Risk factors for schizophrenia.
Schizophrenia (SZ) is a complex disorder without a single cause but with multiple etiologies. Monozygotic twin studies suggesting high discordant rates provide evidence for epigenetic mechanisms among the factors that result in increased susceptibility. Among the different epigenetic modifications in mammals, DNA methylation mediated by DNA methyltransferases (DNMTs) is the most-well studied. Studies on post-mortem brain samples and blood samples of SZ patients revealed altered levels of most DNMTs. In addition, some recent studies also reported disease-associated SNPs in the DNMT genes. While the effects of dysregulation of DNMTs are beginning to be understood, many unanswered questions remain. Here, we review the current evidences that shed light on the relationship between DNMT dysregulation and SZ, and suggest the possible strategies to address some of the unanswered questions.
- DNA methyltransferases
- DNA methylation
- Abnormal neurogenesis
Schizophrenia (SZ) is a severe and chronic mental disorder with an incidence of ∼1%, affecting ∼20 million people worldwide . The main symptoms of SZ include hallucination, delusion, abnormal disorganized behavior, disorganized speech, disturbances of emotions such as marked apathy, etc. The disorder is associated with considerable disability and can affect educational and occupational performance with 2–3 times increased likelihood of death earlier than the general population .
SZ is a complex disorder with no single causative factor but with multiple etiologies (Table 1). The five main factors that are believed to result in increased risk are: physical and chemical changes in brain , pregnancy or birth complications , childhood trauma , genetic  -and epigenetic . Among these, a high risk among first-degree relatives compared to the general population and increased risk in monozygotic than dizygotic twins suggest genetic factors . However, the observed concordance rates (∼50%) in monozygotic twins that were much lesser than expected for a purely genetic risk (nearly 100%) suggest the contribution of epigenetic mechanisms to SZ .
|1||Physical and chemical changes in brain.||Subtle structural changes have been observed in post-mortem brain samples of SZ patients. Imbalances in neurotransmitters such as dopamine and glutamate have been linked to SZ.|||
|2||Pregnancy or birth complications||Low birth weight, infection during pregnancy, asphyxia, premature labour, maternal obesity diagnosis in pregnancy, etc. have been associated with SZ in the offspring.|||
|3||Childhood trauma||There is an increased risk to experience SZ if there is death or permanent separation of one or both parents.|||
|4||Genetic||The risk in identical twins (1 in 2) is four times higher than non-identical twins (1 in 8). These risks are much higher than for general population (1 in 100).||[6, 7]|
|5||Epigenetic||Monozygotic twins show only 45–50% concordance.||[8, 9]|
Recent data based on brain imaging and molecular-genetic studies suggest that SZ is a form of neurodevelopmental disorder . The neurodevelopmental hypothesis for SZ suggests pathological neurodevelopment during first and second trimesters of pregnancy results in altered neuronal circuits which in turn result in psychosis in adolescents or young adults when exposed to increased biological or psychological stress. Evidences in support of this hypothesis comes from genetic studies that identified affected genes and risk factors during perinatal life that may disrupt the normal process of neurodevelopment. In addition, studies over the past 20 years showed that in comparison with controls, SZ patients after the onset exhibit accelerated aging-related loss of brain tissue . Specifically, the patients show increased age-related reduction in the proportion of grey matter compared with controls . These findings suggest that altered neurodevelopment may underlie the processes associated with SZ.
