Excess Translation of Epigenetic Regulators Contributes to Fragile X Syndrome and Is Alleviated by Brd4 Inhibition

Abstract

Fragile X syndrome (FXS) is a leading genetic cause of intellectual disability and autism. FXS results from the loss of function of fragile X mental retardation protein (FMRP), which represses translation of target transcripts. Most of the well-characterized target transcripts of FMRP are synaptic proteins, yet targeting these proteins has not provided effective treatments. We examined a group of FMRP targets that encode transcriptional regulators, particularly chromatin-associated proteins. Loss of FMRP in mice results in widespread changes in chromatin regulation and aberrant gene expression. To determine if targeting epigenetic factors could reverse phenotypes associated with the disorder, we focused on Brd4, a BET protein and chromatin reader targeted by FMRP. Inhibition of Brd4 function alleviated many of the phenotypes associated with FXS. We conclude that loss of FMRP results in significant epigenetic misregulation and that targeting transcription via epigenetic regulators like Brd4 may provide new treatments for FXS.

Keywords: Brd4; FMRP; FXS; chromatin; histones.

Figure 1. FMRP targets include chromatin-associated proteins

(A) Model of FMRP function in neurons. (B) Example of FMRP HITS-CLIP analysis and FMRP targets. (C) PANTHER gene ontology analysis of processes enriched in FMRP HITS-CLIP transcripts. (D) Western blot of chromatin-associated targets of FMRP from WT or Fmr1 KO cultured cortical neurons. (E) Western blot of histone modifications from WT or KO neurons. (F) H3K4me3 staining in neurons transfected with GFP and Fmr1 shRNA. (G, H) Quantification of H3K4me3 in transfected cell relative to the closest neighboring cell for immature (G) and mature neurons (H). For (G), n = 74, 41, and 51 neurons respectively for Luciferase shRNA, Fmr1 shRNA1, Fmr1 shRNA2 from 5 biological replicates. For (H), n = 47, 22, and 44 neurons from Luciferase, Fmr1 shRNA1, and Fmr1 shRNA2 from 4 biological replicates. *p<0.05, ***p<0.001, one-way ANOVA with post-hoc t-test. Scale bar is 10μm. Graphs show mean ± SEM. See also Figure S1, Table S1, and Table S4 for ANOVA values.

Figure 2. ChIP-sequencing reveals increased H3K4me3 and H3K27ac in Fmr1 KO neurons

(A) H3K4me3 ChIP-sequencing gene tracks for WT and KO neurons. (B, C) H3K4me3 TSS occupancy (B) and RPKM values (C) for all genes in WT and KO neurons. (D) Method for in vivo ChIP-sequencing using cerebellar neurons from NeuroD1 bacTRAP mice. (E) H3K4me3 gene tracks for WT and Fmr1 KO mice. (F, G) H3K4me3 TSS occupancy (F) and RPKM values (G) for all genes in WT and KO cerebellar neurons from NeuroD1 bacTRAP mice. G is average of n = 3 replicates. ***p<0.001, two-sided paired t test. See also Figure S2.

Figure 3. RNA-sequencing in Fmr1 KO neurons

(A) Gene tracks and average RPKM for Nr4a1 and Shank2 in WT and Fmr1 KO cultured neurons. n = 3 replicates. (B) Z scores of significantly changed genes in Fmr1 KO neurons. (C) Gene ontology analysis of upregulated genes. (D) Overlap of KO misregulated genes with ASD linked genes. (E) Number and percentage of target genes of chromatin-associated proteins found in the FMRP HITS-CLIP analysis that overlap with KO misregulated genes. (F) Significance of the overlap. (G) Venn diagram of KO misregulated genes and target genes of all chromatin-associated proteins that show significant overlap with KO genes. (H) Number and percentage of target genes of chromatin-associated proteins that overlap with KO misregulated genes also linked to ASD. ***p < 0.001, DESeq adjusted p value. Overlap p values, hypergeometric test. Graphs in A show mean ± SEM. See also Figure S3 and Table S2.

Figure 4. Brd4 expression in WT and Fmr1 KO

(A, B) Brd4 (A) or FMRP (B) protein expression at increasing days in vitro (DIV) in WT neurons. (C) Brd4 protein expression at increasing DIV in KO neurons. Blots are representative of 3 biological replicates. (D) Brd4 staining in WT and KO neurons at 10 DIV. (E) Quantification of Brd4 staining in WT and KO neurons. n = 87 and 69 neurons for WT and KO from 3 biological replicates. (F, G) Brd4 (F) and FMRP (G) protein expression in WT and KO cortical tissue. Representative of 3 biological replicates. (H) Brd4 staining of 1 month old WT and KO sagittal brain slices. (I, J) High magnification (I) and quantification (J) of Brd4 staining in cortex of WT and KO mouse brains. n = 137 and 176 neurons for WT and KO from 3 brains each. Scale bars are 10μm in D and I, and 1mm in H. ***p < 0.001, unpaired t test. a.u., arbitrary units. Graphs show mean ± SEM. See also Figure S4.

