Histone H3 serotonylation dynamics in dorsal raphe nucleus contribute to stress- and antidepressant-mediated gene expression and behavior

Abstract

Background: Major depressive disorder (MDD), along with related mood disorders, is a debilitating illness that affects millions of individuals worldwide. While chronic stress increases incidence levels of mood disorders, stress-mediated disruptions in brain function that precipitate these illnesses remain elusive. Serotonin-associated antidepressants (ADs) remain the first line of therapy for many with depressive symptoms, yet low remission rates and delays between treatment and symptomatic alleviation have prompted skepticism regarding precise roles for serotonin in the precipitation of mood disorders. Our group recently demonstrated that serotonin epigenetically modifies histone proteins (H3K4me3Q5ser) to regulate transcriptional permissiveness in brain. However, this phenomenon has not yet been explored following stress and/or AD exposures.

Methods: We employed a combination of genome-wide and biochemical analyses in dorsal raphe nucleus (DRN) of male and female mice exposed to chronic social defeat stress to examine the impact of stress exposures on H3K4me3Q5ser dynamics, as well as associations between the mark and stress-induced gene expression. We additionally assessed stress-induced regulation of H3K4me3Q5ser following AD exposures, and employed viral-mediated gene therapy to reduce H3K4me3Q5ser levels in DRN and examine the impact on stress-associated gene expression and behavior.

Results: We found that H3K4me3Q5ser plays important roles in stress-mediated transcriptional plasticity. Chronically stressed mice displayed dysregulated H3K4me3Q5ser dynamics in DRN, with both AD- and viral-mediated disruption of these dynamics proving sufficient to rescue stress-mediated gene expression and behavior.

Conclusions: These findings establish a neurotransmission-independent role for serotonin in stress-/AD-associated transcriptional and behavioral plasticity in DRN.

Keywords: ChIP-seq; Chronic social defeat stress; RNA-seq; antidepressants; dorsal raphe nucleus; histone serotonylation.

Fig. 1.. Chronic social stress in both males and female mice results in altered gene expression programs in DRN.

(A) Social interaction ratio of control vs. stress-susceptible vs. resilient male mice (n = 8 in controls, 12 in susceptible and resilient groups). Data were analyzed using a one-way ANOVA, with significant main effects (p<0.0001, F_2,29 = 53.44). Tukey’s multiple comparisons test revealed significant differences between control vs. susceptible mice (p<0.0001) and susceptible vs. resilient mice (p<0.0001). (B) Clustering of control, susceptible and resilient groups for 1,502 differentially expressed (DE) genes (susceptible vs. control; n = 7–8/group, FDR<0.05). (C) Example GO Biological Process and DisGeNET pathway enrichment (FDR<0.05) for the PCGs differentially expressed (at FDR<0.1) in susceptible vs. control males. Dashed line indicates significance via adjusted p-value. (D_Left) Social interaction time of control vs. socially defeated female mice (n = 10/group). Data were analyzed via Student’s two-tailed t-test with significant differences observed between defeated vs. control mice (p=0.0021, t₁₈ = 3.582). (D_Right) Defensive score of control vs. socially defeated female mice. Data were analyzed via Mann-Whitney test (unpaired) with significant differences observed between defeated vs. control mice (p=0.0034, U=14.50). (E) Clustering of defeat and control groups for 234 DE genes (defeat vs. control; n = 5–6/group, FDR< 0.05). (F) Example GO Biological Process and DisGeNET pathway enrichment (FDR<0.05) for the PCGs differentially expressed in defeat vs. control females (at FDR< 0.1). Dashed line indicates significance via FDR. For all bar graphs, data presented as mean ± SEM.

Fig. 2.. Chronic social stress promotes altered H3 serotonylation dynamics in DRN.