2. Epigenetic mechanisms
As mentioned above, evidence on the contribution of epigenetic mechanisms in SZ comes from monozygotic twin studies wherein the concordance rates are only ∼50%. This low concordance rate suggests the interplay of genes and environment resulting in SZ. Because of this interplay, the epigenetic mechanisms have been suggested to be among the etiological factors . Epigenetic mechanisms are defined as processes that can alter the patterns of gene expression without causing a change in the DNA sequence . These mechanisms operate at the levels of transcription, mRNA stability and translation (Table 2). At the level of transcription, mammalian genes can be regulated by covalent modifications of the DNA , modifications of N-terminal tails of histones , microRNAs , circular RNAs  and long noncoding RNAs . A number of modifications of RNA have been reported to influence mRNA stability and efficiency of translation. These modifications and their roles are described elsewhere [22, 23]. Because of epigenetic differences, genetically identical cells in a multicellular organism express different sets of genes that confer cell type – specific identity and function . The most well studied epigenetic modification is methylation of the 5th carbon in the cytosine residues in the genomic DNA, often referred to as cytosine methylation. This modification mostly occurs in the CpG dinucleotides because of the maintenance mechanism in a post-replicative manner involving hemi-methylated DNA [see below]. As such, DNA methylation is often used as a synonym to CpG methylation in mammals. A family of enzymes, referred to as DNA methyltransferases (DNMTs) are responsible for establishment and maintenance of DNA methylation . Several studies that focused on the relationship between DNA methylation and gene expression showed an inverse correlation, meaning that DNA methylation is often associated with repressed state of the promoters . In case of histones, the lysines in the N-terminal tails of core histones can either be acetylated or methylated. These modifications occur on the same lysine residues and are therefore mutually exclusive . Whereas histone lysine acetylation is always associated with gene expression, histone methylation is associated with either expression or silencing depending on the residues involved . For example, methylation at lysine 9 of histone H3 (H3-K9) or H3-K27 is associated with silencing. On the other hand, H3-K4 or H3-K36 methylation is associated with gene expression. Histone methyltransferases and histone acetyltransferases are two families of enzymes for imparting the two covalent modifications of the N-terminal tails of the core histones . As in case of DNA methylation, histone marks are also heritable. The covalently modified nucleosomes from the parental chromatin are segregated equally among the two daughter DNA molecules so that additional nucleosomes containing histone marks identical to the parental nucleosomes are assembled . Both DNA methylation and histone modifications are reversible involving different categories of enzymes and processes. The machinery of DNA methylation and demethylation is described in the next section [Section 2.1]. With regard to the histone modifications, histone demethylases (HDMs), histone methyltransferases (HMTs), histone acetyltransferases (HATs) and histone deacetylases (HDACs) together play a role in erasure and establishment of histone modification marks . HDMs remove methyl groups from the lysines of the core histones so that the unmethylated lysines can be acetylated by HATs. HDACs, on the other hand, remove acetyl groups from the acetylated lysines so that the same residues can be methylated by HMTs.
Apart from covalent modifications of the genome, long noncoding RNAs (lnc RNAs), microRNAs (miRNAs) and circular RNAs (circRNAs) also play an important role in regulating gene expression. Of these, circRNAs and miRNAs regulate expression at post-transcriptional levels whereas lncRNAs can regulate at both transcriptional and post-transcriptional levels. Lnc RNAs are ≥200 nucleotides, do not encode any protein and regulate genes at the both transcriptional and post-transcriptional levels . At the level of transcription, lncRNAs either can promote histone modifications and chromatin condensation or recruit transcription factors to facilitate gene expression or evict transcription factors and result in gene repression. In addition, lncRNAs are also known to influence alternative splicing, polysome recruitment to enable translation, act as decoys for microRNAs (miRNAs) and regulate mRNA stability. The miRNAs, on the other hand cause translational repression of the target mRNAs. Each miRNA is ∼22 bases long and can recognize multiple targets having a few mismatches at their 3′-ends . In cases, where there is no mismatch, miRNA can induce degradation of the target mRNA sequence . CircRNAs are generated by back-splicing or non-colinear splicing of pre-mRNA molecules and may include both exonic and intronic sequences . In addition to competing with canonical splicing and controlling the levels of the corresponding protein-coding mRNAs, circRNAs can also act as protein decoys or miRNA sponges to regulate gene expression .
