Acetylation of histone H3 lysine 14 (H3K14ac) and H4 dual lysines - either K5 and K12, or K8 and K16 - seems to be crucial for Brd4 binding to acetylated histone peptides in vitro
]. Whether the bromodomains of Brd4 also interact with acetyl-lysine in a nonhistone protein has recently been addressed by the identification of Brd4 as a bromodomain-containing protein that interacts with the RelA subunit of nuclear factor (NF)κB in an acetylation-dependent manner [36
]. Both bromodomains of Brd4 are necessary for interacting with K310-acetylated RelA and for tumor necrosis factor (TNF)α-induced transcription of selective NFκB target genes, such as those encoding E-selectin and TNFα, but not A20. It is likely that association with acetylated RelA enhances the interaction between Brd4 and acetylated histone H3 and H4, which appears to be intrinsically weak and thus requires cooperative binding through multiple protein-protein (Brd4-activator and Brd4-histone) and protein-DNA (activator-DNA) interactions [3
]. This represents the commitment step for Brd4-regulated target gene transcription mediated by a transcriptional activator (, left). Although subsequent recruitment of Mediator by Brd4 is likely necessary for Pol II entry to the promoter region and ensuing CTD phosphorylation (, center), it is presently unknown how this step occurs in the context of chromatin. Very often, Pol II remains bound at the promoter region in an inactive state until stress-induced signals reactivate its elongation activity, allowing Pol II to pass through the pausing site.
A broader role of Brd4 in transcriptional regulation has likewise been addressed by knockdown studies and also with gene profiling and chromatin immunoprecipitation (ChIP) analysis of a large set of stress response genes and their chromatin marks. Zhou and colleagues [9
] demonstrated that Brd4-mediated recruitment of P-TEFb to selective G1 genes encoding JunB, c-Myc, Cdk7, and Cyclin D1 begins in mid- and late anaphase, but prior to nuclear envelope formation and the entry of Pol II, TFIIB, and TFIIF, and continues into early G1 phase of the cell cycle. Interestingly, Brd4 association with mitotic chromosomes can occur at different stages of mitosis, depending on the cell type, and seems to be regulated by undefined cellular factors or covalent modifications. This mitotic association of selective G1 genes has been independently observed by the Ozato group, who elegantly illustrated by microarray analysis and quantitative RT-PCR (reverse transcription polymerase chain reaction) that many G1 gene products, including Cyclin D1, D2, E1, and E2, and phosphorylation of pRb, are downregulated in Brd4-knockdown NIH3T3 cells and mouse embryonic fibroblasts [10
], and that Brd4 marking of M/G1 genes increases significantly at telophase, coinciding with increased acetylation of histone H3 and H4 in these genes as well as P-TEFb recruitment and de novo
M/G1 gene transcription [11
The general significance of Brd4-recruited P-TEFb has been further revealed by another study using ChIP analysis of many primary response genes showing that, upon lipopolysaccharide stimulation of Toll-like receptor signaling, paused Pol II with Ser5 phosphorylation promptly resumes elongation following Ser2 phosphorylation of its CTD that is mediated by P-TEFb, which is recruited by Brd4 via recognition of induced acetylation at histone H4K5/8/12 in most primary response gene promoters [37
]. Nevertheless, an independent study performed on the FOSL1
gene, which encodes the Fra1 subunit of AP-1 complexes, reported that MOF (males absent on the first)-acetylated H4K16 triggered by serum-induced PIM1 phosphorylation of H3 serine 10 (H3S10ph) are both required for Brd4 binding to targeted chromatin, which in turn recruits P-TEFb to the FOSL1
]. P-TEFb then induces Ser2 phosphorylation, which reactivates Pol II elongation activity (, right). It appears that H3K9ac is needed for efficient H3S10ph to occur. While both reports clearly indicate the importance of Brd4 in recruiting P-TEFb to poised Pol II in stress-induced gene activation, the nature of modified histone tails involved in the recruitment of Brd4 - that is, acetylated H4K5/8/12 versus H3K9acS10ph-induced H4K16ac - is different. This is possibly due to dynamic binding of different histone acetyltransferases (HATs) and histone deacetylases (HDACs) surrounding the targeted loci at the time of induction. A recent genome-wide binding survey of five HATs [p300, CREB-binding protein (CBP), p300/CBP-associated factor (PCAF), MOF, and Tip60] and four HDACs (HDAC1, HDAC2, HDAC3, and HDAC6) by high-throughput sequencing of antibody-pulled down ChIP fragments (ChIP-Seq) isolated from resting and induced primary human CD4+
T cells surprisingly revealed that both HATs and HDACs co-reside in most active genes and also in many repressed but inducible genes [39
]. Since H3K9/14 and H4K5/8/12/16 can be acetylated by various HATs (p300/CBP for H3K14 and H4K5/8, PCAF for H3K9/14, Tip60 for H3K14 and H4K5/8/12/16, and MOF for H4K16) (reviewed in [40
]), the distinct patterns of acetylated residues identified in Brd4 recruitment probably reflect specific HATs and HDACs available in the surveyed cell types and the induced signaling pathways. This intriguing possibility needs to be explored in the future.
