Review | Open | Published:
BDNF DNA methylation changes as a biomarker of psychiatric disorders: literature review and open access database analysis
Behavioral and Brain Functionsvolume 12, Article number: 17 (2016)
Brain-derived neurotrophic factor (BDNF) plays an important role in nervous system development and function and it is well established that BDNF is involved in the pathogenesis of a wide range of psychiatric disorders. Recently, numerous studies have associated the DNA methylation level of BDNF promoters with certain psychiatric phenotypes. In this review, we summarize data from current literature as well as from our own analysis with respect to the correlation of BDNF methylation changes with psychiatric disorders and address questions about whether DNA methylation related to the BDNF can be useful as biomarker for specific neuropsychiatric disorders.
Brain-derived neurotrophic factor (BDNF) is a member of the neurotrophin family which plays an important role in neural differentiation, survival of nerve cells, neurite outgrowth, and synaptic plasticity. BDNF has been shown to regulate the development, plasticity and survival of dopaminergic, cholinergic and serotonergic neurons. Also, it regulates glutamatergic neurotransmitter release and promotes the development of GABAergic neurons. BDNF is widely expressed throughout the mammalian brain, including the cerebral cortex, hippocampus, basal forebrain, striatum, hypothalamus, brainstem, limbic structures and cerebellum . This makes BDNF a key factor in learning and memory, reward-related processes, cognitive function and circuit formation.
The human BDNF has a complex gene structure, consisting of 11 exons (I–V, Vh, VI-VIII, VIIIh, IX), 9 of which (exon I–VII, IX) contain functional promoters. Exons II, III, IV, V, Vh, VI, and VIIIh do not have a translation start site so translation of these exons starts from the ATG of exon IX. All BDNF mRNAs contain the sequence for the pro-BDNF protein, encoded by exon IX. The use of translation start sites in exons I, VII, and VIII can lead to the pre-proBDNF proteins with longer N-termini. Alternative splice sites are situated in exons II, V, VI and in exon IX. In addition exon IX contains two alternative polyadenylation sites . The use of different splice sites leads to the formation of numerous BDNF transcripts variants that determine a tissue-specific BDNF expression regulation as well as the regulation in responses to environmental stimuli and signaling events [2, 3]. mRNAs transcribed from the non-protein-coding antiBDNF gene and forming duplexes with BDNF mRNAs may play an important role in the regulation of BDNF expression . Additionally, BDNF expression is regulated at the posttranscriptional level by enzymatic cleavage of the pro-BDNF protein into a mature BDNF protein. Pro-BDNF and BDNF can interact with two distinct transmembrane receptors, the p75 neurotropin receptor (p75NTR) and the tropomyosin-related kinase receptor B (TrkB). While pro-BDNF binds preferentially to p75NTR, inducing neuronal apoptosis and long-term depression, mature BDNF binds to TrkB and promotes a downstream signal cascade, leading to neuronal differentiation and survival, neurite outgrowth as well as synaptic plasticity [3, 4].
The BDNF protein has been extensively studied as an important factor involved in the pathogenesis of a wide range of psychiatric disorders, such as major depression, schizophrenia and bipolar disorders. The use of several animal models have also implicated BDNF in anxiety-like behaviors and have shown a decreased BDNF expression in response to different types of stressors [5, 6]. A Val66Met polymorphism (rs6265) of BDNF has been associated with different psychiatric conditions. This polymorphism is related to intracellular traffic, synaptic location, secretion of BDNF as well as poorer working memory performance, reduced cerebellar and hippocampal volumes, and cognitive ability [7, 8]. The polymorphism has been demonstrated to modulate a range of clinical features in schizophrenia patients. Higher frequency of the Met allele has also been associated with depression, anxiety, anorexia, bulimia nervosa and suicidal behavior [9, 10].
Numerous studies have addressed the association between psychological disturbances and levels of BDNF protein in serum or plasma. There are several sources of BDNF in blood: BDNF might be secreted by platelets, immune and vascular endothelial cells [11, 12]. BDNF is also able to overcome the blood–brain barrier . With the correlations reported between BDNF levels in the brain and blood, altered BDNF levels has been established as characteristic for several psychological disorders. Low serum BDNF protein levels have been found in patients with depression, schizophrenia, anxiety and borderline personality disorder [14–17]. While certain treatments with antidepressants increased levels of BDNF in serum , a pre-treated level of BDNF has been shown to possibly predict the response to antidepressant treatment and has been correlated with depression rating improvements during therapy . However, such clinical parameters as age of onset, the severity of symptoms as well as clinical effectiveness of the treatment were not reflected by BDNF levels in blood . Additionally, proBDNF and the mature BDNF form cannot be discriminated by measurement in blood although they affect neuronal cells differently . Since alterations of BDNF levels in blood associate with several psychological disorders, it cannot be considered as a specific biomarker for certain disease.
Epigenetic alterations as DNA methylation, histone modifications and non-coding RNAs are considered to be strongly associated with pathogenesis of psychiatric disorders (reviewed in [19, 20]) with a large number of studies having examined the association between the BDNF methylation level and certain psychological diagnoses. Epigenetic mechanisms have been shown to be very important for BDNF expression regulation [3, 4, 21, 22]. DNA methylation-related chromatin remodeling of BDNF regulatory regions may play critical roles in regulating gene transcription in response to neuronal activity [23, 24]. Epigenetic modifications, on the other hand, may potentially provide robust and stable biomarkers of disease activity. In this review, we have summarized the information about individual CpG sites and CpG regions previously tested in BDNF promoters and shown to be connected to psychiatric disorders. Using open access databases, we have analyzed the DNA methylation level of BDNF promoters in brain samples of individuals with psychiatric disorders. This review addresses the question of whether BDNF methylation changes might be regarded as suitable biomarkers of a specific disorder.
