1. Glia Dysfunction in Autism
The role of glia dysfunction, particularly Bergmann Glia in glutamate removal, is well described in autism
[1]. Single-cell RNA sequencing revealed that autism-associated transcriptome alterations in specific cortical cell types are related to “synaptic signaling of upper-layer excitatory neurons” and microglia
[2]. A large whole-genome study of postmortem brain samples also indicated that DNA methylation alterations associated with autism are involved in the immune system, synaptic signaling, and neuronal regulation and are highly correlated with the affected genes in patients with chromosome 15q duplication and H3K27 acetylation
[3].
Some important microglia genes, such as
TREM2 (as the microglia innate immune receptor gene involved in synapse pruning) are also linked to autism pathogenesis. In mice, the lack of expression of
TREM2 is associated with autism-like behavior and, in humans, a reduced
TREM2 protein level correlates with the severity of autism symptoms
[4]. Additionally, decreased expression of
TREM2 is associated with increased expression of
TNFA, a pro-inflammatory cytokine, and
NOS2 (nitric oxide synthase 2) in mice
[5]. Interestingly, sodium valproate (an epigenetic drug that inhibits HDACs) decreases
TNF-α and
NOS2 expression levels
[6], hinting at an opportunity for autism epigenetic therapy using HDAC inhibitors. Experimental evidence indicates that
TREM2 is also regulated by microRNAs. In this regard, as it is known that the up-regulation of miRNA-34a (an NF-κB-sensitive miRNA) targets
TREM2 and down-regulates its expression in microglia cells
[7], increased expression of miRNA34 a/b/c was also shown in cortical tubers of patients with tuberous sclerosis, an autism spectrum disease
[8]. There is also evidence that
TREM2 expression is regulated by DNA methylation. For example, DNA hypomethylation of
TREM2 intron 1, which is associated with its increased expression, was shown in the blood cells of patients with SCZ and Alzheimer’s disease
[9][10]. On the other hand, increased DNA methylation of CpG sites located upstream of the
TREM2 transcription start site is reported in the superior temporal gyrus of patients with Alzheimer’s disease
[11]. However, in the hippocampus of patients with Alzheimer’s disease, the higher levels of DNA methylation were reported to be due to the enrichment of 5-hydroxymethycytosine associated with upregulation of
TREM2 expression
[12]. Considering these data, further study of the epigenetic dysregulation of
TREM2 is warranted in autism.
Methyl-CpG binding protein 2 (
MECP2) is another important gene in the pathogenesis of autism spectrum syndrome, specifically in Rett syndrome. In general, Rett syndrome is due to the mutation of
MECP2 located in chromosome X. The disease appears mostly in females, as males affected by this mutation usually die shortly after birth. In addition to its mutation, promoter DNA hypermethylation of
MECP2, associated with its reduced protein expression, was shown in the frontal cortex of male autistic patients
[13]. Based on recent data, while neuronal
MECP2 expression is more than that observed in astrocytes, in males, a higher DNA methylation level of
MECP2 regulatory regions is associated with reduced expression of
MECP2 in astrocytes
[14]. This supports the idea that astrocytic DNA hypermethylation of
MECP2 may be a mechanism for disease pathogenesis in male autistic patients. In this regard, previous animal studies have shown that the re-expression of astrocytic
MECP2 in globally
MECP2-deficient mice improves their behavioral and molecular aberrations
[15]. Furthermore, as microglia pathology due to
MECP2 dysfunction was later proposed as the leading cause of Rett syndrome and autism pathogenesis
[16], it has been shown that MECP2 regulates the expression of “microglia genes in response to inflammatory stimuli”
[17].
With the involvement of microglia, it is not surprising that the immune system and complement proteins, such as C1q, C3, and C4, as well as TGFB2, which contribute to synapse pruning during brain maturation
[18], are among the key players in autism pathogenesis
[19][20] and in other major mental diseases, such as SCZ
[21][22]. Relatedly, whole-genome DNA methylation analysis uncovered epigenetic dysregulation of several complement genes such as
C1Q, C3, and
ITGB2 (
C3R), as well as several other inflammatory genes (e.g.,
TNF-α,
IRF8, and
SPI1) in postmortem brain samples of patients with autism
[23]. Therefore, these findings (as summarized in
Table 1) call for more studies on the astroglia-mediated epigenetic dysregulation of complement genes in autism.
