Epigenetic processes are characterized by their ability to regulate gene expression by altering the chromatin structure or its associated proteins in a manner independent of alterations in DNA sequences. Epigenetic modifications can be categorized into two groups as DNA methylation and hydroxymethylation, histone modifications.
DNA methylation was first identified in the early 1950s as the best understood modifications among the epigenetic processes (1, 2). 5-methylcytosine (5mC) is often referred to as the fifth base of the DNA code. DNA methylation involves covalent attachment of a methyl group (CH3) to the fifth position of carbon in the cytosine within CG dinucleotides with resultant formation of 5mC. The symmetrical CG dinucleotides are also called as CpG, due to the presence of phosphodiester bond between cytosine and guanine. The human genome contains short lengths of DNA (~1,000 bp) in which CpG is commonly located (~1 per 10 bp) in unmethylated form and referred as CpG islands; they commonly overlap with the transcription start sites (TSSs) of genes. In human DNA, 5mC is present in approximately 1.5% of the whole genome and CpG base pairs are 5-fold enriched in CpG islands than other regions of the genome (3, 4). CpG islands have the following salient features. In the human genome, more than 60% of mammalian gene promoters are enriched with CpG islands. The CpG islands are generally hypomethylated in every tissue during all stages of development. In the human brain, the methylation of CpG dinucleotides fluctuates largely across the cerebral cortex, cerebellum, and pons (5). DNA methylation levels are gradually enhanced in the human cerebral cortex at many genomic loci at the time of maturation and aging. In addition, both hypermethylation at HOXA1, PGR, and SYK loci and hypomethylation at S100A2 loci are observed in cortical neurons of aged samples (6). This bidirectional methylation variation during lifespan indicates that methylation levels are regulated dynamically in differentiated neurons.
The role of DNA methylation in gene expression depends on the genomic CpG location. Methylation at gene promoter or enhancer region is inversely correlated with gene expression (7, 8) whereas gene body methylation is positively associated with transcription (9–11). The repression of transcription by DNA methylation is explained by two mechanisms, a) methylated cytosine in CpG dinucleotide is identified by 5-methylcytosine-binding transcription factor (MeCP2), that recruit co-repressor protein complexes such as mSin3A and histone deacetylases thereby causing gene repression (12), b) hypermethylation blocks the recognition of transcription factor NRF1 binding sites and result in transcription repression (13). It is not surprising to note that DNA methylation controls several processes including gene expression in X-chromosome inactivation (14), genomic imprinting (15, 16), long-term memory formation (17), carcinogenesis (18), and virus, transposons or retrovirus silencing (19) to ensure genomic stability (20).
The transfer of methyl groups to the DNA from donor S-adenosyl-L-methionine (SAM) is catalyzed by a group of enzymes referred to as DNA methyltransferases (DNMTs). The DNMT1, DNMT2, and DNMT3 enzymes are the key members of the DNMT family (21). During development, tissue-specific DNA methylation patterns are established by
Demethylation of DNA can be achieved by either passive or active ways. DNA methylation marks are maintained in a
The single base pair changes in the DNA sequence, which can influence the levels of gene expression, are called an expression quantitative trait loci (eQTLs). Methylation quantitative trait loci (meQTLs) are the individual DNA sequence variation at specific loci that can cause changes in DNA methylation patterns of CpG sites (Fig. 1B). The meQTLs have been found to overlap with eQTLs and exhibit similar biological mechanism by which the DNA sequence variation affects both expression and methylation. The presence of meQTLs at regulatory sequences can result in altered binding of transcription factors thereby leading to changes in transcription events. Importantly, the meQTL can have an impact on gene regulation based on developmental stage and environment. Traditionally, single nucleotide polymorphism (SNP) associations are known to linked to differences in inheritance to a range of diseases. Similarly, recent researches have shown that meQTLs can also act as biomarkers for the diagnosis of various diseases.
Analysis of DNA methylation aimed at quantifying the total amount of 5-methylcytosine in the genome is termed as global methylation analysis and elucidation of the methylation status of the specific gene loci is called as local methylation analysis. The bisulfite sequencing is the benchmark for the local methylation analysis, although the amplification of long DNA fragments is hard due to DNA fragmentation. Restriction enzyme-based assay depends on the usage of the methylation-sensitive enzyme,
The process of aging is associated with numerous alterations at the cellular and molecular levels, such as cell senescence, telomere shortening, stem cell exhaustion, and gene expression changes. Epigenomic remodeling is also changed during the lifespan, suggesting that changes in DNA methylation form a vital element of the aging process.