A role of DNA methylation in gene expression
Epigenetic changes are gene expression and phenotype alterations heritable through cell division that are not caused by modifications in DNA sequences . The main epigenetic mechanisms include DNA methylation, histone modifications and non-coding RNAs. DNA methylation is the first described and the most studied epigenetic modification . Mammalian DNA methylation occurs predominantly in the nucleotide sequence 5′CpG3′. There are about 28 million CpG sites in the haploid human genome. CpG islands are defined as those sequences that have a length greater than 200 bp, a CpG content of at least 50 % and a CpG frequency greater than 0.6 of observed number to the expected number of CpGs . The promoter regions of 60–70 % of all human genes contain CpG islands . Traditionally high density CpG islands have been considered mostly hypomethylated [27, 28], however more recent studies have shown that dense CpG islands, depending on cell type, might be predominately hypermethylated . A DNA methylation pattern is established and maintained by DNA methyltransferases (DNMT) namely de novo DNMT3A and DNMT3B and the maintenance DNMT1 . DNA methylation affects gene expression, involving multiple mechanisms. Gene expression silencing is mediated by methylated DNA by attracting transcriptionally repressive methyl-CpG-binding proteins (MBPs). In turn, these proteins recruit histone deacytelases (HDACs) and chromatin-remodeling complexes, such as an NCoR-SMRT complex (nuclear receptor corepressor and silencing mediator for retinoid and thyroid hormone receptors), a NuRD (Nucleosome Remodeling Deacetylase) complex, which results in gene repression. The DNA methylation of a gene enhancer may also lead to gene repression. On the other hand, unmethylated CpG sites are bound by methyl-sensitive transcription factors, CXXC domain-containing activator complexes, thereby contributing to gene expression (reviewed in ). Unmethylated CpG sites might also serve as transcriptional factors “landing lights”, marking gene promoter regions which distinguishes them from the transcriptionally irrelevant intergenic chromatin . Gene body DNA methylation may have a positive influence on gene transcription  and intragenic DNA methylation may affect alternative gene splicing . Other interesting findings are the correlation between the density of gene-body DNA methylation and replication timing , and the influence of 5′UTR and 3′UTR DNA methylation on the elongation and termination of transcription .
Initially DNA methylation has been associated with prevention of particular gene expression. However, recent studies have introduced a more complicated impact of DNA methylation on gene expression regulation with several studies having identified a poor correlation between methylation levels of some genes and their expression level . Differential DNA methylation of CpG islands associated with repressed genes has been found between somatic cells  which might imply stochastic DNA methylation of repressed genes. A few studies indicated that binding of transcription factors may precede DNA methylation changes in the enhancer, suggesting an inactive role of DNA methylation in enhancer activity . Thus, DNA methylation is a complex epigenetic modification, which does not always determine gene expression activity. Several studies align with the idea that DNA methylation changes might be triggered by an antisense transcription, changes in histone modification and chromatin protein activity and thus might be a consequence rather than a reason for gene regulation [34–36].
Dynamic changes in DNA methylation are considered important mechanisms in the developmental regulation of gene expression. Embryonic stem cells are pluripotent and characterized by hypomethylation of CpG islands. During differentiation, genes essential for cell specification remain unmethylated while opposite genes, specific for other cell line development, are kept methylated. According to this notion, somatic cells demonstrate different DNA methylation patterns of a number of genes. A genome-wide analysis, comparing the DNA methylation profile across brain and blood, reveals highly tissue-specific differences in DNA methylation between different cortical regions, cerebellum and blood. It has also been shown that tissue-specific differentially methylated regions (TS-DMR) across the cerebellum and frontal cortex are associated with stable gene expression differences . One of most interesting findings was an over-representation of intragenic CpG islands and an under-representation of promoter-associated CpG islands among TS-DMR. Low density CpG promoters were characterized by widespread tissue-specific DNA methylation across brain regions and blood, in comparison with CpG-rich promoters . An intriguing recent study has also demonstrated a significant underrepresentation of promoter regions and an overrepresentation of CpG shores and shelves, gene bodies, as well as underrepresentation of CpG-rich promoters among fetal brain DMRs . Another whole genome DNA methylation study has shown that similar pathways are affected in the brain and blood of Parkinson’s patients. Differently methylated regions between the blood and brain have been predominately presented in high-methylation fractions which are associated with gene bodies and intragenic regions .
5-hydroxymethylcytosine (5hmC), derived from oxidation of 5-methylcytosine (5mC) by the Ten-Eleven Translocation (TET) enzymes, is now considered a new epigenetic DNA modification with relevant roles in regulating DNA demethylation and transcription. 5hmC is generally associated with transcribed genes promoters and bodies, positively correlated with transcription levels and detected in the mammalian genome in all cell types, with the highest content present in the brain .
The role of DNA methylation in the regulation of Bdnf expression was actively investigated by Martinowich and colleagues . The authors found a three-four times higher level of the Bdnf exon IV transcript in Dmnt1 mutant mice in comparison with controls. It was also demonstrated that the activation of Bdnf transcription is regulated by neuronal activity. Enhanced transcription of the Bdnf exon IV promoter was observed in mouse embryonic cortical cells treated with 50 mM KCL, which is known to simulate neuronal activity by activation of voltage-sensitive calcium channels leading to calcium influx and membrane depolarization. More importantly, the authors identified that the region upstream of the Bdnf promoter IV transcriptional start site contains Ca2+-responsive elements, namely the calcium-responsive element 1 (CaRE1), an upstream stimulatory factor-binding site (E-box) and a cyclic adenosine monophosphate (cAMP) response element (CRE), which overlap with several CpG sites. A site-specific DNA methylation in combination with a luciferase activity test showed that DNA methylation of some of these CpG sites can significantly inhibit the Bdnf promoter IV activity induced previously by membrane depolarization. Further experiments showed that the methylation level of several CpG sites in the Bdnf exon IV was significantly lower in KCl-treated culture of mouse E14 cortical cells compared with control culture. This confirms that the methylation level of the Bdnf promoter IV can be changed upon depolarization. Subsequent chromatin immunoprecipitation analysis revealed that methyl-CpG-binding protein MeCP2 (transcription repressor) is more tightly associated with methylated DNA within the Bdnf exon IV than CRE-binding protein (transcription activator). Coimmunoprecipitation assay demonstrated that the histone deacytelase HDAC1 and corepressor mSin3A also associated with the Bdnf promoter. All three proteins dissociated from the promoter upon membrane depolarization suggesting that Bdnf expression activation requires the dissociation of the MeCP2-HDAC-mSin3A repression complex. All together, these findings indicated that Bdnf expression level is determined by the DNA methylation pattern and chromatin modifications which, in turn, can be regulated by membrane depolarization. In a latter study, the regulation of human BDNF transcription by membrane depolarization was confirmed . The authors indicated that neuronal activity-regulated transcription of human BDNF promoter I depends primarily on the novel asymmetric E-box-like element, PasRE (basic helix-loop-helix (bHLH)-PAS transcription factor response element), which is bound by the bHLH-PAS transcription factors ARNT2 (aryl hydrocarbon receptor nuclear translocator 2) and NPAS4 (neuronal PAS domain protein 4). While neuronal activity-regulated transcription of the BDNF promoter IV is regulated predominately by CRE, PasRE elements and the upstream stimulatory factor binding element (UBE). The CRE and PasRE elements overlap with CpG sites (Fig. 1), however a separate study is necessary to investigate if DNA methylation of these CpG sites can influence transcription of the BDNF promoters following membrane depolarization.
