BRD4 Connects Enhancer Remodeling to Senescence Immune Surveillance

Abstract

Oncogene-induced senescence is a potent barrier to tumorigenesis that limits cellular expansion following certain oncogenic events. Senescent cells display a repressive chromatin configuration thought to stably silence proliferation-promoting genes while simultaneously activating an unusual form of immune surveillance involving a secretory program referred to as the senescence-associated secretory phenotype (SASP). Here, we demonstrate that senescence also involves a global remodeling of the enhancer landscape with recruitment of the chromatin reader BRD4 to newly activated super-enhancers adjacent to key SASP genes. Transcriptional profiling and functional studies indicate that BRD4 is required for the SASP and downstream paracrine signaling. Consequently, BRD4 inhibition disrupts immune cell-mediated targeting and elimination of premalignant senescent cells in vitro and in vivo Our results identify a critical role for BRD4-bound super-enhancers in senescence immune surveillance and in the proper execution of a tumor-suppressive program.

Significance: This study reveals how cells undergoing oncogene-induced senescence acquire a distinctive enhancer landscape that includes formation of super-enhancers adjacent to immune-modulatory genes required for paracrine immune activation. This process links BRD4 and super-enhancers to a tumor-suppressive immune surveillance program that can be disrupted by small molecule inhibitors of the bromo and extra terminal domain family of proteins. Cancer Discov; 6(6); 612-29. ©2016 AACR.See related commentary by Vizioli and Adams, p. 576This article is highlighted in the In This Issue feature, p. 561.

Figure 1. OIS is accompanied by global remodeling of enhancers

(A) Heat maps showing H3K27Ac ChIP-Seq signals over the 32,986 H3K27Ac-enriched union enhancers identified in proliferating, quiescent and senescent IMR90 cells. Rows correspond to ±5Kb regions across the midpoint of each H3K27Ac-enriched union enhancer, ranked by increased H3K27Ac signal in senescent cells versus proliferating cells. Color shading corresponds to the H3K27Ac ChIP-Seq read count in each region. (B and C) Heat maps illustrating average read counts (normalized for total number of reads per region) of H3K27Ac signals over ± 5-Kb regions centered around 31,731 typical enhancers (TEs) (B) and 1,255 super-enhancers (SEs) (C) in the indicated conditions. Senescence-activated or -inactivated enhancers marked by brackets were defined as H3K27Ac-enriched union enhancer regions exhibiting a greater or less than two fold change in H3K27Ac signals in senescence versus proliferating cells, respectively. (D and E) Box plots showing relative changes in H3K27Ac signals in the indicated enhancer regions from quiescent (blue) and senescent (red) cells when compared to proliferating counterparts. Senescence-activated or -inactivated TEs (D) and SEs (E) were defined as described above. Fold changes with log2 scale (y-axis) were calculated by dividing H3K27Ac tag counts from quiescent or senescent conditions by H3K27Ac tag counts from proliferating condition. Significance was determined using a two-tailed t test. (F–I) H3K27Ac ChIP-Seq occupancy profiles at representative loci of senescence-activated (F and H) and senescence-inactivated (G and I) TEs (F and G) or SEs (H and I). Black bars above gene tracks denote TEs or SEs. Grey color on IL37 gene indicates it is not substantially expressed based on RNA-Seq data.

Figure 2. Enhancer remodeling during OIS is associated with a unique gene expression program