Brd4-mediated recruitment of P-TEFb is also important for transcription of viral promoters, as recently demonstrated in several HPV-harboring human cell lines, in which HPV-encoded E2 protein is able to suppress viral transcription by displacing P-TEFb from Brd4-bound HPV chromatin owing to competition between high-affinity E2 and low-affinity P-TEFb for the same binding site at the CTM of Brd4 [42
]. This result provides an additional mechanism for the E2-Brd4 silencing complex that efficiently inhibits HPV chromatin transcription by blocking TFIID and Pol II recruitment to the viral promoter in a bromodomain-dependent manner [16
]. In HPV it is likely that both Brd4-regulated initiation and elongation steps of transcription are simultaneously targeted by E2 for efficient suppression of viral gene expression, consistent with a central role of Brd4 in both cellular and viral gene transcription.
Recent structural determination of several BET bromodomains further provides atomic details for bromodomain/acetyl-lysine interactions. While an earlier study reporting that homodimer formation of BD1 derived from human Brd2 creates a negatively charged secondary binding pocket at the dimer interface may help to determine binding specificity provided by the acetyl-lysine-binding pocket of BD1 [43
], dimerization of mouse Brd4 BD1 or BD2 was not seen in solution or in crystal structures [44
]. Both BD1 and BD2 of mouse Brd4 exhibit the classical bromodomain fold consisting of four α-helices (αZ
, and αC
) and two interconnecting ZA and BC loops. Charged amino acid residues (acidic in BD2, basic in BD1) surrounding the acetyl-lysine-binding pocket may enhance the binding specificity of BD1 and BD2. An amino-terminal extension of BD1 that helps stabilize the bromodomain fold is also evident in the structure. It is noted that the binding affinities of BD2 for K390-acetylated Cyclin T1 peptide and H4K5ac or H4K16ac peptide are rather similar (Kd ~110 μM), indicating the possibility of switching of BD2 binding from acetylated chromatin to acetylated P-TEFb [44
]. As predicted [3
], cooperative binding between Brd4 and a DNA-binding factor is generally required to stabilize Brd4 association with chromatin, as now shown by the low affinity measured for the interaction between BD1/BD2 and acetylated histone tails, which is mediated by a few hydrogen bonds and sparse hydrophobic contacts [44
The cooperative nature of bromodomain binding to acetyl-lysine has also been demonstrated by isothermal calorimetry measuring differences in the binding affinities of BD1 and BD2 of mouse Brdt for histone H3 and H4 peptides with varying degrees of acetylation at distinct lysine residues [45
]. Surprisingly, BD2 fails to bind tetra- and mono-acetylated H4 peptides, whereas BD1 binds as efficiently as an amino-terminal portion of Brdt protein containing both BD1 and BD2 to tetra-acetylated H4 peptide (Kd ~28.0 μM). Clearly, BD1 and BD2 are not equivalent binders and each has unique features and binding affinities for acetyl-lysine residues in both histones and nonhistone proteins. A compilation of binding studies performed with various BET proteins or their derived bromodomains and acetylated lysines in histone H3 and H4 at different positions is summarized in . Clearly, cooperative binding to acetylated histone tails requires diacetylated lysines with properly spaced residues. A crystal structure of mouse Brdt BD1 in complex with diacetylated H4K5acK8ac peptide revealed an unexpected association of two acetylation marks on the same histone tail with a single bromodomain [45
]. BD1 binding to K5ac is prototypical and K8ac-BD1 interaction is mediated primarily through hydrophobic contacts, mimicking overall a two-pronged plug with K5ac inserted deep into the acetyl-lysine-binding pocket and K8ac bent diagonally towards K5ac. The structure of BD1 binding to the diacetylated K5ac-GG-K8ac peptide seems to be conserved among several other bromodomains, although these are yet to be experimentally illustrated.
Summary of binding studies performed with various BET family proteins or their derived bromodomains with acetylated histone H3 and H4 peptides in vitro or with chromatin in live cells