortem brain of schizophrenia patients [71–78]. However, variability in these results highlights the contribution of the diverse genotypes and tissues studied [for review see 79]. It remains unclear if the responsible oxidative stress was due to environmental factors or was of genetic origin, preventing the affected brain areas from reacting adequately to oxidative stress. We propose that a primary genetic defect of GSH synthesis is at the origin of the failure of antioxidant defenses in schizophrenia. This implies the involvement of a critical neurodevelopmental component in schizophrenia when compared with neurodegenerative disorders. Indeed, there is also increasing evidence for the involvement of oxidative stress-induced cellular damage in the pathogenesis of various neurodegenerative diseases such as Parkinson's, Alzheimer's and Huntington's. However, in these cases, ROS/RNS increase and GSH depletion appears to be a downstream consequences of other primary causes (such as mitochondrial complex I dysfunction in Parkinson's disease, amyloid-β peptide toxicity in Alzheimer's disease, and huntingtin-related mitochondrial dysfunction in Huntington's disease) [70].

An association between schizophrenia and a trinucleotide repeat polymorphism in the key gene responsible for GSH synthesis has been recently demonstrated, which suggests a genetic origin for the dysregulation of the redox system seen in the disease [21]. Indeed, patients suffering from schizophrenia present a brain deficit in the GSH system which is of genetic origin: (1) GSH levels in the brain and cerebrospinal fluid are decreased [80–82]; (2) glutamate cysteine ligase (GCL) activity and GSH synthesis are decreased in patients’ fibroblasts under oxidative stress conditions [21], and (3) allelic variants of the key GSH-synthesizing enzyme the GCL-modulatory subunit (GCLM) [20] and catalytic subunit (GCLC) [21] are associated with the disease. In particular, in 2 case-control studies with a total of 570 patients and 797 controls, a GAG trinucleotide with 7, 8 or 9 repeat polymorphisms in the GCLC gene showed a significant intergroup difference regarding the overall genotype distribution [21]: the GCLC genotypes 7/7 and 7/9 are more frequent in controls (‘low risk’ genotypes), while 8/7, 8/8, 8/9 and 9/9 are more frequent in patients (‘high risk’ genotypes). This polymorphism has functional consequences: the high-risk genotypes had lower GCL activity, GCLC protein expression and GSH content than subjects with low risk. Interestingly, the high-risk genotype is present in 36–40% of patients and is 3 times more frequent in patients. This is consistent with the decreased GSH levels in the cerebrospinal fluid and medial prefrontal cortex in vivo [80, 82], as well as in post-mortem striatum [81]. Furthermore, high-risk genotype patients have lower fibroblast GSH levels and higher plasmatic free oxidized cysteine levels than low-risk ones (Gysin et al., in preparation), pointing to generalized oxidative systemic conditions [76, 83].

Taken together, these results provide evidence that polymorphisms in the key GSH-synthesizing genes are associated with schizophrenia, leading to a redox dysregulation favoring oxidative and nitrosative stress consequences. These results inspired the development of the ‘glutathione hypothesis’ [84]: brain deficits in the GSH system would lead to both a functional and a structural disconnectivity, which could be a basis of the disease etiology. Moreover, results gathered in experimental models, revealed that a decrease in GSH, particularly during development, induces morphological [85, 86], electrophysiological [87, 88] and behavioral [89–91] anomalies analogous to those observed in the disease (see ‘Developmental animal models with redox dysregulation’, below), thus providing additional support to the hypothesis.

We thus propose that a redox/antioxidant dysregulation due to GSH deficit could represent a vulnerability factor in the early phase of brain development in schizophrenia. Combined with other genetic and environmental factors, it could favor the development of the disease [84]. Life event stresses, through hypothalamic-pituitaryadrenal axis stimulation, induce substantial dopamine release [92–94]. This could result, when combined with GSH deficit, in an increase in ROS and thus in oxidative damage to lipids, proteins and DNA [95], leading during brain development and maturation to progressive structural and functional disconnectivity.

