Phosphorylation by JNK Switches BRD4 Functions

SUMMARY

BRD4, a key regulator with pleiotropic functions, plays crucial roles in cancers and cellular stress responses. It exhibits dual functionality: chromatin-bound BRD4 regulates remodeling through its acetyltransferase (HAT) activity, while promoter-associated BRD4 regulates transcription through its kinase activity. Notably, chromatin-bound BRD4 lacks kinase activity, and RNA Pol II-bound BRD4 exhibits no HAT activity. This study unveils one mechanism underlying BRD4’s functional switch. In response to diverse stimuli, JNK-mediated phosphorylation of human BRD4 at Thr1186 and Thr1212 triggers its transient release from chromatin, disrupting its HAT activity and potentiating its kinase activity. Released BRD4 directly interacts with and phosphorylates Pol II, PTEFb, and c-MYC, thereby promoting transcription of target genes involved in immune and inflammatory responses. JNK-mediated BRD4 functional switching induces CD8 expression in thymocytes and epithelial-to-mesenchymal transition (EMT) in prostate cancer cells. These findings elucidate the mechanism by which BRD4 transitions from a chromatin regulator to a transcriptional activator.

Graphical Abstract

INTRODUCTION

Bromodomain 4 (BRD4), a BET family nuclear protein, is broadly expressed and implicated in numerous cancers and diseases; it is being actively investigated as a therapeutic target^1–4. BRD4 is functionally pleiotropic, regulating chromatin structure, transcription and RNA splicing^5–9. During cellular homeostasis, BRD4 is primarily chromatin-bound^10,11. BRD4 has intrinsic HAT activity that acetylates H3K122, leading to nucleosome eviction and localized chromatin decompaction^5,12,13. It also has intrinsic kinase activity that directly phosphorylates the Pol II CTD, MYC, TAF7 and PTEFb, leading to transcription initiation, pause release and transcription elongation^14,15. Thus, BRD4 regulates chromatin architecture and transcription.

Although BRD4’s pleiotropic functions have been characterized individually, little is known about the signals and mechanisms that regulate its transitions between functional states. The switching of BRD4 from a chromatin regulator to a transcription factor suggests the existence of a signaling mechanism(s) that triggers its release from chromatin. Although BRD4 is mostly chromatin-bound during interphase, it is transiently released from chromatin at the end of mitosis and in response to stresses, such as UV^16–19. Upon its release from chromatin, BRD4 recruits chromatin regulators and transcription factors from the nucleoplasm to chromatin, leading to transcription initiation^19–22. BRD4 then travels with the transcription elongation complex to regulate alternative splicing^8,9. BRD4 release from chromatin is necessary for proper regulation of gene expression.

The mechanism regulating BRD4 release is not known but has been correlated with activation of the c-Jun N-terminal kinase (JNK)¹⁶. The JNK subfamily of kinases consists of three closely related paralogs JNK1, JNK2 and JNK3; JNK1 and JNK2 are widely expressed and generally target the same substrates. The JNK kinases are part of the mitogen activated protein kinase (MAPK) signaling pathway and are activated in response to a wide range of abiotic and biotic stresses, cellular and physiological processes, such as cell cycle, immunological and inflammatory responses and tumorigenesis²³.

BRD4 undergoes various post-translational modifications, including phosphorylation by CK2, JAK2 and CDK1 kinases^21,24,25 which enhance its chromatin binding. No modification has been directly linked to its release from chromatin. We hypothesized that post-translational modifications regulate BRD4’s release and transitions between functional states. Given the correlation between JNK activation and release of BRD4 from chromatin in response to stress¹⁶, we hypothesized that JNK phosphorylates BRD4 to mediate its functional switch from a chromatin modifier to a transcription regulator. We report that activation of the MAPK/JNK pathway results in JNK-mediated phosphorylation of BRD4 at three distinct sites within the C-terminal domain of BRD4. This phosphorylation triggers the release of BRD4 from chromatin, its recruitment to and transcriptional activation of critical inflammatory genes. Following its release from chromatin, BRD4 is dephosphorylated by the PP4 phosphatase, enhancing its interaction with RNA Pol II to activate transcription.

The toggling of BRD4 between chromatin-bound and released forms determines its function. When bound to chromatin, BRD4’s HAT is active while its kinase activity is inhibited by the underlying nucleosomes. JNK-mediated release of BRD4 from chromatin leads to a cessation of its HAT activity and reactivation of its kinase activity, resulting in increased phosphorylation of CDK9Thr186, Pol II CTDSer2 and MYCThr58, and a significant increase in the expression of over 1000 genes, including immune and inflammatory response pathways. Both EMT induction and thymocyte activation depend on JNK-mediated release of BRD4 from chromatin. Thus, JNK phosphorylation of BRD4 is a critical mechanism that switches BRD4 function from a chromatin regulator to a transcriptional regulator.

RESULTS

JNK directly binds to and phosphorylates BRD4

BRD4 and JNK constitutively co-localize in situ, as assessed by proximity ligation assays (PLA) and co-immunoprecipitation from HeLa nuclear extracts, indicating that they exist in a complex (Fig. 1A, B). They directly interact, as demonstrated in pull-down assays with purified recombinant proteins (Fig. 1C). In in vitro pull-down assays, JNK1 bound the N-terminal and C-terminal halves of BRD4 independently (Supplementary Fig. 1A), indicating it makes robust contacts with both regions of BRD4. JNK substrates have a consensus docking motif where JNK binds but phosphorylates at a distal site²³. A consensus JNK docking motif occurs in the N-terminal domain of BRD4 between aa616/615 and aa622/623 and is highly conserved in mouse and human BRD4 (Supplementary Fig. 1B).

Fig. 1: JNK directly interacts with and phosphorylates BRD4.

Fig 1A: BRD4 colocalizes with kinase active JNK. Proximity Ligation Assays (PLAs) with anti-BRD4 and anti-pJNK on fixed HCT116 cells. Negative control; anti-nucleolin and anti-BRD4. (Scale bars, 20 μM)

Fig 1B: JNK co-immunoprecipitates with BRD4. BRD4 was immunoprecipitated from HeLa nuclear extract using anti-BRD4 and immunoblotted with anti-JNK.

Fig 1C: BRD4 binds JNK directly. Recombinant JNK1 (0.1, and 0.2μg) was pulled down with 0.5μg rBRD4 immobilized on Flag-beads.

Fig 1D: JNK phosphorylates BRD4. Upper: map of BRD4 and deletion mutants. Lower: autoradiograph of kinase assays with GST-JNK1 and BRD4 WT or deletion mutants.

Fig 1E: JNK phosphorylation sites on BRD4. Upper: JNK consensus phosphorylation sites located on human/mouse BRD4. Lower: autoradiograph of kinase assays with His-JNK1 and BRD4 WT or the point mutants.

Fig 1F: BRD4 is phosphorylated at Thr1186 and Thr1212 JNK activation. HCT116 cells weretreated with anisomycin, heat shock, LPS treatment or UV stress. BRD4 phosphorylation was assessed by immunoblotting (upper) and densitometric quantification (lower).

Fig 1G: BRD4’s interaction with JNK is abrogated by phosphorylation. Co-IP of JNK with BRD4 following anisomycin treatment of WT and 3A-BRD4 expressing HCT116 cells.

Also see Figs. S1 and S2

In in vitro kinase assays, both JNK1 and JNK2 robustly phosphorylated BRD4 which was inhibited by the JNK inhibitor SP600125 (Supplementary Fig. 1C). JNK target site(s) on BRD4 mapped to the C-terminus (Fig. 1D). BRD4 did not phosphorylate JNK (Fig. 1D). Thus, JNK directly binds to and phosphorylates BRD4 at its C-terminus.

The sites on BRD4 phosphorylated by JNK mapped to a segment between aa 1141 and 1340 that contains three consensus JNK phosphorylation sequences, (Pro)-x-Ser/Thr-Pro²³, conserved in both mouse and human BRD4 (Supplementary Fig. 1D; Fig. 1E upper panel). We generated Ser/Thr to Alanine point mutations at Ser1153, Thr1222 and Thr1248 in full length mouse BRD4 as either single, double, or triple mutations (Fig. 1E, lower panel). Only mutation of all three potential phosphorylation sites abrogated phosphorylation by JNK. Phosphorylation of the single point mutants was minimally reduced; phosphorylation of the double mutations was markedly reduced. BRD4 autophosphorylation was not affected by these mutations (Fig. 1E left lanes).

To better characterize JNK phosphorylation of BRD4, we developed highly specific antibodies targeted against either phosphorylated Thr1186/1222 or Thr1212/1248 of human/mouse BRD4 (Supplementary Fig. 2A, 2B). Using these antibodies, we confirmed that both JNK1 and JNK2 robustly phosphorylated WT-BRD4, but not the triple mutant (Supplementary Fig 2C). Neither Thr1222 or Thr1248 was phosphorylated by either CK2 or CDK1, the only other Ser/Thr kinases known to affect BRD4’s interaction with chromatin (Supplementary Fig 2C). BRD4 is not phosphorylated by other terminal MAPKs, p38α or ERK1 (Supplementary Fig 2D; ²⁶).

To determine if JNK kinase activation increases BRD4 pThr1222 and pThr1248 in vivo, HCT116 cells were treated with a JNK agonist (anisomycin), heat shock, lipopolysaccharide (LPS), UV or cisplatin (Fig. 1F; Supplementary Fig. 2E). Whole cell extracts were immunoblotted with anti-pJNK antibody and anti-BRD4 pT1186/pT1212 antibodies to monitor JNK activation and BRD4 phosphorylation, respectively. BRD4 pThr1186 and pThr1212 levels increased concomitantly with pJNK levels. Thus, JNK activation by multiple stimuli triggers increased phosphorylation of BRD4.

