Angioplasty induces epigenomic remodeling in injured arteries

(A) Bean plot showing genome-wide distribution of BRD4 or histone mark ChIPseq peak values. P-values from Wilcox test are presented above the plots. (B) Scatter plot showing injury-induced change in binding density (ChIPseq reads) of BRD4 or a histone mark. Red and blue indicates increase and decrease, respectively, with a twofold cutoff.

ChIPseq was performed as described for Fig 1. Integrative genomics viewer (IGV) tracks show comparison of normalized ChIPseq peaks between injured arteries (+, light color) and uninjured sham-control arteries (−, dark color). Boxes highlight the regions where the binding of H3K27ac and BRD4 increased after arterial injury. Non-specific input confirms low background noise. (A, B, C) IGV profiles of ChIPseq peaks illustrating loci-specific binding density of chromatin marks. (D) Schematic proposition of H3K27me3 and H3K27ac redistribution between anti-proliferative (e.g., P57) and pro-proliferative genes (e.g., Uhrf1 and Ccnd1), based on the artery tissue ChIPseq analysis. Note: The cartoon does not represent the true genomic locations of Uhrf1 and Ccnd1.

(A, B) BRD4 ChIPseq peaks at Ezh2 (B is the zoom-in version of A). Boxes highlight one enhancer region at the distal end of Ezh2 promoter and another within an intron. Genomic coordinates of Box-I: chr4 142018866-142019623; Box-II: chr4 142017383-142017905. Mean coverage of ChIPseq peaks (indicative of BRD4 binding dencity) within each box is labeled, with an arrow pointing to the box. Note the H3K4me1 signal is low relative to H3K27ac of the same scale. (C) Effect of Cas9-mediated enhancer deletion (genome editing) on EZH2 expression in rat primary aortic SMCs. sg, small guide RNA. The pair of sgRNAs flank an Ezh2 intronic region, that is, +1,161 bp to +218 bp from the transcription start site. Quantification: Mean ± SEM; n = 3 independent experiments; paired t test, *P < 0.05. (D) Disruption of EZH2 expression through a non-genome-editing, dead Cas9 (dCas9)-facilitated approach. Rat primary aortic SMCs were used. The same sgRNAs (as that in C) were used. Quantification: Mean ± SEM; n = 3 independent experiments; paired t test, *P < 0.05. (E, F) Effect of BRD4 silencing on EZH2 expression. BRD2, BRD3, or BRD4 was silenced with their specific siRNAs (validated in our recent reports) (Wang et al, 2015; Zhang et al, 2019). Cultured rat aortic SMCs were starved for 6 h before transfection with the siRNA for BRD2, 3, or 4 overnight. The cells recovered (without transfection reagents) for 24 and 48 h before RNA and protein extraction, respectively. EZH2 protein and mRNA were measured with Western blot and qRT-PCR (normalized by ΔΔCT-log₂) assays. Quantification: Mean ± SEM; n = 3 independent experiments; one-way ANOVA with Bonferroni test, *P < 0.05 compared with the scrambled-sequence siRNA control. (G) Schematic depicting BRD4 and its co-localization with H3K27ac at Ezh2 that promote Ezh2 transcription.

Source data are available for this figure.

(A) Cartoon of mouse common femoral artery where wire injury was made to induce IH. (B) Diagram indicating the time line for tamoxifen feeding, wire injury, and tissue collection. (C) Immunofluorescence confirming tamoxifen-induced BRD4 KO in mouse arteries. Scale bar: 50 μm. (D) Comparison of IH between WT (Brd4^fl/fl) and smooth muscle cell-specific BRD4 KO (Brd4^fl/fl; Myh11-CreER^T2) mice. Neointima thickness is demarcated by arrow heads. IH is normalized as intima/media area (I/M) ratio. Scale bar: 50 μm. (E) Quantification: Mean ± SEM; n = 4–7 mice, as indicated by the data points in scatter plots. Statistics: one-way ANOVA with Bonferroni test; **P < 0.01, ***P < 0.001; r.u., relative unit.

Tamoxifen-induced BRD4 KO and wire injury were performed as described for Fig 5. (A, B, C, D) Immunofluorescence shows comparison of EZH2 (A, B) or its catalytic product H3K27me3 (C, D) between WT and BRD4 conditional KO mice. Neointima is demarcated by arrow heads. Fluorescence intensity was normalized to cell number (DAPI-stained nuclei). Scale bar: 50 μm. Quantification: Mean ± SEM; n = 5 mice as indicated by the data points in scatter plots. Statistics: nonparametric Mann–Whitney test following Shapiro–Wilk normality determination, *P < 0.05, **P < 0.01; r.u., relative unit.