As GSH is the main non-protein cellular redox regulator, protecting against cell damage due to ROS, this deficit would be particularly damaging in brain regions rich in dopamine (e.g. prefrontal cortex), whose metabolism generates ROS. This mechanism could be responsible for morphological alterations such as anomalies of dendritic spines and of parvalbumin-positive inhibitory interneurones in prefrontal cortex [46, 49, 96].

A GSH deficit would also depress NMDA (glutamate) receptor responses [97], a phenomenon known to be involved in perturbations of sensory and cognitive functions in schizophrenia [64], as demonstrated by the psychotomimetic action of the NMDA antagonist phencyclidine [98]. Indeed, GSH potentiates the glutamate response of the NMDA receptor (NMDAR) through interaction at the redox site [97]. This action could be depressed in case of a GSH deficit, leading to effects similar to those induced by phencyclidine. In summary, the framework of the ‘glutathione hypothesis’ can integrate both dopamine and glutamate theories.

The etiological hub of GSH deficit can have many causes via an interaction between genetic and environmental factors [84] (fig. 1). Besides the GSH regulatory genes described above, some other genetic factors identified as implicated in schizophrenia could also lead to a redox imbalance and an oxidative stress. Indeed, a positive association with schizophrenia has been found for a SNP in PRODH which increases the proline oxidase (PRODH) activity, reported to promote ROS generation [99, 100]. On the other hand, various environmental insults known to be schizophrenia risk factors all lead to a GSH deficit: viral infections [24], inflammation, toxic or traumatic insults and stress during gestation or birth or childhood, psycho-social stress and perhaps even diet and post-natal exposure to toxins [25–27]. It is thus likely that such insults, particularly when combined with a genetically deficient redox system, will cause oxidative stress and damaging peroxidations. Impacts during early development may become apparent only in adulthood. Exposure to oxidative stress at various developmental stages affects at least 2 essential cerebral processes that are dysfunctional in schizophrenia (fig. 1): (1) reductions in parvalbumin (PV) fast spiking GABAergic interneurons (FSGI) [46] known to be crucial for brain oscillatory activity [101], and (2) deficient myelination [102].

The NMDAR, which is essential for synaptic plasticity, learning and memory, possesses a redox site which modulates its activity: it is depressed under oxidizing conditions and thus hypoactive when GSH is low [97]. Antagonists of the NMDAR (phencyclidine, ketamine) are known to induce psychotic states in normal subjects and worsen the symptoms of patients [98]. At the cellular level, the prefrontal cortex FSGI show a decrease of PV and GAD67 in post-mortem brains of patients [46]. The same result is obtained in animal models under low GSH conditions [85] or after treatment with NMDAR antagonists [103, 104, 105]. It thus appears that GSH deficit induces an impaired function in FSGI, particularly during brain development. This NMDAR hypofunction induced FSGI defect is mediated by activation of NAPDH oxidases [106, 107]. The latter also produces ROS, which will not be sufficiently reduced when GSH is low. The FSGI are critically involved in the functional cortical circuitry responsible for synchronization and gamma band EEG oscillations during cognitive tasks [59, 101, 108]. Their impairment could potentially lead to decreased synchronization and γ-oscillation power and to cognitive deficits both in patients and in animal models. This chain of events is likely to be causally involved in the generation of schizophrenia phenotype.

In addition, oxidative stress is likely to affect the development of progenitor cells in the central nervous system, and the precursors of oligodendrocytes are particularly sensitive to redox balance. A tendency toward the oxidative side of the balance favors differentiation over proliferation, leading to a deficit in oligodendrocytes and to anomalies of myelination [109, 110]. The development of appropriate levels of myelin is affected in the schizophrenic brain and the resulting errors in conduction speed of action potentials is likely to contribute to the deficits in connectivity and synchronization in diverse pathways which would underlie the cognitive and negative symptoms.

As noted above, schizophrenia is a multifaceted disorder, with evidence concerning its onset and etiology pointing to a complex interplay of genetic, environmental and developmental factors. Several approaches have been taken to develop animal models of schizophrenia [for review see 111, 112]. These include:

• Specific pharmacological or genetic manipulations that aim at modeling a particular aspect of the pathophysiology observed in schizophrenia in order to assess the consequences of these defects. These are applicable to post-pubertal and chronic stages of the disease.