We next asked whether JNK’s interaction with BRD4 is dependent on its ability to phosphorylate BRD4. HCT116 cells were transfected with Flag-tagged WT or the triple mutant (3A-BRD4) and treated with anisomycin. Flag-BRD4 was immunoprecipitated from whole cell extracts and immunoblotted to measure co-immunoprecipitated JNK. The activation of JNK substantially decreased its interaction with WT-BRD4. In contrast, JNK binding to 3A-BRD4 was maintained independent of JNK activation status (Fig. 1G). These results indicate that JNK phosphorylation of BRD4 mediates its dissociation from JNK.

Phosphorylation of BRD4 by JNK triggers its release from chromatin

To determine whether JNK activation mediated release of BRD4 from chromatin, HCT116 cells were treated with anisomycin to activate JNK. Control cells were pre-treated with the inhibitor SP600125 prior to anisomycin treatment (Fig. 2A). Cell extracts were separated into chromatin-free (CF) and chromatin-bound (CB) protein fractions using differential salt extractions at 150mM and 300mM NaCl respectively²⁷ and immunoblotted for BRD4 and histone H3. Prior to anisomycin treatment, BRD4 was found primarily in the chromatin-bound fraction. JNK activation resulted in a dramatic re-localization of BRD4 to the chromatin-free fraction. Conversely, transfection of HCT116 cells with exogenous dominant negative JNK1 (JNK1 APF) or JNK2 (JNK2 APF) kinase-dead mutants prevented BRD4 release from chromatin in response to anisomycin treatment (Fig. 2B), confirming the role of JNK in releasing BRD4 from chromatin. Importantly, JNK interacts with BRD4 predominantly in the chromatin-bound fraction, unlike Pol II which interacts with both chromatin-bound and free BRD4 (Supplementary Fig. 3A).

Figure 2: JNK phosphorylation of BRD4 releases it from chromatin.

Fig. 2A: BRD4 is released from the chromatin upon JNK activation by anisomycin. Immunoblots of chromatin-free (CF) and chromatin-bound (CB) BRD4 in HCT116 cells following treatment with anisomycin.

Fig 2B: Inhibition of JNK kinase blocks BRD4’s release from chromatin. Immunoblots of chromatin-free (CF) and chromatin-bound (CB) BRD4 in HCT116 cells transfected with WT JNK1 or JNK1 and dominant negative kinase mutants JNK1/JNK2 (APF) individually or in combination, followed by anisomycin treatment.

Fig. 2C: BRD4 is released from the chromatin upon JNK activation by heat shock. Left: Immunoblots of chromatin-free (CF) and chromatin-bound (CB) BRD4 in HCT116 cells grown at 37°C or heat shocked at 42°C for 15 minutes in the presence or absence of JNK inhibitor SP600125. Right: Immunoblots showing pJNK levels under the above conditions.

Fig 2D: JNK preferentially phosphorylates BRD4 bound to mononucleosomes. Anti-BRD4 pT1212 and anti-BRD4 immunoblots of kinase assays with recombinant JNK1 and BRD4 after preincubating BRD4 with or without assembled mononucleosomes (MN) for 10 or 20 min.

Fig 2E: Mutation of BRD4 phosphorylation sites 4 prevents BRD4 release from chromatin. Left: Immunoblots of chromatin-free (CF) and chromatin-bound (CB) BRD4 in HCT116 cells transfected with WT or 3A-BRD4 and subjected to heat shock treatment. Right; Immunoblots showing pJNK levels following heat shock in WT and 3A-BRD4 expressing cells.

Fig. 2F: JNK activation results in global loss of chromatin-bound BRD4. Total BRD4 peaks detected in BRD4 ChIP-seq of control untreated and anisomycin treated DLD1 BRD4-IAA7 cells expressing endogenous BRD4 or exogenous WT or 3A-BRD4 following auxin treatment.

Fig 2G: Loss of JNK-phosphorylated BRD4 from chromatin is widespread. Distribution of BRD4 ChIP-seq peaks across the genomes of cells described in 2F.

Fig 2H: BRD4 ChIP-seq peak distribution profiles across the gene body of representative FDPS, HMGCR and INSIG1 genes in control untreated and anisomycin treated DLD1 BRD4-IAA7 cells expressing exogenous WT or 3A-BRD4 following auxin treatment.

Densitometric quantifications of CF:CB BRD4 ratio are shown below immunoblots; anti-histone H3 immunoblot monitors purity of CB and CF fractions.

Also see Fig. S3

We next tested the effect of physiological activation of JNK by heat shock²⁸. Incubation of HCT116 cells at 42°C for 20 min activated JNK, as measured by pJNK levels, and released BRD4 from chromatin (Fig. 2C). The JNK inhibitor, SP600125, blocked this release. Other BET family proteins, BRD2 and BRD3, lack JNK phosphorylation sites. Accordingly, JNK activation did not release them from chromatin (Supplementary Fig. 3B). These results demonstrate that activation of JNK kinase by either chemical or physiological stimulation mediates the dissociation of BRD4 from chromatin, establishing a direct role for JNK in controlling BRD4 chromatin binding.

The finding that JNK phosphorylates BRD4 and releases it from chromatin led us to hypothesize that JNK preferentially phosphorylates chromatin-bound BRD4. We tested this in vitro by preincubating BRD4 with mononucleosomes for increasing times, followed by the addition of JNK. JNK phosphorylation of BRD4 increased with longer nucleosome preincubation (Fig. 2D). Importantly, JNK phosphorylation of a BRD4 mutant (ΔB1B2) which lacks bromodomains and is unable to bind nucleosomes was unaffected by nucleosome preincubation (SupplementaryFig.3C). Thus, JNK preferentially phosphorylates chromatin-bound BRD4 in vitro.

To extend these findings in vivo, HCT116 cells transfected with Flag tagged WT or 3A-BRD4 were subjected to heat shock (Fig. 2E); cell extracts were fractionated into chromatin-bound and chromatin-free fractions. Strikingly, Flag immunoblots of the chromatin-free and chromatin bound fractions revealed that whereas WT-BRD4 was largely released from chromatin following heat shock, 3A-BRD4 - which cannot be phosphorylated by JNK - remained chromatin bound. (Fig. 2E; left panel). Since pJNK levels in both WT and 3A-BRD4 transfected cells were comparable following heat shock, the difference in chromatin association of BRD4 is not due to differential JNK activation (Fig. 2E; right panel). These results demonstrate that JNK phosphorylation controls BRD4 release from chromatin.

The above results did not distinguish whether the release of BRD4 was genome-wide or restricted to specific genomic regions. Therefore, we next investigated the effects of JNK phosphorylation of BRD4 on its chromatin binding through BRD4 ChIP-Seq. Exogenous WT or mutant 3A-BRD4 were expressed in a DLD1 colon carcinoma cell line (DLD1 BRD4-IAA7) where endogenous BRD4 could be rapidly degraded through an auxin-inducible degron system (Supplementary Fig. 3D, left panel). DLD1 BRD4-IAA7 cells were transfected with WT-BRD4, 3A-BRD4 or empty vector control and treated with anisomycin. One hour before harvesting, cells transfected with WT or 3A-BRD4 were treated with auxin. The expression of only exogenous BRD4 in the transfected cells and endogenous BRD4 in the control cells was confirmed by immunoblotting (Supplementary. Fig. 3D, right panel). Endogenous BRD4 in the control cells allowed us to examine the effects of anisomycin on endogenous BRD4.

JNK-mediated phosphorylation of both endogenous and exogenous WT-BRD4 led to a loss of ~80% of total BRD4 ChIP-seq peaks (Fig. 2F). In contrast, anisomycin did not reduce BRD4 ChIP-seq peaks significantly in 3A-BRD4 expressing cells. An analysis of BRD4 peaks across three representative genes - FDPS, HMGCR, and INSIG1 - illustrates the pattern of BRD4 loss following JNK activation in cells expressing WT-BRD4, but not those expressing 3A-BRD4 (Fig. 2H, Supplementary Fig. 3E). A similar loss of BRD4 peaks is seen in cells expressing endogenous BRD4 (Supplementary Fig. 3F, 3G). Release of BRD4 from chromatin extends across the 5’ UTR, promoter, exons, introns, 3’UTR, enhancers and superenhancers and distal intergenic (Fig. 2G, Supplementary Fig. 3H). Similar results were observed in a metagene analysis (Supplementary Fig 3I). Together, these data show that phosphorylation of BRD4 by JNK activation causes its global release from chromatin.

JNK mediated release of BRD4 from chromatin terminates nucleosome dissociation by BRD4

We next investigated how JNK phosphorylation of BRD4 affects its interaction with nucleosomes. Flag tagged WT or 3A-BRD4 were incubated in the presence or absence of JNK, and then used to pull down in vitro reconstituted mononucleosomes. While unphosphorylated WT-BRD4 bound to nucleosomes, phosphorylated BRD4 did not. 3A-BRD4 efficiently pulled down nucleosomes under either condition. Therefore, phosphorylated WT-BRD4 cannot stably associate with nucleosomes (Fig. 3A).

Figure 3: JNK-mediated BRD4 release from chromatin disrupts its nucleosome clearance function.

Fig 3A: BRD4 binding to mononucleosomes is abrogated upon phosphorylation by JNK. Anti-histone H3 immunoblot of assembled mononucleosomes pulled down by recombinant WT or 3A-BRD4 that was either unphosphorylated or pre-phosphorylated (*) by JNK and immobilized on Flag-beads.

Fig 3B: JNK phosphorylation of BRD4 inhibits H3K122 acetylation. Anti-histone H3K122ac immunoblot of assembled mononucleosomes subjected to an in vitro HAT assay with recombinant WT or 3A-BRD4 that was either unphosphorylated or pre-phosphorylated (*) by JNK.

Fig 3C: H3K122 acetylation is reduced in vivo upon JNK activation. Immunoblots of whole cell extracts (WCEs) of HCT116 cells that were untreated (control) or treated with either DMSO (Mock) or Anisomycin.

