Genetic variation in DNMT3B and increased global DNA methylation is associated with suicide attempts in psychiatric patients

Authors


Corresponding author: Dr Therese M. Murphy, Department of Psychiatry and Mental Health Research & Education and Research centre, St. Vincent's University Hospital, and School of Medicine & Medical Science, University College Dublin, Elm Park, Dublin 4, Ireland. E-mail: murphyth@tcd.ie

Abstract

Recently, a significant epigenetic component in the pathology of suicide has been realized. Here we investigate candidate functional SNPs in epigenetic-regulatory genes, DNMT1 and DNMT3B, for association with suicide attempt (SA) among patients with co-existing psychiatric illness. In addition, global DNA methylation levels [5-methyl cytosine (5-mC%)] between SA and psychiatric controls were quantified using the Methylflash Methylated DNA Quantification Kit. DNA was obtained from blood of 79 suicide attempters and 80 non-attempters, assessed for DSM-IV Axis I disorders. Functional SNPs were selected for each gene (DNMT1; n = 7, DNMT3B; n = 10), and genotyped. A SNP (rs2424932) residing in the 3′ UTR of the DNMT3B gene was associated with SA compared with a non-attempter control group (P = 0.001; Chi-squared test, Bonferroni adjusted P value = 0.02). Moreover, haplotype analysis identified a DNMT3B haplotype which differed between cases and controls, however this association did not hold after Bonferroni correction (P = 0.01, Bonferroni adjusted P value = 0.56). Global methylation analysis showed that psychiatric patients with a history of SA had significantly higher levels of global DNA methylation compared with controls (P = 0.018, Student's t-test). In conclusion, this is the first report investigating polymorphisms in DNMT genes and global DNA methylation quantification in SA risk. Preliminary findings suggest that allelic variability in DNMT3B may be relevant to the underlying diathesis for suicidal acts and our findings support the hypothesis that aberrant DNA methylation profiles may contribute to the biology of suicidal acts. Thus, analysis of global DNA hypermethylation in blood may represent a biomarker for increased SA risk in psychiatric patients.

Suicide is a worldwide public health problem and a leading cause of death among young people in developed countries. Every year, almost one million people die from suicide globally with suicide attempts (SAs) up to 20 times more frequent than completed suicide (World Health Organization 2012). The risk for suicidal acts is multi-factorial, and consists of a range of biological, psychiatric, psychosocial, interpersonal and cultural risk factors.

In addition to classical genetic abnormalities, an epigenetic component in the pathology of suicide has been realized. Epigenetics can be defined as the mechanisms that initiate and maintain heritable patterns of gene expression without altering the sequence of the genome (Holliday 1987). There are several layers of epigenetic complexity including histone modifications, chromatin remodelling and DNA methylation, the latter being the most thoroughly studied to date (Esteller 2006). DNA methylation refers to the addition of a methyl group to the carbon at position 5 of the cytosine ring, resulting in 5-methylcytosine (5-mC) (Razin & Riggs 1980) and is a key regulatory mechanism of the genome, playing a central role in diverse biological processes. In mammalian cells, a family of DNA methyltransferase (DNMT1, DNMT3a, DNMT3b and DNMT3L) enzymes is responsible for establishing and faithfully maintaining DNA methylation patterns. DNMT1 is a maintenance enzyme which binds methyl groups to hemi-methylated DNA during DNA replication. DNMT3a and DNMT3b are de novo DNMT enzymes, which establish methylation patterns during embryonic development, genomic imprinting and X chromosome inactivation (Chédin 2011; Okano et al. 1999). DNMT3L interacts with and is critical for DNMT3a/3b methyltransferase activity (Chen et al. 2005).

