ABSTRACT Follicular lymphoma (FL) is an indolent but incurable malignancy of germinal center B (GCB)-cells, and is the second most common form of non-Hodgkin lymphoma (NHL). Mutations of CREBBP, which encodes a lysine acetyltransferase (KAT) protein, occur in ~60% of FL and arise early during disease evolution. We found that conditional knock-out (KO) of Crebbp in murine models and can cooperate with Bcl2 over-expression to drive B- cell lymphoma, but that these tumors had key differences to FL. Furthermore, FL tumors have a predominance of missense mutations within the CREBBP KAT domain and infrequently harbor nonsense/frameshift mutations that are modeled by KO. We therefore modeled the most frequent CREBBP KAT domain mutation, R1446C, using CRISPR editing of a CREBBP wild-type (WT) lymphoma cell line and found that CREBBP-R1446C mutation has a more profound epigenetic effect than biallelic knock-out (KO). Regions with reduced H3K27Ac were enriched for loci that are normally bound by both CREBBP and BCL6 in germinal center B-cells, suggesting that the loss H3K27Ac may be linked to failure of CREBBP to oppose BCL6-dependent HDAC3 activity. Notably, BCL6 also recruits EZH2 and loss of H3K27Ac in CREBBP-R1446C cells was accompanied by a gain of the EZH2-catalyzed H3K27me3 mark, suggesting that enhancer inactivation via loss of H3K27Ac is followed by enhancer decommissioning through addition of H3K27me3. Furthermore, the CREBBP mutation phenotype could be partially rescued by either HDAC3 or EZH2 inhibition, and was further enhanced by combination of HDAC3 and EZH2 inhibitors. These results suggest that CREBBP KAT domain mutations suppress histone acetylation through a dominant repressive mechanism, and that epigenetic crosstalk between CREBBP and EZH2 may play a prominent role in regulating the epigenetic landscape of B-cell lymphoma. We have extended upon these observations by performing biochemical and structural analysis of major CREBBP mutational hotspots and observed key differences between the R1446C/H alleles and the next most common hotspots at Y1482 and Y1503. These amino acids all reside within the catalytic pocket of CREBBP, but R1446 mediates interaction with the CoA portion of the acetyl donor, acetyl-CoA, while Y1482 and Y1503 mediate interaction with the acetyl group. While R1446 mutations maintain some catalytic activity, the Y1482 and Y1503 mutations are catalytically dead. Therefore, these mutations are likely to have different functional consequences. In this proposal, we aim to (i) characterize the mechanism of dominant epigenetic repression by CREBBP KAT domain mutations and differences in function between KAT domain hotspot mutations, and (ii) define the role of epigenetic crosstalk between CREBBP and EZH2 in B-cell lymphoma, and the consequences for response to targeted agents.

Follicular lymphoma (FL) is the second most common form of non-Hodgkin’s lymphoma, and is characterized by recurrent mutations in genes that encode histone modifying enzymes. Mutations in the histone acetyltransferase, CREBBP, are present in approximately two thirds of cases and are acquired as an early event in disease evolution. Here we aim to investigate the molecular mechanisms by which CREBBP mutations alter the epigenome of FL and cooperate with other histone modifying enzymes such as EP300 and EZH2.