Methylation of the promoter of brain-derived neurotrophic factor exon IV and antidepressant response in major depression

Article metrics

Insufficient response to antidepressant pharmacotherapy with monoaminergic drugs is a major problem in the therapy of major depression (MD), occurring in up to two-thirds of depressed patients. To date, there is no biological marker reliably determining patients at high risk to treatment failure. Here, we report the first preliminary evidence that patients showing hypomethylation of the promoter region of the brain-derived neurotrophic factor (BDNF) gene are unlikely to benefit from antidepressant pharmacotherapy.

Several lines of evidence have linked the neurotrophin BDNF with both the pathophysiology of depression and the mode of action of antidepressants.1 Effective antidepressive treatment seems to increase peripheral BDNF levels.2, 3 We have shown earlier that the non-increase of BDNF in serum4 or plasma5 during the first week of antidepressant treatment predicts the final non-response and non-remission with high sensitivity, suggesting that early changes in peripheral BDNF may constitute or reflect a necessary prerequisite for final treatment response. For the complex regulation of BDNF expression by methylation of DNA and histone modifications, exon IV has a dominant role.6, 7 However, almost all evidence regarding these effects has been derived from animal studies. So far, clinical studies in the field showed that (i) methylation profiles of bdnf exon I can distinguish between depressed patients and healthy controls,8 and (ii) responders to treatment with antidepressants display a decrease in bdnf exon IV-related trimethylation at histone H3.9

In order to assess whether DNA methylation of bdnf exon IV promoter in patients with MD affects antidepressant treatment response, we determined the methylation status in leukocytes of 13 cytosine–phosphate–guanine (CpG) sites within the bdnf exon IV promoter10 in the same sample of patients with MD, which has served for the investigation of the impact of early peripheral BDNF changes on final response to antidepressant pharmacotherapy (see also Supplementary information).4, 5

The intent-to-treat sample consisted of 46 patients with MD;4, 5 owing to the missing methylation data at baseline (BL) in seven patients, the final sample consisted of 39 patients (20/19 female/male participants; mean age±s.d.=44.9±12.7; mean Hamilton Depression Rating Scale-21 items version sum score at BL±s.d.=20.4±4.5; mean time to end point (EP)±s.d.=4.2 weeks; range 2–6 weeks). Of the 13 investigated CpG sites, 12 yielded sufficient measurements; the CpG at position +42 was dropped from the analysis. The BL methylation status at CpG position −87 predicted antidepressant non-response: final non-responders had a significantly lower methylated C-fraction than final responders (Figure 1a). Patients without methylation at CpG site −87 had a significantly higher risk for non-response than those with any methylation (Figure 1b). There was no significant interaction of gender or class of antidepressant with the association of CpG-87 methylation status and response at EP (CpG-87* gender: χ2: 1.53; P=0.217; CpG-87*class of antidepressant: χ2: 1.66; P=0.437). These results were also found for final non-remission with a similar effect size (data not shown).

Figure 1

(a) DNA methylation at bdnf exon IV promoter CpG position −87 (relative to first nucleotide of exon IV) significantly differs between final responders and non-responders (Mann–Whitney U: 297; P=0.003; Padj=0.03; Cohen’s d: 0.33). (b) Dichotomized methylation levels (no methylation vs any methylation) show a high predictive value for final non-response. (Fisher’s test: OR; no response if not methylated) =9.00, 95% CI=1.887–42.92; P=0.005). (c) Plasma BDNF levels drop in the unmethylated group during first week of treatment, whereas it increases in the group showing any methylation at bdnf exon IV promoter—CpG position −87. (t-test: t=1.82; df=26; P=0.08) numbers given in the figure represent mean (±s.d.). BDNF plasma levels were only available in a subgroup of the participants (n=28 for this analysis). (d) DNA methylation of the bdnf exon IV promoter does not significantly change during antidepressant treatment (two-way analysis of variance (ANOVA): CpG: F=39.46; P<0.0001; time: F=1.38; P=0.25; CpG*time: F=1.00; P=0.46). DNA methylation data were missing from one participant at end point; therefore, this analysis is based on the subgroup with DNA methylation data at all time points (n=38). (e) In vitro changes of luminescence of a bdnf exon IV promoter construct subcloned in a pGL4 luciferase reporter plasmid and transfected into SH-SY5Y neuroblastoma cell line. Some constructs were artificially methylated using sss1-methylase and S-adenosyl-methionine. Incubation with fluoxetine or venlafaxine for 48 h leads to a decrease of reporter gene activity only in those experiments with an unmethylated construct (repeated measurements ANOVA: fluoxetine unmethylated: F=9.24; df=3; P=0.0019; fluoxetine methylated: F=0.45; df=3; P=0.72; venlafaxine unmethylated: F=7.15; df=3; P=0.0029; venlafaxine methylated: F=0.25; df=3; P=0.86; P-values given in the figure are derived from Dunnett’s multiple comparison test).