Recently, studies have identified a prominent role for aberrant promoter methylation driven gene inactivation in brain tissue of suicide victims (Ernst et al. 2009; Mcgowan et al. 2009; Vire et al. 2006). In addition, studies have shown altered DNMT1 and DNMT3B gene expression in the brains of schizophrenia patients and suicide victims, respectively, relative to controls (Valinluck & Sowers 2007; Vire et al. 2006). Poulter and colleagues showed that increased DNMT3B mRNA and protein expression correlated with increased promoter methylation of a GABAA receptor subunit gene, suggesting that aberrant DNMT3B expression may result in aberrant promoter methylation of neurobiological genes important for suicidal behaviour. Previously, polymorphisms in the DNMTs have been linked to increased risk of cancer (Liu et al. 2008) and human intelligence (Haggarty et al. 2010), suggesting that variation at these genetic loci may have a role in disease risk and cognitive function variation. To date, there are no studies investigating the relationship between polymorphisms in the DNMT genes, global DNA methylation levels and SA in psychiatric patients.

Here, we test the hypothesis that functional single nucleotide polymorphisms (SNPs) in epigenetic-regulatory genes, DNMT1 and DNMT3B, may be associated with SA among patients with co-existing psychiatric illness (suicide attempters vs. non-attempter psychiatric controls). In addition, in order to evaluate the functional consequence of DNMT polymorphisms and differences in global DNA methylation levels between SA and psychiatric controls, global methylation quantification was performed in our cohort of psychiatric patients.

Materials and methods

Clinical sample collection

The study consisted of 159 psychiatric patients recruited as part of the Ireland North/South Urban/Rural Epidemiologic (INSURE) collaborative project as described previously (Bannan N 2002; Murphy et al. 2011). Briefly, newly referred patients to six Community Psychiatric Clinics on the island of Ireland were invited to take part in a clinical and molecular genetics study of suicidal behaviour in major psychiatric disorders over a year long recruitment period. In total, 159 patients gave written informed consent for this study. Patients were interviewed for Axis I Psychiatric Disorders using the Structured Clinical Interview for DSM-IV (American Psychiatric Association 2000). History of attempted suicide was recorded using the Columbia University Suicide History Questionnaire (Brodsky et al. 2001), which incorporates the Scale for Suicide Ideation (Beck et al. 1988), and the Suicide Intent Scale (Beck et al. 1975). A SA was defined as a completed act of self-harm with at least some expressed intent to die (scoring at least 1 on item 10 of Suicide Intent Scale (Beck et al. 1975). Seventy-nine (47.4% male, 52.6% female; mean age: 33.6 years) had a history of SA and 80 (61.3% male, 38.7% female; mean age: 40.9 years) had not. Childhood abuse was defined as the presence or absence of a history of sexual and/or physical abuse under the age of 16. Further clinical variables from clinical and demographic assessments are outlined in Table 1. All clinical research interviews were performed by a trained research psychiatrist or research psychiatric nurse. All patients gave written informed consent to take part in this study. Ethics approval was obtained for the study from the St Vincent's Healthcare Group Ethics and Medical Research Committee.

Table 1. Clinical features and DSM-IV psychiatric diagnosis of study population
 Attempter (n = 79) n (%*)Non-Attempter (n = 80) n (%*)P value
  • *

    Percentages of variables are calculated from all valid cases/controls only. In certain cases, information could not be obtained from patients' charts.

Sex0.113
Male37 (47.4)49 (61.3) 
Female49 (52.6)31 (38.7) 
Mean Age (Yrs)33.5740.90.001
Family history of SA30 (39.5)13 (16.5)0.003
Smoker55 (80.9)45 (57.7)0.005
Child abuse39 (50)26 (32.5)0.038
Axis I diagnosis
Substance dependence24 (30.4)19 (23.8)0.446
Alcohol dependence39 (49.4)31 (38.8)0.235
Psychotic disorders6 (8.3)6 (7.7)1
Bipolar disorder5 (6.9)3 (3.8)0.481
MDD51 (65.4)38 (47.5)0.035
Anxiety disorder32 (41)22 (27.5)0.104
Eating disorder1 (1.3)0 (0)0.497