Several studies in rats have demonstrated that stressful environment conditions, such as an early-life adverse experience or maltreatment, may induce long-lasting changes in the methylation level of the Bdnf promoter IV which is associated with the lower Bdnf expression level in prefrontal cortex. Moreover, the offspring derived from maltreated-females showed hypermethylation of the Bdnf in the prefrontal cortex and hippocampus, suggesting that DNA methylation modifications might be inherited across generations [42, 43].
DNA methylation of BDNF in psychiatric disorders
DNA methylation alterations of BDNF, in connection to various psychiatric disorders, have been extensively studied in last several years (Table 1). Firstly, BDNF was studied by applying the enrichment microarray analysis; Mill and colleagues  tested the methylation level of several BDNF regions. They did not reveal any difference in the methylation level in patients with schizophrenia and bipolar disorders in comparison with controls. However, they identified an influence of rs6265 genotype, situated in exon IX on the methylation level of the surrounding region. Depending on the allele (C or T) of the rs6265, an additional CpG site can emerge in the region. CC (Val homozygotes) genotype carriers demonstrated a significant increase in methylation levels of four nearby CpG sites in comparison with CT and TT (Met homozygotes) genotype carriers. This is especially interesting, as several studies have shown the opposite effect of the Val/Val and Met/Met genotype on the schizophrenic and non-psychotic psychiatric disease phenotype . In the recent study of Chagnon and colleagues, a significantly higher methylation level of one BDNF region (exon VI, 3 CpG sites) was observed in older women with anxiety and/or depression compared with controls . This difference was more pronounced in CT genotype carriers of rs6265 in comparison with the CC genotype carriers. TT carriers were not found among both patients and controls. The higher DNA methylation in women with anxiety/depression compared with healthy controls was confirmed in a second small CT genotype carriers’ cohort (eight subjects). No difference was detected for CC and TT carriers. A comprehensive study assessing the whole BDNF methylation levels in a large cohort of CC, CT and TT carriers will be necessary to elucidate the impact of rs6265 on the methylation level of distal BDNF regions.
Keller and colleagues were the first to investigate the BDNF DNA methylation changes in suicide victims in comparison with controls . Using bisulfite pyrosequencing and direct bisulfite sequencing, EpiTYPER as a confirmation analyses, these authors analyzed the methylation levels of CpG sites at BDNF promoter/exon IV in the Wernicke area of the brain. A significant increase of the average methylation level of 4 CpG sites located downstream of TSS (+10, +16, +25, +28) (Fig. 1b) and methylation levels of two separate CpG sites (+10, +25) was found in suicide completers. Moreover, they also indicated an association between the average methylation level of these 4 CpG sites and the BDNF transcript IV levels.
Fuchikami and colleagues were the first to examine the possibility of DNA methylation changes as a biomarker of major depression. Applying the EpiTYPER technique, these authors determined the methylation level of two CpG islands associated with promoter I and promoter IV of BDNF in major depressive disorder (MDD) patients in comparison with healthy individuals. The methylation levels of 29 out of 35 CpG sites inside of promoter I (Fig. 1a) were significantly different between patients and controls, although different CpG sites demonstrated multidirectional methylation changes. 21 CpG sites showed increased methylation levels in controls, while the other 8 CpG sites showed increased methylation levels in depression patients .
In a study of major depressed patients with suicidal behavior, higher BDNF promoter VI methylation levels were found to be strongly associated with a previous history of suicidal attempts, suicidal ideation and less improvement during antidepressant treatment, independent of the antidepressant type . As BDNF methylation status was not measured after the treatment in this study, it is not possible to conclude if treatment outcomes might be reflected by DNA methylation status. In two subsequent studies, authors found a significant association between the BDNF promoter VI methylation levels and late-life depression  as well as depression related to breast cancer . It is worth mentioning, that both studies did not reveal any association between the methylation level of the analyzed region and rs6265 genotype.
Kleimann and colleagues, for the first time, examined an effect of electroconvulsive therapy (ECT) on DNA methylation of BDNF  in treatment-resistant major depressive patients. The difference in the methylation level of BDNF promoter I was presented between remitters/responders and non-remitters/non-responders over the whole series of ECT treatment. It is difficult to say if the found difference was conditioned by only ECT since there was no information about BDNF methylation level before ECT and most patients were previously treated by antidepressants and/or antipsychotics.
By means of bisulfite pyrosequencing, Igekame and colleagues found a significant increase (about 1 %) of the methylation level of one CpG site in the promoter I of BDNF in schizophrenia patients compared with controls  (Fig. 1a). Only in the male group of patients the additional CpG site was approximately 2 % higher methylated. It should be noted that the differences were not significant after multiple corrections and the sample size is not large enough to estimate a methylation level difference as low as 1–2 %. Authors did not reveal any difference in the methylation level of BDNF promoter IV.
On the contrary, Kordi-Tamandani and colleagues revealed that the methylated allele frequency of the BDNF promoter IV was lower in the schizophrenia patients, than in controls. According to the methylation data, the relative expression level of BDNF was significantly higher in schizophrenia patients than in controls .