• Disruptions of normal brain development and maturation, focusing on those which lead to behavioral impairments related to schizophrenia that only appear after puberty.

• Animal models with obstetrical complications and prenatal maternal infections, two conditions known to increase the risk of schizophrenia.

However, none of these models by itself addresses in a comprehensive manner the complexity and heterogeneity of schizophrenia and its multiple stages of development. Integration of results obtained from models of these different elements are needed to determine the conditions and defects that can produce the various symptoms of schizophrenia. It is becoming apparent that several different defects independently or in combination can converge to provoke similar behavioral dysfunctions related to schizophrenia. There is thus a strong need to develop new models that combine several manipulations (e.g. combining genetic or pharmacological manipulations with a developmental environmental factor). A unifying pathogenesis concept was proposed [31]: ‘genetic susceptibility in concert with particular stressors during development, may lead to a critical threshold that, when crossed, produces the clinical syndrome at a later stage in life.’

We review here results concerning 2 animal models which explore such convergence of both genetic and environmental risk factors during development, based on impairment of GSH synthesis, redox dysregulation and increased oxidative stress.

We have established a pharmacological model in rats based on inducing transient redox dysregulation during development involving specific inhibition of GSH synthesis with t-butyl sulfoximide (BSO) leading to a 50–60% decrease of brain GSH levels from postnatal days (PN) 5 to 16. Alone or combined with oxidative stress (induced by a blockade of dopamine uptake with GBR12909 leading to high levels of extracellular dopamine and thus to ROS production), this treatment leads to following morphological, electrophysiological and behavioral anomalies:

• In prefrontal cortex neurons, we observed a decrease in dendritic spine density [86, 113], as well as in parvalbumin immunoreactivity [85]. These observations are similar to those reported in the brain of schizophrenia patients [49, 96].

• Memory and sensory integration are perturbed (later in female rats than in males [89–91]), reproducing some of the cognitive deficits observed in schizophrenia.

• In rat hippocampal slices, GSH depletion impairs NMDA-dependent synaptic plasticity [87]. In neuronal cultures, while dopamine enhanced NMDA responses in control, it depressed them in GSH-depleted neurons. Antagonist of D2-Rs prevented this depression, a mechanism contributing to the efficacy of antipsychotics [88].

All of these anomalies are quite similar to those reported in schizophrenia and show that an insult imposed in the developmental period of PN 5–16 has long-term behavioral consequences. This pharmacological model has, however, some technical limitations, in particular bound to the fact that the period at which BSO can be applied systemically is restricted by its transitory permeability across the blood brain barrier.

This preclinical animal model permits exploration of how interaction between this susceptibility gene and environmental insults during brain development will result in impaired neuronal integrity and connectivity, setting off a cascade of events that extend to adult life. As discussed above, the fact that the GCLM gene has allelic variations associated with schizophrenia in patients [20] indicates that the GCLM^-/- mouse is a useful model. Its GSH level is low (20% of wild type) throughout development, rendering it at permanent risk for oxidative stress (note that a knockout of the gene coding the catalytic subunit GCLC is lethal in mice). At the other end, environmental stress, through hypothalamic-pituitary-adrenal axis stimulation, induces substantial dopamine release [92–94]. This would result in an augmentation in ROS and thus further increase oxidative stress. We thus investigated an animal model which involves GCLM as a risk gene causing redox dysregulation and employ hyperdopaminergia as an environmental stressor which can be applied at various stages during neural development. GCLM^-/- mice showed selective and region-specific anomalies in the GABAergic system. As in the BSO-treated rats, PV immunoreactive interneurons in GCLM^-/- mice were particularly affected.