PowerPoint slide

Furthermore, patients with an unmethylated CpG site −87 showed a marked, but statistically not significant (P=0.08), decrease of plasma BDNF levels during the first week of treatment (Figure 1c). DNA methylation of the 12 investigated CpG sites in bdnf exon IV promoter did not change significantly during antidepressant treatment between BL and EP (Figure 1d).

For further analysis, we cloned the bdnf exon IV promoter fragment into a pGL4 luciferase expression vector and tested the effect of methylation of the whole vector on antidepressant-induced changes in luciferase expression in a neuroblastoma cell line (SH-SY5Y) (see also Supplementary information). We found a significant decrease of luciferase expression after 48-h incubation with fluoxetine or venlafaxine in the unmethylated fragment only, whereas antidepressants did not change expression levels when the promoter fragment was methylated (Figure 1e).

Our results are supported by a strong neurobiological framework, linking antidepressant response to the ability of an antidepressant to increase BDNF expression. Apart from the well-known pathway via 3′,5′-cyclic adenosine monophosphate (cAMP)-mediated activation of cAMP response element-binding protein, antidepressants can also increase BDNF expression via phosphorylation of methyl-CpG-binding protein 2 (MeCP2) which—in its unphosphorylated form—binds to the promoter and forms a repressor complex, but dissociates from the DNA upon phosphorylation.11 Thus, as methylation of the promoter is a prerequisite for specific MeCP2 binding, our results suggest that this mechanism of antidepressant action on BDNF can only be active in carriers of methylation at the relevant CpG site within the promoter. Antidepressant treatment did not significantly affect the methylation at bdnf promoter IV; thus, changes of the methylation status in this DNA region seem not being involved in the response to antidepressant treatment. Our results derived from a clinical study in MD patients were corroborated by an established in vitro system analysis; nevertheless, they are preliminary and must be confirmed by independent samples with consistent medication treatment and alternative methods. Additional effort is required to disentangle the relation between MD, methylation of bdnf exon IV promoter and MeCP2 dissociation following antidepressant treatment, as this finally might lead to biomarker-guided treatment of MD.


  1. 1

    Krishnan V, Nestler E . Nature 2008; 455: 894–902.

  2. 2

    Sen S, Duman R, Sanacora G . Biol Psychiatry 2008; 64: 527–532.

  3. 3

    Molendijk ML, Bus BA, Spinhoven P, Penninx BW, Kenis G, Prickaerts J et al. Mol Psychiatry 2011; 16: 1088–1095.

  4. 4

    Tadić A, Wagner S, Schlicht KF, Peetz D, Borysenko L, Dreimüller N et al. Prog Neuropsychopharmacol Biol Psychiatry 2011; 35: 415–420.

  5. 5

    Dreimüller N, Schlicht KF, Wagner S, Peetz D, Borysenko L, Hiemke C et al. Neuropharmacology 2012; 62: 264–269.

  6. 6

    Hing B, Davidson S, Lear M, Brren G, Quinn J, McGuffin P et al. Biol Psychiatry 2012; 71: 618–626.

  7. 7

    Tsankova NM, Berton O, Renthal W, Kumar A, Neve RL, Nestler EJ . Nat Neurosci. 2006; 9: 519–525.

  8. 8

    Fuchikami M, Morinobu S, Segawa M, Okamoto Y, Yamawaki S, Ozaki N et al. PLoS One 2011; 6: e23881.

  9. 9

    Lopez JP, Mamdani F, Labonte B, Beaulieu MM, Yang JP, Berlim MT et al. Mol Psychiatry 2013; 18: 398–399.

  10. 10

    Keller S, Sarchiapone M, Zarrilli F, Videtic A, Ferraro A, Carli V et al. Arch Gen Psychiatry. 2010; 67: 258–267.

  11. 11

    Hutchinson AN, Deng JV, Cohen S, West AE . J Neurosci 2012; 32: 14355–14363.

Download references


This work was supported by in-house research funds of the University Medical Centre Mainz and Hannover Medical School. Parts of the doctoral thesis of Konrad F. Schlicht are incorporated into this report.

Author information

Correspondence to A Tadić.

Ethics declarations

Competing interests

AT, KL, SB and HF are designated as inventors of the European patent number 12171541.1–2404 ‘Method for predicting response or non-response to a monoaminergic antidepressant’.

Additional information

Supplementary Information accompanies the paper on the Molecular Psychiatry website

Supplementary information

PowerPoint slides

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Tadić, A., Müller-Engling, L., Schlicht, K. et al. Methylation of the promoter of brain-derived neurotrophic factor exon IV and antidepressant response in major depression. Mol Psychiatry 19, 281–283 (2014) doi:10.1038/mp.2013.58

Download citation

Further reading