Gene and SNP selection

Two important epigenetic-control enzymes (DNMT1 & DNMT3B), with evidence of altered gene expression in suicide or a related psychiatric illness were chosen for genotypic analysis. Potentially functional SNPs (pfSNPs) were selected for each gene (DNMT1; n = 7, DNMT3B; n = 10) using a combination of freely available online resources (TAMAL, PfSNP & NCBI DbSNP). TAMAL software (Hemminger et al. 2006) integrates SNP data from current versions of other online resources (including HapMap, dbSNP, Affymetrix, Perlegen and the UCSC Genome Browser). SNP results were filtered by the type of mutation caused, gene location, or predicted functionality. TAMAL was used to identify haplotype tagging SNPs in both genes. Potentially Functional SNP software integrates over 40 algorithms and bioinformatics resources to annotate the potential functionality of SNPs, using published reports, sequence motifs and genetic approaches (Wang et al. 2011). dbSNP http://www.ncbi.nlm.nih.gov/projects/SNP/ and the UCSC Genome Browser http://genome.ucsc.edu/ were used to gather European genotype frequencies and to verify SNP genomic location. Candidate SNP preference was given to tagging SNPs which were also potentially functional, followed by pfSNPs in linkage disequilibrium (LD) with a tagging SNP. Tagging SNPs causing synonymous or non-synonymous mutations were selected next. The remaining SNPs were then chosen from the pfSNP results according to potential function and gene location. Candidate functional SNPs (including their location in the gene, variation, functionality and previous evidence of association with psychiatric illness) are illustrated in Table 2.

Table 2. Summary of candidate SNPs
Gene symbol/rs numberRegionVariationMinor allelePredicted functionality of SNPAssociation with a psychiatric illness (PMID)
  1. ESE, exonic splicing enhancer; GWAS, genome wide association study; MiR, microRNA; TF, transcription factor.

DNMT1
rs2288349IntronA/GATagging SNP, in LD with rs759920 (a pfSNP) 
rs721186ExonC/TTTagging SNP, synonymous mutation 
rs759920IntronA/GGCreates/disrupts splicing regulatory element 
rs8101626IntronA/GGDisrupts a splicing regulatory element 
rs2290684IntronA/GADisrupts/creates splicing regulatory element 
rs2116940Downstream of 3′ UTRA/GGDisrupts Brn-2 TF binding site creates DBP TF binding site creates v-Myb TF binding site 
rs8112895Promoter CpG islandA/GGDisrupts a splicing regulatory element 
DNMT3B
rs2889703IntronA/CADisrupts splicing regulatory element, promoter, TF binding site 
rs998382IntronC/TCIn LD with pfSNP rs4911110, Regulatory region 
rs6058891ExonC/TCSynonymous mutation, disrupts/creates exonic splicing 
rs2424922ExonC/TCSynonymous creates an ESE site, disrupts 5′ splicing site 
rs24249323′ UTRA/GATagging SNP, creates GR TF binding site, creates HSF TF binding site 
rs6119954IntronA/GAIn LD with pfSNP rs2235760PMID: 20128888
rs60588963′ UTRC/TTDisrupts a MiR site, disrupts CP2 TF binding site, creates a GR TF binding site, creates a HSF TF binding site 
rs4911256IntronA/GGTagging SNP 
rs4911259IntronG/TGCreates a splicing regulatory element, GWAS 
rs6058894IntronC/TTDisrupts a splicing regulation site 

Genotyping

DNA was extracted from patient blood samples using standard techniques. Flanking SNP sequences were obtained from DbSNP and all genotyping assays were designed and validated by KBioscience (Hertfordshire, UK). Genotyping for SNP analysis was performed by KBioscience using the competitive allele specific PCR (KASP) chemistry coupled with a FRET-based genotyping system (http://www.kbioscience.co.uk/reagents/KASP/KASP.html). Genotyping data was viewed using SNPviewer (KBioscience), allowing graphical viewing of the clusters that group the allele calls.