In anterior cingulate of GCLM^-/- mice, concomitantly to an increase in oxidative stress as revealed by 8-oxo-dG (marker of DNA oxidation), the developmental expression of PV was impaired at PN 10, but normalized at PN 20. Additional stress (GBR treatment) during postnatal development (from PN 10–20) prevents this normalization at PN 20 [114]. Moreover, myelination is also impaired as revealed by a weaker myelin basic protein immunolabelling intensity and thinner myelin basic protein immunoreactivity profiles [114].

In ventral but not dorsal hippocampus of adult GCLM^-/- mice, oxidative stress marker 8-hydroxy-2-deoxyguanosine was increased while PV immunoreactivity of GABA interneurons and kainate-induced γ-oscillations were reduced. These effects were severe in the dentate gyrus and CA3 region but not CA1. Furthermore, GCLM^{– /–} had no impairment in dorsal hippocampus-related spatial learning and memory (rewarded alternation and Morris water maze) while they display novelty-induced hyperactivity, reduced anxiety, alterations in social behavior and deficiency in object memory, all tasks related to ventral hippocampus [114, 115].

Altogether, these observations confirm that PV immunoreactive interneurons are particularly sensitive to a GSH deficit but their vulnerability depends on brain region and correlates with the level of oxidative stress. This also supports the notion that PV immunoreactive fast-spiking interneurons are highly vulnerable to oxidative stress [116]. As noted above, patients with schizophrenia are characterized by decreases in PV containing GABAergic interneurons that are crucially involved in the generation of high-frequency oscillations. Moreover the synchronization of such oscillatory activity which is at the basis of neural activity coordination during perceptual and cognitive processes is also impaired in schizophrenia [108, for review see 117]. The impairment of PV interneurons and neural synchronization in GCLM^-/- mice suggests that GSH deficit and redox dysregulation underly cognitive and behavioral anomalies observed in schizophrenia, at least in high-risk GCLC genotype patients.

The myelination anomalies observed in GCLM^-/- mice are consistent with the impairment of oligodendroglia-mediated myelination in schizophrenia as evidenced from gene expression profiling, neurocytochemical and neuroimaging studies [54, 118, 119]. A deficit in myelination would influence the axonal conduction velocity and thus prevent precise synchronizations. It also would have an impact on the association pathways essential for intermodal sensory integration and the ‘binding’ process [117], underlying the cognitive and negative symptoms. As cortical myelination extends until late adolescence for the temporal and prefrontal regions, its deficit could be related to the delayed onset of the disease in early adulthood. As discussed above, intracellular redox state appears to be a necessary and sufficient modulator of the balance between self-renewal and differentiation in dividing oligodendrocyte-type-2 astrocyte progenitor cells [109, 110]. More specifically, cells that are more oxidized tend to differentiate, whereas those that are more reduced undergo self-renewal. Therefore the redox dysregulation observed in schizophrenia may lead to myelination perturbation through oligodendrocytes mitogenic signaling disruption [109].

N-acetyl cysteine (NAC) is a commercial drug approved as an add-on treatment for bronchitis and as an antidote in paracetamol intoxication. Recently, biological effects of NAC have been studied in order to explore potential additional clinical indications, such as graft rejection [120], cystic fibrosis [121], chronic obstructive pulmonary disease [122], arthritis [120], some forms of cancer [120], neurodegenerative disorders [123–125] and cocaine and heroin dependency [126, for review see 127]. Various NAC characteristics suggest it is a very promising candidate in the context of a potential GSH-redox dysregulation linked to schizophrenia, through its influence on the GSH system as well as through a direct antioxidant effect. These characteristics can be defined as follows:

• Induction of in vivo biosynthesis of GSH: cysteine, an NAC metabolite, is essential for GSH synthesis. Availability of cysteine is therefore a crucial factor for constitution of adequate intracerebral reserves of GSH, considering glutathione itself does not cross the blood-brain barrier.

• Gene expression modulation by oxidative stress: NAC plays an important modulating role in expression of genes linked to oxidative stress through its effect on transcription factors such as NF-κB and AP1 [128].

• Antioxidant effect of NACNAC reduces concentration of free radicals and other oxidants through direct inactivation of reactive oxygenated compounds through the molecule's free thiol group and formation of NAC-disulfide as the final product.