(A) Picture illustrating intraluminal infusion of lentivirus to express a gene in the balloon-injured rat carotid artery wall. A cannula (yellow device) connected to a syringe was used to inject lentivirus to the carotid artery lumen for infusion into the injured artery wall. (B) Gain of function. EZH overexpression (EZH1 or EZH2 each in fusion with GFP) in the injured rat carotid artery wall was accomplished via intraluminal infusion of lentivirus. Arteries were collected at post-injury day 14 for histology. Immunostained cross sections indicate increase of H3K27me3 in EZH-overexpressing arteries versus Lenti-GFP controls, predominantly in the neointima layer. A, adventitia. M, media. N, neointima. Scale bar: 50 μm. Quantification: Mean ± SEM; n = 3 rats; one-way ANOVA with Bonferroni test, *P < 0.05. No significance between Lenti-EZH1 and Lenti-GFP. (C) EZH gain-of-function exacerbates IH (measured as I/M ratio). Neointima is demarcated between arrow heads. Scale bar: 50 μm. Quantification: Mean ± SEM; n = 4–5 rats; one-way ANOVA with Bonferroni test, *P < 0.05. (D) Picture depicting perivascular application of pan-EZH inhibitor UNC1999 dispersed in a thermosensitive hydrogel. (E) EZH loss of function (inhibition) mitigates IH. Arteries were collected at post-injury day 14. Scale bar: 50 μm. Quantification: Mean ± SEM; n = 4–5 rats; unpaired t test, *P < 0.05.

(A, B, C, D) Loss of function. Silencing efficiency is indicated by Western blots (A). For proliferation (B) and migration (C, D) assays, cells were harvested at 72 h or imaged at 24 h after PDGF-BB stimulation, respectively. Scr, scrambled. NS, not significant. (E, F, G, H) Gain of function. EZH1 or EZH2 overexpression is shown in Western blots (E). * marks the recombinant protein in fusion with GFP; the lower band is endogenous protein. For proliferation (F) and migration (G, H) assays, cells were harvested at 72 h or imaged at 24 h after PDGF-BB stimulation, respectively. Quantification: Mean ± SEM; n = 3 independent experiments. Statistics: one-way ANOVA with Bonferroni test, *P < 0.05.

Source data are available for this figure.

MOVAS cells were pretreated with the pan-EZH1/2 inhibitor UNC1999 (5 μM) for 2 h, or transduced with lentivirus to silence or overexpress EZH1 or EZH2. Starved cells were stimulated with PDGF-BB (final 20 ng/ml) for 24 or 48 h before harvest for qRT-PCR or Western blot assay, respectively. Quantification: Mean ± SEM; n = 3 independent experiments. Statistics: one-way ANOVA with Bonferroni test, *P < 0.05, **P < 0.01. (A, B, C) Effect of pan-EZH1/2 inhibition on the expression of P57, cyclin-D1, and UHRF1. (D, E, F) Effect of EZH1 or EZH2 silencing on the expression of P57, cyclin-D1, and UHRF1. (G, H, I) Effect of increasing EZH1 or EZH2 on the expression of P57, cyclin-D1, and UHRF1. (J) ChIP-qPCR indicating H3K27ac or H3K4me1 binding at Ezh2 or Uhrf1. qPCR data were normalized to IgG control.

Source data are available for this figure.

(A, B) Decrease in UHRF1 due to tamoxifen-induced BRD4 KO in wire-injured mouse femoral arteries. Neointima is demarcated between arrow heads. Scale bar: 50 μm. Fluorescence intensity was normalized to cell number. Quantification: Mean ± SEM; n = 5 mice. Statistics: nonparametric Mann–Whitney test following Shapiro–Wilk normality determination, *P < 0.05; r.u., relative unit. (C, D) Increase of UHRF1 after EZH1 or EZH2 overexpression in angioplasty-injured rat carotid arteries. Neointima is demarcated between arrow heads. Scale bar: 50 μm. Fluorescence intensity was normalized to cell number. Quantification: Mean ± SEM; n = 3 rats. Statistics: nonparametric Mann–Whitney test following Shapiro–Wilk normality determination, *P < 0.05 compared with GFP control; r.u., relative unit.