Table 1. Genes linked to non-neuronal brain cell function and supporting evidence indicating their epigenetic dysregulation in mental diseases.
2. Glia Dysfunction in Schizophrenia and Bipolar Disorder
Several lines of evidence indicate that inflammation and inflammation-induced oxidative stress, as well as many insidious/dormant infectious elements, alter the expression status and epigenetic landscapes of brain cells; this evidence is reviewed here and elsewhere
[48][49]. One of the best examples of this phenomenon concerns the impact of maternal immune activation on brain cells’ (and, in particular, microglia) gene expression and epigenetic status in conjunction with the pathogenesis of mental diseases
[50], as discussed in more detail in the following sections. In addition, analyses of the brain gene expression data in publicly available datasets reveal expression alterations of genes related to cortical astrocytes both in SCZ and in bipolar disorder (BD)
[51]. Other human postmortem brain studies also revealed that the altered expression of genes that are important to glia or astrocyte functions (e.g.,
SLC1A2 and
TGFB2) is linked to psychiatric phenotypes
[29]. Interestingly, as the expression of astrocytes’ glutamate transporter,
GLT-1 (
SLC1A2) exhibits >100% and 70% increases in the postmortem brains of patients with SCZ and psychotic BD, respectively
[29]. The use of ceftriaxone (an antibiotic that selectively enhances
GLT-1 expression) could reduce prepulse inhibition (which is also reduced in SCZ patients) in rats, which could be reversed by dihydrokainate (DHK), an antagonist of
GLT-1 [52]. Other research findings indicate that
GLT-1 expression is regulated by diverse epigenetic mechanisms
[30][31]. For instance, while miR-218 downregulates astrocytic
GLT-1 expression
[32] and the hypo-expression of miR-218 increases susceptibility to stress, its reduced expression has been observed in the medial prefrontal cortices of patients with depression and suicide
[53][54]. Regarding
TGFB2, while its expression is increased in the postmortem brains of patients with SCZ and psychotic BD, due to its promoter DNA hypomethylation
[29], other studies have shown that
TGFB2 is over-expressed in the neurons of patients with Alzheimer’s disease
[55][56]. It is also the only cytokine that is increased in the cerebrospinal fluid of these patients
[57]. In vitro studies indicate that the expression of
TGFB2 is induced by toxic amyloid betas in both glial and neuronal cells. In turn, the increased
TGFB2 binds to the extracellular domain of amyloid beta precursor protein and triggers a neuronal cell death pathway in Alzheimer’s disease. Interestingly, the degrees of
TGFB2-induced cell death are larger in cells expressing a familial AD-related mutant
APP than in those expressing wild-type
APP [57][58]. Together, these data suggest the potential roles of
GLT-1 and
TGFB2 epigenetic alterations in the pathogenesis of neuropsychiatric diseases, indicating that they are legitimate targets for therapeutic interventions
[59].
Other genes that are mainly expressed by astrocytes and glial cells (e.g.,
S100B, the S100 calcium-binding protein B) are also linked to SCZ pathogenesis in GWAS analysis. Moreover, just as a higher level of the S100B protein is reported in the blood cells of SCZ patients
[35], an increased serum level of
S100B was also reported in BD patients
[60]. While
S100B promotes hippocampal synaptogenesis after traumatic brain injury
[61], there is experimental evidence that its expression is regulated by DNA methylation
[36].