Initial studies reported that there are approximately 45,000 CpG islands per haploid genome in humans and 37,000 in the mouse (41). Approximately half of the mammalian gene promoters are linked with one or more CpG islands and are often free of methylation (42–44). The examination of CpG islands in genomic DNA of human blood, muscle, and spleen showed differential methylation at
The changes in DNA methylation in blood have been reflected in brain regions and other tissues. In general, most of the tissues follow the pattern of rapid changes in DNA methylation in early life followed by a gradual decrease later in life. The analysis of human postmortem brain from adult subjects of different ages has led to the identification of such changes. The analysis of CpG methylation in the human prefrontal cortex (PFC) of fetal and elder samples with a focus on 50 promoter regions showed that both sex and age play important roles in DNA methylation at CpG islands on the X chromosome. Importantly, NNAT (neuronatin) expression in the PFC development drops significantly, as it progresses from the fetal to adult stage whereas DNA methylation showed the reverse pattern. Genes involved in schizophrenia and autism (DRD2, NOS1, NRXN1, and SOX10) showed active age-relevant methylation variation during development from fetal to postnatal life although the study was unable to differentiate between methyl and 5-hmC which are abundantly present in the brain (46). The chronological age measures the time passed since birth whereas biological age measures the biological state of an individual. Biological age is loosely defined and used to estimate tissue and organ functional decline, risk of disease linked with age, morbidity, and mortality. Biological age is also termed as physiological age, organismal age or phenotypic age. The biological aging of the same tissues can vary across individuals with the same chronological age. The DNA methylation analysis studies involving various regions of the brain at different ages demonstrate a correlation with chronological age. The examination of the DNA methylation in 387 human brain regions aged from 1 to 102 years revealed that DNA methylation sites (such as PIPOX, DPP8, RHBDD1, FLJ21839, PTGER3 genes) are highly associated with chronological age (47). The above studies are also supported by the finding that alterations in DNA methylation at clusters of CpG sites across multiple tissues including blood and brain are usually related to aging. The DNA methylation patterns of blood mimic the brain during aging (48). Previous reports showed a global decline in DNA methylation, whereas site-specific examination depicted a higher variability in methylation levels in blood cells with aging. These studies also suggest that the loss of regulation of maintenance of DNA methylation occurs over the years. Recent genome-wide investigations in newborns and centenarians demonstrate a more specific pattern; the majority of intergenic CpG sites had a hypomethylation mark, whereas most of the CpG islands remain hypermethylated with aging (49).
The application of machine learning methods to predict biological age based on site-specific DNA methylation alteration has led to the proposal of the term “epigenetic clock” (50, 51). The epigenetic clock predicts the age of human cells, tissues, and organs across individuals based on methylation at specific CpG sites (52). The epigenetic clock predicts biological age more efficiently than chronological age. Hannum’s clock based on methylation at 71 CpGs was analyzed in adult human whole blood to determine the biological age. The Horvath’s clock is based on methylation status at 353 CpGs tested in various sources of cells, tissues, and organs across the lifespan. The age-associated methylation of CpGs profiled in blood, kidney, and skeletal muscle samples hold also true in the brain. Interestingly, CpGs linked with tissue-specific gene expression are hidden from common methylation changes with age (53). The methylation levels at CpG sites of MLH1, TP53, somatostatin (SST), KLF14, methyl-CpG binding domain protein 4 (MBD4), NEK4, JAKMIP3, STEAP2, and ELOVL2 have been reported to be significantly connected with the chronological age of individuals, proving the occurrence of aging-related methylation changes in an expected pattern (51). In addition, genes with promoter hypermethylation have been reported to be enriched in Polycomb group targets (50). Most recently, Levine
Epigenetic drift refers to a large number of CpG methylation changes that act as a biomarker for age within an individual but are not uniform across the individuals. Epigenetic drift has been connected to a multitude of age-related status that includes cell type or organ, neurodegenerative diseases, cancer, gender, body mass index (BMI), lifestyle, and demographic variables (55). The 30 anatomic tissues of super centenarians (people who have attained the age of 110 years or older) showed that cerebellum ages more slowly than other parts of the human body. Interestingly, the genetic and transcriptional studies concluded that RNA helicase genes highly expressed in the cerebellum slows the aging rate of supercentenarians (56). Methylation data from human cortex predicted age-related neurodegeneration. Further, accelerated epigenetic age has been reported as heritable and to exhibit a significant association with amyloid load, diffuse plaques, and decline in working memory (57).