(A) Heat map of unsupervised hierarchical clustering from RNA-Seq data. (B–E) Gene Set Enrichment Analyses (GSEA) comparing the expression of genes associated with senescence-activated or senescent inactivated typical enhancers (TEs) or super enhancers (SEs) between proliferating and senescent cells. Genes with the closest transcriptional start site (TSS) from H3K27Ac binding sites were considered as enhancer-associated genes. Each signature was defined from top 200 senescence-activated (B) and -inactivated (C) TEs, or 198 senescence-activated (D) and 191 inactivated (E) SEs. Genes were rank ordered according to their fold change between proliferating and senescent conditions, as determined by their averaged RNA-Seq reads across three biological replicates per condition. NES, normalized enrichment score; FDR, false-discovery rate. (F–I) GREAT gene ontology (GO) analyses of the genes associated with highly confident H3K27Ac binding sites for each of the indicated class of enhancers, activated (F and H) or inactivated (G and I) during senescence. Enhancers with tag counts >40 in senescent (F and H) or proliferating conditions (G and I) were considered as highly confident regions for H3K27Ac binding. (J–M) Heat maps showing normalized expression of genes associated with the indicated senescence-activated or senescent inactivated TEs and SEs in proliferating, quiescent and senescent cells. The list of genes for each panel was extracted from GO analyses of Fig. 2F–2I.

Figure 3. BRD4 is recruited to senescence-activated SEs associated with SASP genes

(A) BRD4 ChIP-Seq enrichment meta-profiles in proliferating (black), quiescent (blue) and senescent (red) IMR90 cells, representing the average read counts per 20-bp bin across a 5-Kb window centered on all 32,986 union H3K27Ac enhancers. (B) Immunoblot analysis of BRD4 and β-actin (loading control) in whole cell lysates from proliferating, quiescent and senescent cells. (C and D) Scatterplots of absolute BRD4 signals (tag counts) at senescence-activated TEs (C) or SEs (D) in proliferating (x-axis) and senescent (y-axis) cells. The 1,307 TEs and 37 SEs showing a greater than 0.5 fold change (log2 scale) in BRD4 signals in senescence versus proliferating cells are marked in red. (E and F) Box plots of absolute BRD4 signals at the indicated enhancer loci in proliferating (white), quiescent (blue) or senescent (red) cells. P-values were calculated by comparing absolute BRD4 ChIP-Seq signals at 1,307 BRD4-gained TEs (E) or 37 BRD4-gained SEs (F) in proliferating cells versus quiescent or senescent counterparts using a two-tailed t-test. Normalized tag counts of the BRD4 ChIP-Seq is provided in Table S2. (G) ChIP-Seq occupancy profiles for BRD4 (blue) and H3K27Ac (red) at IL1A/IL1B and IL8 locus at indicated conditions. Black bars above gene tracks denote SEs. The IL1A and IL1B H3K27Ac plots are data from Fig. 1H. (H) GREAT GO analysis of 37 BRD4-gained SEs. (I) Heat map showing the relative expression of genes associated with the 37 BRD4-gained senescence-activated SEs in proliferating, quiescent or senescent IMR90 cells. (J) GSEA of the genes associated with 37 BRD4-gained SEs with proliferating and senescent conditions. Genes associated with BRD4 binding sites were defined as described in Fig. 2. (K) Motif analysis of 37 BRD4-gained SEs for putative transcription factor binding sites. A 1-Kb region centered on BRD4 peaks was used for motif discovery. Motifs with p-value less than 10⁻¹² were considered as significant candidates.