Fig 3D: In vivo H3K122 acetylation by BRD4 is regulated by JNK. Immunoblots of WCEs of HCT116 cells that were transfected with WT or 3A-BRD4 and treated with or without anisomycin.

Fig. 3E: In vivo H3K122 acetylation is regulated by JNK kinase activity. Immunoblots of WCE’s of HCT116 cells that were transfected with WT JNK1, JNK2 or their respective dominant negative mutants (JNK APF), either individually or together and treated with or without anisomycin. Densitometric quantification of H3K122ac levels is shown below.

Fig 3F: Nucleosome clearance activity by BRD4 is controlled by JNK phosphorylation. Autoradiograph of an in vitro nucleosome clearance assay showing dissociation of assembled mononucleosomes by unphosphorylated or JNK pre-phosphorylated (*) WT or 3A-BRD4 upon being subjected to a HAT assay in the presence or absence of AcCoA.

Also see Fig. S4

BRD4 mediates chromatin decompaction through its HAT activity by acetylating nucleosomal histone H3 at Lysine122⁵. Accordingly, phosphorylated WT-BRD4 was unable to acetylate nucleosomes at H3K122 (Fig. 3B). However, phosphorylated WT-BRD4 bound and acetylated free histone H3 in an in vitro HAT assay (Supplementary Fig. 4A and 4B), demonstrating that JNK phosphorylation of BRD4 does not inhibit its HAT catalytic activity, but rather prevents it from binding nucleosomes. JNK activation in vivo in HCT116 cells led to substantially reduced H3K122ac levels (Fig. 3C). Furthermore, expression of exogenous WT or 3A-BRD4 in HCT116 cells led to elevated H3K122ac levels which were blocked by JNK activation in cells expressing WT-BRD4, but not 3A-BRD4 (Fig. 3D). Conversely, the reduction of in vivo H3K122ac levels upon JNK activation was blocked by the exogenous expression of JNK1/JNK2 APF dominant negative mutants (Fig. 3E). Thus, JNK phosphorylation of BRD4 prevents its interaction with and subsequent acetylation of nucleosomes in vitro and in vivo but does not inhibit BRD4 HAT catalytic activity.

To confirm that JNK phosphorylation of BRD4 abrogated its ability to dissociate nucleosomes, an in vitro nucleosome clearance assay was done with WT or 3A-BRD4 that had been subjected to a kinase assay with or without JNK (Fig. 3F). Unphosphorylated, but not phosphorylated, WT-BRD4 efficiently dissociated nucleosomes and released free DNA. In contrast, 3A-BRD4 dissociated nucleosomes even after a kinase assay with JNK. Therefore, JNK phosphorylation abrogated BRD4 HAT-mediated nucleosome clearance activity.

Release of BRD4 from chromatin activates its kinase activity.

Since BRD4 has intrinsic kinase activity⁶, we asked if JNK-mediated phosphorylation affects BRD4’s ability to phosphorylate its substrates. In in vitro kinase assays, phosphorylation of BRD4 by JNK did not affect BRD4’s phosphorylation of Pol II CTD, CDK9/PTEFb or MYC. (JNK does not phosphorylate these proteins (Supplementary Fig.5A)). Surprisingly, following anisomycin treatment of HCT116 cells, JNK-mediated phosphorylation of BRD4 led to increased phosphorylation of the BRD4-specific targets CDK9 Thr186, Pol II CTD Ser2 and MYC Thr58 (Fig. 4A). Inhibition of JNK kinase activity with a JNK specific peptide inhibitor, D-JNK-1, blocked both phosphorylation of BRD4 and enhanced phosphorylation of BRD4 substrates as well as the decrease in H3K122ac levels (Fig. 4A). Further, exogenous expression of JNK dominant negative kinase mutants, JNK1 APF and JNK2 APF, similarly blocked increased BRD4 kinase activity upon JNK activation, confirming the role of active JNK in activating BRD4 kinase (Fig. 4B).

Figure 4: JNK-mediated BRD4 release from chromatin activates BRD4 kinase.

Fig. 4A: JNK activation induces phosphorylation of BRD4 kinase substrates. Immunoblots of WCE’s of HCT116 cells that were treated with DMSO (mock), anisomycin or anisomycin with JNK peptide inhibitor D-JNK1.

Fig 4B: Blocking JNK activity inhibits induction of BRD4 kinase. Immunoblots of WCE’s of HCT116 cells that were transfected with Flag-tagged WT JNK1, JNK2 or respective dominant negative mutants (JNK APF), either individually or together and treated with or without anisomycin.

Fig 4C: JNK phosphorylation of BRD4 is necessary for induction of BRD4 kinase. Immunoblots of WCE’s of HCT116 cells transfected with WT or 3A-BRD4 and subjected to heat shock.

Fig 4D: BRD4 phosphorylation regulates MYC stability. Immunofluorescence images showing MYC levels in HCT116 cells transfected with either WT or 3A-BRD4 or empty vector (control) and subjected to heat shock. (Scale bars, 25 μM)

Also see Fig. S5

Heat shock of HCT116 cells expressing WT or 3A-BRD4 substantially increased CDK9 pThr186, Pol II CTD pSer2 and MYC pThr58 levels in WT-BRD4, but not 3A-BRD4, expressing cells. (Fig. 4C). Phosphorylation of MYC at Thr58 by BRD4 leads to its degradation²⁶. Accordingly, JNK activation by heat shock in DLD1 BRD4-IAA7 cells where endogenous BRD4 was deleted and replaced with WT or 3A-BRD4 led to a marked loss of MYC in WT-BRD4, but not 3A-BRD4, expressing cells, demonstrating that JNK-mediated stimulation of BRD4 kinase activity depends on the presence of JNK phosphorylation sites on BRD4 (Fig. 4D). The increased phosphorylation of BRD4 substrates following JNK activation suggested that the release of BRD4 from chromatin enhances its interaction with those substrates. Indeed, anisomycin enhanced the interaction of BRD4 with Pol II CTD, CDK9 and MYC from cell lysates of HCT116 cells (Fig. 5B, Sup. Fig. 5B). JNK-phosphorylated BRD4 phosphorylated CDK9 at Thr186, which activates CDK9 kinase (Figs. 4A-C)^7,29. While both BRD4 and PTEFb phosphorylate Pol II CTD Ser2, PTEFb also phosphorylates CTD Ser5 and Ser7^30,31. Accordingly, Pol II CTD pSer2, pSer5 and pSer7 levels increased in anisomycin treated HCT116 cells (Supplementary Fig. 5C). Therefore, JNK-mediated release of BRD4 from chromatin switches its function, reactivating its kinase activity and increasing interaction with, and phosphorylation of, its transcriptional partners.

Figure 5: JNK phosphorylation toggles BRD4 enzymatic activities and is reversed by PP4 phosphatase.

Fig. 5A: BRD4 kinase and HAT activities are cross regulated by its substrates. Top: Autoradiograph of an in vitro kinase assay with BRD4 and Pol II CTD in the presence or absence of assembled mononucleosomes. Bottom: Immunoblots of an in vitro HAT assay with BRD4 and histone H3 in the presence or absence of Pol II CTD.

Fig 5B: JNK activation enhances the interaction between BRD4 and its kinase substrates. Immunoblots showing co-immunoprecipitated total and T1212 phosphorylated BRD4 from HCT116 cells treated with or without anisomycin. Top: BRD4 co-immunoprecipitated with Pol II CTD. Bottom: BRD4 co-immunoprecipitated with CDK9.

Fig 5C: JNK-mediated phosphorylation of BRD4 is transient. Immunoblots of WCE’s of HCT116 cells grown under normal conditions, subjected to heat shock or heat shocked and then rescued for 20 min.

Fig 5D: Inhibition of phosphatases enhances phosphorylated BRD4 levels. Immunoblots of WCE’s of HCT116 cells that were treated with or without anisomycin alone or anisomycin with phosphatase inhibitor, nodularin.

Fig 5E: Phosphatase PP4 dephosphorylates JNK-phosphorylated BRD4. Immunoblots of WCE’s of HCT116 cells that were transfected with either control, PP2Ac or PP4c SiRNA and treated with or without anisomycin.

Fig 5F: BRD4’s interaction with Pol II CTD is modulated by PP4. Immunoblots showing total and pT1212 BRD4 co-immunoprecipitated with Pol II CTD from HCT116 cells transfected with either control or PP4c SiRNA and treated with anisomycin.

BRD4 enzymatic activities are cross regulated.

The release of BRD4 from chromatin ceases its HAT function while activating its kinase activity, suggesting that these activities are negatively cross regulated. Indeed, in an in vitro BRD4 kinase assay with Pol II CTD, increasing amounts of mono-nucleosomes substantially inhibited both BRD4 trans-phosphorylation of Pol II CTD and autophosphorylation (Fig. 5A; upper panel). Conversely, in a BRD4 HAT assay with histone H3 as substrate, the addition of Pol II CTD inhibited BRD4 HAT activity (Fig. 5A; lower panel). Thus, the enzymatic activities of BRD4 are negatively cross-regulated by its substrates.

PP4 phosphatase dephosphorylates BRD4, promoting its interaction with RNA Pol II

Surprisingly, unlike CDK9 and MYC, Pol II CTD only interacted with the non-phosphorylated form of BRD4 following anisomycin treatment (Fig. 5B, Supplementary Fig. 5B). Therefore, we investigated the fate of phosphorylated BRD4 following heat shock and recovery. pThr1186 and pThr1212 BRD4 levels significantly increased upon heat shock, concomitant with JNK activation and returned to baseline levels within 20 min of recovery (Figs. 1F, 5C). Total BRD4 levels remained unchanged indicating that phosphorylation of BRD4 at Thr1186 and Thr1212 does not lead to its degradation.