In the study of D’Addario, a 7 % increase of DNA methylation at the region of BDNF promoter I (Fig. 1a) was observed in bipolar disorder (BD) II patients, but not BD I, compared with controls . This was in correlation with the significantly decreased BDNF expression in BD II subjects. It should be taken in account that these data do not reflect the initial DNA methylation levels of BD patients, since part of the patients enrolled in the study were maintained on a diverse antidepressant treatment for at least 1 month. Interestingly, the authors showed, that BD patients on antidepressant treatment revealed a higher BDNF methylation level compared with antidepressants-free patients. However, the patients treated by valproate and lithium demonstrated a significantly decreased BDNF methylation level, in comparison with the control level.
In the subsequent study, these authors extended the BDNF promoter I methylation level analysis to the group of patients with major depression on stable pharmacologic treatment in comparison with healthy individuals . In concordance with the previous study, a significantly increased methylation level of the analyzed region inside of BDNF promoter I was observed in MDD patients together with a significant reduction of BDNF expression. These authors also showed that MDD patients treated only by antidepressant drugs (i.e., selective serotonin and selective norepinephrine reuptake inhibitors) had a 10 % higher methylation level of the BDNF promoter in comparison with patients treated by combinatory therapy of antidepressants and mood stabilizers .
However, in a further study which was performed on a larger cohort of MDD, BD I and BD II patients, these authors were not able to show the effect of mood stabilizers on DNA methylation . Patients treated by lithium and valproate tend to demonstrate BDNF methylation levels close to controls in comparison with patients treated by other antidepressant agents such as selective serotonin reuptake inhibitors, serotonin norepinephrine reuptake inhibitors and atypical antipsychotics. MDD subjects and BD II patients showed a significantly higher methylation level of the analyzed region in comparison with BD I subjects .
In the absence of an initial DNA methylation status for the patients, the results of these studies are difficult to interpret. It is not possible to discriminate if these described changes in the methylation level among different diagnostic groups are associated with disease status or whether they are connected with various pharmacological treatments. It was noted that BD I patients were mostly treated by mood stabilizers, while inside of the BD II and MDD groups, patients obtained a different therapy . Thus, the BDNF promoter methylation analysis in drug-naive MDD, BD patients is necessary in order to make a reliable conclusion about whether this parameter might be used as biomarker.
Interesting results have been obtained in a study by Tadic and colleagues where the correlation between DNA methylation of BDNF promoter IV (Fig. 1b) in major depressive patients and antidepressant treatment response was assessed . Of the 12 analyzed CpG sites, the methylation level of one CpG site was about 2 % lower in non-responders compared to responders. Following this, in vitro experiments showed a decreased luciferase expression of a vector containing an unmethylated fragment of BDNF promoter IV, in response to both SSRI and the SNRI treatment. These findings were connected with the ability of antidepressants to phosphorylate MeCP2, which leads to its dissociation from a promoter and consequently to the activation of expression. Moreover, these authors suggested that the DNA methylation status may play a significant role in the binding of MeCP2 to the promoter region and described that the mechanism of antidepressant action on BDNF can only be active in carriers of methylated allele at the specific CpG site. However, it is difficult to say that this conclusion was supported by experimental in vivo data, as both responders and non-responders demonstrated a strongly hypomethylated (4–6 %) status of the analyzed region.
DNA methylation levels of the BDNF promoter I and promoter IV (Fig. 1) were studied in patients with borderline personality disorder (BDP), whose disease phenotype is closely related to the depression and suicide phenotype. Using the high resolution melt (HRM) analysis, Perround and colleagues observed that regions in the BDNF promoter I and promoter IV had an almost 8 and 18 % higher methylation level respectively for BDP patients . Moreover, a larger number of childhood maltreatment was significantly associated with a higher methylation status of BDNF promoters (mean percentage at both regions). Another very interesting result of this study is that during the intensive dialectical behavior therapy (I-DBT) non-responders showed increased BDNF methylation levels, while responders showed a decrease in BDNF methylation status, in some cases comparable to the methylation level of controls. This suggests that BDNF methylation changes might be relevant for treatment response prediction. The limitation of this study is that most of the subjects were on antidepressant treatment before I-DBT, that way the determined methylation level of BDP patients before treatment does not directly reflect a baseline methylation level of such patients.
An additional study of the influence of childhood maternal care on BDNF methylation level was performed by Unternaehrer and colleagues . Authors showed greater whole blood DNA methylation in the low versus high maternal care group in a CpG island situated within the BDNF exon VI. More importantly, authors investigated differential blood cell distribution as a potential factor in connection with maternal care and DNA methylation of BDNF. It should be considered as an example of a standard experimental set up for such types of studies due to the blood cell specific DNA methylation pattern.
Thaler and colleagues explored DNA methylation changes of the BDNF promoter IV (Fig. 1b) in women with bulimic eating syndromes . These women displayed increases in the methylation level at specific CpG sites, especially in cases when the bulimic syndromes were complicated by borderline personality disorder or history of severe childhood abuse. It is interesting that majority of found CpG sites were binding sites for various transcriptional factors. As no multiple corrections were applied, additional studies will be required to confirm the obtained differences.
In the databases, the DNA methylation level was assessed at over 485,000 CpG sites using the Illumina Infinium Human Methylation450 Bead Chip. 25 probes corresponding to 12 CpG sites situated in the BDNF promoter I/exon I (Fig. 1a) and to 13 CpG sites situated in the BDNF promoter IV/exon IV were included in the analysis (Fig. 1b). The methylation level of each promoter was calculated by averaging the methylation levels of the corresponding CpG sites. Distribution normality for all variables was checked using Kolmogorov–Smirnov test. Because of non-Gaussian distribution, statistical comparisons of methylation levels between patients and controls were performed using the non-parametric Mann–Whitney test, although in the text and in the tables values are presented as mean ± SEM. A Bonferroni correction was used to adjust for multiple comparisons. Statistical analyses were performed using GraphPad Prism5 (GraphPad) and the statistical software R (http://www.r-project.org). A significance level of a = 0.05 or less was considered significant. The statistical power analysis for the described studies and the open access databases was performed with a web browser program “Post-hoc Power Calculator” (http://clincalc.com/Stats/Power.aspx#1).