Global methylation analysis

A total of 144 psychiatric patients (n = 73 attempters; n = 70 non-attempters) had DNA remaining of sufficient quantity to quantify DNA methylation levels. Global DNA methylation status in patient DNA samples was performed using the Methylflash Methylated DNA Quantification Kit (Epigentek, Farmingdale, NY, USA). Briefly, sample DNA is bound to high DNA affinity strip wells. Methylated DNA is detected using capture and detection antibodies to 5-methyl cytosine (5-mC) and then quantified colorimetrical by reading absorbance at 450 nm using a Multiskan EX Microplate Photometer (Thermo Fisher Scientific Inc, Waltham, MA, USA). The amount of methylated DNA is proportional to the OD intensity measured. The absolute amount of methylated DNA was quantified as per protocol using a standard curve, plotting the OD values vs. 5 serial dilution of control methylated DNA (0.5–10 ng).

Statistical analysis

A non-parametric Mann–Whitney test was used to calculate the difference in age between cases and controls. For all other comparisons of clinical and demographic variables a Fisher's exact test or Chi-squared test of association was used. All statistical analysis was performed on SPSS (PASW statistics 18, Chicago, IL, USA) and genotypic associations verified using SNP and Variation Suite (SVS) 7 software. For all tests, significance was ascribed at P < 0.05.

Deviations from Hardy Weinberg Equilibrium (HWE) were calculated using the exact test implemented by SVS 7 software (Bozeman, MT, USA; http://www.goldenhelix.com/SNP_Variation/svstrial.html). Allelic/genotypic tests of association examining the relationship between patient's alleles or genotype, at each of the 17 genetic loci, and SA were performed using Pearson's Chi-square test or Fisher's exact test (when counts are low). Binary logistical regression analyses were performed to evaluate the contribution of individual SNPs in the prediction of SA, correcting for potential confounders such as age, gender and certain psychiatric disorders and to interrogate associated SNPs for potential Genotype x E (history of childhood abuse) interactions. Analysis of G×E interactions were applied to SNPs with evidence of association in the allelic or genotypic tests of association (n = 6) and not examined for all candidate SNPs listed in this study. Bonferroni adjusted P values were obtained from SVS 7 software. The Bonferroni adjustment applied by SVS 7 software multiplies each individual P value by the number of times a test was performed. This value, which can be regarded as ultraconservative (Turic et al. 2005), estimates the probability this test would have obtained the same value by chance at least once from all the times this test was performed (SVS manual, Golden Helix http://www.goldenhelix.com). Bonferroni adjusted P values were also calculated for association between SNPs shown to be nominally associated with SA only (n = 6) and global methylation levels.

Haplotype analysis was performed on SNPs in high LD using SVS 7 software. SVS 7 haplotype block detection option was utilized to detect haplotype blocks in our SNP data using the (Gabriel et al. 2002) method. Haplotype frequency estimations were determined using the Composite Haplotype Method of haplotype frequency estimation (Weir 1996).

A Student's t-test was used to examine the difference in global DNA methylation levels (5-mC% content) between cases and controls, smokers and non-smokers and to examine association between 5-mC content and other psychiatric disorders. Logistic regression analysis was performed to evaluate the contribution of global DNA methylation levels to a prediction model of SA, correcting for potential confounders (age, gender and smoking). One-way analysis of variance (anova) was used to examine the difference in global DNA methylation levels and genotype at the 6 susceptibility genetic loci. In addition, the Tukey post-hoc test was used to perform multiple comparisons to assess the significance of differences among means.

Results

Clinical and psychological features of the study population

The mean age of the suicide attempter (n = 79) and non-attempter (n = 80) groups was significantly different [33.57 years vs. 40.9 yrs respectively (Z = −3.31, P = 0.001)]. A positive first degree family history of SA was significantly more frequent in the attempter group (Pearson Chi Square = 4.30, df = 1, P = 0.003). Suicide attempters were more likely to have a history of childhood abuse (Pearson Chi Square = 4.43, df = 1, P = 0.04), a diagnosis of major depressive disorder (MDD; Pearson Chi Square = 9.12, df = 1, P = 0.04) and be a smoker (Pearson Chi Squared = 8.01, df = 1, P = 0.005). No other clinical variables were significantly different between the two groups (Table 1).