The rapid loss of phosphorylated BRD4 (Fig. 5C) led us to identify candidate phosphatases. Exogenous Flag tagged WT or 3A-BRD4 was expressed in the DLD BRD4-IAA7 cells, depleted of endogenous BRD4 with auxin, and subjected to heat shock, followed by fractionation of chromatin-free or bound BRD4. Mass spec analysis of immunoprecipitated BRD4 fractions (Supplementary Fig. 5D) identified phosphatases PP4 and PP2A co-immunoprecipitating with WT-BRD4 in the chromatin-free fraction, but not the chromatin-bound fraction. Neither phosphatase co-immunoprecipitated with 3A-BRD4. The association of the phosphatases with chromatin-free, but not chromatin-bound, BRD4 was confirmed in HCT116 cells (Supplementary Fig. 5E).

Interestingly, the number of proteins interacting with BRD4 in the chromatin-free fractions following JNK activation increased significantly, with a corresponding decrease in the chromatin-bound protein fractions (Supplementary Figs. 5D and 5E). In contrast, no significant change was observed with 3A-BRD4 (Supplementary Fig 5D). Therefore, phosphorylation of BRD4 by JNK results in extensive changes in the BRD4 protein interactome.

The de-phosphorylation of BRD4 by PP4 and PP2A was assessed in vitro and in vivo. In an in vitro assay, BRD4 pre-phosphorylated by JNK was dephosphorylated by both PP2A and PP4 (Supplementary Fig. 5F). Dephosphorylation by either phosphatase was blocked by the inhibitor Nodularin. In vivo, Nodularin enhanced phosphorylation of BRD4 following anisomycin treatment of HCT116 cells (Fig. 5D). To directly validate the roles of PP4 and PP2A in dephosphorylating BRD4, catalytic subunits of PP4 or PP2A were knocked down with siRNA (Fig. 5E). Subsequent JNK activation substantially increased BRD4 phosphorylation in cells depleted of PP4. PP2A depletion had only a modest effect. (Fig. 5E). Thus, in vivo, PP4 is the primary BRD4 phosphatase. Strikingly, PP4 phosphatase depletion substantially reduced the interaction of BRD4 with Pol II, confirming that PP4 de-phosphorylation of BRD4 is a prerequisite for its interaction with Pol II (Fig. 5F).

JNK-mediated BRD4 release from chromatin activates transcription of inflammatory and immune genes.

The increase in phosphorylated Pol II CTD levels following BRD4 release from chromatin (Fig. 4A-C, Supplementary Fig. 5C) predicted a concomitant increase in transcription. To examine this, HCT116 cells were transfected with either WT, 3A-BRD4 or empty vector control, treated with anisomycin to activate JNK and subjected to RNA-seq analysis of total RNA (Fig. 6A). Anisomycin significantly increased transcription in cells transfected with WT-BRD4 (Fig. 6A left panel, Supplementary Fig. 6A). In contrast, anisomycin only minimally affected transcription in 3A-BRD4-transfected cells (Fig. 6A middle panel, Supplementary Fig. 6A). A direct comparison of the transcriptomes of cells expressing 3A-BRD4 or WT-BRD4 following anisomycin confirmed that mutation of the JNK phosphorylation sites in 3A-BRD4 eliminated transcriptional activation induced by WT-BRD4 (Fig. 6A; right panel). Thus, activated transcription for a significant subset of genes depends on JNK-mediated phosphorylation and release of BRD4 from chromatin. In the absence of JNK activation, over-expression of either WT-BRD4 or 3A-BRD4 minimally affected transcription, indicating that over-expression does not perturb the homeostatic RNA profile (Supplementary Fig. 6A).

Figure 6: JNK-mediated BRD4 release from chromatin activates transcription.

Fig. 6A: JNK activation enhances expression of BRD4 regulated genes. Volcano plots showing differential gene expression observed in RNA-seq analysis of WT and 3A-BRD4 expressing HCT116 cells after anisomycin treatment.

Fig 6B: Inflammatory and immune response pathways are enriched among the BRD4 regulated genes induced by JNK activation. GO analysis of genes induced in anisomycin treated WT-BRD4 expressing cells relative to control HCT116 cells.

Fig 6C: Induction of key inflammatory and immune response genes depend on JNK phosphorylation of BRD4. Quantitative RT-PCR of cDNA from anisomycin treated WT and 3A-BRD4 expressing cells relative to control cells. Error bars, s.e.m. (n=3 independent experiments; *P <0.001 by two tailed Student’s t tests)

Fig 6D: JNK activation leads to increased BRD4-RNA Pol II interaction at BRD4 regulated inflammatory and immune response genes. Sequential-ChIP assays showing Pol II and Pol II- bound BRD4 at the promoter and gene body regions of CCL20, CXCL1, BIRC3 and control MYC genes. Error bars, s.e.m. (n=4 technical replicates from 2 independent experiments; *P <0.05 by two tailed Student’s t tests)

Also see Fig. S6

Strikingly, genes induced by JNK-activation of BRD4 were enriched for biological pathways involved in inflammatory and immune responses (Fig. 6B). JNK activation of WT-BRD4, but not 3A-BRD4, significantly increased expression of CCL20, CXCL1, CXCL2, TNFAIP3, NFκBIA and BIRC3, as assessed by quantitative RT-PCR (Fig. 6C). KEGG pathway analysis similarly mapped infection and inflammatory disease pathways (Supplementary Fig. 6C).

We considered two possibilities for the increased transcription: release of phosphorylated BRD4 either 1) de-repressed transcription of these genes or 2) directly activated their transcription. To distinguish these possibilities, we performed sequential-ChIP (Seq-ChIP) analysis and monitored the extent of BRD4’s interaction with chromatin-bound Pol II (Fig. 6D). Chromatin from anisomycin-treated or control HCT116 cells transfected with WT-BRD4 was first immunoprecipitated with anti-Pol II antibody (ChIP1). A fraction of the ChIP1 immunoprecipitate was used to assess Pol II levels at the promoter and gene body of BRD4 regulated genes. To assess the levels of BRD4 bound to Pol II at the same gene loci, the remaining ChIP1 immunoprecipitate was used in a second ChIP with anti-BRD4 antibody (ChIP2). The amount of DNA bound by both BRD4 and Pol II was determined by quantitative RT-PCR at the promoters and gene bodies of CCL20, CXCL1 and BIRC3, genes that are highly induced by JNK-mediated release of BRD4 from chromatin. Since transcription at the Myc gene was unaffected by activation of BRD4, it served as a control. Remarkably, whereas Pol II binding at the promoters did not increase following JNK activation, BRD4 binding to Pol II significantly increased; no significant change in BRD4 bound to Pol II was observed at the Myc locus. Binding of both Pol II and BRD4 within the gene body increased following JNK activation, concomitant with increased transcription of CCL20, CXCL1 and BIRC3 genes (Fig. 6D). Thus, JNK-mediated BRD4 release from chromatin increases transcription at specific genes through its enhanced interaction with Pol II.

BET bromodomain inhibitors such as JQ1 also release BRD4 from chromatin, mostly repressing transcription³². A comparison of the transcripts affected by JNK-mediated BRD4 release and JQ1 following either anisomycin or JQ1 treatment of HCT116 cells revealed that only 4.7% of transcripts affected by JQ1 treatment overlapped with those affected by anisomycin (Supplementary Fig. 6D). Importantly, whereas anisomycin increased BRD4 binding to both Pol II and CDK9 (Fig. 5B), JQ1 decreased it (Supplementary Fig. 6E). Further, BRD4 returns to chromatin when JNK activation is ceased, unlike JQ1 treatment (Supplementary Fig. 6F). Thus, release of BRD4 from chromatin mediated by JNK or JQ1 has distinct biological consequences.

BRD4 also regulates alternative splicing independent of its chromatin and transcription regulatory functions by interacting with the splicing machinery and cohesin^8,9,33. While anisomycin significantly increased alternative splicing in WT-BRD4 expressing cells, no increase was observed in the 3A-BRD4 expressing cells (Table 1). All classes of alternative splicing were affected proportionately. These effects on splicing are consistent with JNK-mediated BRD4 activation of transcription.

Table 1:

JNK phosphorylation mediated BRD4 release from chromatin regulates alternative splicing. Number of differential splice events/genes in HCT116 cells expressing WT or 3A-BRD4 relative to control cells upon anisomycin treatment [false discovery rate (FDR) < 0.05].

	WT-BRD4 vs. Control		3A-BRD4 vs. Control		WT-BRD4 vs. 3A-BRD4		Control +Anis. vs. Control -Anis.
Splicing type	Events	Genes	Events	Genes	Events	Genes	Events	Genes
SE	893	753	185	178	928	789	313	272
MXE	75	67	21	20	78	67	27	25
A5SS	84	73	28	25	73	66	64	55
A3SS	77	67	32	30	59	57	37	28
RI	28	25	6	6	23	21	13	11
Total	1157	916	272	247	1161	949	454	371

BRD4 phosphorylation and chromatin release correlate with thymocyte activation and EMT

The induction of immune response genes upon JNK-mediated BRD4 activation (Figs. 6A-C) suggested a role in immune responses. Therefore, we examined whether JNK-mediated thymocyte activation correlated with BRD4 phosphorylation and release from chromatin. Primary thymocytes from B6 mice were stimulated with PMA/Ionomycin to activate the MAPK/JNK pathway³⁴; activation was confirmed by the increased expression of CD8 and CD69, markers of thymocyte activation. (Fig. 7A; left panel). Following treatment, both pJNK and BRD4 pThr1186 and pThr1212 increased. Importantly, increased BRD4 was detected in the chromatin-free fraction of the activated thymocytes (Fig. 7). Thus, MAPK/JNK-mediated thymocyte activation leads to BRD4 phosphorylation and its release from chromatin.

Figure 7: BRD4 phosphorylation and chromatin release are correlated with thymocyte activation and EMT.

Fig. 7A: Thymocyte activation correlates with JNK phosphorylation of BRD4. Left panels: Flow cytometry profiles of thymocytes activated by 0.3ng PMA/0.3μg Ionomycin or by 10ng PMA/3.75μg Ionomycin. FACS analysis of CD4/CD8 (Upper) and CD69 expression (Lower). Right panel: Immunoblots of WCE’s from unstimulated and stimulated thymocytes. Densitometric quantification of relative BRD4 phosphorylation levels is shown below.