According to results of previous studies both promoter I and IV of BDNF demonstrated strongly hypomethylated levels in all analyzed cohorts (Table 3). Frontal cortex methylation levels of both promoters I and IV were lower in schizophrenia subjects compared to controls (E-GEOD-61431, E-GEOD-61380, E-GEOD-61107 databases). Cerebellum methylation levels of promoter IV were lower in schizophrenia patients, at the same time cerebellum methylation levels of promoter I were similar in patients and controls. After applying multiple corrections, a significant difference in the methylation level of BDNF promoter I (p = 0.0023, Mann–Whitney test; E-GEOD-61380 database,) and of BDNF promoter IV (p = 0.0015, Mann–Whitney test; E-GEOD-61107 database) was found between schizophrenia patients and healthy individuals (Table 3). However, the power analysis indicated that the E-GEOD-61380 database is not powerful enough to determine a 0.3 % methylation level difference.
We did not observe any difference between MDD patients and controls in the methylation levels of BDNF promoters I and IV in the frontal cortex neuronal and glial cells (E-GEOD-41826). We separately analyzed several CpG sites which were determined as significantly differently methylated in previous studies (Fig. 1). The difference was found in the methylation level of cg14589148 (p = 0.0147, Mann–Whitney test; E-GEOD-61380 database), cg25457956 and cg10022526 (p = 0.0396 and p = 0.0227, Mann–Whitney test; E-GEOD-61431 database, cerebellum), however these data did not remain significant after multiple corrections.
The variety and complexity of psychiatric disorders is very high and biomarkers able to reliably differentiate characteristics of certain similar disorders, determine severity and treatment efficiency which could be very valuable. Taking into the consideration the inaccessibility of the target tissue—the brain—in psychiatric disorders, an ideal biomarker should be easily peripherally measured but at the same time it should reflect disease-associated alterations in the target tissue. DNA methylation is a stable and heritable modification and can be reliably measured from a small amount of material regardless of the storage conditions. These remarkable characteristics make DNA methylation a robust biomarker of disease activity. Numerous cancer studies have proven that DNA methylation changes allow manifestations of different pathological stages of disease and monitoring of treatment efficiency (reviewed in ).
At present, a growing body of data has associated BDNF promoters’ methylation level with development of various neuropsychiatric disorders. Most of the studies have analyzed different regions within promoter I and promoter IV. As it is demonstrated in Fig. 1a, the promoter I region is differently methylated in major depression and bipolar disorders patients in the studies done by D’Addario. These partly overlap with the region differently methylated in BPD patients of the Perroud’s study and with one CpG site differently methylated in major depression patients of the Fuchikami’s study. Similarly, within promoter IV, the CpG sites are differently methylated in suicide completers of Keller’s study, in patients with BDP and BN of the Taler’s study, and in major depressive patients in Tadic’s study, which partially overlap with the differently methylated region in BPD patients in Perroud’ study (Fig. 1b). These overlapping regions within BDNF promoter I and promoter IV may be of special interest as possible biomarker of psychiatric diseases, taking into consideration that several independent studies reported about the differences in the methylation level of these regions. Although the methylation levels’ alteration of these regions associated with multiple disorders do not allow the consideration of these methylation changes as reliable biomarkers for certain psychiatric disturbances.
Most of the studies have been performed using peripheral blood cells. The main question which appears is to what extent DNA methylation changes in blood can reflect the changes in the target tissue—the brain; a highly tissue-specified BDNF expression implies variation of BDNF promoter methylation level in various tissues. However, it has been reported that there might not be an exact correlation between gene DNA methylation and its expression level . Moreover, all studies were concentrated on the methylation levels of BDNF promoters, which are CpG rich. It is well established that differently methylated regions across the brain and blood are primarily located in intragenic regions and are underrepresented in promoter and regulatory gene regions. CpG rich promoters were characterized by less tissue-specific DNA methylation across brain regions and blood [37, 39]. Recently, a strong correlation between the ventral prefrontal cortex and quadriceps for the methylation levels of BDNF promoter I has been found . Similar epigenetic changes, particularly DNA methylation, in the brain and peripheral tissues in connection to psychiatric disorders might be initiated by environmental factors such as maternal stress or diet during early prenatal development or even have an inherited character. In the postnatal period, mothering behavior, early life traumas, environmental stressors factors and hormone dissonance may be responsible for the widespread epigenetic alterations associated with the psychiatric phenotype [19, 42, 43]. Keller and colleagues demonstrated that BDNF promoters’ DNA methylation changes in brain are associated with major depression and suicide. In one of the brain open-access databases, we revealed a significant but minor (1.3 %) difference in the average methylation level of BDNF promoter IV in schizophrenia patients.
Most of the changes found in the BDNF methylation levels between patients and healthy individuals are very subtle. The question about whether these differences may initiate changes in the BDNF expression levels has been partly addressed in previous studies. Thus, Keller and colleagues indicated much lower BDNF transcript IV levels in samples showing 20 to 30 % methylation of 4 CpG sites in BDNF promoter IV compared with samples showing 3 to 5 % methylation . In the study of D’Addario and colleagues, a 9 % alteration in the BDNF promoter I methylation level between BDII patients and controls was matched with changes in the BDNF expression level in the blood cells . It may suggest that the 9–15 % difference of the BDNF methylation level is functionally relevant and can induce changes in the BDNF transcripts level. In turn, Perroud and colleagues were not able to find any correlation between the BDNF promoters’ methylation and BDNF protein levels, although an 8 and 18 % methylation level difference was found respectively in BDNF promoter I and promoter IV between BDP patients and controls . The discordance, in the association of the BDNF protein level with the DNA methylation level, might be conditioned by other regulators of BDNF expression: particularly histone modifications, miRNAs, antiBDNF transcripts, formation of dsRNA duplexes with BDNF transcripts, alternative splicing, and posttranscriptional cleavage. Recent studies have introduced the idea that DNA methylation may not even have an active influence on gene expression, but instead it might reflect gene expression regulation, mediated by histone modifications or chromatin structure . Additionally, DNA hydroxymethylation, which cannot be distinguished from DNA methylation by bisulfite treatment, can impact BDNF expression in a manner opposite to DNA methylation. Therefore, a complex analysis of several factors determining BDNF expression is necessary to make the conclusion of whether the small changes in the BDNF methylation level may affect the BDNF expression changes both on transcript and protein levels.