Another line of evidence in support of the role of astroglia in SCZ is the existence of D2-like receptors in astrocytes. While astroglia account for almost one-third of
DRD2 binding sites in the brain cortex, and
DRD3 is also expressed in astrocytes, mice deficient for this D2-like receptor or that are treated with a DRD3 antagonist do not show astroglia inflammatory activity in response to LPS (lipopolysaccharide) challenge. It should be noted that, although microglia do not express
DRD3, in
DRD3 deficient mice, the expression of
Fizz1, an anti-inflammatory protein, is increased in glial cells (both in vitro and in vivo). This also attenuates microglial activation in response to LPS challenge
[62]. The fact that commonly used antipsychotic drugs block DRD2-like receptors, and that the long-term use of olanzapine alters
DRD2 promoter DNA methylation levels
[63], suggests that the effects of DRD2-like antagonists in SCZ treatment could be due to the inhibition of astroglia’s inflammatory activity, mediated in part by
DRD2 epigenetic modifications.
Human major histocompatibility complex (
MHC) genes are among other genes associated with microglia functions that are involved in SCZ pathogenesis in GWAS analyses
[64][65]. MHC class I is involved in complement-mediated synaptic pruning
[3] and exhibits reduced expression in the brains of SCZ patients
[37]. Additionally, it has been shown that glia overactivity mediated by complement
C4A (one of the genes of MHC III) and the increased expression of
C4A may have deleterious effects in SCZ
[27][66]. Notably, in a study of humanized glial chimeric mice, it was shown that mice with glial cells produced from the iPSC of patients with childhood-onset SCZ, exhibited premature glia migration into the cortex and reduced expansion of white matter and its hypomyelination compared to the mice with glia from the normal controls. This was associated with a delay in astrocytic differentiation and abnormality in astrocytic morphology, as well as reduced prepulse inhibition, increased anxiety, and sleep problems. Additionally, the cultured glial progenitor cells from SCZ patients exhibited aberrant expression of genes linked to glial differentiation as well as synapse-associated genes in the RNA-seq analysis, suggesting that the observed glial pathology originates from these cells
[67]. In reference to potential clinical applications of these findings, it is noteworthy that as an exaggerated synapse pruning has been repeatedly reported in adolescents, particularly in SCZ patients
[68], which could be mitigated by minocycline
[69][70]; thus it is not surprising that the inhibition of microglia activity by minocycline is effective in the treatment of negative symptoms of SCZ in randomized double-blind studies
[71][72].
In addition to the relation between genetic variations of the complement system and SCZ
[66], there is also evidence that non-genetic alterations of the activity of complement system are associated with SCZ. For example, as summarized in
Table 1, increased
C4 and
C1q levels were reported in the prefrontal cortices of patients with SCZ
[24] and the blood cells of antipsychotic-naive first-episode SCZ patients
[28], as well as those with chronic SCZ and in individuals at high risk for psychosis; meanwhile, increased
C3 levels were also shown in the latter group
[73]. There are also reports of increased levels of
C3a, C5a, and
C5b-9 in drug-free patients with bipolar disorder
[74], and of increased expression of
C1q, C4, and factor B in the peripheral blood mononuclear cells of chronic BD patients
[75]. Epigenetic analysis of different elements of the complement system in other mental diseases revealed those genes of the complement system that are linked to glial activity and are subjects of epigenetic dysregulation (
Table 1). For instance, in whole-genome DNA methylation analysis, the epigenetic dysregulation of
C1q, C3, and
ITGB2 (
C3R) was reported in autism
[23]. The DNA hypomethylation of
C3 associated with its increased expression was also shown in the postmortem brains of patients with Alzheimer’s disease
[25][26]. Furthermore, DNA methylation alterations affecting
C4A and
C4B expression were reported in a genome-wide DNA methylation analysis of patients with Attention-Deficit/Hyperactivity Disorder
[76].
