Crebbp Mutations Disrupt Dynamic Enhancer Acetylation in B-Cells, Enabling HDAC3 to Drive Lymphomagenesis

Somatic mutations in CREBBP occur frequently in germinal center derived lymphomas such as DLBCL and FL. However whether or how these mutations might contribute to lymphomagenesis is still largely unknown. Most CREBBP mutations are predicted to result in loss of function since they target the histone acetyltransferase (HAT) domain or give rise to premature stop codon prior to the HAT domain. Here, we show that Crebbp shRNA knockdown (KD) accelerated lymphomagenesis in VavP-Bcl2 transgenic mice, a model that recapitulates human GC-derived lymphomas. The median time to lymphoma onset in VavP-Bcl2/Crebbp^KD mice was 114 days, significantly shorter than control VavP-Bcl2/GFP mice (193 days, p=0.04). Histopathology revealed that VavP-Bcl2/Crebbp^KD lymphomas were more aggressive and widely disseminated than VavP-Bcl2/GFP lymphomas.

CREBBP can regulate gene enhancer function through H3K27 acetylation. ChIP-seq in VavP-Bcl2/Crebbp^KD lymphoma cells revealed significant reduction of H3K27ac peaks compared to control lymphoma cells (N = 1717, Kolmogorov-Smirnov test, p<2.2E-16). Loss of H3K27ac was markedly skewed towards enhancers. We observed similar loss of enhancer H3K27ac in human DLBCL cells after CREBBP shRNA KD. Enhancer H3K27ac loss was significantly associated with repression of nearby genes in both murine (FDR q=0.044) and human lymphoma cells (FDR q=0). RNA-seq performed in three independent FL or DLBCL patient cohorts revealed a characteristic CREBBP mutant gene expression signature featuring prominent transcriptional repression (p=1.32E-14, p=0.001, and p=0.0002 respectively). Notably, the human patient CREBBP mutant signature was highly enriched in murine and human cell line CREBBP shRNA profiles (FDR=0, GSEA), indicating that CREBBP KD signature was highly similar to CREBBP mutant signature in humans. Functional analysis of the CREBBP mutant/KD signature showed significant enrichment of GC exit pathways including genes induced by CD40, IRF4 and plasma cell differentiation; as well as immune response processes including antigen processing and presentation, such as MHC class II genes (BH-adjusted p<0.05).

To better understand mechanism we performed an integrative analysis of CREBBP signatures against databases of B-cell transcription factor and epigenome profiles. This analysis yielded significant enrichment (BH-adjusted p<0.05) for i) enhancers bound by the BCL6 transcriptional repressor and its SMRT/HDAC3 corepressor complex, ii) enhancers that are normally deacetylated in GC B-cells, and iii) genes induced by BCL6 siRNA. This is notable because in normal GCs BCL6 represses enhancers by recruiting SMRT/HDAC3 complexes to deacetylate H3K27. Hence our data suggest that CREBBP is a counteracting HAT to BCL6/SMRT/HDAC3. Indeed, conditional knockout of Hdac3 in GC B-cells in mice resulted in impaired GC formation and a transcriptional signature featuring upregulation of the same genes that are repressed by CREBBP KD (GSEA FDR=0). Moreover, CREBBP KD in DLBCL cells resulted in H3K27ac loss at BCL6/SMRT/HDAC3 regulated enhancers, including those nearby CDKN1A, NFATC1, FOXP1, and MHC II genes, such as HLA-DQA1 and HLA-DRB5. CREBBP KD also resulted in silencing of these genes. Since we show HDAC3 is the opposing HDAC to CREBBP then we reasoned that CREBBP mutant DLBCLs might be especially dependent on HDAC3. Indeed we observed that HDAC3 shRNA resulted in profound suppression of CREBBP mutant DLBCL cells in vitro and in vivo (DLBCL xenografts in mice, p=0.005), whereas CREBBP WT cell lines were barely affected by HDAC3 KD.

The opposing effects of BCL6/SMRT/HDAC3 and CREBBP on MHC class II could have implications for immune surveillance. Accordingly CREBBP KD induced significant loss of cell surface HLA-DR molecules (p<0.05), and these cells exhibited up to 90% less capability to stimulate T-cell response in allogeneic mixed lymphocyte reaction experiments. The loss of MHC class II molecules and T-cell response was rescued when CREBBP loss of function cells were exposed to a specific HDAC3 inhibitor. In summary, CREBBP mutations drive lymphomagenesis by enabling unopposed suppression of enhancers by BCL6/SMRT/HDAC3 complexes, resulting in a repressive transcriptional programming that disrupts GC exit and evades immune surveillance. HDAC3 targeted therapy may rescue these effects and serve as a precision approach for CREBBP mutant lymphomas.