Despite a relatively large number of studies, the use of the BDNF methylation level as a biomarker of psychiatric disorders still needs considerable further research to become reality. The main reason is the lack of confirmatory research for each particular disease. Another factor is the use of a variety of techniques for methylation level determination (Table 1), which hampers the analysis of different areas within the BDNF promoters without providing a correct comparison of the results of different studies for the same disorder. Although, methylation-specific real-time PCR and high resolution melt analysis are very sensitive methods with the ability to detect as little as 0.1–1 % of methylated DNA [63, 64], they cannot provide individual CpG sites’ resolution, allowing the determination of the average methylation level of a whole region. Primer design, PCR bias or the presence of too many CpG sites in the analyzed region might affect the obtained results. The bisulfite sequencing-based techniques widely applied in the studies differ also in their coverage and sensitivity. Direct bisulfite sequencing can accurately detect an intermediate difference (≥20 %) in methylation level while with pyrosequencing a minor difference of 5 % can be reliably detected while the resolution of cloning bisulfite sequencing is dependent on the number of sequenced clones [65, 66]. Furthermore, the inclusion of medicated and treated patients in the studies does not allow for specification of whether the observed DNA methylation changes are related to the specific disease phenotype or are a consequence of medical or psychiatric treatment. It is known that certain agents, applied for psychiatric therapy, cause epigenetic modifications . Insufficient attention has been paid to BDNF methylation level investigations in the brain. So far only Keller’s study and our analysis indicate a partial correlation of the BDNF methylation level in the brain to the development of psychiatric disorders. However, we did not find any difference in the methylation level of the CpG site (cg11241206), differently methylated between controls and suicide completers in Keller’s study. Measured methylation levels of others CpG sites (Fig. 1), previously described as differently methylated in the studies, using blood cells did not show any correlation after multiple corrections. The restrictions of the method did not allow us to completely assess the methylation level of the regions analyzed in previous studies using brain samples as Infinium HumanMethylation450 BeadChip Kit covers far from all CpG sites in BDNF promoter I and IV (Fig. 1). A detailed analysis in the brain with a method, determining the methylation levels of all CpG sites, is required. Both Keller’s and our study covered only certain brain areas (frontal cortex, cerebellum, Wernicke area), while BDNF DNA methylation changes in such brain regions as hippocampus, nucleus accumbens, amygdala [3, 68, 69] may be essential for psychiatric disorders development.
Summing up the results of this research provides a good reason to believe that the methylation levels of certain regions within BDNF promoters can be considered as biomarker of specific diseases. Several independent studies on different patients’ cohorts with a standardized method for the methylation level analysis are necessary to confirm this suggestion. Applying the most modern approaches for the analysis of methylation will allow highlighting of the role of both DNA methylation and DNA hydroxymethylation (primarily occurring in the neuronal cells) with regard to the development of mental illnesses. Considering the easy accessibility of blood samples, it is the DNA methylation level of blood cells that might be used in clinical practice for accurate disease diagnosis and treatment-efficacy monitoring. The question of DNA methylation discrepancy within various blood cell lines  and possible differences of cell composition between psychiatric patients and healthy individuals  should also be taken in account. Modern techniques of cell sorting and counting will help to smooth over such differences. Considering the fact that most changes in DNA methylation linked to mental illness are fairly small in terms of percentage, it is necessary to use a high-resolution method that can reliably detect small differences. We suggest that through further investigations, an unequivocal answer if the BDNF DNA methylation level is a suitable psychiatric disorder biomarker may be provided.
aryl hydrocarbon receptor nuclear translocator 2
brain-derived neurotrophic factor
borderline personality disorder
calcium-responsive element 1
cAMP response element
high resolution melt
dialectical behavior therapy
major depressive disorder
methylation-specific real-time PCR
nuclear receptor corepressor
neuronal PAS domain protein 4
nucleosome remodeling deacetylase
p75 neurotropin receptor
basic helix-loop-helix transcription factor response element
silencing mediator for retinoid and thyroid hormone receptors
serotonin norepinephrine reuptake inhibitors
selective serotonin reuptake inhibitors
ten-eleven translocation enzymes
tropomyosin-related kinase receptor B
tissue-specific differentially methylated region
upstream stimulatory factor binding element
Murer MG, Yan Q, Raisman-Vozari R. Brain-derived neurotrophic factor in the control human brain, and in Alzheimer’s disease and Parkinson’s disease. Prog Neurobiol. 2001;63(1):71–124.
Pruunsild P, et al. Dissecting the human BDNF locus: bidirectional transcription, complex splicing, and multiple promoters. Genomics. 2007;90(3):397–406.
Boulle F, et al. Epigenetic regulation of the BDNF gene: implications for psychiatric disorders. Mol Psychiatry. 2012;17(6):584–96.
Mitchelmore C, Gede L. Brain derived neurotrophic factor: epigenetic regulation in psychiatric disorders. Brain Res. 2014;1586:162–72.
Tsankova NM, et al. Sustained hippocampal chromatin regulation in a mouse model of depression and antidepressant action. Nat Neurosci. 2006;9(4):519–25.
Jha S, et al. Enriched environment treatment reverses depression-like behavior and restores reduced hippocampal neurogenesis and protein levels of brain-derived neurotrophic factor in mice lacking its expression through promoter IV. Transl Psychiatry. 2011;1:e40. doi:10.1038/tp.2011.33.
Lee Y, et al. Association between the BDNF Val66Met polymorphism and chronicity of depression. Psychiatry Investig. 2013;10(1):56–61.
Brooks SJ, et al. BDNF polymorphisms are linked to poorer working memory performance, reduced cerebellar and hippocampal volumes and differences in prefrontal cortex in a Swedish elderly population. PLoS ONE. 2014;9(1):e82707. doi:10.1371/journal.pone.0082707.