As epigenetic alterations are frequently reported in SCZ and autism and these diseases are more common in males, it is important to note that, in females, one of the X chromosomes is subject to random inactivation by DNA methylation. Hence, if the activity of any gene in one of the X chromosomes is imbalanced due to inherited or de novo mutations, in a female subject, half of the neighboring cells can work normally, partially balancing the tissue functions. For example,
SRPX2 (regulated by the complement C1q) which is localized chromosome X and is involved in language and cognitive development
[77], exhibits expression reduced by almost 20% in the postmortem brains of SCZ patients
[29]. Although the
SRPX2 gene codes a neuronal protein, C1q binds to
SRPX2, inhibiting synapse eliminations
[78]. Thus, a close cooperation between
SRPX2 and this complement is required for the fine tuning of synapse pruning in normal brain development. In cancer research, it has been shown that DNA methylation regulates
SRPX2 expression levels
[79]. Therefore, DNA methylation alteration of
SRPX2 could be an interesting subject for further studies in SCZ, as well as in autism and dyspraxia, which are both more prevalent in males than in females. There is also a correlation between the expression of
MECP2, a methyl CpG binding protein, and
SRPX2 expression
[80], which warrants further research.
Other evidence related to astroglia epigenetic alterations in mental diseases comes from imprinted genes in which one copy of the parental alleles (in autosomes) is inactivated by DNA methylation. In this regard, whole-genome DNA methylation analysis for rare epigenetic variations identified that the NDN gene, which is highly expressed in astrocytes, was linked to SCZ as well as to autism pathogenesis
[38]. This gene is exclusively expressed from the paternal allele and is in the Prader-Willi syndrome deletion region implicated in autism pathogenesis
[81].
3. Astroglia Pathology and Dysfunction in Depression
In addition to SCZ and BD, there is evidence for astroglia dysfunctions in depression. For example, whole-transcriptome analysis using RNA-seq of human postmortem brain samples from drug-free individuals with MDD (major depressive disorder) and suicide revealed deficits in genes related to microglial and astrocytic cell functions
[82]. Aberrant DNA methylation patterns specific to astrocytes were also shown in the prefrontal cortices of postmortem brain samples of patients with depression
[83]. Another study reported the upregulation of astroglia’s potassium channel gene (
Kir4.1 or
KCNJ10) and reduced
GLT-1 (
SLC1A2) activity (which removes ~90% of extracellular/synapse glutamate) and increased neuronal bursting activity of the lateral habenula as key factors in the induction of depression-like behaviors
[33][34]. A recent study revealed that DNA methylation regulates
KCNJ10 expression in astrocytes
[39]. Aberrant DNA methylation of
NMDAR (more specifically, the hypermethylation of the
GRIN2A subunit) was also reported in the hippocampus and prefrontal cortex of MDD patients
[41]. However, in SCZ patients, DNA hypomethylation of
GRIN2B was shown in blood cells
[84]. Interestingly, ketamine, which is used to treat MDD, decreases neuronal bursting activity
[85] by blocking glial
NMDAR in the lateral habenula, which is considered to be the brain’s “antireward” center
[33][34]. Nevertheless, in rats, ketamine’s effects on depressive-like behavior was attributed to its activity in the regulation of astrocytic
GLT-1, and also through BDNF-TrkB signaling
[86]. It has also been shown that ketamine alleviates DNA hypermethylation of
BDNF in the medial prefrontal cortex and hippocampus in a mouse model of PTSD
[87]. Furthermore, while
BDNF and its receptor NTRK2 play key roles in astrocytes’ maturation and functions
[88], DNA hypermethylation of
NTRK2 and its reduced expression was reported in the postmortem brains of patients who died by suicide
[40].
HMGB1 is another microglia-associated gene involved in depression
[42][89]. Animal studies have shown that unpredictable chronic stress can lead to microglia activation in the hippocampus and depressive-like symptoms
[90]. This type of stress could increase
HMGB1 expression in the hippocampal microglia, and the infusion of
HMGB1 into the mice hippocampus could also induce depression
[43]. Interestingly, the activation of the microglia, along with depressive symptoms, could be prevented by minocycline or imipramine
[90]. While
HMGB1 is a well-known marker of inflammation, increased expression of
HMGB1, associated with its promoter DNA’s methylation alteration, was reported in cardiac progenitor cells following hypoxia and metabolic diseases
[44][45], suggesting that DNA methylation is a mechanism for
HMGB1 regulation. However, in brain cells,
HMGB1 expression is also regulated by HDAC4&5 and miR-129
[46][47], the latter of which was shown to regulate neuronal migration in mice brains
[91].