Seminal studies conducted by Lister
In summary, differentially methylated genes identified from three epigenetic clocks were associated with various biological processes and molecular functions resulting in control of the aging process (Table 1). It is hypothesized that further experimental investigations in addition to the present statistical model and analysis might possibly formulate ways to retard biological aging by focusing on the age-linked DNA methylation levels.
In the last decade, the importance of DNA methylation in the functioning of the central nervous system has gained immense attention (61–68). Numerous studies have investigated patterns of DNA methylation at various cellular stages such as stem cells and terminally differentiated cells like neurons and glia. A few research groups have focused their attention on pathological stages of neurodegenerative diseases. In the subsequent sections, we aim to discuss recent findings of DNA methylomes in Alzheimer’s disease, Parkinson’s disease, ALS/FTD, Huntington Disease, and Multiple sclerosis.
Presently, there are 46.8 million dementia cases worldwide and the number is expected to rise to 74.7 million in 2030. Health care burden of dementia in 2015 exceeded
Simultaneous studies published by two independent groups have provided the first complete DNA methylome of AD. These two independent studies found four unique hypermethylated genes i.e. ANK1, RHBDF2, RPL13, and CDH23 (67, 68). Jager
To address these limitations, Sanchez-Mut
In support of prior AD methylome studies, Yu
To further explore whether methylation at CpG sites located on AD susceptibility gene regions is associated with Aβ burden, the study involving 740 brain samples by Illumina 450K array reported that the observed methylation changes are independent of AD-associated genetic variants in BIN1, CLU, MS4A6A, ABCA7, CD2AP, and APOE loci. The higher methylation of CpGs in ABCA7, CD2AP, CLU, and MS4A6A loci were stated to be associated with Aβ plaque (79).
Studies investigating the role of DNA methylome in neurodegenerative diseases like AD are still at the commencement stage. Further biochemical and
In conclusion, many methylome studies in AD brain have demonstrated that DNA methylation changes at a few common gene loci play a critical role. These studies are encouraging since DMR in ANK1, RHBDF2, Bin1, and ABCA7 loci are linked with the formation of Aβ plaque. Identification of the implications of DNA methylation changes at identified CpG loci on cellular signaling pathways needs to be investigated in the future.
Parkinson’s disease (PD) is identified as the most frequently occurring neurodegenerative disorder after AD. Approximately, 7 to 10 million people worldwide are predicted to be affected by PD. Parkinson’s disease is diagnosed by the early symptoms of bradykinesia, muscular rigidity, rest tremor, postural and gait impairment, and late symptoms include postural instability, freezing of gait, falls, dysphagia, and speech defects (80). Pathologically, PD is characterized by the formation of intraneuronal inclusions called Lewy bodies, composed of α-synuclein (α-syn), along with degeneration of dopaminergic neurons mainly in the substantia nigra. Familial cases of PD, directly affecting ~15% of all cases have been reported to be caused by mutations in the LRRK2, PARK7, PINK1, PRKN, or SNCA genes (81). PD develops from a complex interaction of genetics and environment. PD is now considered as a slowly progressive neurodegenerative disorder that initiates years before the appearance of symptoms. DNA methylation patterns are modified by environmental factors and can be inherited through cell division. Possibly, environmental factors may cause alterations in DNA methylation patterns, which determine the individual genetic susceptibility to PD. In addition to the acquired mutations in one gene or a group of genes involved in the progress of several disorders, DNA methylation alterations have been found to contribute essentially to their development. Therefore, it is not surprising to note that deregulation of DNA methylation can be critical for the onset of PD.