Genetic variants in DMNT1 and DNMT3B genes and association with SA in psychiatric patients

The results of allelic and genotypic association tests between SNPs and SA are illustrated in Table 3. No association between DNMT1 polymorphisms (n = 7) and SA were identified. In the DNMT3B gene, however, we identified a SNP (rs2424932) residing in the 3′ UTR of the DNMT3B gene, located within a transcription factor binding site (TFBS), which showed evidence of allelic and genotypic association with SA compared with a non-attempter control group (Pearson Chi Squared = 9.43 , df = 1, P = 0.001, Bonferroni adjusted P value = 0.02; P = 0.004, Pearson Chi Squared = 10.81, df = 2, Bonferroni adjusted P value = 0.06, respectively). Controlling for potential confounders this SNP was a significant predictor of SA [odds ratio (OR) = 1.91; 95% confidence interval (95% CI) 1.10–3.33, P = 0.02]. Patients who had a recessive genotype (A/A) were almost twice as likely to have a history of attempted suicide as those who did not.

Table 3. Association of candidate markers
Gene symbolrs numberGenotype frequencySignificance (p)
DD n (%) Dd n (%) dd n (%) Cases¥DD n (%) Dd n (%) dd n (%) Controls¥AllelicGenotypic
  1. P value obtained by a Pearson's Chi-square or Fisher's exact test.

  2. Cases, suicide attempters; controls, non-attempter psychiatric controls; D, major allele; d, minor allele; ¥, percentages of variables are calculated from valid cases and controls only.

  3. *P ≤ 0.05. **P ≤ 0.01; ***P ≤ 0.001.

DNMT1rs228834922 (28.6) 39 (50.6) 16 (20.8)21 (26.6) 10 (12.8) 2 (2.6)0.8090.962
 rs72118675 (96.2) 3 (1.3) 0 (0)79 (98.8) 1 (1.3) (0)0.3660.364
 rs75992015 (14.8) 41 (40.5) 22 (22.7)15 (15.2) 41 (51.3) 24 (23.3)0.8390.970
 rs810162617 (22.1) 22 (28.6) 28 (49.4)18 (23.1) 22 (28.2) 38 (48.7)0.9040.989
 rs229068422 (28.9) 39 (51.3) 15 (19.7)24 (31.2) 29 (50.6) 14 (18.2)0.7400.944
 rs211694063 (80.8) 15 (19.2) 0 (0)64 (83.1) 13 (16.9) 0 (0)0.7180.835
 rs811289566 (82.5) 14 (17.5) 0 (0)62 (80.5) 15 (19.5) 0 (0)0.7620.838
DNMT3Brs288970327 (34.6) 38 (48.7) 13 (16.7)14 (17.7) 51 (64.6) 14 (17.7)0.1100.049*
 rs99838231 (40.8) 37 (48.7) 8 (10.5)17 (21.3) 51 (63.8) 12 (15)0.031*0.030*
 rs605889126 (33.3) 39 (50) 13 (16.7)13 (16.5) 52 (65.8) 14 (17.7)0.1110.045*
 rs242492226 (33.8) 38 (49.4) 13 (16.9)14 (17.7) 51 (64.6) 14 (17.7)0.1350.064
 rs242493217 (21.8) 43 (55.1) 18 (23.1)34 (23.1) 38 (48.1) 7 (8.9)0.001***0.004**
 rs611995461 (79.2) 14 (18.2) 2 (2.5)58 (74.4) 17 (21.8) 3 (3.8)0.4270.758
 rs605889666 (84.6) 10 (12.8) 2 (2.6)71 (89.9) 8 (10.1) 0 (0)0.1750.372
 rs491125623 (29.5) 39 (50) 16 (20.5)10 (12.7) 50 (63.3) 19 (24.1)0.0710.035*
 rs491125937 (48.1) 32 (41.6) 8 (10.4)18 (23.1) 48 (61.5) 12 (15.4)0.007**0.005**
 rs605889465 (84.4) 10 (13) 1 (2.6)71 (89.9) 8 (10.1) 0 (0)0.1650.371

