Toward a BETter grasp of acetyl-lysine readers

In this issue of Blood, Stonestrom et al describe the unanticipated complexity of the distinct yet overlapping activities of acetyl-lysine–binding bromodomain and extraterminal motif (BET) proteins bromodomain-containing 2-4 (BRD2-4) in erythropoiesis, in the context of rising interests in BET pharmacologic inhibitors.¹ 

Epigenetic modifications dictate chromatin structure and affect gene expression. It has been long established that aberrant activity of the enzymes controlling the deposition, recognition, and removal of these chromatin marks often leads to cancer. Indeed, since the first report of a link between the acetyltransferase CREB-binding protein (CBP), the oncoprotein E1A, and cell proliferation/differentiation,²  large numbers of translocations and mutations involving epigenetic modifiers have been reported in various cancers.³ 

Lysine acetylation is a major posttranslational histone modification, associated with chromatin opening and positive regulation of gene expression. While attention was initially focused on writers (CBP, p300) and erasers (histone deacetylases) of this modification as potential drug targets, epigenetic readers of acetylation have recently been in the limelight. The BET motif family includes ubiquitously expressed BRD2, BRD3, and BRD4.⁴  These proteins possess 2 highly conserved tandem bromodomains mediating binding to acetylated histones and transcriptional regulators; BETs also interact with components of the basal transcription machinery.⁵  Among their targets are genes controlling progression of the cell cycle. BETs are thus functionally and mechanistically associated with the control of gene expression and cell growth.

Involvement of BETs in the pathophysiology of various cancers came as no surprise. As chromosomal translocation partners or cofactors of transcriptional complexes nucleated by oncogenic proteins, BRD2/3/4 have appeared as suitable pharmacologic targets in a range of malignancies such as carcinomas, leukemias, and prostate cancer.⁶  Although protein/protein interactions are notoriously difficult to disrupt, recent technological advances have led to the development of numerous BET inhibitors, including the acetyl-lysine mimetics JQ1 and I-BET, currently being evaluated in clinical trials.⁷  These molecules interact with the highly conserved acetyl-lysine–binding pockets of BETs, leading to efficient disruption of their interaction with acetylated targets.

BET inhibitors have therefore been identified as promising therapeutic molecules. However, it is becoming clear that not all BET targets are sensitive to BET inhibitors and that BETs may operate through distinct mechanisms; yet there has been little study of the functional differences between BETs. What are the active targets of individual BETs? Do BETs have overlapping functions? Why do BET inhibitors exhibit restricted activity on some gene targets? It is essential to start answering these questions to be able to fully understand the biology of BETs, dissect the mechanisms underlying the activities of current inhibitors, and develop new inhibitors specifically targeting biologically relevant BETs.

To better understand epigenetic and transcriptional regulation, Gerd Blobel’s group recently studied the role of BETs in the activity of the key transcription factor, GATA1, in erythropoiesis.⁸  They demonstrated that BETs bind acetylated GATA1 and established a functional relationship between BETs and GATA1 association with its chromatin targets.⁸  Through genome-wide chromatin immunoprecipitation (ChIP) and transcriptomic analyses combined with pan vs selective inhibition of BETs, Stonestrom et al now further dissect the mechanisms of action of BETs in erythropoiesis.¹ 

A well-established model of erythroid differentiation relying on inducible expression of GATA1 (G1ER cells)⁹  allowed the authors to test BET activities in absence or presence of GATA1 and to compare the effects of JQ1 inhibition vs depletion of individual BETs. They uncover a complex landscape of BET chromatin occupancy and transcriptional activities supporting the association of GATA1 with chromatin and the regulation of its target genes.

At the heart of this study is the discovery that individual BETs use distinct mechanisms of action in erythropoiesis, suggesting functional interplay. Indeed, BRD3 was present at the greatest number of GATA1-occupied sites and was strongly influenced by the presence of GATA1. In contrast, recruitment of BRD4 and BRD2 correlated less well with GATA1 occupancy, with BRD4 binding strongly linked to histone acetylation levels. Therefore, it was unexpected that CRISPR-Cas9–mediated BRD3 deletion produced no measurable phenotype, suggesting functional compensation by other BETs. Given the high conservation of structural motifs among BETs, it is now important to identify what determines the specific and overlapping activities of BETs and understand the role of their bromodomains in chromatin vs transcription factor binding.

The authors then probed the functional relationships between BETs, GATA1 chromatin occupancy, and transcriptional activation. Pulsed application of JQ1 not only showed destabilization of GATA1 binding to different extents genome-wide, but also highlighted BET transcriptional activities post-GATA1 chromatin occupancy, presumably through interactions with the transcriptional machinery.

To establish reliable relationships between BETs and GATA1 targets, the authors repeated transcriptome analyses in G1ER cells using spike-in controls: remarkably, only 220 genes were activated more than twofold in response to GATA1 induction, whereas 5094 genes were repressed. Comprehensive analyses of gene expression combined with ChIP data led to 4 major conclusions: (1) BET inhibition overwhelmingly reduced transcription activation with little effects on inhibition by GATA1, (2) not all BET-bound genes were JQ1-sensitive, (3) gene activation by GATA1 predicted JQ1 sensitivity much better than BRD4 occupancy (thereby raising the question of how reliable the so-called BRD4-bound superenhancers¹⁰  are in predicting a response to BET inhibition), and (4) BETs and GATA1 chromatin binding as well as sensitivity of GATA1 to JQ1 all seemed to correlate with co-occupancy by GATA1 protein partners. Overall, this study offers a platform to identify distinct classes of GATA1 targets and to dissect the molecular mechanisms underlying JQ1 sensitivity likely to be influenced by the multiprotein complexes that include GATA1.

Finally, despite its high degree of colocalization, depletion of BRD3 had little effect on GATA1 activity unless in the context of BRD2 depletion. Moreover, BRD3 overexpression partially restored the defects caused by BRD2 loss. The ability of BETs to partially compensate for each other needs to be taken into consideration when linking pharmacologic BET inhibition to any given BET protein.

In conclusion, this thorough analysis paves the way for further molecular dissection of the activities of these critical epigenetic readers and should motivate similar studies in other cellular systems. It is essential to fully describe the biology of BETs for an informed optimization of inhibitors targeting specific oncogenic mechanisms.