BRD4 and Cancer: going beyond transcriptional regulation

Transcription initiation begins with the recruitment of RNA-PolII on the preinitiation complex (PIC) at the gene promoter, followed by RNA-PolII phosphorylation at Serine 5 and stabilization of the RNA-PolII-promoter interaction. PIC assembly is widely influenced by ENHs and controlled by TFs and other transcription regulatory proteins [25]. Mediator of RNA-PolII transcription (Mediator) is a large complex with modular organization which transduces signals from TFs and activators at the ENHs to promoters, timing PIC formation and transcription initiation [26‐28]. BRD4 has been isolated as preferred cofactor of the Mediator complex [29‐31]. BRD4 and the Mediator complex have been shown to largely co-localize at active ENHs and in particular at the level of super-ENHs-SE and their genomic localization is disrupted by BET inhibition (Fig. 1a, b). Indeed, early after BETi treatment Mediator detaches by the set of regulatory elements to which is bound and its displacement tracked closely with the sensitivity of gene expression to BETi [7, 29]. Beside highlighting the relevance of BRD4 in mediating the crosstalk between promoters and distal ENHs these data indicate that inhibition of Mediator is one of the central mechanisms underneath BETi cytotoxicity.

Many TFs and chromatin remodeling proteins have been shown to interact with BRD4 [6, 32, 33]. Since TFs are mainly associated with ENHs the interplay with these proteins is crucial for BRD4-ENHs interaction and to define BRD4 selectivity for target genes. TFs recruit acetyltransferases like p300/CBP, which in turn promote nucleosome and/or non histonic proteins (like the same TFs) acetylation. Therefore, recruitment of BRD4 to active ENHs is likely mediated by interaction with acetylated histones as well as direct binding to ENH-associated TFs [6]. Indeed, a recent model proposed that the simultaneous binding of BRD4 to histones and TFs would stabilize TFs binding to ENHs maintaining elevated the transcriptional activity of these elements. The interaction of BRD4 with TFs and histone modifiers can be either mediated by BDs and therefore dependent on the acetylation status of TFs [32, 34, 35] or by other BRD4 domains, thus being not affected by BETi treatment [33, 36, 37]. It has been shown that the ET domain of BRD4 is involved in transcriptional regulation by interacting with histone modifiers including the arginine demethylase JMJD6 [36], the lysine methyltransferase NSD3 and the nucleosome remodeling enzymes SWIF/SNF and CDH4 [37]. Through these interactions BRD4 facilitates transcriptional activation by de-compacting chromatin and accelerating messenger RNA synthesis. Using a yeast purified FLAG-tagged BRD4, Wu and colleagues performed an unbiased biochemical screening looking for BRD4 interactors [33]. Beside identifying consolidated BRD4 partners and many chromatin modifiers, these Authors demonstrated that BRD4 associates in a BD independent mode with a selected group of TFs including p53, YY1, AP2, c-JUN, C/EBPα and β and the heterodimer c-MYC/MAX. BRD4 interaction with p53 is mediated by two regions outside the BDs and in part regulated by casein kinase (CK2) mediated phosphorylation. CK2 phosphorylates BRD4 in an acidic region juxtaposed to BDs enabling BRD4 loading on DNA and interaction with p53. The selective binding with certain TFs as well as the possible modulation of these interactions by signaling pathways, have major implication in dictating BRD4 specificity for its target genes and may strongly contribute to BRD4 context-dependent functions. Indeed, Najafova and colleagues, studying osteoblast differentiation, showed that BRD4 is enriched at putative osteoblast specific distal ENHs where BRD4 colocalizes specifically with a restricted set of TFs including C/EBPb, TEAD1, FOSL2 and JUND [22]. Also in acute myeloid leukemia, BRD4 has been shown to co-localize and cooperate with hematopoietic TFs (including PU.1, FLI1, ERG, C/EBPα, C/EBPβ, and MYB) in conjunction with the lysine acetyltransferase p300/CBP to support lineage-specific transcriptional circuits in this disease [38]. BRD4 has been also shown as necessary coactivator of the inflammatory transcriptional program driven by NF-κB by binding to acetylated RelA [32, 34] and of a diacetylated form of TWIST in triple negative breast cancer. Beside controlling protein coding genes, BRD4 also regulates transcription of ENH-associated RNA (eRNAs) [39]. Many active ENHs have been shown to be site of active transcription for short (less than 200 nt) and bidirectional RNAs, which in turn assist ENHs in promoting target genes expression [40, 41]. Inhibition of BRD4 by BETi treatment has been shown to consistently repress the expression of a pool of eRNAs transcribed from BRD4 enriched ENHs. Treatment with BETi disassociates BRD4 from target ENHs abolishing BRD4-RNAPolII interaction at these sites reducing RNA-PolII recruitment with consequent inhibition of eRNA synthesis [39]. Promoting ENHs active transcription, BRD4 fosters ENHs activity not only through an architectural remodeling.