An additional five SNPs (rs2889703, rs6058891, rs4911256 (associated with SA in a genotypic model only), rs998382 and rs4911259 (associated in both an allelic and genotypic model) were nominally associated with SA (Table 3), with rs4911256 showing the strongest association with SA. Haplotype analysis showed that four of these SNPs (rs998382, rs6058891, rs4911256, rs4911259) were in high LD (Fig. 1) and identified a DNMT3B haplotype (rs998382 (T), rs6058891 (T), rs2424922 (T), rs4911256 (A), rs4911259 (T)) which differed between cases and controls, however this association did not research significance after stringent multiple testing correction (Pearson Chi-Squared = 6.018, df = 1, P = 0.01; Bonferroni adjusted P value = 0.56). Deviations from HWE were observed in the case of rs998382 (P = 0.05, Fisher's exact test).

Figure 1.

LD -plot and haplotype blocks. Haplotype blocks generated by SVS software based on correlation between markers are illustrated in black. A 5 marker haplotype block highlighted in red was nominally associated with suicide attempters.

Examining association with psychiatric illness

These six SNPs of interest were tested for possible association (Genotypic and Allelic) with Axis I psychiatric disorders illustrated in Table 1. Rs2424932 was found to be significantly associated with substance dependence (Pearson Chi-Squared = 9.113, df = 2, P = 0.011; Fisher's exact test), whereby an A/A genotype was more frequent in psychiatric patients with a history of substance dependence. All other SNPs showed no evidence of association with any psychiatric disorder.

Genetic and environmental interaction analysis

To evaluate further the association between susceptibility genotypes only and childhood abuse, logistic regression was used to model risk as a function of the genotype models associated with SA at each locus, childhood abuse and their interaction. No evidence of Gene x history of childhood abuse interaction was identified.

Global methylation quantification in suicide attempters vs. psychiatric controls

Global methylation quantification was performed to establish if (1) global DNA methylation levels of suicide attempters differ from non-attempter psychiatric controls and (2) if susceptibility DNMT3B polymorphisms (associated with SA), correlates with hyper or hypo global methylation levels. Strikingly, global methylation (5-mC%) analysis showed that psychiatric patients with a history of SA had significantly higher levels of global DNA methylation compared with controls (M ± SD: Cases (1.11 ± .59) Controls (0.89 ± .47); t(141) = −2.39, P = 0.02, Student t-test, Fig. 2). Global methylation quantification was not associated with age (P = 0.80), gender (P = 0.35), smoking (P = 0.30), a history of childhood abuse (P = 0.40) or any other Axis I psychiatric diagnosis in our cohort (P > 0.05). Controlling for potential confounders (age, gender and smoking) global 5-mC% was a significant predictor of SA (OR = 2.1; 95% CI 1.01–4.24, P = 0.05). Our model (5-mC% as an independent predicator and controlling for age, gender and smoking) correctly classified 68.7% of cases overall, with a sensitivity of 67.6% and a specificity of 69.8%.

Figure 2.

Sample plot comparison of global methylation levels between cases and controls. Psychiatric patients with a history of suicide attempts (suicide attempters) had significantly higher levels of global DNA methylation (mean 5-mC%) compared with non-attempter controls (P = 0.018; Student's t-test). Circles represent outliers.