Fig. 7B: Thymocyte activation is correlated with JNK mediated release of BRD4 from chromatin. Immunoblot of chromatin-free (CF) and chromatin-bound (CB) BRD4 in thymocytes unstimulated or stimulated as described above. Densitometric quantification of CF:CB BRD4 ratio is shown below. Anti-histone H3 immunoblot monitors purity of separation. Immunoblots of WCE’s from PC3 cells at day 0 and day 5 of treatment with or without EMT-inducing media supplement.

Fig. 7D: EMT induction correlates with JNK-mediated release of BRD4 from chromatin. Immunoblot of chromatin-free (CF) and chromatin-bound (CB) BRD4 in PC3 cells after five days of treatment with or without (control) EMT-inducing media.

Fig. 7E: EMT induction and BRD4 phosphorylation are both dependent on JNK activity. Immunoblots of WCE’s from PC3 cells that were treated, or not, with EMT-inducing media alone or in combination with JNK peptide inhibitor D-JNK-1.

Fig. 7F: EMT induction and expression of EMT regulators is dependent on BRD4 phosphorylation. Immunoblots of WCE’s from PC3 cells that were treated, or not, with EMT-inducing media and transfected with WT-BRD4, 3A-BRD4 or empty vector control on day 3 of treatment.

BRD4 plays a critical role in the epithelial to mesenchymal transition (EMT) during cancer metastasis by inducing required transcription factors³⁵. Since JNK activation is required during EMT induction³⁶, we asked whether induction of EMT correlates with JNK-mediated BRD4 phosphorylation and release from chromatin. PC-3 prostate cancer cells were cultured in an EMT inducing medium (Fig. 7C). After 5 days, expression of the mesenchymal cell marker Vimentin was markedly elevated while expression of the epithelial cell marker E-cadherin was abrogated, confirming the induction of EMT (Fig. 7C). Remarkably, the levels of both BRD4 pThr1186 and pThr1212 were dramatically increased in parallel with EMT induction (Fig. 7C). Further, the mesenchymal-like cells displayed a substantial decrease in chromatin-bound BRD4 with a corresponding increase in chromatin-free BRD4 (Fig. 7D). Importantly, the JNK inhibitor D-JNK-1 blocked JNK activation, EMT induction and BRD4 phosphorylation (Fig. 7E).

Since EMT requires the transcription factors SNAIL, TWIST and SLUG, we tested the requirement for BRD4 phosphorylation by JNK in their expression. PC-3 cells were transfected with WT or 3A-BRD4 on the third day after EMT induction and the levels of BRD4 pThr1212, SNAIL, TWIST and SLUG were monitored (Fig. 7F). Cells expressing WT-BRD4, but not 3A-BRD4, had markedly elevated levels of SNAIL, TWIST and SLUG. (Fig. 7F). Strikingly, the level of BRD4 pThr1212 correlated with expression of EMT transcription factors and EMT induction. Thus, JNK phosphorylation of BRD4 and its release from chromatin mediates the induction of EMT.

DISCUSSION

BRD4 is a chromatin remodeler that becomes a transcriptional activator in response to external stimuli. Its functional toggling is governed by the JNK/MAPK pathway, which is activated by numerous stimuli that include both stress and homeostatic biological processes (e.g. cell cycle, innate immune responses). In response to activation, JNK phosphorylates BRD4 at Thr1186 and Thr1212, leading to its release from chromatin. De-phosphorylation of BRD4 by PP4 phosphatase enhances its binding to, and phosphorylation of, its transcription factor substrates, thereby activating transcription. JNK-mediated switching of BRD4 to a transcriptional activator induces genes in inflammatory and immune response pathways, and those required for epithelial to mesenchymal transition (EMT). Release of BRD4 from chromatin by JNK phosphorylation is a distinct mechanism to regulate the toggling of its functions.

Although BRD4 binds chromatin through its two N-terminal bromodomains, phosphorylation at its C-terminal end by JNK dissociates BRD4 from chromatin. The JNK target sites on BRD4, Thr1186 and Thr1212, are in a highly conserved region of the C-terminus and are not known to be phosphorylated by any other kinase. Determining the biophysical mechanism by which JNK-mediated phosphorylation of BRD4 leads to release of its N-terminal bromodomains from chromatin is complicated by lack of a well-defined 3D structure of BRD4, which has an intrinsically disordered C-terminus. We recently reported that BRD4 exists as a dimer³⁷ which is predicted to be anti-parallel. Such an anti-parallel dimer would bring the C-terminal phosphorylated domain of BRD4 in proximity with the N-terminal bromodomains, perhaps facilitating dissociation of BRD4 from chromatin.

JNK-mediated BRD4 release from chromatin occurs in response to a wide range of stimuli (e.g. chemical, biotic, abiotic) and is not restricted to a specific stress or biological processes. Critically, JNK-mediated release of BRD4 from chromatin is genome-wide and uniform across genic, intergenic, and distal enhancer regions, but activates only a subset of genes involved in inflammation. Thus, JNK phosphorylation has a global effect on BRD4 binding to chromatin, but a localized effect on its transcriptional activation. The regulatory mechanisms that localize BRD4 to inflammatory genes remain to be determined.

BRD4 has intrinsic HAT activity which regulates local chromatin architecture. Our current findings demonstrate that JNK phosphorylation-mediated release of BRD4 from chromatin abrogates its chromatin remodeling function by disrupting its nucleosome clearance activity, identifying a mechanism that transiently regulates its HAT activity. We speculate that JNK-mediated disruption in BRD4 HAT activity functions as a mechanism to briefly halt expression of genes that depend on BRD4 HAT activity such as Myc, Fos and Aurora B ⁵.

Dephosphorylation by PP4 increases BRD4’s interaction with and phosphorylation of its kinase substrates, leading to increased transcription of its target genes by: 1) phosphorylating CDK9 at Thr186, essential for recruiting and activating PTEFb/CDK9 kinase²⁹; (2) increasing phosphorylation of the RNA Pol II CTD, necessary for transcription start, pause release and elongation^30,31 3) increasing phosphorylation of MYC Thr58, leading to MYC degradation and transcriptional pause release^26,38.

These findings address several outstanding questions about the role of BRD4 in transcription: How does BRD4 recruit and activate CDK9/PTEFb from the nucleoplasm? How is BRD4 phosphorylation of Pol II CTD activated? What role does BRD4 phosphorylation mediated MYC degradation play in transcription? We answer these questions by showing that JNK-mediated release of BRD4 from chromatin enhances its interaction with its kinase substrates, their recruitment to promoters and increased transcription.

JNK activates ~1000 BRD4-dependent genes, of which the most highly induced are those involved in inflammatory, immune and stress response pathways, demonstrating a mechanistic link between BRD4 and JNK in regulating these pathways^25,39,40. Among those pathways are those leading to thymocyte activation and EMT. Transcription factors critical for EMT are activated following JNK phosphorylation of BRD4. Interestingly, both JNK activation and BRD4 regulation of EMT during cancer progression and metastasis have been reported^35,36.

Although both JNK and BET inhibitor JQ1 release BRD4 from chromatin, JNK-mediated release enhances BRD4’s association with RNA Pol II, whereas JQ1 does not. Thus, targeted phosphorylation of BRD4 by JNK is physiological and increases transcription of a subset of genes; JQ1 globally decreases transcription⁴¹.

The switching of BRD4 enzymatic activities from a HAT to a kinase and the resulting functional transition from a chromatin remodeler to a transcriptional activator raises the question of why such a switch is necessary. Our current results demonstrate that BRD4 enzymatic activities are cross-regulated by its substrates – BRD4 kinase activity is suppressed by nucleosomes while its HAT activity is suppressed by Pol II CTD and by MYC²⁶. This cross-regulation provides a mechanism that ensures that each of BRD4’s pleiotropic activities is only functional at the right time and place.

We propose a model (Fig. 8) wherein BRD4 is homeostatically chromatin bound with an HAT active that locally remodels chromatin, but with a suppressed kinase activity. In response to stimuli such as stress, JNK is activated to phosphorylate BRD4, transiently releasing it from chromatin and liberating its kinase activity while suppressing its HAT activity. PP4 phosphatase dephosphorylates BRD4, which then activates transcription of stress/inflammatory response and EMT inducing genes. A portion of dephosphorylated BRD4 not involved in transcriptional activation returns to chromatin to resume its function as a chromatin remodeler.

Figure 8:

Model of BRD4-JNK interaction and the switching of BRD4 functions. BRD4 primarily functions as a chromatin regulator by acetylating H3K122 and dissociating nucleosomes. Upon activation, JNK phosphorylates BRD4 releasing it from chromatin and activating its kinase. Chromatin-free BRD4 is then dephosphorylated by PP4, enhancing its interaction with and phosphorylation of, Pol II CTD, PTEFb and MYC, thereby activating transcription at specific genes. A portion of dephosphorylated BRD4 returns to chromatin to renew its chromatin regulatory function.

In summary, our findings establish a distinct mechanism for the switching of BRD4 functions from a chromatin remodeler to a transcriptional activator. Future studies on therapeutic targeting of BRD4 need to consider its pleiotropic functions and their cross-regulation.

Limitations of the study.

Overexpression of BRD4 potentially amplified JNK-mediated expression of target genes. Although preliminary experiments used SP600125, which inhibits kinases other than JNK⁴², D-JNK1 or kinase-dead mutants were used in subsequent experiments. The mechanisms that recruit BRD4 released from chromatin to target genes or determine the relative proportion of BRD4 returning to chromatin, remain to be identified.