RNA-PolII pausing at promoter-proximal regions and its release into active transcription are a major rate limiting steps of gene expression [25]. Release of RNA-PolII into active transcription is mediated by the positive transcription elongation factor-b (P-TEFb) complex, comprising cyclin T1 (CCNT1) and cyclin-dependent kinase 9 (CDK9). In its active form, P-TEFb complex interacts with TFs and cofactors, phosphorylates pausing factors (like the Negative Elongation Factor complex NELF) displacing their binding from chromatin, and phosphorylates the carboxy-terminal domain (CTD) of RNA-PolII at Serine 2 activating transcription [25]. Among the 3 ubiquitously expressed BRD proteins, BRD4 is the only one able to interact with p-TEFb through its C-terminal domain [14, 16, 42‐45]. Interaction with BRD4 prevents P-TEFb binding to the inhibitory ribonucleoprotein complex 7SK/HEXIM that sequesters P-TEFb in its inactive form [43, 45]. Accumulation of BRD4 on hyper-acetylated and transcriptionally active TSSs serves as docking sites for P-TEFb which in turns facilitates activation of RNA-PolII and its release into active elongation. Recent evidence indicates that also ENH-associated BRD4, through its interaction with JMJD6, influences P-TEFb activity promoting elongation [36]. JMJD6 colocalizes with BRD4 selectively on a definite set of active distal ENHs characterized by high levels of H3K4Me1 and H3K27Ac and named anti-pause ENHs. BRD4 mediates JMJD6 recruitment on these sites that in turns 1. demethylates H4 arginine 3 mono- and dimethylated (H4R3me1, H4R3me2) which are associated with transcription repression, 2. demethylates the 5′-methyl-phosphate cap of 7SK RNA causing its degradation and promoting local activation of P-TEFb. Long range chromatin looping mediates the effect of ENH-associated BRD4/JMJD6/P-TEFb complex on RNA-PolII release at TSS of target genes. Nucleosomes can also impede RNA-PolII progression along the gene body reducing transcript elongation and gene expression [36]. A recent report by Kanno and colleagues showed that BRD4 facilitates RNA-PolII transition along the gene body in a P-TEFb independent manner, further promoting transcript elongation [39]. These Authors showed that BRD4 beside being enriched on active promoters and ENHs is bound along the body of highly transcribed genes. Gene expression alterations following BRD4 inhibition either by BETi or BRD4 silencing largely correlate with BRD4 localization along gene bodies. This indicates that the removal of BRD4 from these genes has deleterious consequences for the RNA-PolII processivity. BRD4 follows RNA-PolII during elongation since abrogation of transcription by Actinomycin-b dramatically reduces the amount of BRD4 bound along target genes. Based on this evidence the Authors proposed that BRD4 functions as chaperone that taking advantage of its ability to interact with acetylated histones facilitates the passage of RNA-PolII through hyperacetylated nucleosomes [39]. Linked to the effect of BRD4 on transcription elongation, Donato et al. recently proposed a compensatory mechanism that may affect BETi gene target selectivity. They indicated that BETi unsensitive genes can compensate the impairment of transcription elongation induced by BRD4 inhibition by increasing RNA-PolII recruitment at TSS and transcription initiation. This is not possible for BRD4-addicted genes which have already maximized the amount of RNA-PolII loaded at TSS, to sustain the high rate of required transcription. The block of transcription elongation imposed by BRD4 inhibition impedes promoter clearance and further recruitment of RNA-PolII conferring selective susceptibility to BETi to highly expressed genes [46].