• Protection of nerve cells: numerous studies, both in vitro and in vivo, have shown that administration of NAC protects nervous cells against free radicals [120, 128].

In a double-blind, placebo-controlled study, NAC has proven to be efficient at improving schizophrenia symptoms. Indeed, NAC, as an add-on treatment to antipsychotics, decreased negative symptoms and reduced side effects (akathisia) in a cohort of 140 chronic patients [129]. Moreover, this GSH precursor is also effective in improving mismatch negativity [130], an auditory related, NMDA-dependent evoked potential typically impaired in schizophrenia [62, 131]. This is encouraging since present antipsychotic treatments are rather ineffective against cognitive and negative symptoms and have no effect on certain biomarkers like mismatch negativity, a pre-attentional component which is proposed to gate some cognitive and functional modules.

It is of interest to note that the high-risk GAG trinucleotide polymorphism is also associated with bipolar patients [Gysin et al., unpubl. observations], but not with major depression, supporting the view that various genetic anomalies are common to several psychoses. This is consistent with the observation that NAC supplementation improves bipolar patients [132], and is consistent with the concept of a psychosis continuum, as proposed by Crow [133].

Early intervention in psychosis has become an important focus of interest in psychiatry [134, 135]. Prospective studies conducted in first episode psychosis patients [136–139] have identified long delays between onset of psychotic symptoms and initiation of adequate treatment and lack of specificity in treatment of the early phase of psychosis (fig. 2). Two additional concepts emerged from this research: first, that full-blown psychosis is preceded by a ‘prodromal phase’ [140–143], and second, that the first few years after onset of the major symptoms constitute a ‘critical phase’ where outcome is more likely to be influenced [144]. However, the diagnosis of the prodromal phase still relies exclusively on clinical assessment [145] with limited specificity and hence a high rate of false positives, which raises important ethical issues when designing therapeutic strategies [146, 147] and underlines the need for valid biomarkers.

Our present knowledge indicates that the maximal efficacy of our treatments would be in the early psychosis and prodromal phase before redox dysregulation/ oxidative stress has done major damage. Considering NAC has negligible side effects, its efficacy in early psychosis and prodromal phase will be a first step towards identifying pharmacological agents that are much more acceptable to patients and may therefore improve adherence to treatment. Moreover, NAC supplementation has very promising potential in children and adolescents who suffer from neurodevelopmental psychotic disorders, because neuroprotection could be crucial at the critical ages of adolescence when pathological processes are interfering with ongoing brain development. Finally, the presence of a GSH-redox dysregulation or its genetic correlates may prove to be a useful marker in the frame of early detection of schizophrenia.

Disturbances in single-carbon metabolism appear to be related to a variety of neuropsychiatric disorders, covering a broad spectrum that includes depression [148], autism [149] and psychosis [150, 151]. Indeed, the enzymes and metabolites of the methionine and folate cyle are associated with schizophrenia [152–155]. However, we do not know yet whether an observed disturbance is a primary event that is fundamentally related to the pathogenesis or a secondary phenomenon reflecting a nonpathogenic mechanism.

Interestingly, methionine given per os has been shown to be the only amino acid that exacerbates the psychotic symptoms in schizophrenic patients [154]. Experimental methionine loading brings about various effects on the single-carbon cycle as it lowers serum folate concentration [156], induces oxidative stress [157], and lowers the amino acid cysteine [158], the rate-limiting precursor in the GSH synthesis. The exacerbation of psychosis could thus be the consequences of an aggravation of the impairment of the GSH deficit and redox dyregulation hub, at least in the high-risk GCLC genotype schizophrenia patients. Moreover, GSH is a cofactor for the function of methionine adenosyltransferase (MAT), which is a sensitive target for oxidation, and MAT activity is therefore strongly dependent on cellular GSH levels [159]. MAT has been reported to be significantly underactive in red blood cells and brains of schizophrenic patients [160]. In addition, GSH deficit, through methionine and the transmethylation pathway could contribute to the dysregulation of DNA methylation thus affecting epigenetic processes (fig. 3). Indeed, under oxidative stress conditions, methionine synthase is inactivated, allowing homocysteine to be shunted into the transsulfuration pathway in order to favour GSH synthesis [161]. As GSH synthesis is impaired in the high-risk genotype, both transmethylation and transsulfuration pathways will be depressed, leading to perturbations of the DNA methylation process and increase of homocysteine levels often observed in schizophrenia [28, 162, 163].