Genome-wide association studies (GWAS) identified several PD risk loci in the cerebellum and frontal cortex of post mortem PD brains, namely PARK16, GPNMB, and STX1B genes that were associated with differential DNA methylation (82). To investigate DNA methylation alterations in PD, an epigenome-wide association study (EWAS) was performed. The examination of cortex and putamen regions from PD patients using Illumina 27K array showed hypomethylation of the CYP2E1 gene and up-regulation of mRNA levels of CYP2E1 in PD cortex samples (63). Masliah
In summary, since there exists only a few human brain methylome studies in PD and differentially methylated genes commonly found among different studies are not many, future studies must focus on the association of DNA methylation variation with the occurrence of pathological hallmarks of PD.
ALS and FTD are two lethal neurodegenerative diseases with an occurrence of 30,000 cases alone in the US leading to comorbidity in about 50% of cases. FTD and ALS are investigated as a part of a common spectrum and share similar neurodegenerative pathways (85). Symptoms of FTD include dementia, commonly portrayed as behavioral, personality, and speech problems caused by degeneration of frontal and temporal cortical neurons. ALS patients exhibit reduced control over voluntary movements caused by progressive degeneration of lower motor neurons and loss of upper motor neurons resulting in muscle wasting. FTD is the second most prevalent dementia after AD. The finding of numerous mutations in two dozen genes explained the involvement of several pathways in the development of ALS/FTD, while the pinpointing of hexanucleotide repeat expansion (HRE) in C9orf72 accounted for up to 80% of familial ALS-FTD (86).
The HRE mutation in C9orf72 is hypothesized as the cause of disease pathogenesis involving three mechanisms: 1) The HRE mutation in C9orf72 may cause a reduction in the expression of its mRNA in the allele harboring hexanucleotide repeat (haploinsufficiency); 2) RNA foci with transcribed G4C2 repeats may cause sequestration of RNA-binding protein(s); and 3) RANT (repeat-associated non-ATG mediated translation) aggregates may lead to accumulation of toxic RNA and protein pathologies. Aberrant effects of unstable DNA elements targeting the genome have been reported to be neutralized by suppression of the incorrect gene expression via CpG methylation. C9orf72 promoter is hypermethylated in 30% of mutation carriers in
Recent methylome study carried out in ALS/FTD with few patient samples has potentially demonstrated the beneficial effects of CpG methylation in ALS/FTD. It is believed that further studies with a large number of samples might implicate the association of methylation with other mutated genes in ALS/FTD.
In western countries, Huntington’s disease (HD) has been reported in 10.6–13.7 person per 100,000 people. The affected individuals display cognitive decline with sleep disturbances, weight loss, and emotional disturbance followed by involuntary movement disorder (Chorea). The disorder can typically manifest between infancy and senescence (90). HD is a progressive autosomal dominant neurological disorder caused by CAG trinucleotide repeat expansion of the huntingtin (HTT) gene. The CAG repeat lengths of ≥ 40 units are translated into polyglutamine strand of the pathogenic HTT protein. Mutant HTT is resistant to protein turnover and highly penetrant resulting in cellular toxicity and neurodegeneration (91). Approximately, 60–70% of the variability in the age of onset identified in HD cases showed an inverse correlation with CAG repeat length. Another ~30% of the variability in the age of onset may be contributed to genetic, epigenetic, and environmental factors (92). Therefore, alteration in DNA methylation is considered to affect HTT gene expression by changing the transcription of the HTT promoter.
Adenosine A2A receptor (ADORA2A), a G-protein-coupled receptor is abundantly expressed in basal ganglia, while its expression levels are markedly decreased in HD. Examination of putamen of 10 HD patients showed hypermethylation of CpG islands and decreased 5-hmC in the 5′UTR region of ADORA2A. Thus, hypermethylation of ADORA2A is the molecular mechanism behind the decreased expression of A2AR in HD (93). EWAS (epigenome-wide association study) analyses in the cortex of small HD samples revealed the lowest evidence of alterations in HD-related DNA methylation. However, DNA methylation may be linked with the age of disease onset in HD cortex samples. Intriguingly, the researchers observed a site-specific decrease in the differences in DNA methylation at the HTT proximal promoter specifically at CTCF-binding site; the CTCF site displayed increased occupancy in cortex tissue, which is responsible for cortex-specific HTT expression (92). However, these studies had several limitations; 1) it is possible that HD-relevant DNA methylation alterations are present at sites beyond the analyzed sites, 2) as CAG repeat expansion leads to HD mutation; it is possible that the expanded HD CAG repeats in human cortex will be influenced by aberrant non-CpG methylation, and 3) a small number of diseased samples.