2.1 DNA methylation and demethylation machinery
Of the different epigenetic mechanisms influencing gene expression described above, DNA methylation-mediated regulation of gene expression is the most-well studied. DNA methylation is established and maintained by DNMT family of enzymes whereas different mechanisms exist for demethylation (Figure 1A). Of the four members of DNMTs that facilitate DNA methylation, DNMT3L does not have an active methyltransferase (catalytic) domain. DNMT3A and 3B are
2.2 DNA methylation studies in schizophrenia
Initial studies on DNA methylation differences between SZ patients and controls, and among discordant monozygotic twins focused on candidate genes identified by genetic studies. For example, Abdolmaleky et al.  by using DNA from frontal lobes of post-mortem brain samples showed ∼50% increased methylation in the RELN promoter. Subsequent DNA methylation studies focused on genes involved in Dopaminergic , GABAergic , Glutamatergic , serotonergic pathways  of neurotransmission and genes such as BDNF . These studies used DNAs either post-mortem brain samples or peripheral blood lymphocytes. However, the data did not always yield consistent reports. For example, in case of BDNF promoter IV, decreased DNA methylation was observed in peripheral blood in a study by Kordi et al.  whereas, Ikegame et al.  and Ümit Sertan Çöpoğlu et al.  reported no change in the methylation levels in the same tissue. Subsequent studies which used genome-wide methylation analysis identified many genes showing statistically significant differences in DNA methylation, but the effective values or the degree of methylation differences observed were not large enough to demonstrate a biological effect such as altered expression. For example, in one of the first studies, Mill et al.  by using microarrays identified genes RPL39 and WDR18 with increased methylation in the promoter upstream regions of 8% and 3%, respectively. Studies conducted after these observations used a variety of technologies such as Methylated DNA Immunoprecipitation (MeDIP) – sequencing and Illumina-27 K and 450 K arrays and reported differentially methylated sequences with low effective values. Importantly these studies identified genes with little or no overlap among the top gene hits corresponding to the most significant differentially methylated sites . Nevertheless, some of these genome-wide studies also identified methylation differences in candidate genes such as COMT , GAD1, RELN  and BDNF . Although these genome-wide studies did not yield common genes with significant differences in DNA methylation, bioinformatic analyses revealed common pathways. For example, methylome data using the blood DNAs revealed the involvement of functioning of the immune system . This in turn is in agreement with the genome-wide association studies that identified immune-related genes including the major histocompatibility locus . Another common pathway identified in both blood– and DNA- based studies is the neurodevelopmental processes . The DNA methylation studies were also extended to study the effects on gene expression. In one such study, Liu
2.3 Dysregulated DNMTs in schizophrenia
Epigenetic processes and epigenetic modifications are tightly controlled to enable normal mammalian development. In this context, the presence of aberrant DNA methylation patterns affecting the candidate genes suggests the possibility of the role of dysregulation of epigenetic machinery in SZ. Investigations on dysregulation of DNA methylation machinery in SZ dates back to 2005 when Veldic
Taken together, there is reasonable argument for DNMT1 and DNMT3A and, possibly DNMT3B overexpression as risk factors for SZ. However, the information on the number of genes dysregulated due to DNMT1 overexpression was limited only three (
2.4 Models of dysregulated DNMTs
In the light of reports suggesting increased DNMT1 and/or DNMT3A levels as risk factors for SZ, it is important to understand the effects of their overexpression on neurodevelopment. Unfortunately, overexpression of DNMT1 results in mid-gestational lethality in mice  making it impossible to generate animal models with constitutive overexpression. In addition, reduction of DNMT1 protein levels, but not its absence, appears to be an essential step for differentiation . In this context, it is also difficult to generate mice conditional alleles of
In order to study the effects of DNMT1 overexpression on neurogenesis, D’Aiuto
In a recent study, Saxena
In conclusion, molecular details that connect DNMT1 overexpression with abnormal neurogenesis are beginning to emerge. With the availability of genome-wide methylation and transcriptome analysis methods, it is now possible to investigate the effects of DNMT1 overexpression in post-mortem brain samples of SZ patients. However, this effort requires an understanding on the incidence of DNMT1 overexpression in these samples. Of particular interest is to compare the effects of overexpression of DNMT1 or DNMT3A or both during the process of neuronal differentiation and the nature of the altered transcript levels. Whether the genes affected are only related to SZ or other neuropsychiatric disorders or neurodevelopmental disorders is an important question that needs to be addressed. Such information is useful to explore the contribution of epigenetic mechanisms in a wider spectrum of neurological disorders. In addition, improvement in the methods for generating genetically modified ESCs, their differentiation into specific types of neurons and development of brain organoids should help advance our understanding of the relationship between dysregulation of DNA methyltransferases and neurodevelopmental disorders such as schizophrenia.
Work in KNM’s laboratory is supported by grants from Science and Engineering Research Board, Department of Biotechnology and Birla Institute of Technology and Science Pilani. PAM received fellowship from a project funded by the Department of Biotechnology and later from Centre for Human Disease Research (BITS Pilani). KNM received OPERA award (BITS Pilani) and partial funding from the Centre for Human Disease Research.
Conflict of interest
The authors declare no conflicts of interest.