Epigenetic Regulation of BDNF Expression According to Antidepressant Response

LETTERS TO THE EDITOR Epigenetic regulation of BDNF expression according to antidepressant response

Molecular Psychiatry (2013) 18, 398--399; doi:10.1038/mp.2012.38; The BDNF gene has distinct splice variants, each regulated by a published online 1 May 2012 specific promoter region, that determine tissue-specific regulation of expression.3 Of these variants, BDNF-IV is the most commonly studied, and its expression changes have been associated with Several lines of evidence support the role of brain-derived behavioral phenotypes, psychiatric disorders and epigenetic neurotrophic factor (BDNF) in the pathophysiology and pharma- modifications.4 The promoter region in exon-IV contains specific cotherapy of depression.1 The neurotrophin hypothesis of binding sites for the cyclic-AMP-responsive-element-binding depression postulates that stress and depression are associated protein5 and the methyl-CpG-binding-protein-2,6 making it a with decreased BDNF expression, which can be reversed by preferential candidate for epigenetic regulation. antidepressant treatment.2 The goal of the present study was to Tsankova et al.7 reported that mice exposed to chronic social investigate the effect of antidepressant treatment on the defeat stress displayed lower levels of BDNF-IV associated with a epigenetic regulation of BDNF in major depressive disorder (MDD). significant increase in histone H3 lysine 27 trimethylation

a 2.0 Peripheral BDNF b 15 H3K27me3 T0 1.5 T0 *** T8 T8 *** 10 *** 1.0 *** Qty Mean Qty Mean 5 0.5 (H3K27Me3) (BDNF/B-Actin)

0.0 0 All subjects NRES RES All subjects NRES RES

(N=25) (N=10) (N=15) (N=25) (N=10) (N=15)

c 2.0 Peripheral BDNF NRES d 15 H3K27me3 NRES RES RES 1.5 * 10

1.0 ** Qty Mean Qty Mean 5 0.5 (H3K27Me3) (BDNF/B-Actin)