No association between DNMT3B polymorphisms rs2424932 and rs49211529 and hyper/hypo global methylation were observed (rs2424932; F2,138 = 0.526, P = 0.59; rs49211529; F2,137 = 1.63, P = 0.2; Fig. 3a,f). In contrast, rs998382, rs6058891, rs4911256 and rs2889703 were associated with global DNA methylation, however this association did not reach statistically significance after stringent multiple testing correction (rs998382; F2,137 = 3.04, P = 0.05, Bonferroni adjusted P value = 0.30; rs6058891; F2,138 = 4.24, P = 0.02, Bonferroni adjusted P value = 0.10; rs4911256; F2,138 = 4.62, P = 0.01, Bonferroni adjusted P value = 0.06; rs2889703 F2,138 = 4.23, P = 0.02, Bonferroni adjusted P value = 0.10; Fig. 3). Post hoc tests showed that individuals with a T/T, T/T, A/A and C/C genotype at the rs998382, rs6058891, rs4911256 and rs2889703 loci, respectively had higher mean 5-mC% compared to individuals with an alternative genotype (Fig. 3b–e). Interestingly, these genotypes at the rs998382, rs6058891, rs4911256 and rs2889703 loci were nominally associated with SA compared with psychiatric controls, suggesting that psychiatric patients with these susceptibility genotypes have hypermethylation of their genome and are more likely to have a history of SA.

Figure 3.

Sample plot comparison between global methylation levels and DNMT3B susceptibility genotypes. Genotypes of DNMT3B SNPs, rs24242932, rs998382, rs2889703, rs6058891, rs4911256 and rs4911256 and their corresponding mean global methylation levels (5-mC) are illustrated (a, b, c, d, e and f, respectively). DNMT3B SNPs rs998382, rs6058891, rs4911256 and rs2889703 were associated with global DNA methylation (P < 0.05, one-way ANOVA). Bfadjusted P = Bonferroni adjust P value. Associations did not hold for stringent multiple testing. Circles represent outliers.

Discussion

Utilizing a unique clinical and biological data set of suicide attempters and non-attempter psychiatric controls, we have identified a number of genetic variants associated with SA in an important epigenetic machinery gene, DNMT3B and documented, for the first time, differential blood global DNA methylation levels between SA and controls. Although preliminary, the findings here are consistent with a model that epigenetic mechanisms play an important role in the pathogenesis of suicidal acts.

DNMT1 is a key maintenance methyltransferase enzyme responsible for copying pre-existing methylation patterns onto newly replicated DNA strands during cell divisions (Kinney & Pradhan 2011). Aberrant expression of the DNMT1 gene had previously been associated with schizophrenia (Zhubi et al. 2009). In this study no association between DNMT1 polymorphisms (n = 7) and SA was found. In addition, no association between DNMT1 polymorphisms and a diagnosis of schizophrenia in our cohort was identified. However, there were a limited number of individuals with a diagnosis of schizophrenia and/or other psychotic disorders in our patient cohort (n = 12).

DNMT3b is a de novo methyltransferase, responsible for establishing methylation patterns during early embryogenesis (Chédin 2011). Interestingly, increased DNMT3b mRNA and protein expression have been found in several regions of suicide/MDD brains (Poulter et al. 2008). A functional DNMT3B polymorphism was found to be significantly associated and five additional DNMT3B polymorphisms nominally associated with SA in this study. These loci include splicing regulation sites, TFBSs and other regulatory regions, and thus could potentially influence expression of the DNMT3B gene and hence it's DNA methylation activity. These SNPs have not previously been associated with disease. The most strongly associated SNP, rs2424932, is present in the 3′UTR (an important regulatory region) of the DNMT3B gene. In addition, the A/G polymorphism at this locus can create alternative TFBSs, hence it is plausible to suggest that this SNP could potentially alter DNMT3B gene regulation. Aberrant DNMT3B expression in the brains of suicide victims has previously been shown to correlate with promoter methylation of an important neurobiological gene (Poulter et al. 2008), hence it is possible that functional SNPs, which are likely to influence expression of the DNMT3B gene, may be associated with aberrant promoter DNA methylation profiles of genes important for suicidal behaviour. RNA was not available for our patients; hence the functional consequence of the SNPs identified in this study on DNMT3B gene expression could not be accessed directly.