A most encouraging feature of single-carbon metabolism is its potential modification by natural means, such as B vitamins and antioxidants. In one case report, cobalamine treatment alleviated psychotic symptoms [152]. However, this clinical effect diminished with time, and the metabolic abnormality was thus not wholly cobalamin dependent. In a double-blind, placebo-controlled trial, methylfolate supplementation significantly improved clinical and social recovery among both depressed and schizophrenic patients [164].

Another potential alternative and adjunctive to current antipsychotic treatments is the use of certain polyunsaturated fatty acids (PUFA), the omega-3 and omega-6, which play key roles in brain structure and function but which must be derived from dietary sources [165]. A high dietary ratio of omega-6, found in soft margarine, most vegetable oils and animal fats, to omega-3, found principally in oily fish and seafood, has been linked with vulnerability to many disorders of physical and mental health [166].

In our context, GSH deficit and redox dysregulaton in schizophrenia could lead to oxidative stress and ROS-mediated injury as supported by increased lipid peroxidation products and reduced membrane PUFAs. Decrease in membrane phospholipids in blood cells of psychotic patients [167, 168] and fibroblasts from drug-naïve patients [169] and in post-mortem brain [170] were indeed reported. It has been also suggested that peripheral membrane anomalies correlate with abnormal central phospholipid metabolism in first-episode and chronic schizophrenia patients [171, 172]. Recently, a microarray and proteomic study on post-mortem brain showed anomalies of mitochondrial function and oxidative stress pathways in schizophrenia [76]. Mitochondrial dysfunction in schizophrenia has also been observed [74, 173]. As main ROS producers, mitochondria are particularly susceptible to oxidative damage. Since the brain is highly vulnerable to oxidative damage because of its high oxygen consumption, its high content of oxidizable PUFAs and the presence of redox-active metals (Cu, Fe), a deficit in GSH could be particularly damaging to the neuronal function.

There is increasing evidence that dietary supplementation with omega-3 fatty acids may be beneficial in psychiatric conditions [174]. This evidence includes randomized controlled trials in conditions such as schizophrenia, depression and borderline personality disorder [175–178]. However, recent meta-analyses of these studies show little evidence of a robust clinically relevant effect of omega-3 PUFA in schizophrenia, while the most convincing evidence for beneficial effects of omega-3 PUFA is to be found in depression [179]. Moreover, supplementation with omega-3 PUFA and vitamins C and E appear to exacerbate the positive symptoms in a subgroup of schizophrenia patients with low plasma PUFA [180]. These puzzling results might be explained by assuming an intrinsic GSH deficit of genetic origin in these patients. Indeed, antioxidants such as vitamins C and E might become pro-oxidants in an oxidizing environment [181, 182] as they require GSH to be reduced and regenerated [70]. The same argument can be applied for the short-term effect of the above-described cobalamine treatment. Thus, nutritional approachs must take into account the genetic and epigenetic background of individual patients. Nutrigenetics research will offer a strong foundation for future clinical investigations towards alternative treatment and prevention of psychiatric diseases.

Redox dysregulation may constitute a hub where genetic and environmental vulnerability factors converge, and their timing in brain development is likely to play a decisive role in the phenotype of schizophrenia patients. In experimental models, such redox dysregulation induces anomalies strikingly similar to those observed in patients. A treatment restoring redox balance, deprived of side-effects, yields improvements in chronic patients. Its application in early psychosis and prodrome, intended to halt pathological developmental processes, is promising. The proposed mechanisms should provide biomarkers for early detection, paving the way for prevention perspectives in which nutrigenetics would play a primordial role.