0.0 0 T0 T8 T0 T8

e BDNF f %Δ HAMD Scores g %Δ HAMD Scores 2.5 vs H3K27me3 40 vs Δ BDNF 40 vs Δ H3K27me3 2.0 30 30 1.5 20 20 1.0 Qty Mean HAMD Scores 0.5 10 HAMD Scores 10 Δ Δ (BDNF/B-Actin) 0.0 0 0 % 0.0 0.5 1.0 1.5 0.5 1.0 1.5 2.0 2.5 3.0 % 0.0 0.5 1.0 Qty Mean Δ BDNF Δ H3K27me3 (H3K27Me3/B-Actin) Figure 1. Epigenetic regulation of brain-derived neurotrophic factor (BDNF). (a, c) qRT-PCR AQ values of total BDNF and two endogenous controls (b-Actin and GAPDH). (b, d) Histone H3 lysine 27 trimethylation (H3K27me3) levels at BDNF IV promoter normalized to input DNA and b-Actin as controls. (e) Correlation between total BDNF and H3K27me3 levels. (f) Correlation between change in depression severity and change in BDNF expression. (g) Correlation between change in depression severity and change in H3K27me3 expression. Depressed patients (N ¼ 25) were classified into responders (RES) and non-responders (NRES). The asterisks refer to P-values (*Pp0.05; **Pp0.01; ***Pp0.001). Variance bars represent s.e.m. Letters to the Editor 399 (H3K27me3), a modification associated with transcriptional repres- expression through alterations of promoter-IV H3K27me3 levels. Our sion. Long-term imipramine treatment reversed BDNF-IV down- results suggest that changes in H3K27 methylation state of the regulation to baseline levels.7 Studies by our group in the BDNF promoter IV and BDNF expression levels in peripheral tissue postmortem brain of depressed subjects with or without history are biomarker correlates of antidepressant response. Finally, these of antidepressant treatment compared with controls showed an findings are preliminary and await confirmation by larger samples increased expression of BDNF-IV and a decrease of H3K27 and alternative designs. Moreover, additional work is necessary to trimethylation levels in subjects treated with antidepressants only.8 better understand the relationship between epigenetic modifica- In order to investigate the epigenetic regulation of BDNF in tions and the exon-specific regulation of BDNF, as this has the MDD patients according to antidepressant treatment, we con- potential to lead to new therapeutic options for MDD. ducted a prospective study in 25 treatment-naive MDD patients. All patients had Hamilton Rating Scale for Depression (HAM-D) scores X24 at baseline (N ¼ 25, X ¼ 29.4±1.2). All participants gave written informed consent for this study, which was approved CONFLICT OF INTEREST by our Institutional Review Board. The authors declare no conflict of interest. Subjects were excluded from the study if they had comorbidity with other major psychiatric disorders, if they had positive tests for JP Lopez, F Mamdani, B Labonte, M-M Beaulieu, JP Yang, illicit drugs at any point during the study or if they had general MT Berlim, C Ernst and G Turecki medical illnesses. Patients (12 males and 13 females) were treated McGill Group for Suicide Studies, Douglas Mental Health University with citalopram, starting with an initial dose of 10 mg die, which Institute, McGill University, Montreal, QC, Canada was titrated progressively to a maximum of 60 mg die. All final E-mail: [email protected] doses were within the therapeutic range and blood levels of total BDNF and H3K27me3 were measured at baseline (T0) and after 8 weeks (T8) of treatment. Subject treatment compliance was REFERENCES assessed using high-performance liquid chromatography at the end of the trial. All subjects showed detectable plasma citalopram 1 Dwivedi Y. Neuropsychiatr Dis Treat 2009; 5: 433--449. 2 Molendijk ML, Bus BA, Spinhoven P, Penninx BW, Kenis G, Prickaerts J et al. levels and we observed a significant correlation between citalopram Mol Psychiatry 2011; 16: 1088--1095. dose and plasma concentration (Spearman’s r ¼ 0.54; P ¼ 0.005). 3 Pruunsild P, Kazantseva A, Aid T, Palm K, Timmusk T. Genomics 2007; 90: 397--406. A repeated-measures ANOVA with Bonferroni correction 4 Boulle F, van den Hove DL, Jakob SB, Rutten BP, Hamon M, van Os J et al. Mol revealed that in line with previous findings, the expression of Psychiatry advance online publication, 6 September 2011; doi:10.1038/mp.2011.107. peripheral BDNF mRNA in depressed patients (N ¼ 25) was 5 Tao X, West AE, Chen WG, Corfas G, Greenberg ME. Neuron 2002; 33: significantly elevated after 8 weeks of citalopram treatment 383--395. (FC ¼ 34%; Po0.001; Figure 1a). Subjects were classified into 6 Martinowich K, Hattori D, Wu H, Fouse S, He F, Hu Y et al. Science 2003; 302: responders (RES) and non-responders (NRES), based on changes to 890--893. the HAM-D scores. We defined response as 8-week HAM-D scores 7 Tsankova NM, Berton O, Renthal W, Kumar A, Neve RL, Nestler EJ. Nat Neurosci 2006; 9: 519--525. o9, whereas non-response was defined as 8-week HAM-D scores 8 Chen ES, Ernst C, Turecki G. Int J Neuropsychopharmacol 2011; 14: 427--429. 450% reduction in baseline HAM-D scores (T8; RES ¼ 5.6±0.7; N ¼ 15; NRES ¼ 17.0±1.4; N ¼ 10). The RES group showed increased BDNF levels (T8--T0; FC ¼ 49%; Po0.001; Figure 1a) after treatment, whereas there was no significant difference in the NRES group (T8--T0; FC ¼ 3%; P40.05; Figure 1a). Consistently, RES Linking in vivo brain had higher T8 BDNF levels as compared with NRES (T-test Po0.05; Figure 1c), whereas there was no difference at T0 (T-test P ¼ 0.98; serotonin type 1B receptor Figure 1c). Finally, we found a significant correlation between change in depression severity and change in BDNF expression (Pearson’s r ¼ 0.49; R2 ¼ 0.25; Po0.05; Figure 1f). These findings density to phenotypic indicated a relationship between peripheral BDNF expression and citalopram treatment response. heterogeneity of posttraumatic To investigate the role of chromatin modifications in BDNF expression changes, based on previous findings in rodents,7 we stress symptomatology performed chromatin immunoprecipitation (ChIP) and found a significant decrease in H3K27me3 levels at promoter-IV of the Molecular Psychiatry (2013) 18, 399--401; doi:10.1038/mp.2012.60; BDNF gene after 8 weeks of citalopram treatment in all patients (N ¼ 25) according to a repeated-measures ANOVA with Bonferro- published online 15 May 2012 ni correction (FC ¼ 31%; Po0.001; Figure 1b). However, these results were explained primarily by changes in the RES group (FC ¼ 43%; Po0.001; Figure 1b), and there was no significant Brain serotonergic (5-HT) dysfunction has been linked to the difference in the NRES group (P40.05; Figure 1b). Consistently, we neurobiology of posttraumatic stress disorder (PTSD) in both found a significant difference in H3K27me3 levels at T8 between preclinical and clinical studies.1--3 Animal studies have found that groups (T-test Po0.01; Figure 1d), but no difference at T0 (T-test stress exposure reduces serotonin type 1B (5-HT1B) receptor P40.05; Figure 1d). Furthermore, we found a significant negative function in multiple brain regions, which consequently results in correlation between change in depression severity and change in behaviors that resemble chronic anxiety.4 Recent work from our H3K27me3 expression (Pearson’s r ¼0.63; R2 ¼ 0.39; Po0.01; group has suggested that humans with PTSD have markedly Figure 1g). Finally, total BDNF and H3K27me3 levels were reduced 5-HT1B receptor density in a neural circuit that has been significantly negatively correlated (Pearson’s r ¼0.86; R2 ¼ 0.75; consistently implicated in this disorder, including the caudate, Po0.0001; Figure 1e). amygdala and anterior cingulate cortex.2 However, no study of To our knowledge, this is the first study that translates to human which we are aware has examined how these abnormalities are findings previously reported in an animal model of depression,7 linked to the most up-to-date and refined phenotypic model of reporting evidence suggesting that antidepressants regulate BDNF PTSD symptom dimensionality.

& 2013 Macmillan Publishers Limited Molecular Psychiatry (2013), 398 -- 403