Notaras M, et al. A role for the BDNF gene Val66Met polymorphism in schizophrenia? A comprehensive review. Neurosci Biobehav Rev. 2015;51:15–30.
Sarchiapone M, et al. Association of polymorphism (Val66Met) of brain-derived neurotrophic factor with suicide attempts in depressed patients. Neuropsychobiology. 2008;57(3):139–45.
Rosenfeld RD, et al. Purification and identification of brain-derived neurotrophic factor from human serum. Protein Expr Purif. 1995;6(4):465–71.
Kerschensteiner M, et al. Activated human T cells, B cells, and monocytes produce brain-derived neurotrophic factor in vitro and in inflammatory brain lesions: a neuroprotective role of inflammation? J Exp Med. 1999;189(5):865–70.
Pan W, Banks WA, Fasold MB, Bluth J, Kastin AJ. Transport of brain-derived neurotrophic factor across the blood-brain barrier. Neuropharmacology. 1998;37(12):1553–61.
Molendijk ML, et al. Serum levels of brain-derived neurotrophic factor in major depressive disorder: state-trait issues, clinical features and pharmacological treatment. Mol Psychiatry. 2011;16(11):1088–95.
Nurjono M, et al. A review of brain-derived neurotrophic factor as a candidate biomarker in Schizophrenia. Clin Psychopharmacol Neurosci. 2012;10(2):61–70.
Suliman S, et al. Brain-derived neurotrophic factor (BDNF) protein levels in anxiety disorders: systematic review and meta-regression analysis. Front Integr Neurosci. 2013;7:55.
Koenigsberg HW, et al. Platelet protein kinase C and brain-derived neurotrophic factor levels in borderline personality disorder patients. Psychiatry Res. 2012;199(2):92–7.
Wolkowitz OM, et al. Serum BDNF levels before treatment predict SSRI response in depression. Prog Neuropsychopharmacol Biol Psychiatry. 2011;35(7):1623–30.
Guintivano J, Kaminsky ZA. Role of epigenetic factors in the development of mental illness throughout life. Neurosci Res. 2016;102:56–66.
Pishva E, et al. The epigenome and postnatal environmental influences in psychotic disorders. Soc Psychiatry Psychiatr Epidemiol. 2014;49(3):337–48.
Varendi K, et al. miR-1, miR-10b, miR-155, and miR-191 are novel regulators of BDNF. Cell Mol Life Sci CMLS. 2014;71(22):4443–56.
Ikegame T, et al. DNA methylation of the BDNF gene and its relevance to psychiatric disorders. J Hum Genet. 2013;58(7):434–8.
Martinowich K, et al. DNA methylation-related chromatin remodeling in activity-dependent BDNF gene regulation. Science. 2003;302(5646):890–3.
Vialou V, et al. Epigenetic mechanisms of depression and antidepressant action. Annu Rev Pharmacol Toxicol. 2013;53:59–87.
Egger G, et al. Epigenetics in human disease and prospects for epigenetic therapy. Nature. 2004;429(6990):457–63.
Allis CD, Jenuwein T, Reinberg D. Epigenetics. CSHL Press; 2007
Illingworth RS, Bird AP. CpG islands–’a rough guide’. FEBS Lett. 2009;583(11):1713–20.
Weber M, et al. Distribution, silencing potential and evolutionary impact of promoter DNA methylation in the human genome. Nat Genet. 2007;39(4):457–66.
Zhang Y, et al. DNA methylation analysis of chromosome 21 gene promoters at single base pair and single allele resolution. PLoS Genet. 2009;5(3):e1000438. doi:10.1371/journal.pgen.1000438.
Spruijt CG, Vermeulen M. DNA methylation: old dog, new tricks? Nat Struct Mol Biol. 2014;21(11):949–54.
Ball MP, et al. Targeted and genome-scale strategies reveal gene-body methylation signatures in human cells. Nat Biotechnol. 2009;27(4):361–8.
Maunakea AK, et al. Intragenic DNA methylation modulates alternative splicing by recruiting MeCP2 to promote exon recognition. Cell Res. 2013;23(11):1256–69.
Choi JK, et al. Nucleosome deposition and DNA methylation at coding region boundaries. Genome Biol. 2009;10(9):R89. doi:10.1186/gb-2009-10-9-r89.
Schübeler D. Epigenetic Islands in a genetic ocean. Science. 2012;338(6108):756–7.
Tufarelli C, et al. Transcription of antisense RNA leading to gene silencing and methylation as a novel cause of human genetic disease. Nat Genet. 2003;34(2):157–65.
Lehnertz B, et al. Suv39 h-mediated histone H3 lysine 9 methylation directs DNA methylation to major satellite repeats at pericentric heterochromatin. Curr Biol CB. 2003;13(14):1192–200.
Davies MN, et al. Functional annotation of the human brain methylome identifies tissue-specific epigenetic variation across brain and blood. Genome Biol. 2012;13(6):R43. doi:10.1186/gb-2012-13-6-r43.
Spiers H, et al. Methylomic trajectories across human fetal brain development. Genome Res. 2015;25:338. doi:10.1101/gr.180273.114.
Masliah E, et al. Distinctive patterns of DNA methylation associated with Parkinson disease: identification of concordant epigenetic changes in brain and peripheral blood leukocytes. Epigenetics. 2013;8(10):1030–8.
Wen L, et al. Whole-genome analysis of 5-hydroxymethylcytosine and 5-methylcytosine at base resolution in the human brain. Genome Biol. 2014;15(3):R49. doi:10.1186/gb-2014-15-3-r49.
Pruunsild P, et al. Identification of cis-elements and transcription factors regulating neuronal activity-dependent transcription of human BDNF gene. J Neurosci. 2011;31(9):3295–308.
Roth TL, et al. Lasting epigenetic influence of early-life adversity on the BDNF gene. Biol Psychiatry. 2009;65(9):760–9.
Roth TL, Sweatt JD. Epigenetic marking of the BDNF gene by early-life adverse experiences. Horm Behav. 2011;59(3):315–20.
