It has been known for some time that mammalian Brd proteins bind to acetylated lysine residues in histones of euchromatic nucleosomes (see Introduction). However, no other biochemical activities had been ascribed to these proteins, nor had they been shown to associate with specific endogenous genes. Data presented here establish that the Brd2 and Brd3 proteins are associated in vivo
with the highly acetylated chromatin of transcribed genes, but not with transcriptionally inactive chromatin (). In particular, the Brd2 and Brd3 associated nucleosomes contained histone H4 acetylated on K5 and K12, but not on K16, and histone H3 acetylated on K14, but not on K9 (). Acetylation of K12 and K5 of histone H4 is found in chromatin that packages the entire length of transcribed genes in Saccharomyces cerevisae
(Liu et al., 2005
). Likewise, histone H4 carrying these modifications was found throughout the transcribed cyclin D1 and RPS28 genes in human cells, as were Brd2 and Brd3 (). This property is in stark contrast to the restricted association of PHD domain-containing proteins that recognize histone H3 trimethylated on K4 near promoters (Li et al., 2006
; Pena et al., 2006
; Shi et al., 2006
; Wysocka et al., 2006
). The broad distribution of the Brd proteins along these genes might reflect a largely structural role for these proteins in establishing chromatin domains in which genes are accessible to the transcriptional machinery. However, Brd2 is required for transcription of the cyclin D1 gene (), even thought this gene is also bound by Brd3 (). This observation is difficult to reconcile with such a passive role, and rather suggests that Brd2 participates actively in transcription. Consistent with this view, Brd2 (and Brd3) allowed transcriptional elongation through acetylated nucleosomes in vitro
(). In the case of Brd2, such transcription required binding of the protein to the acetylated lysine residues with which it is known to associate in vivo,
as well as the bromodomains that mediate this interaction (). We therefore propose that Brd2 also facilitates movement of RNA polymerase II on nucleosomal templates during transcription of particular genes in vivo.
Previous studies have implicated several mammalian BET family proteins in the regulation of transcription (Denis et al., 2000
; Kanno et al., 2004
); Schweiger et al., 2006
; (You et al., 2004
) (Schweiger et al., 2006
). For example, Brd4 can stimulate transcription from the human immunodeficiency virus type 1 LTR promoter by facilitating recruitment of the elongation factor pTEF-b (Jang et al., 2005
; Yang et al., 2005
). It was also observed that the binding of Brd4 to chromatin along the entire length of an integrated HIV-1 LTR-luciferase gene is required to establish a similar pattern of association of pTef-b (Jang et al., 2005
). This finding suggests that the Brd4 protein might indirectly regulate elongation of transcription. The results of our in vitro
experiments establish that Brd2 and Brd3 effectively remove nucleosomal barriers to transcription elongation by RNA polymerase II, independently of pTef-b or any other elongation proteins (). These data therefore identify a previously unknown function of these Brd proteins, one that depends on their ability to bind to acetylated histones (). Furthermore, the inhibition of cyclin D1 transcription by RNAi-mediated knockdown of Brd2 () provides the first example of transcriptional regulation of an endogenous gene in its natural genomic and chromatin context by a member of the Brd/BET protein family.
In the chromatin transcription assay, Brd2 and Brd3 were at least as effective in stimulating transcription through acetylated nucleosomes as FACT, which has been shown to bind to H2A and H2B, and to function as a histone chaperone (Orphanides et al., 1999
). Brd2 interacts primarily with histone H4 acetylated on K5 and K12 (Kanno et al., 2004
) and the former interaction is necessary for its function in transcription elongation (). In view of their different binding specificities, it is unlikely that Brd2 or Brd3 stimulate elongation through nucleosomes by the same mechanism of FACT. Nevertheless, their ability to replace FACT in the chromatin transcription assay () suggested that that these Brd proteins are also histone chaperones. Exactly how Brd proteins structurally alter and/or remove nucleosomes to allow passage of RNA polymerase remains to be established. Previous studies have shown that the ATP-dependent chromatin remodeling enzymes such as RSC can remove nucleosomal barriers to allow passage of elongating RNA polymerase II in an acetylation dependent manner. Moreover, SWI/SNF has been shown to displace acetylated nucleosomes and this activity requires its’ bromodomain (Carey et al., 2006
; Chandy et al., 2006
). All our transcription reactions contain an ATP-dependent chromatin assembly and remodeling factor RSF. Consequently, it is possible that in this simplified system the Brd proteins altered nucleosomes in conjunction with RSF mimicking cooperation with any of the several chromatin remodeling enzymes present in cells. Alternatively, Brd proteins bound to acetylated nucleosomes might act in concert with a translocating RNA polymerase II molecule to drive nucleosome removal or restructuring.
It is likely that Brd proteins also contribute to the induction and maintenance of euchromatin (Mattsson et al., 2002
). In contrast to other bromodomain proteins, Brd2 (and Brd4) remain bound to chromatin during mitosis, a property that suggests that they contribute to epigenetic inheritance (Dey et al., 2003
; Kanno et al., 2004
). Furthermore, extensive “ChIP on CHIP” experiments have shown that the yeast BET protein Bdf1 is bound to hyper-acetylated transcribed chromatin that is specifically non-acetylated at lysine 16 of histone H4 (H4K16) (Kurdistani et al., 2004
). When acetylated, this residue provides a docking site for the heterochromatic protein Sir3 (Ladurner et al., 2003
). It is therefore believed that the deacetylation of H4K16 in heterochromatin results in the replacement of Sir3 with Bdf1, and the subsequent induction of euchromatin (Kurdistani et al., 2004
). In fact, the loss of Bdf1 leads to Sir3 mediated spreading of telomeric heterochromatin (Ladurner et al., 2003
). The observation that the nucleosomes to which Brd2 and Brd3 are bound in human cells specifically lacked histone H4 containing K16Ac () suggests that this distinguishing feature of euchromatin, and its recognition by Brd proteins, is likely to be conserved among eukaryotes.
In these studies, we examined the association of Brd2 and Brd3 with eight endogenous genes. Of these, Brd2 bound to three and its depletion by RNAi decreased the transcription of one (cyclin D1) ( and ). However, Brd3 was also bound to all the genes associated with Brd2. These observations suggest some degree of redundancy in the functions of Brd2 and Brd3. This property underscores the need for systematic genome wide approaches, to catalog genes bound and regulated by Brd proteins. Such information will facilitate efforts to establish the physiological roles played by these proteins, and their mechanism(s) of action in vivo.