In addition to function as scaffolding platform for a variety of TFs and transcription related proteins, BRD4 has been shown to directly affect RNA-PolII activity by its intrinsic kinase and acetyl-transferase activity [47‐49].

BRD4 directly stimulates RNA-PolII CTD domain phosphorylation at Serine 2 triggering productive transcription [49, 50]. Noticeably, BRD4-mediated RNA-PolII phosphorylation results in the specific activation of Topoisomerase I which associates with RNA-PolII during transcription progression favoring DNA unwinding and RNA-PolII transition [51]. Furthermore, BRD4 has been shown to phosphorylate CDK9 (of the P-TEFb complex) regulating its enzymatic activity. As histone acetyltransferase (HAT) BRD4 has been shown to acetylate histones H3 and H4 with a distinct pattern, different from those of classical HATs. BRD4 acetylates H3 K122 resulting in nucleosome eviction and chromatin deconstruction leading to increased transcription [47].

Considered the centrality and the multifunctional role that BRD4 plays in transcription regulation it is not surprising that BETi represent an attracting strategy in cancer therapy. However, while the successful employment of these drugs is moving faster from selected MYC-addicted hematological malignancies to solid cancers, questions about cancer specific susceptibility to these drugs still remain. Gene expression profiles showed that BRD4 is expressed in a broad range of somatic cells. As well, ChIP seq data showed that BRD4 occupies many ENHs and promoters also in non-cancer cells and generally contributes to gene expression (Table 1) [6]. Thus, how inhibition of a global chromatin regulator as BRD4 may results in a such cancer selective targeting, it remains at least in part unclear. The recent discovery of SEs suggests that BETi cancer specificity may be a matter of quantity [7, 52]. SEs are large clusters of ENHs closely spaced in genome that possess unique functional properties distinguishing them from regular ENHs [52]. These regions show a density of transcription factors and regulators binding as well as a quantity of H3K27Ac and H3K3Me1 that surmount of one order of magnitude the one detected on regular ENHs. In cancer, SEs are central to maintain cell identity and are deputy to drive the expression of specific oncogenes to which cancer cells become highly addicted. Genome-wide studies indicated that SEs mark lineage specific TFs and oncogenes in a wide range of cancer type, revealing that these structures are acquired during tumorigenesis and are unique to cancer cells while largely absent in untrasformed cells [7, 53, 54].

Thus, shortcutting the functional organization of these structures causes a dramatic drop in oncogenes expression and consequential cancer cells growth inhibition or death.

In the attempt of defining the molecular mechanisms beyond the cancer cytotoxicity of BETi in Multiple Myeloma (MM), Loven and colleagues showed that BRD4, together with Mediator, and P-TEFb is heavily accumulated on SEs and that BETi causes a preferential loss of BRD4 at these elements. Inhibition of BRD4 then resulted in loss of SEs function with consequent transcription elongation defects and block of oncogenes expression including c-MYC or other SEs associated MM relevant genes such as IGLL5, IRF4 and XBP1 [7]. At SEs, BRD4, together with Mediator, acts as the catalyst of the assembly of a cooperative and synergistic high density transcriptional complexes changing structure, dynamics and function of chromatin. Sabari and colleagues recently investigated the biophysical nature of SEs showing that these elements exhibit properties of liquid-like condensates. They showed that accumulation of BRD4 and Mediator at SEs forms nuclear puncta and gives rise to phase-separated droplets that compartmentalize and concentrate the transcription apparatus favoring protein crosstalk and interplay [55]. In such environment, inhibition of BRD4 has greater consequences that on regular ENHs conferring to SE-associated genes a higher magnitude of susceptibility to BETi and explaining at least in part the cancer selective activity of these drugs (Fig. 1b).