Figure 4. BRD4 bromodomain is critical for expression of SASP genes

(A) Heat map of supervised hierarchical clustering from RNA-Seq data. Senescence gain- and loss- signatures correspond to top 100 genes that display more than two fold change in senescent (but not in quiescent) cells as compared to proliferating condition. Expression values (rpkm) of combined 200 senescent-signature genes in the indicated conditions were subjected to clustering based on one minus Pearson correlation test. (B) Supervised principal component analysis (PCA) of BRD4 gene expression signature for the indicated conditions. BRD4 signature was defined by identifying senescent-activated genes that displayed more than two fold decrease with shBRD4 expression. (C) GSEA on the RNA-Seq data of BRD4 signature genes. 210 gene sets available from the Molecular Signature Database (size of gene set; min=15, max=500) were used in the analysis. From 210 gene sets tested, only 24 gene sets were considered as statistically significant (FDR<25%). Similar gene sets were collapsed as representative groups, including immune system/inflammation-related pathways (red), NF-κB signaling pathway (blue) and Ras-signaling pathway (green). (D) GSEA of 37 BRD4-gained SE signature following shRen or shBRD4 expression. (E) Area-proportional Venn diagram showing overlap between senescence-activated genes regulated by BRD4 (red) or p65 (blue). (F, G) qRT-PCR analyses of the indicated SASP factors in proliferating (P) or senescent (S) cells expressing shRNAs targeting Renilla (neutral control), BRD4 or p65, or treated with vehicle (veh) or JQ1. Shown are representative senescence-activated SASP genes exhibiting BRD4 and p65 co-dependency (F), or only BRD4-dependency (G). See Supplementary Figs. S3C and S3D for additional examples. Data are represented as mean ± standard deviation (SD) of two biological replicates. (H) H3K27Ac ChIP-Seq occupancy profiles showing gain of H3K27Ac-marked SEs associated with the indicated SASP genes in two senescence conditions triggered by different stimuli. The OIS H3K27Ac mark plots are data from Fig. 1H. (I) qRT-PCR analyses of the indicated SASP factors in proliferating (P) or senescent (S) cells expressing shRNAs targeting Renilla (neutral control) or BRD4, or treated with vehicle (veh) or JQ1. Senescence was induced by etoposide treatment. See Supplementary Fig. S5C for additional senescence-activated SASP genes exhibiting BRD4-dependency in the context of etoposide-induced senescence. Data are represented as mean ± standard deviation (SD) of two biological replicates.

Figure 5. BRD4 is indispensable for the production of senescence-associated secretome but not for the cell cycle arrest

(A) Immunoblot analyses of the indicated proteins in whole cell lysates of proliferating (P) or senescent (S) IMR90 cells expressing shRNAs against BRD4 (shBRD4) or Renilla (shRen, neutral control). Senescence was induced by H-Ras^G12D expression. β-actin was used as loading control. (B) Quantification of SA-β-gal staining in proliferating (P) or senescent (S) IMR90 cells expressing the indicated shRNAs, or treated with JQ1 or vehicle (veh). Data are presented as means ± SD of two independent experiments. (C and D) Quantification of BrdU incorporation (C) and SAHF formation (D) in proliferating (P) or senescent (S) IMR90 cells expressing the indicated shRNAs, or treated with JQ1 or vehicle (veh). Data are presented as means ± SD of two independent experiments. (E) Colony formation assays documenting the impact of inactivating BRD4 on the ability of H-Ras^V12-expressing IMR90 cells to grow at low density. Shown are representative crystal violet stainings of 6-cm plates 2 weeks after plating. Proliferating (P) and senescent (S) cells expressing a tandem Rb and p53 shRNA (S/shRb+p53; expected to bypass senescence) were used as control. (F and G) Cytokine array analysis of conditioned media analysis from proliferating (P) or senescent (S) IMR90 cells expressing the indicated shRNAs (F) and relative quantification (G). Data are presented as means ± SD of two independent experiments.

Figure 6. BRD4 mediates paracrine senescence and SASP-triggered immune cell activation in vitro

(A) Proliferation of naïve IMR90 cells, untreated or subjected to conditioned media (CM) from proliferating (P) or senescent (S) IMR90 cells expressing the indicated shRNAs, as a measure of SASP-induced paracrine senescence. (B) Representative micrographs of the same cell populations as in (A) stained for SA-β-gal activity after a 7-day exposure to the indicated CM. (C) Quantification of the percentage of naïve IMR90 cells staining positive for SA-β-gal activity following a 7-day exposure to CM from proliferating (P) or senescent (S) cells expressing the indicated shRNAs. Data correspond to the mean ± SD of two independent experiments. (D) qRT-PCR analysis of SASP gene expression in naïve IMR90 cells exposed to the indicated CM. Data correspond to the mean ± SD of two independent experiments. (E) qRT-PCR analysis of M1 (left panels) or M2 (right panels) polarization markers in human monocyte-derived macrophages exposed to CM from proliferating (P) or senescent (S) cells expressing the indicated shRNAs. (F) Representative micrographs of co-cultures of NK cells (red) with proliferating (P) or senescent (S) IMR90 cells expressing the indicated GFP-linked shRNAs. Note the rapid NK cell-mediated elimination of senescent cells expressing the neutral shRNA against Ren (shRen). (G) Quantification of NK cell “targeting” of IMR90 cells. NK cell attraction to the indicated IMR90 cell populations was quantified at the indicated time points by automated detection of the total IMR90 cell area covered by NK cells. (H) Quantification of IMR90 viability over time in NK-IMR90 co-cultures. The viability of IMR90 cells expressing the indicated GFP-linked shRNAs was quantified at the indicated time points by automated high-throughput detection of GFP positive signals (see Supplemental Experimental Procedures for additional details), and expressed normalized to GFP values at time point zero. Data correspond to means ± SEM of 2 biological replicates, 4 adjacent fields per well were analyzed per replicate, per condition.