Figure. 5. Effects of JQ1 on KO neurons

(A, B) Volcano plot of JQ1-regulated genes in WT (A) and KO (B) neurons. (C) Gene ontology groups disrupted in KO that are oppositely regulated after JQ1 treatment of KO. (D) JQ1 downregulated genes significantly overlap with KO upregulated genes and vice versa. (E) JQ1 downregulated genes do not significantly overlap with FMRP target genes. (F) Gene tracks in WT and KO neurons with JQ1 treatment. (G) Heat map and box plot of z scores of up and downregulated KO genes with JQ1 treatment. (H, J) Box plot of the effects of JQ1 on the relative RPKM values of genes that are up (H) and down (J) in KO neurons. (I, K) JQ1-induced changes in KO neurons for genes that are up (I) and down (K) in KO neurons. n = 3 biological replicates. ***p > 0.001, ANOVA with post hoc paired two-sided t-test. Overlap p values, hypergeometric test. Graphs in G show mean ± SEM. See also Figure S5, Table S3, and Table S4 for ANOVA values.

Figure 6. Effects of JQ1 on FXS phenotypes

(A, B) Dendrites (A) and spine number (B) of WT or KO neurons transfected with GFP and treated with JQ1 for 24 hours. n = 68, 66, 80, 77 for WT, KO, WT+JQ1, KO+JQ1 from 2 biological replicates. (C) Behavioral testing paradigm. (D, H) Total distance traveled over 1 hour in open field for high dose cohort (D) or medium dose older cohort (H). (E) Examples of marble burying assay. (F and I) Number of marbles fully buried in 15 minutes for high dose cohort (E) or medium dose older cohort (H). (G and J) Social interaction testing as measured by the relative preference for a mouse or Lego for high dose (G) or medium dose older cohort (J). (K) Novel object recognition testing one day after initial exposure to objects. For high dose cohort n = 8 to 10 mice per condition. For medium dose older cohort older subset n = 6 to 8 mice per condition. Scale bar is 10μm. *p < 0.05, ***p < 0.001, two-way ANOVA with post hoc paired two-sided Student’s t-test. Graphs show mean ± SEM. See also Figure S6 and Table S4 for ANOVA values.

Figure 7. Combinatorial treatment targeting active Brd4 alleviates FXS phenotypes

(A, B) Nr4a1 (A) and Shank2 (B) transcripts in sorted neurons transfected with Brd4 or Brd4-SSS492ESE. n = 5. (C, D) Dendrites (C) and spine number (D) of WT neurons transfected with GFP and Brd4, a phospho-mimic Brd4 (Brd4-SSS492ESE), or a phospho-mutant form of Brd4 (Brd4-S492A). n = 3 to 12 biological replicates with 5 to 10 neurons counted per replicate. (E, F) Dendrites (E) and spine number (F) of WT neurons transfected with GFP and Brd4 with either the WT human sequence surrounding amino acid 116 or the autism mutation of a proline deletion at position 116. n = 7 biological replicates per construct with 5 to 10 neurons counted per replicate. (G) Model of Brd4 activation and drug targets. (H, I) Nr4a1 (H) and Shank2 (I) transcript levels are decreased in response to combined low dose JQ1 and CX-4945, but not with either drug alone at low doses. n = 3 to 5 replicates per condition. (J) Low dose JQ1 + CX-4945 reverses the increase in spine number in KO neurons. N = 2 to 5 biological replicates. (K) Total distance over 1 hour in open field. (L) Number of marbles buried in 15 minutes. (M) Social interaction testing as measured by the relative preference for a mouse or Lego. N = 16 to 26 mice per condition. (N) Model of FMRP function in transcriptional regulation in neurons. FMRP targets include chromatin-associated proteins in addition to synaptic proteins. Loss of FMRP in FXS results in misregulation of chromatin-associated proteins, histone modifications, and transcriptional output. Brd4 inhibition alleviates transcriptional and phenotypic changes in FXS. Scale bar is 10μm. *p < 0.05, ***p < 0.001. For A-F: one-way ANOVA, for H-M: two-way ANOVA, with post hoc paired two-sided Student’s t-test. a.u., arbitrary units. Graphs show mean ± SEM. See also Figure S7 and Table S4 for ANOVA values.