STAR METHODS

Experimental model and study participant details

Cell Lines and Culture

HCT116, DLD1 Colorectal adenocarcinoma and PC3 prostatic adenocarcinoma cells were acquired from ATCC. DLD1-BRD4-IAA7 cells were a gift from Dr. Ali Shilatifard, Northwestern University, Chicago, IL. and are as described previously ⁴³. HCT116, DLD1 and DLD1-BRD4-IAA cells were grown in DMEM media with 10% FBS at 37°C and 7.5% CO2. PC3 cells were grown in F-12K 10% FBS at 37°C and 5% CO2. Cell lines were tested for mycoplasma contamination. Drosophila SF9 cells were grown at 27°C in TNM-FH insect medium (BD Biosciences Pharmingen).

Method details

Plasmid constructs.

Murine 6X His-BRD4-Flag WT and ΔN, ΔC, ΔB1, ΔB2 and ΔB1B2 mutants are as described previously ⁶. The BRD4 F1, F2, F3 and F4 fragments are also as described previously ²⁶. The murine BRD4-Flag WT, S1153A, T1222A, T1248A, S1153A+T1222A, T1222A+T1248A and the 3A-BRD4 mutants was generated as point mutants using the TagMaster^® Site-Directed Mutagenesis Kit (GM Biosciences) with the BRD4 WT construct in the pFASTbac1 insect vector as template. The human WT and 3A-BRD4 mutant were similarly generated in the mammalian pCDNA5 vector. The pDNA3 plasmids bearing Flag-WT JNK1, JNK2, and Flag-JNK1 APF, JNK2 APF dominant negative mutants were a gift from the Roger Davis lab (Addgene plasmids 13798, 13755, 13846 and 13761 respectively).

Recombinant proteins.

Flag-tagged BRD4 and BRD4 mutants were purified from insect cells as described earlier ^6,26. Flag-tagged BRD4 F1-F4 fragments were expressed and purified from E.coli cells (BL21 strain, Novagen) by inducing expression using 0.7mM IPTG overnight at 18°C and purified using FLAG beads (Sigma). GST-Pol II CTD was purified similarly and purified using GST Sepharose beads (Sigma). All purified proteins were concentrated using microcon size exclusion columns (Millipore), recovered in HKEG buffer (20mM Hepes, pH 7.9, 100mM KCl, 0.2mM EDTA, 20% vol/vol Glycerol) and stored frozen at −80°C. Highly purified (> 99%) human histones H3 and H4 substrates were obtained from New England Biolabs as recombinant proteins expressed in E.coli; their purity was confirmed by the company through Mass Spectrometry Analysis (ESI-TOF MS) and peptide sequencing. Purified recombinant human His-MYC (>95% purity), expressed in E.coli, was purchased from Raybiotech. Purified recombinant kinase active human GST-tagged JNK1 and JNK2 was purchased from SignalChem. Purified recombinant kinase active human His-tagged JNK1 was purchased from Thermo Fisher Scientific. Purified recombinant active human protein Phosphatase 2Ac was purchased from Cayman chemical and Purified recombinant active human protein Phosphatase 4 (PPP4C) was purchased from MyBioSource, Inc.

Development of custom phospho-BRD4 antibody

Custom phospho-specific BRD4 antibodies were designed and purified by Vivitude (Gardner, MA, USA). Briefly, non-phosphorylated and phosphorylated peptides targeting Thr1186 BRD4 (Ac-CQKQEPKTPVAPK-amide, Ac-CQKQEPK(pT)PVAPK-amide) and Thr1212 BRD4 (Ac-CVQKHPTTPS-amide, Ac-CVQKHPT(pT)PS-amide) were synthesized. Three immunizations of two rabbits (New Zealand White - SPF) was done using each peptide, with preimmune serum collected as controls. Anti-serum was collected from each rabbit post-immunization and tested through ELISA/immunoblots. Anti-serum showing the highest titers/affinity was double affinity purified to yield phospho-specific antibodies. Affinity purified antibodies were further tested for specificity and affinity as detailed in supplementary Fig. 2.

In situ Proximity Ligation assays

For PLA, approximately 10⁴ HCT116 cells were grown overnight in μ-Slide Angiogenesis (I-bidi). PLA was conducted using the Duolink^® In Situ PLA^® Kit (Sigma) according to the manufacturer’s protocol. The primary antibodies used were as follows: anti-BRD4 rabbit monoclonal antibody (Bethyl; [BL-149–2H5]) (1:100 dilution), anti-phospho JNK mouse monoclonal antibody (G-7, Santa cruz biotechnology) (1:100 dilution), and anti-Nucleolin (sc-8031, Santa cruz biotechnology) (1:100 dilution). Cells were observed with Zeiss LSM880 Multi-Photon Confocal Microscope.

Heat shock, UV and LPS treatment of cells

HCT116 cells were subjected to heat shock by incubating them at 42°C for 15 minutes with pre-warmed media and then immediately harvesting them. Cells were subjected to UV treatment in a Stratalinker ultraviolet (UV) cross-linker (Stratagene) by exposing them to 80 J/m² of UV without culture media followed by recovery in media for 30min before harvesting the cells. Cells were treated with 1μg/ml Lipopolysaccharide (Thermo Fisher Scientific) for 30 minutes.

Non-chromatin and chromatin-bound protein fractionation

Treated and untreated cells were subjected to the REAP protocol as described by Suzuki and colleagues ⁴⁴ to collect the nuclear fraction. The nuclear pellet was resuspended in approx. 3x volume of ice-cold buffer A (10mM HEPES pH 7.9, 1.5mM MgCl2, 10mM KCl, 1mM DTT, 0.5mM PMSF and 1x protease inhibitor cocktail) with 0.5% NP-40 and 75mM NaCl, incubated on ice for 10min, followed by centrifugation at 5000g, 4°C for 5 min. The above steps were repeated twice, and the supernatants pooled to be used as the non-chromatin bound protein fraction. The remaining pellet was resuspended in approx. 4x volume high salt buffer (10mM HEPES pH7.9, 20% glycerol, 350mM NaCl, 1.5mM MgCl2, 0.4mM EDTA, 0.5% NP40, 1mM DTT, 0.5mM PMSF and 1x protease inhibitor cocktail) and rotated at 4°C for 30 min. The samples were then centrifuged at 12000g at 40C for 10 min to collect the supernatant to be used as the chromatin-bound protein fraction. While making chromatin-bound and nucleoplasmic BRD4 and associated proteins fractions for mass spec analysis, the chromatin bound fraction was made by incubating the chromatin pellet in buffer A with 500 μM JQ1 overnight and 25 U/mL Benzonase^® endonuclease (Sigma) for 1hr instead of the high salt buffer. Similarly, extracts made for immunoprecipitating Pol II and CDK9 with BRD4 were made by incubating the chromatin pellet in buffer A with 25 U/mL Benzonase^® endonuclease (Sigma) for 1hr and 100 U of MNase/ml for 15 minutes instead of the high salt buffer.

Transient transfections

Transient transfections of the mammalian expression plasmid constructs containing BRD4, 3A-BRD4, JNK1, JNK2, JNK1APF and JNK2 APF were done using Lipofectamine (Invitrogen) and harvested 18 hrs post transfection except were mentioned otherwise. Under these conditions, cell viability and growth were not affected by the transfections. When testing for histone acetylation in vivo, cells were treated with 5 mM Sodium butyrate during transfection to inhibit HDAC activity. Transfected cells were treated with 30 μg/ml Anisomycin (A) for two hours and/or 500 μM auxin for 1 hr before harvesting where indicated in the figure legends. Whole cell extracts were made from the cells and analyzed by immunoblotting using specific antibodies as indicated in the figure legends and equal amounts loaded on gels based on total protein quantities.

siRNA transfections

siRNA transfections were done using the DharmaFECT 1 transfection reagent (Horizon Discovery Biosciences), as per the manufacturer’s instructions. PP2Ac and PP4c knockdown were done by transfection of 50 nM siRNA (ON-TARGETplus smartpool human PP2Ac and PP4c siRNA or siGENOME nontargeting siRNA pool; Horizon Discovery Biosciences) into HCT116 cells. Whole cell extracts were made from the cells 48 h after transfection and used for immunoblotting.

Immunofluorescence analysis

HCT116 cells transfected with or without WT or 3A-BRD4 were subjected to heat shock at 42°C for 15 minutes and plated on cover slips. The cells were fixed with 4% paraformaldeyhde, permeabilized with 0.5% Triton-X, labeled with anti-MYC antibody (Abcam; Y69) and stained with DRAQ5^™ (Thermo Scientific). Confocal images were acquired using a Zeiss LSM510 META confocal microscope.

Co-Immunoprecipitations and in vitro binding assays

To co-immunoprecipitate specific proteins from cell extracts as detailed in the figure legends, magnetic beads (Dynabeads Protein A; ThermoFisher Scientific) were coated with 5μg of the bait antibody and incubated with the cell extract for 3hr at 4°C. The beads were then washed thrice with 50mM Tris (pH 8.0), 200 mM NaCl, and 0.2% NP-40. Bound proteins were separated on SDS PAGE gels and immunoblotted with specific antibodies as mentioned in the figures. For In vitro binding assays, Flag-BRD4 was pre-incubated for 1hr with M2 Flag-agarose (Sigma), beads and then incubated overnight at 4°C with recombinant purified JNK. The beads were washed twice with 50mM Tris (pH 8.0), 150 mM NaCl, and 0.2% NP-40 and immunoblotted with antibodies against JNK and BRD4. All immunoblot analyses were performed using secondary antibodies from Li-Cor and the Odyssey^™ infrared scanner.

HAT assays

HAT assays were done as described previously ⁵ with minor changes. Purified BRD4 (500 ng) was incubated with 1 μg of histone substrate (New England Biolabs), 0.6 mM unlabeled Acetyl CoA (Sigma) in the presence of HAT buffer (50 mM Tris pH 8.0, 1 mM DTT, 0.1 mM EDTA and 25% V/V glycerol) at 30°C for 30 min. The reaction was stopped with SDS sample buffer, and the samples were run on 15% SDS-polyacrylamide gels that were immunoblotted with the appropriate antibody as mentioned in the figure legends.

