An HDAC-dependent epigenetic mechanism that enhances the efficacy of the antidepressant drug fluoxetine

Abstract

Depression is a prevalent and debilitating psychiatric illnesses. However, currently prescribed antidepressant drugs are only efficacious in a limited group of patients. Studies on Balb/c mice suggested that histone deacetylase (HDAC) inhibition may enhance the efficacy of the widely-prescribed antidepressant drug fluoxetine. This study shows that reducing HDAC activity in fluoxetine-treated Balb/c mice leads to robust antidepressant and anxiolytic effects. While reducing the activity of class I HDACs 1 and 3 led to antidepressant effects, additional class II HDAC inhibition was necessary to exert anxiolytic effects. In fluoxetine-treated mice, HDAC inhibitors increased enrichment of acetylated histone H4 protein and RNA polymerase II at promotor 3 of the brain-derived neurotrophic factor (Bdnf) gene and increased Bdnf transcription from this promotor. Reducing Bdnf-stimulated tropomyosin kinase B receptor activation in fluoxetine-treated mice with low HDAC activity abolished the behavioral effects of fluoxetine, suggesting that the HDAC-triggered epigenetic stimulation of Bdnf expression is critical for therapeutic efficacy.

Figure 1. Adolescent fluoxetine potentiates the enrichment of acH4K12 histone and RNA Polymerase II at Bdnf promotor 3 and increases Bdnf transcript variant 3 expression in the forebrain neocortex of IMS mice.

(a) acH4K12 ChIP targeting Bdnf promoters 1 to 5. (b) Pol II ChIP targeting promoters 1 to 3. (c) Real-time PCR measures of Bdnf transcript variants 1 to 3. Data (mean ± sem of 7 animals/group) are from SFR and IMS mice, and IMS mice treated with fluoxetine during adolescence in the absence (IMS-F) or presence of pCPA (IMS-pCPA-F). Data were compared by ANOVA (acH4K12 ChIP: P1: F(_3,27) = 0.94, p = 0.44; P2:F(_3,27) = 3.53, p = 0.03; P3: F(_3,27) = 4.46, p = 0.013; P4: F(_3,27) = 0.853, p = 0.48; P5: F(_3,27) = 1.19, p = 0.335; Pol II ChIP: P1: F(_3,27) = 1.78, p = 0.18; P2: F(_3,27) = 2.26, p = 0.11; P3: F_(3,27) = 9.37 p = 0.0003); Bdnf mRNA: P1: F(_3,27) = 1.15, p = 0.35; P2: F(_3,27) = 2.47, p = 0.1; P3: F(_3,27) = 16.04, p = 0.0001)) and statistical differences were resolved post hoc (Tukey Kramer multiple comparisons) as indicated. (d) Behavior of IMS, fluoxetine-treated IMS mice, and fluoxetine-treated IMS mice co-treated with pCPA (IMS-F pCPA) or Ana-12 (IMS-F Ana12) in the Forced Swim Test (FST). (e) Behavior of the same groups of mice in the Elevated Plus Maze (EPM). Data are mean ± sem of 7 animals/group. Significant differences revealed by ANOVA (FST: F(_4,34) = 7.64, p = 0.0003; EPM: F(_4,34) = 5.76, p = 0.0014) were resolved post hoc (Tukey Kramer multiple comparisons) as indicated.

Figure 2. HDAC inhibitors enhance the effects of adolescent fluoxetine in SFR mice.

(a) Behavior of non-treated SFR mice, SFR mice treated with fluoxetine only (F) or co-treated with MS-275 (MS-F), NaB (NaB-F), or TSA (TSA-F) in the FST. (b) Behavior of the same groups of mice in the EPM. (c) Behavior in the Light/Dark Exploration test (L/D test). Statistical differences were determined using ANOVA (FST: F(_4,40) = 4.902, p = 0.0025 (n = 8/group); EPM: F(_4,35) = 7.296, p = 0.0003 (n = 7/group); L/D test: F(_4,35) = 3.119, p = 0.03 (n = 7/group)) and resolved post hoc (Tukey Kramer multiple comparisons) as indicated. (d) Enrichment of acH4K12 at Bdnf promotors 1 to 3. (e) Enrichment of RNA Polymerase II at Bdnf promotors 1 to 3. (f) Real-time RT-PCR measures of Bdnf transcript variants 1 to 3. Data in d–f were obtained from forebrain neocortex of SFR mice treated with fluoxetine alone or co-treated with MS-275, NaB, or TSA during adolescence and they were compared to corresponding data obtained from non-treated SFR mice that are also shown in Fig. 1. Significant differences revealed by ANOVA (acH4K12 ChIP: P1 F(_4,23) = 2.587,p = 0.07; P2 F(_4,20) = 5.104, p = 0.0085; P3 F(_4,21) = 7.565, p = 0.001; Pol II ChIP: P1 F(_4,18) = 1.784,p = 0.1882; P2 F(_4,20) = 2.515, p = 0.0854; P3 F(_4,25) = 5.803, p = 0.0026)) were resolve post hoc as indicated. For real-time RT-PCR measures of Bdnf mRNA levels, Kruskal Wallis non-parametric ANOVA revealed significant differences that were resolved using Dunn's multiple comparison tests as indicated.

Figure 3. The effects of adult fluoxetine in SFR mice in the presence and absence of HDAC inhibitors.

(a) Behavior of non-treated SFR mice, SFR mice treated with fluoxetine only (F) or co-treated with MS-275 (MS-F) or TSA (TSA-F) in the FST. (b) Corresponding EPM and (c) L/D test behavior. Statistical differences were determined using ANOVA (FST: F(_3,32) = 5.684, p = 0.0033 (n = 8/group); EPM: F(_3,32) = 6.689, p = 0.0016 (n = 8/group); L/D test: F(_3,26) = 3.119, p = 0.03 (n = 7–8/group)) and resolved post hoc (Tukey Kramer multiple comparisons) as indicated. (d) The effects of adult fluoxetine on acH4K12 enrichment at promotors 1 to 5 the Bdnf gene in the presence and absence of HDAC inhibitors. (e) Corresponding Pol II enrichment at promotors 1 to 3 and (f), Bdnf transcript variants 1 to 3 mRNA expression. Data are mean ± sem of measures from 5–6 animals per group. ANOVA revealed significant differences for acH4K12 levels only at Bdnf promotor 3 (F(_{3, 22}) = 4.328, p = 0.0174) that were resolved post hoc as indicated. For the Pol II densities, significant differences revealed by ANOVA (F(_3,23) = 4.965, p0.0098) were found only for Bdnf promotor 3, and post hoc Tukey Kramer multiple comparisons resolved these differences for fluoxetine-treated mice that were co-treated with MS-275 or TSA. For real-time RT-PCR measures of Bdnf transcript variant 1 to 3 expression, ANOVA revealed no significant differences in expression of transcript variants 1 (p = 0.25) and 2 (p = 0.07) between groups, but significant differences revealed for transcript variant 3 (F(_3,20) = 3.19, p = 0.0049)) were resolved post hoc as indicated.