Mill J, et al. Epigenomic profiling reveals DNA-methylation changes associated with major psychosis. Am J Hum Genet. 2008;82(3):696–711.
Chagnon YC, et al. DNA methylation and single nucleotide variants in the brain-derived neurotrophic factor (BDNF) and oxytocin receptor (OXTR) genes are associated with anxiety/depression in older women. Genet Aging. 2015;6:230.
Keller S, et al. Increased BDNF promoter methylation in the Wernicke area of suicide subjects. Arch Gen Psychiatry. 2010;67(3):258–67.
Fuchikami M, et al. DNA methylation profiles of the brain-derived neurotrophic factor (BDNF) gene as a potent diagnostic biomarker in major depression. PLoS ONE. 2011;6(8):e23881. doi:10.1371/journal.pone.0023881.
Kang H-J, et al. BDNF promoter methylation and suicidal behavior in depressive patients. J Affect Disord. 2013;151(2):679–85.
Kang HJ, et al. Longitudinal associations between BDNF promoter methylation and late-life depression. Neurobiol Aging. 2015;36(4):1764. doi:10.1016/j.neurobiolaging.2014.12.035.
Kang H-J, et al. A longitudinal study of BDNF promoter methylation and depression in breast cancer. Psychiatry Investig. 2015;12(4):523–31.
Kleimann A, et al. BDNF serum levels and promoter methylation of BDNF exon I, IV and VI in depressed patients receiving electroconvulsive therapy. J Neural Transm. 2014;122(6):925–8.
Ikegame T, et al. DNA methylation analysis of BDNF gene promoters in peripheral blood cells of schizophrenia patients. Neurosci Res. 2013;77(4):208–14.
Kordi-Tamandani DM, et al. DNA methylation and expression profiles of the brain-derived neurotrophic factor (BDNF) and dopamine transporter (DAT1) genes in patients with schizophrenia. Mol Biol Rep. 2012;39(12):10889–93.
D’Addario C, et al. Selective DNA methylation of BDNF promoter in bipolar disorder: differences among patients with BDI and BDII. Neuropsychopharmacol. 2012;37(7):1647–55.
D’Addario C, et al. Epigenetic modulation of BDNF gene in patients with major depressive disorder. Biol Psychiatry. 2013;73(2):e6–7. doi:10.1016/j.biopsych.2012.07.009.
Dell’Osso B, et al. Epigenetic modulation of BDNF gene: differences in DNA methylation between unipolar and bipolar patients. J Affect Disord. 2014;166:330–3.
Tadić A, et al. Methylation of the promoter of brain-derived neurotrophic factor exon IV and antidepressant response in major depression. Mol Psychiatry. 2014;19(3):281–3.
Perroud N, et al. Response to psychotherapy in borderline personality disorder and methylation status of the BDNF gene. Transl Psychiatry. 2013;3:e207. doi:10.1038/tp.2012.140.
Unternaehrer E, et al. Childhood maternal care is associated with DNA methylation of the genes for brain-derived neurotrophic factor (BDNF) and oxytocin receptor (OXTR) in peripheral blood cells in adult men and women. Stress. 2015;18(4):451–61.
Thaler L, et al. Methylation of BDNF in women with bulimic eating syndromes: associations with childhood abuse and borderline personality disorder. Prog Neuropsychopharmacol Biol Psychiatry. 2014;54:43–9.
Mikeska T, Craig JM. DNA methylation biomarkers: cancer and beyond. Genes. 2014;5(3):821–64.
Stenz L, et al. BDNF promoter I methylation correlates between post-mortem human peripheral and brain tissues. Neurosci Res. 2015;91:1–7.
Wojdacz TK, Dobrovic A. Methylation-sensitive high resolution melting (MS-HRM): a new approach for sensitive and high-throughput assessment of methylation. Nucleic Acids Res. 2007;35(6):e41. doi:10.1093/nar/gkm013.
Shen L, Waterland RA. Methods of DNA methylation analysis. Curr Opin Clin Nutr Metab Care. 2007;10(5):576–81.
Lewin J, et al. Quantitative DNA methylation analysis based on four-dye trace data from direct sequencing of PCR amplificates. Bioinformatics. 2004;20(17):3005–12.
Uhlmann K, et al. Evaluation of a potential epigenetic biomarker by quantitative methyl-single nucleotide polymorphism analysis. Electrophoresis. 2002;23(24):4072–9.
Milutinovic S, et al. Valproate induces widespread epigenetic reprogramming which involves demethylation of specific genes. Carcinogenesis. 2007;28(3):560–71.
Deltheil T, et al. Behavioral and serotonergic consequences of decreasing or increasing hippocampus brain-derived neurotrophic factor protein levels in mice. Neuropharmacology. 2008;55(6):1006–14.
Berton O, et al. Essential role of BDNF in the mesolimbic dopamine pathway in social defeat stress. Science. 2006;311(5762):864–8.
Reinius LE, et al. Differential DNA methylation in purified human blood cells: implications for cell lineage and studies on disease susceptibility. PLoS ONE. 2012;7(7):e41361. doi:10.1371/journal.pone.0041361.
Maes M, et al. Evidence for a systemic immune activation during depression: results of leukocyte enumeration by flow cytometry in conjunction with monoclonal antibody staining. Psychol Med. 1992;22(1):45–53.
GYZ and HBS conceived this review. GYZ drafted the manuscript, performed analyses, interpreted data. HC participated in data analyses. HBS provided critical revision of the manuscript. All authors read and approved the final manuscript.
We are grateful to Christina Zhukovsky for proofreading the article.
Availability of data and materials
The datasets (E-GEOD-61107, E-GEOD-61380, E-GEOD-61431, E-GEOD-41826) supporting the conclusions of this article are available in the ArrayExpress Archive of Functional Genomics Data (http://www.ebi.ac.uk/arrayexpress).
The authors declare that they have no competing interests.
HS was supported by the Swedish Research Council and the Swedish Brain Foundation. Hao Cao was supported by the China Research Council. The funders had no role in the design of the study and collection, analysis, and interpretation of data and in writing the manuscript.