In vitro Nucleosome assembly

Purified 5S rDNA (208bp) was procured from New England Biolabs (NEB). Unmodified recombinant human nucleosomes were assembled on the 5S rDNA with purified human histone H2A/H2B dimers and histone H3/H4 tetramers using the EpiMark Nucleosome Assembly kit (NEB) following the manufacturer’s instructions.

Nucleosome eviction assays

For in vitro eviction assays, 5S rDNA (7.7 pmoles) was end labelled with 20 pmoles of γ³²P labeled ATP using T4 polynucleotide kinase (NEB) according to the manufacturer’s protocol. Nucleosomes were assembled in vitro on the radiolabeled 5S rDNA as described above. These labeled nucleosomes were subjected to HAT assays under specific conditions as described in the figure legend. The release of free DNA upon nucleosome eviction was observed by running the samples through a 6% polyacrylamide gel under native conditions and exposing it to a phosphorimager screen.

In vitro Kinase assays

In vitro kinase assays with recombinant proteins were performed in 20 μl of 50mM Tris (pH 7.5), 5mM DTT, 5mM MnCl₂, 5mM MgCl₂ with 10μCi of γ³²P ATP (6000Ci/mM) and/or 40μM ATP where indicated in the figure legend. The kinase reactions for incubated for 1 hour at 30°C, following which the proteins were resolved by SDS-PAGE and the extent of phosphorylation quantitated by a phosphorimager. When phosphorylation was determined by immunoblotting as indicated in the figures, kinase assays were performed with unlabeled ATP.

Mass spectrometric analysis

Flag tagged BRD4 and associated proteins purified from chromatin-bound and nucleoplasmic cell protein fractions were immunoprecipitated using M2 Flag-agarose (Sigma) beads.

Digestion of on-bead IP proteins:

Each sample was treated with 100μL 100mM HEPES pH 8, 50μL each of reducing and alkylating solutions provided with the Thermo EasyPep Kit (Thermo A40006) and incubated in the dark for 1hr. The samples were then treated with 10μL 50ng/μL trypsin/LysC and incubated at 37°C overnight. Following digestion, 175μL of sample was transferred to a new tube and treated with 10μL of 10μg/μL TMTpro (Thermo A52045) reagent and incubated for 1hr at 25 ⁰C with shaking. Excess TMTpro was quenched with 50μL of 5% hydroxylamine, 20% Formic acid for 10min. Samples were cleaned using Oasis HLB μ-elution plates (Waters) and eluted peptides were dried.

LC/MS analysis of IP peptides:

Peptides were resuspended in 50μL of 0.1% FA and 5μL was analyzed using a Dionex U3000 RSLC in front of an Orbitrap Eclipse (Thermo) equipped with an EasySpray ion source. All MS injections employed the TopSpeed method with a cycle time of 3 second that consisted of the following: Spray voltage was 1800V and ion transfer temperature of 275 ⁰C. MS1 scans were acquired in the Orbitrap with resolution of 120,000, AGC of 4e5 ions, and max injection time of 50ms, mass range of 375–1600 m/z; MS2 scans were acquired in the Orbitrap using with resolution of 15,000, AGC of 5e4, max injection time of 22ms, HCD energy of 30%, isolation width of 1.6Da, intensity threshold of 2.5e4 and charges 2–6 for MS2 selection.

Database search and post-processing analysis:

All MS files were searched with Proteome Discoverer 2.4 using the Sequest node. Data was searched against the Uniprot Human database from Feb 2020 using a full tryptic digest, 2 max missed cleavages, minimum peptide length of 6 amino acids and maximum peptide length of 40 amino acids, an MS1 mass tolerance of 10 ppm, MS2 mass tolerance of 0.02 Da, variable oxidation on methionine (+15.995 Da), variable protein N-terminus modifications of acetyl (+42.001 Da), Met-loss (−131.404 Da), Met-loss+acetyl (−89.030 Da) and fixed modifications of carbamidomethyl on cysteine (+57.021). Percolator was used for FDR analysis.

RNA Preparation

RNA was isolated from HCT116 cells that were transfected with empty vector (NT), BRD4 WT (WT) or 3A-BRD4 (TM) and treated with DMSO or 30 μg/ml Anisomycin (A) for two hours using the Micro RNeasy-Plus Kit following the manufacturers protocol (Qiagen). RNA was similarly isolated from HCT116 cells that were treated with DMSO or 500nM JQ1 for 18 hours.

RNA-seq analysis

cDNA Libraries were prepared using the Illumina^® stranded mRNA Prep ligation kit according to the manufacturer’s instructions. from RNA isolated as described above. For all cell line datasets, RNA-seq reads were aligned to human reference genome hg38 using STAR aligner 2.7.8a. Raw read counts were obtained using htseq-count 0.11.4 and normalized for further analysis using the built-in normalization algorithms of DESeq2. We used rMATS 4.1.2 for detecting alternative splicing events. The significantly differentially spliced events were defined as the FDR adjusted p-value < 0.05 and the absolute value of inclusion level difference (IncLevelDifference) > 0.1. RNA from the DMSO/JQ1 treated HCT116 cells was similarly processed and the dataset was compared with the dataset from the anisomycin treated cells.

mRNA quantification by qRT-PCR

Quantitative Real time PCR was performed as described earlier ⁵. Reactions were performed in triplicate using cDNA with the SYBR green Supermix (Bio-Rad) mastermix and 0.4 M of each primer. Primer sequences are shown in supplementary table 1.

ChiP-seq analysis

DLD1-BRD4-IAA7 cells that were transfected with empty vector (EV), BRD4 WT (WT) or 3A-BRD4 (TM) and treated with DMSO or 30 μg/ml Anisomycin (A) for two hours. Cells transfected with WT or 3A-BRD4, but not empty vector, were further treated with 500 μM auxin (IAA; Abcam) for 1 hour to deplete endogenous BRD4. ChIP was performed on these cells using the SimpleChIP^® Plus Enzymatic Chromatin IP Kit (Cell Signaling Technology) as per the manufacturer’s instructions with the AbFlex^® BRD4 antibody (Active Motif). ChIP-seq libraries were prepared using SWIFT_Accel-NGS 2S DNA Library Prep and were sequenced on the NovaSeq 6000SP platform. For ChIP-seq data analysis, sequencing reads were aligned to human reference genome hg38 using Bowtie2–2.3.4.1. The duplicated reads were removed, and only uniquely mapped reads were used for peak identification. ChIP-seq peaks were called using MACS 3.0.0 with these parameters -t for target sample, -c for input sample and -q 0.05 for q-value. The super-enhancers were identified from the SEDB database (https://academic.oup.com/nar/article/47/D1/D235/5146197) and enhancers from the EnhancerAtlas database (http://enhanceratlas.org).

Sequential ChiP assays

Sequential ChIP assays for RNA Pol II and Pol II bound BRD4 were done using the Re-ChIP-IT^® kit (Active Motif) as per the manufacturer’s instructions. Briefly, chromatin from HCT116 cells was prepared using the kit and subjected to immunoprecipitation using anti-RNA Pol II antibody (Active Motif) for the first ChIP. The immunoprecipitated material was then subjected to a second ChIP using anti-BRD4 antibody (Active Motif). DNA from both the ChIPs was then purified and subjected to qRT-PCR using specific primers.

EMT induction

EMT induction in PC3 cells was done by using the StemXVivo EMT Inducing Media Supplement (100X) (Bio-Techne Corporation) using the manufacturer’s instructions. Briefly, ~0.5 × 10⁶ PC3 cells in a 10 cm plate were grown in 6ml media with 1X EMT inducing media supplement. Three days after plating, the media was replaced with fresh media containing the supplement (1X). The cells were harvested for immunoblotting five days after plating. Transfections with WT or 3A-BRD4 were done 24 hr before harvesting the cells were indicated in Fig. 7.

In vitro Thymocyte activation

Total thymocytes from B6 mice were stimulated with PMA (Sigma) and ionomycin (Sigma) either weakly (0.3ng PMA/0.3μg Ionomycin) or strongly (10ng PMA/3.75μg Ionomycin) for two hours. Cells were harvested, stained with fluorochrome-conjugated antibodies with the following specificities: CD4 (GK1.5), CD8α (53–6-7) and CD69 (H1.2F3) for 40 min at 4°C. Stained samples were analyzed on a LSRFortessa (BD Biosciences). Dead cells were excluded by forward light-scatter gating and propidium iodide staining. Data were analyzed using FlowJo v.10.6.2 software.