It is impulsive to reach a conclusion on the role of DNA methylation in HD based on the findings of a couple of studies. Well-designed methylome studies in HD cases might yield novel therapeutic target for developing a treatment strategy for HD.
Multiple sclerosis (MS) is an autoimmune, chronic inflammatory, and demyelinating disease of the central nervous system. MS affects approximately 2.5 million people worldwide, especially young adults and women. The physical symptoms include fatigue, mobility challenges, bladder or bowel problems, tremor, vision problems, problems with coordination, and cognitive difficulties (94). MS is characterized by the presence of lesions in the CNS with focal destruction of myelin, which is considered as a pathological hallmark of MS. However, the surrounding normal-appearing white matter (NAWM) shows the abnormalities such as edema, axonal damage, and reactive gliosis which contributes to progressive brain atrophy (95). GWAS have identified MS susceptibility loci; however, their functional importance related to MS pathogenesis is still unexplained. The factors such as the occurrence of SNPs in monozygotic twins is relatively low. High susceptibility of women, geographic risk of developing MS, and the influence of migration on disease onset suggest that the epigenetic changes play important roles in the pathogenesis of MS.
Earlier reports revealed increased citrullination of histone H3 and acetyl-H3 in NAWM of MS brain (96, 97) and mentioned the prospect that DNA methylation alterations might regulate gene expression. Interestingly, NAWM derived from MS brain analysis showed hypermethylation of oligodendrocyte and neuronal function-related genes BCL2L2, HAGHL, and NDRG1 and reduced transcript levels. In addition, hypomethylation was observed in cysteine proteases such as LGMN and CTSZ along with an increase in their transcript levels (98).
This review presents a comprehensive summary of the human brain DNA methylome during aging and neurological diseases. Neurodegenerative diseases are multifactorial and associated with age, environmental, and genetic factors. The human DNA brain methylome studies have opened a new door for understanding the CNS diseases. The recent development of techniques assists in finding accurate levels of different DNA methylation modification throughout the genome. The reported studies suggest the followings. 1) The DNA methylation alterations happen initially in the disease process, 2) DNA methylation alterations alone or in combination with disease-specific SNPs can enhance disease susceptibility (Fig. 1C), and 3) DNA methylation changes can be correlated with misfolded proteins of the neurodegenerative diseases in specific brain regions. The key features of human DNA methylome studies are discussed in Table 2. The identified gene loci may potentially help in understanding the cause and early diagnosis of CNS diseases, and development of small molecules for treatment. Although, the DNA methylome studies in neurodegenerative diseases are in their beginning stage; they are providing a novel path towards new research direction. The
This work was supported by the 2019 Research Fund of the University of Seoul.
The authors have no conflicting interests.
DNA methylation and its consequences. (A) DNA methylation and TET mediated DNA demethylation. Cytosine (C) is methylated at 5th carbon of the pyrimidine ring by DNA methyltransferases (DNMT) to form 5-methylcytosine (5-mC). Ten-Eleven Translocation (TET1-3) enzymes sequentially act on 5mC to generate 5 hydroxymethylcytosine (5-hmC), 5-formylcytosine (5-fC), and finally 5-carboxylcytosine (5-caC). The 5fC and 5caC are excised directly by thymine DNA glycosylase (TDG). The resulting abasic sites, which are generated by TDG since it has the ability to catalyze the glycosidic bond between the base and deoxyribose sugar of DNA, are eventually replaced with unmethylated cytosines by base excision repair (BER). (B) Relationship between SNP, CpG site, eQTLs, mQTLs, and neurodegenerative diseases. Genome-Wide Association Studies (GWAS) have identified several Single-Nucleotide Polymorphisms (SNPs) which indicate differences in the inheritance of diseases. The single base pair changes in the DNA sequence can affect the gene expression levels and referred to as expression quantitative trait loci (eQTLs). Methylation quantitative trait loci (meQTLs) are the individual DNA sequence variation at specific loci that can cause changes in DNA methylation patterns of CpG sites. Significant correlation of methylation mark with gene expression is termed as expression Quantitative Trait Methylations (eQTMs). Recent studies have shown that eQTLs, meQTLs, and eQTMS are linked with neurodegenerative diseases. (C) Differential CpG methylation and its association with neurodegenerative diseases.