(A) Western blotting analysis of H3K4me3Q5ser in DRN (24hr post-SI testing) of control vs. stress-susceptible vs. resilient male mice (n = 6–10/group). Data were analyzed using a one-way ANOVA, with significant main effects observed (p=0.0002, F_2,23 = 12.43). Tukey’s multiple comparisons test revealed significant differences between susceptible vs. resilient mice (p=0.0001), and an a posteriori Student’s two-tailed t-test revealed a significant difference between control vs. susceptible mice (p=0.05, t₁₄ = 2.117). Total histone H3 levels were used as a loading control. (B) Analysis of H3K4me3Q5ser in DRN of control vs. socially defeated female mice (n = 5/group). Data were analyzed via a Student’s two-tailed t-test with a significant difference observed between defeated vs. control mice (p=0.0473, t₈ = 2.341). Total histone H3 levels were used as a loading control. (C) Analysis of H3K4me3Q5ser in DRN from human postmortem brain of MDD individuals with or without antidepressants onboard at their time of death vs. respective demographically matched controls (n = 5/group). Data were analyzed via Student’s two-tailed t-tests (individual MDD groups vs. matched controls) with significant differences observed between MDD –AD’s vs. age/sex matched controls (p=0.0166, t₈ = 3.020). Total histone H3 levels were used as a loading control. For all western blotting graphs, */#p <0.05, ***p <0.001. Data presented as mean ± SEM. A.U., arbitrary units, normalized to respective controls. (D) Venn diagram depicting the degree of overlap between H3K4me3Q5ser enriched PCGs (peaks >5 fold-enriched over input, FDR<0.05) in control male vs. female DRN (n = 3/group, 3–4 samples pooled per n). Odds ratio (OR) and respective p-value of overlap is provided. (E) Representative IGV genome-browser tracks for two sex-specific loci (Xist in female DRN and Kdm5d in male DRN) displaying sex specific enrichment of permissive H3K4me3Q5ser vs. respective inputs. (F) Venn diagram depicting the degree of overlap between male vs. female PCGs displaying differential enrichment for H3K4me3Q5ser in DRN as a consequence of CSDS (male susceptible vs. control, and female defeated vs. control; n = 3/group, 3–4 samples pooled per n, FDR<0.05, Log2FC ≥1.5 or ≤−1.5). Odds ratio (OR) and respective p-value of overlap is provided. (G) Selected GWAS catalog, DisGeNET and GO Biological Process pathways for PCGs displaying overlapping (male vs. female; 1,382 PCGs) differential enrichment for H3K4me3Q5ser as a consequence of CSDS (FDR<0.05). (H) Venn diagram depicting the degree of overlap between male stress-susceptible vs. control and male stress-resilient vs. stress-susceptible PCGs displaying altered H3K4me3Q5ser enrichment in DRN (n = 3/group, 3–4 samples pooled per n), FDR<0.05, Log2FC ≥1.5 or ≤−1.5). Odds ratio (OR) and respective p-value of overlap is provided. (I) Selected GWAS catalog, DisGeNET and GO Biological Process pathways for PCGs displaying overlapping and reversed (90.6% of overlapping PCGs) differential enrichment for H3K4me3Q5ser in male stress-susceptible vs. control and male stress-resilient vs. stress-susceptible comparisons (FDR<0.05). See Fig. S7A–C for uncropped blots.

Fig. 3.. Chronic fluoxetine treatments rescue behavioral deficits and stress-induced H3K4me3Q5ser dynamics in DRN in stress-susceptible male mice.