QUANTIFICATION AND STATISTICAL ANALYSIS

All experiments were performed a minimum of two times; where repeated more often, it is so indicated in the figure legend. Statistics in this study were presented as mean ± SE. Error bars represented SE in triplicate experiments if not mentioned otherwise. Statistical significance for comparisons was generally assessed by Student’s t test. p values below 0.05 were marked by one asterisk, while two asterisks indicate p value < 0.01 and three asterisks indicate p value < 0.001.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit monoclonal anti- BRD4	Bethyl	Cat# A700–004; RRID: AB_2631885
Rabbit polyclonal anti-BRD4	Santa Cruz biotechnology	Cat# sc-48772; RRID: AB_2065729
Mouse monoclonal anti-BRD4	Sigma	Cat# AMAB90841; RRID: AB_2665685
Mouse monoclonal anti-BRD3	Santa Cruz biotechnology	Cat# sc-81202; RRID: AB_1119692
Rabbit monoclonal anti-BRD2	abcam	Cat# ab139690; RRID: AB_2737409
Mouse monoclonal anti-DYKDDDDK tag	GenScript	Cat# A00187; RRID: AB_1720813
Mouse monoclonal anti- Nucleolin (C23, clone MS-3)	Santa Cruz biotechnology	Cat# sc-8031; RRID: AB_670271
Mouse monoclonal anti-JNK	Santa Cruz biotechnology	Cat# sc-7345; RRID: AB_675864
Mouse monoclonal anti-pJNK	Santa Cruz biotechnology	Cat# sc-6254; RRID: AB_628232
Mouse monoclonal anti-CDK9	Santa Cruz biotechnology	Cat# sc-13130; RRID: AB_627245
Mouse monoclonal anti-Pol II	Santa Cruz biotechnology	Cat# sc-56767; 8WG16; RRID: AB_627245
Mouse monoclonal anti-Vimentin	Santa Cruz biotechnology	Cat# sc-373717; RRID: AB_10917747
Rabbit polyclonal anti-CDK9 pThr186	Millipore-sigma	Cat# SAB4504223; N/A
Rat monoclonal anti-Pol II-CTD pSer-5	Millipore	Cat# 04–1572;3E8; RRID: AB_10615822
Rat monoclonal anti-Pol II-CTD pSer-2	Millipore	Cat# 04–1571;3E10; RRID: AB_11212363
Rat monoclonal anti-Pol II-CTD pSer-7	Millipore	Cat# 04–1570;4E12; RRID: AB_10618152
Mouse monoclonal anti-FLAG	Sigma	Cat# F3165; RRID: AB_259529
Rabbit polyclonal anti-β-Tubulin	abcam	Cat# ab6046; RRID: AB_2210370
Rabbit monoclonal anti-MYC	abcam	Cat# ab32072; Y69; RRID: AB_731658
Rabbit monoclonal anti-MYC pThr58	abcam	Cat# ab185655; RRID: AB_2920883
Rabbit polyclonal anti- Histone AcH3K122	abcam	Cat# ab33309; RRID: AB_942262
Rabbit monoclonal anti-Histone H3	Cell Signaling	Cat# 4499; RRID: AB_10544537
Rabbit polyclonal anti- Phospho-c-Jun (Ser63)	Cell Signaling	Cat# 9261; RRID: AB_2130162
Rabbit polyclonal anti- PP2AC	Cell Signaling	Cat# 2038; RRID: AB_2169495
Mouse monoclonal anti-E-Cadherin	Cell Signaling	Cat# 14472; 4A2; RRID: AB_2728770
Rabbit polyclonal anti-PP2AC	Cell Signaling	Cat# 2038, RRID: AB_2169495
Rabbit polyclonal anti-PPP4C	Proteintech	Cat# 10262–1-AP, RRID: AB_2300020
Rabbit recombinant anti-BRD4	Active Motif	Cat# 91301, RRID: AB_2813829
Mouse recombinant anti-RNA Pol II	Active Motif	Cat# 91151, RRID: AB_2793789
Rat monoclonal anti-CD4	Thermo Fisher Sci	Cat# 47–0042-82, RRID: AB_1272183
Rat monoclonal anti-CD8	Thermo Fisher Sci	Cat# MCD0828, RRID: AB_10372364
Hamster monoclonal anti-CD69	Biolegend	Cat# 104508, RRID: AB_313111
Rabbit polyclonal anti-BRD4 pThr1186	Vivitude; This study	Custom; N/A
Rabbit polyclonal anti-BRD4 pThr1212	Vivitude; This study	Custom; N/A
IRDye®680RD Goat anti-mouse IgG Secondary Antibody	LI-COR Biosciences	Cat# P/N: 926–68070; RRID: AB_10956588
IRDye® 800CW Goat anti-Rabbit IgG Secondary Antibody	LI-COR Biosciences	Cat# P/N: 926–32211; RRID: AB_621843
Chemicals, peptides, and recombinant proteins
Phorbol 12-myristate 13-acetate (PMA)	Sigma	Cat# P1585; N/A
Ionomycin	Sigma	Cat# I0634; N/A
Anisomycin	GoldBio	Cat# A-580–25; N/A
SP600125 JNK inhibitor	Selleckchem	Cat# S1460; N/A
D-JNK-1 JNK inhibitor	MedChemExpress	Cat# HY-P0069; N/A
Nodularin inhibitor	Enzo Life Sciences	Cat# ALX-350–061-C100; N/A
Cisplatin	Enzo Life Sciences	Cat# ALX-400–040-M250; N/A
Lipopolysaccharide (LPS)	Thermo Fisher Sci	Cat# 00–4976-03; N/A
(+)-JQ1 inhibitor	BPS Bioscience	Cat# 27401; N/A
Ni-NTA Agarose beads	Qiagen	Cat# 30210; N/A
Anti-DYKDDDDK G1 Affinity Resin	GenScript	Cat# L00432; N/A
Protein A/G Magnetic Beads	Thermo Scientific	Cat# 88803; N/A
Benzonase	Sigma	Cat# E1014; N/A
RNase-Free DNase Set	QIAGEN	Cat# 79254; N/A
Anti-FLAG Peptide	Sigma	Cat# F-3290; N/A
Custom peptide Ac-CQKQEPKTPVAPK-amide	Vivitude; This study	N/A
Custom peptide Ac-CQKQEPK(pT)PVAPK -amide	Vivitude; This study	N/A
Custom peptide Ac-CVQKHPTTPS-amide	Vivitude; This study	N/A
Custom peptide Ac-CVQKHPT(pT)PS-amide	Vivitude; This study	N/A
Histone H3.3 Human, Recombinant	New England Biolabs	Cat# M2507S; N/A
Recombinant Human c-Myc	RayBiotech	Cat# 230–00580-100; N/A
Recombinant human GST-JNK1	SignalChem	Cat# M33–10G-10; N/A
Recombinant human GST-JNK2	SignalChem	Cat# M34–10BG-10; N/A
Recombinant human His-JNK1	Thermo Fisher Sci	Cat# PV3319; N/A
Recombinant human Protein Phosphatase 2A C	Cayman chemical	Cat# 10011237; N/A
Recombinant human Protein Phosphatase PPP4C	MyBioSource, Inc.	Cat# MBS142407; N/A
Critical commercial assays
SimpleChIP Plus Enzymatic Chromatin IP Kit	Cell Signaling Technology	Cat# 9005; N/A
Re-ChIP-IT kit	Active Motif	Cat# 53016; N/A
EpiMark Nucleosome Assembly Kit	New England Biolabs	Cat# E5350S; N/A
RNAeasy Plus mini kit	Qiagen	Cat# 74134; N/A
RNase-Free DNase Set	Qiagen	Cat# 79254; N/A
Duolink® In Situ Red Starter Kit Mouse/Rabbit	Sigma	Cat# DUO92101; N/A
SuperScript™ III First-Strand Synthesis System	Thermo Scientific	Cat# 18080051; N/A
SYBR™ Green PCR Master Mix	Thermo Scientific	Cat# 4309155; N/A
RNA 6000 Nano Kit	Agilent	Cat# 5067–1511; N/A
Quick-Load® Taq 2X Master Mix	New England BioLabs	Cat# M0271L; N/A
Deposited Data
RNA-seq data	This study	GEO; GSE252982
ChIP-seq data	This study	GEO; GSE252982
Mass spec proteomics data	This study	MassIVE; MSV000093873
All original images	This study	http://dx.doi.org/10.17632/mfz7x3zxg6.1
Experimental Models: Cell Lines
PC3 Cell Line	ATCC	RRID: CVCL_0035
HCT116 Cell Lines	ATCC	CCL-247; RRID: CVCL_0291
DLD1 cell line	ATCC	RRID: CVCL_0248
DLD1-BRD4-IAA7 cell line	Dr. Ali Shilatifard; 43	N/A
Recombinant DNA
Mouse WT-BRD4-FLAG in the pFAST Bac1 baculovirus transfer vector	Dr. Dinah Singer; 8	N/A
Murine Flag-His-BRD4 WT (1–1402 amino acids) in the pAcHLT-C baculovirus transfer vector	Dr. Kieko Ozato; 45	N/A
Murine Flag-His-BRD4 ΔN, ΔC, ΔB1, ΔB2 and ΔB1B2 mutants in the pAcHLT-C baculovirus transfer vector	Dr. Keiko Ozato; 45	N/A
Mouse BRD4-FLAG S1153A, T1222A, T1248A, S1153A+T1222A, T1222A+T1248A and the 3A mutants in the pFAST Bac1 baculovirus transfer vector	This study	N/A
Human WT-BRD4-FLAG in the pCDNA5 mammalian vector	Dr. Kornelia Polyak; 46	Addgene plasmid #90331
Human 3A-BRD4-FLAG in the pCDNA5 mammalian vector	This study	N/A
BRD4-FLAG F1, F2, F3 and F4 fragments in the pET11 bacterial vector	Dr. Dinah Singer; 26	N/A
Flag-WT JNK1, JNK2, and Flag-JNK1 APF, JNK2 APF dominant negative mutants in the pCDNA3 mammalian vector	Dr. Roger Davis; 47	Addgene plasmids #13798, #13755, #13846 and #13761
Software and Algorithms
Image studio Lite	Li-Cor	https://www.licor.com/islite
ImageJ	National Institutes of Health; 48	https://imagej.nih.gov/ij/
NIS-Elements 4.5	Nikon	https://www.microscope.healthcare.nikon.com/products/software/nis-elements
Bowtie2	GitHub; 49	https://github.com/BenLangmead/bowtie2
MACS	GitHub; 50	https://github.com/taoliu/MACS
STAR aligner for sequence alignment RNA-seq analysis	GitHub; 51	https://github.com/alexdobin/STAR
rMATS for alternative splicing analysis	SourceForge; 52	http://rnaseq-mats.sourceforge.net
rmats2sashimiplot	GitHub	https://github.com/Xinglab/rmats2sashimiplot
String 11.5	String Consortium	https://string-db.org
db SUPER	Tsinghua University	http://asntech.org/dbsuper
EnhancerAtlas 2.0	Enhanceratlas.org	http://enhanceratlas.org/
Proteome Discoverer 2.4	Thermo Fisher Sci	RRID:SCR_014477

Materials availability

All unique/stable reagents generated in this study are available from the lead contact with a completed Materials Transfer Agreement.