GO analysis using the Enrichr tool for the CpG genes obtained from Horvath’s clock, Hannum’s clock, and DNAm PhenoAge with described methylation alterations in aging
|Gene Ontology||GO Terms||P-value||Genes|
|Biological Process||Negative regulation of programmed cell death||3.0E-04||DDR1;LTK;FOXE3;PRKAA2;GLO1;FHL2;PSEN1;HSP90B1;CASP3;RPS6KA1;EPHB1;NGFR;
|Unsaturated fatty acid biosynthetic process||4.72E-04||ALOX5;ALOX5AP;SCD5;DEGS1;ALOX12|
|Endocrine system development||9.5E-04||WT1;SIX1;GLI2;WNT4;DKK3;STRA6|
|Adrenal gland development||9.5E-04||WT1;WNT4;DKK3;STRA6|
|N-acetylneuraminate metabolic process||9.54E-04||AMDHD2;NPL;ST3GAL1;GNE|
|Metanephric mesenchyme development||1.4E-03||WT1;BASP1;SIX1;WNT4|
|Lipoxin biosynthetic process||2.6E-03||ALOX5;ALOX5AP;ALOX12|
|Response to fatty acid||2.6E-03||GLDC;PDK4;UCP1;NR1H4|
|Molecular Function||G-protein coupled receptor binding||8.2E-04||CALCA;PDCD6IP;USP20;WNT5B;WNT8B;ADM;REEP1;POMC;SFRP1;UCN2;BAMBI;PENK;
|Protein homodimerization activity||9.5E-04||GLDC;HM13;ODC1;NPR3;CAMK2A;PDGFB;CIB2;PVR;CDH9;SYNE1;CD79A;CREB3L3;
|U1 snRNA binding||4.0E-03||RBPMS;SNRPB2;PRPF8|
|Ether hydrolase activity||5.8E-03||EPHX2;EPHX3;ALOX12|
|Aspartic-type endopeptidase activity||1.4E-02||BACE1;CASP3;HM13;PSEN1|
|RNA stem-loop binding||2.1E-02||RBPMS;SNRPB2;LARP6|
GO, gene ontology; The significant terms with multiple genes at the biological process and molecular function characterization levels are shown. P-values show the statistical significance of the enrichment of gene ontology terms with analyzed genes.
Overview of human brain DNA methylome studies involving neurodegenerative diseases
|Disease||No of DMR||Examples of nearest genes||Brain Region||Sample Size||Reference|
|Alzheimer’s Disease||71||ANK1, CDH23, DIP2A, RHBDF2, RPL13, SERPINF1 and SERPINF2||Entorhinal, temporal, and prefrontal cortex||708||67|
|100||ANK1, MIR486, PCBD1, SLC15A4, SIRT6, MEST, MLST8, ZNF512, TMX4||Entorhinalcortex, superior temporal gyrus and prefrontal cortex||104||68|
|5||SORL1, ABCA7, HLA-DRB5, SLC24A4, BIN1||Dorsolateral prefrontal cortex||740||66|
|6||BIN1, CLU, MS4A6A, ABCA7, CD2AP, and APOE||Dorsolateral prefrontal cortex||740||79|
|Parkinson’s Disease||PARK16, GPNMB, STX1B, STBD1||Frontal cortex and Cerebellum||292||82|
|CYP2E1, Catalase, LOC148811, LOC84245||Putamen and cortex||6||63|
|6||FLJ32569, NOTCH4, SLC44A4, GPNMB, KIAA1267, STX1B2, SLC4A11||Frontal cortex and cerebellar regions||292||84|
|Methylated binding site in the HTT proximal promoter for the CTCF transcription factor||Cortex||10||92|
|Multiple Scelerosis||Hypomethylated: 495||AKAP6, CX3CL1, GATA3, LRRC27, MBL2, ARHGAP22, CTSZ, MGAT, BCL2L2, SBF1,||Normal appearing white matter||28||98|
|Hypermethylated: 439||SLC47A2, CRY2, NDRG|
DMR, Differentially Methylated Regions.