(A) Schematic of experimental timeline for CSDS and SI testing in male mice, followed by 30 days of fluoxetine treatments vs. water, additional SI testing and DRN collections. (B) SI ratio of control vs. stress-susceptible vs. stress-resilient male mice, pre- vs. post-30 days of water administration as a vehicle control (n = 10–15/group). Data were analyzed using a two-way repeated measures ANOVA, with significant main effects of stress (p=0.0005, F_2,37 = 9.298) and interaction of stress x time (p=0.0234, F_2,37 = 4.162) observed. Posthoc t-tests with Bonferroni correction revealed significant differences between control versus stress-susceptible mice, pre-treatment (p=0.0003), stress-susceptible versus resilient mice, pre-treatment (p=0.0003), and control versus stress-susceptible mice, post-treatment (p=0.0201 (C) SI ratio of control vs. stress-susceptible vs. resilient male mice, pre- vs. post-30 days of fluoxetine administration (n = 10–19/group). Data were analyzed using a two-way repeated measures ANOVA, with significant main effect of interaction of stress x treatment (p=0.0018, F_2,36 = 7.548). Bonferroni’s multiple comparisons tests revealed significant difference between stress-susceptible mice, pre- vs. post-30 days of fluoxetine administration (p=0.0098). Neither control (p > 0.9999) nor resilient mice (p=0.0917) displayed significant alteration in SI following fluoxetine treatments. Posthoc t-tests with Bonferroni correction revealed significant differences between control versus stress-susceptible mice, pre-treatment (p=0.0111) and stress-susceptible versus resilient mice, pre-treatment (p=0.0066). (D) Analysis of H3K4me3Q5ser/H3 in DRN of control vs. stress-susceptible vs. stress-resilient male mice following 30 days of fluoxetine administration vs. water (n = 10–19/group). Data were analyzed using a two-way ANOVA, with significant main effects of stress (p=0.0289, F_2,71 = 3.725) and stress x fluoxetine (p=0.0420, F_2,71 = 3.316). Sidak’s multiple comparisons tests revealed significant differences between stress-susceptible mice post-30 days of fluoxetine administration vs. stress-susceptible mice post-30 days of water administration (p=0.0094), and Tukey’s multiple comparisons tests revealed significant differences between stress-susceptible mice post-30 days of water administration vs. control mice post-30 days of water administration (p=0.0554), and stress-susceptible mice post-30 days of water administration vs. resilient mice post-30 days of water administration (p=0.0013). To control for samples spread across numerous membranes, GAPDH levels were used as an additional loading control. (E) Peak-centered heatmaps depicting H3K4me3Q5ser enrichment at PCGs displaying significant differential enrichment (FDR<0.05, Log2FC ≥1.0 or ≤−1.0) between SUS FLX vs. SUS H₂O mice for each of the four groups (control H₂O, control FLX, SUS H₂O and SUS FLX). (F) Representative IGV genome-browser tracks for two genic loci (Opa1 and Elmo2) displaying significantly (*) increased enrichment for H3K4me3Q5ser in SUS (H₂O) vs. control (H₂O) mice and rescue following of this increased enrichment in SUS (FLX) vs. SUS (H₂O) animals (respective inputs are included). (G) Venn diagram depicting the degree of overlap between male PCGs displaying protracted differential enrichment of H3K4me3Q5ser in DRN as a consequence of CSDS (susceptible H₂O vs. control H₂O) vs. PCGs displaying regulation of the mark by chronic fluoxetine treatments in susceptible mice (susceptible FLX vs. susceptible H₂O); n = 3/group, 3–4 samples pooled per n, FDR<0.05, Log2FC ≥1.0 or ≤−1.0). Odds ratio (OR) and respective p-value of overlap is provided. (H) Selected GWAS catalog and GO Biological Process pathways for PCGs displaying differential enrichment for H3K4me3Q5ser in male stress-susceptible FLX vs. stress-susceptible H₂O comparisons (FDR<0.05). For all bar graphs, *p < 0.05, **p < 0.01, ***p < 0.001. Data presented as mean +/− SEM. A.U., arbitrary units, normalized to respective controls. See Fig. S7D for uncropped blots.

Fig. 4.. Viral-mediated downregulation of H3 serotonylation in DRN promotes stress resilience and rescues stress-induced gene expression.

(A) Schematic of experimental timeline for male CSDS after intra-DRN viral transduction by empty vector, H3.3 WT or H3.3Q5A vectors, followed by behavioral testing and tissue collections for RNA-seq. (B) Representative IHC/IF images of mouse DRN transduced with a lentivirus expressing H3.3Q5A-HA-EF1-RFP (red fluorescent protein) overlayed with staining for HA and a nuclear co-stain [4′,6-diamidino-2-phenylindole (DAPI)]. (C) SI ratios of GFP, H3.3 WT and H3.3Q5A transduced mice, control vs. CSDS (n = 9–13/group). Data were analyzed using a two-way ANOVA, with significant main effects of stress observed (p=0.0001, F_1,57 = 17.29). Bonferroni’s multiple comparisons tests revealed significant differences between control vs. CSDS groups in GFP (p=0.0310) and H3.3 WT mice (p=0.0474), with no differences observed between control vs. CSDS H3.3Q5A mice. (D) Threshold-free RRHO analyses comparing transcriptional profiles for stress-regulated genes in empty vector and H3.3 WT-transduced DRN (control vs. CSDS) to H3.3Q5A-transduced DRN from CSDS mice (n = 4–9/group), demonstrating that H3.3Q5A significantly reversed gene expression programs observed in response to stress in both control groups. Each pixel represents the overlap between differential transcriptomes, with the significance of overlap of a hypergeometric test color-coded. (E) Selected GWAS catalog, DisGeNET and GO Biological Process pathways for PCGs displaying differentially expressed genes in CSDS empty vs. control empty comparisons and rescue in CSDS H3.3Q5A vs. CSDS Empty comparisons (FDR<0.1). Select enriched pathways are shown (FDR<0.05).