Figure 7. Brd4 inhibition impairs immune surveillance of oncogenic N-Ras expressing senescent cells in vivo

(A) A schematic diagram of the three sleeping beauty constructs used in the hydrodynamic transfection experiments. The blue triangles indicate the sleeping beauty direct and inverted repeats. SB: Sleeping Beauty transposase. (B) Experimental strategy to address the impact of systemic Brd4 inhibition on clearance of senescent hepatocytes. Mice were injected with the indicated constructs through hydrodynamic transfection and treated with I-BET 762 (iBET) or vehicle orally every day (15mg/kg). Livers were harvested at days 6 and 12 and subjected to histological analyses to assess the extent of immune-mediated senescent hepatocyte clearance (see Figs. C–E). The same vector system and timeline was used for hepatocyte-specific Brd4 depletion using shRNAs against Brd4 (see Figs. F–L). Figure adapted from ref. . (C) Representative immunohistochemistry (IHC) stainings for GFP on liver sections harvested from vehicle or iBET-treated mice at day 6 and 12 post-transduction of NRas^G12D-IRES-GFP by hydrodynamic injection. Arrows point to clusters of immune infiltrates. (D) Quantification of the number of GFP positive foci in livers from vehicle or iBET-treated mice at the indicated time points post-injection. Each dot represents the mean numbers of GFP+ foci per mouse (ten 10X fields quantified/mouse) and results are expressed relative to vehicle-day 6 mean value. Means ± SD values for each experimental group are also indicated. (E) Representative H&E staining of livers from vehicle or iBET-treated mice harvested at day 12 post-injection. Arrows point to clusters of immune infiltrates. (F) Representative immunofluorescence (IF) for N-Ras (green) on day 6 (top panels) or day 12 (bottom panels) liver tissue sections from mice injected with transposon-based vectors encoding for oncogenic N-Ras- and indicated shRNAs. Nuclei are counterstained with DAPI. (G) Bar graphs show corresponding quantifications of the number of N-Ras+ cells per field, at day 6 or day 12 post-injections. Data are presented as mean + S.D. (n=4) from thirty 10X fields per mouse. Results are presented as number of GFP positive foci relative to control shRNA on day 6. (H) Immunohistochemistry (IHC) staining for GFP on day 6 liver tissues from mice injected with transposon-based vectors encoding for oncogenic N-Ras and indicated shRNAs. Hematoxylin was used for counterstaining. GFP marks hepatocytes co-expressing N-Ras and the indicated shRNAs. Arrows indicate immune infiltrates surrounding N-Ras/shRen-expressing GFP positive foci. (I) Co-immunofluorescence for N-Ras (green) and CD45 (red) on frozen liver tissue sections harvested on day 6. Nuclei are counterstained with DAPI. (J) Bar graphs show corresponding quantifications of the percentage of N-Ras foci infiltrated by CD45+ cells at day 6. Data are presented as mean + S.D. (n=4) from thirty 10X fields per mouse. (K) Triple-immunofluorescence for N-Ras (green), Cd45 (white) and the SASP factor Mcp1 (red) on frozen liver tissue sections harvested on day 6. Nuclei are counterstained with DAPI. Arrows point to N-Ras/shRen-expressing foci exhibiting induced levels of the secreted chemokine Mcp1.