To date a number of gene × childhood adversity interactions have been reported in psychiatric patients (Caspi et al. 2003, Gillespie et al. 2005, Prathikanti & Weinberger 2005). Childhood trauma (including abuse) has been reported to interact with low expressing 5-HTTLPR genotypes and moderate the risk of suicidal behaviour (Roy et al. 2007). Recently, a link between aberrant epigenetic processes, traumatic life events and psychiatric disorders has been established. For example, aberrant epigenetic silencing of the gene encoding the glucocorticoid receptor was reported in the brains of suicide victims with a history of childhood neglect/abuse (Mcgowan et al. 2009). Here, we investigate a possible moderation effect of DNMT3B susceptibility polymorphisms on childhood abuse and SA risk. No gene × childhood abuse interaction was found. Future studies could investigate the putative interaction of these genetic variants with childhood trauma score, which would include sexual, physical and emotional abuse and neglect. Such a study would provide a more comprehensive assessment of gene × childhood adversity interaction and risk for SA at this locus. In addition, the moderate sample size used in this study may have been under powered to identify small gene by environment effects.

Examination of global methylation levels [the presence of 5-mC among all cytokines and not only among cytosines directly preceding guanines (CpG sites)] found that psychiatric patients with a history of SA had significantly higher levels of 5-mC, suggesting that patients with a history of SA have global hypermethylation of their genome. Moreover, we show that aberrant DNA methylation patterns in patients exhibiting suicidal behaviour can be identified by examining methylation profiles in their blood. Indeed, global methylation analysis in a number of cancers have proposed that examination of 5-mC content in blood could act as a reliable biomarker of certain cancers (Woo & Kim 2012), thus global hypermethylation (as assessed by 5-mC%) may represent a novel biomarker of SA risk in psychiatric patients. Interestingly, individuals with SA susceptibility DNMT3B genotypes (rs998382 (T/T), rs6058891 (T/T), rs4911256 (A/A) and rs2889703 (C/C) genotypes had higher 5-mC levels than individuals with alternative genotypes at each locus. It is important to note these SNPs (rs998382, rs6058891 and rs4911256) are in high LD with each other; hence one or more of these SNPs could be driving the association with 5-mC levels in the blood. As global DNA methylation levels do not reflect gene specific methylation, susceptibility genotypes could also be associated with hypo or hyper DNA methylation of particular genomic locations (e.g. promoter CpG islands), which was not assessed in the current study. Blood 5-mC% levels observed in this study (<3%) are consistent with previous studies examining 5-mC content using an antibody to 5-mC in blood and various other tissues (Kinnally et al. 2011, Li & Liu 2011, Nestor et al. 2012, Woo & Kim 2012).

A number of limitations are apparent in the current study. Firstly, case/control samples were not age-matched, with the SA group having a significantly younger mean age. The SA group also contained a greater number of individuals with a family history of SA and a history of childhood abuse. In addition, rates of MDD diagnoses were different in SA and non-attempter psychiatric controls. However, the genetic variants identified in this study were not associated with either MDD, a family history of SA or abuse. The modest sample size used in this study may have reduced power to detect small genetic effects. Finally, global methylation quantification was performed on DNA derived from whole blood, which includes a heterogeneous mixture of cell types, hence blood cell specific differences in DNA methylation levels could not determined.

In conclusion, these observations suggest that epigenetic mechanisms play an important role in the pathogenesis of suicidal acts in psychiatric patients. Thus, findings presented here warrant replication in larger and independent cohorts of suicide attempters. Moreover, an understanding of the functional consequence of these DNMT3B genetic variants to DNMT3b functionality would be advantageous to advance our understanding of their role in SA. Global DNA methylation analysis suggests that quantification of blood 5-mC levels may represent a novel biomarker for SA risk in psychiatric patients and future investigation in a larger cohort of patients is justified.

Acknowledgments

We acknowledge funding from the Craig-Dobbin Newman Fellowship in Mental Health Research and the Programme for Research in Third-Level Institutions (PRTI) and statistical advice from the Centre for Support and Training in Analysis and Research (CSTAR), University College Dublin. We would like to acknowledge assistance from Aggie Boylan, Hammad Khan and Sharyn Carley. We the author's confirm that there is no conflict of interest to declare.

Ancillary