Evidence for epigenetic dysregulation in lymphomas continues to accumulate. Histone deacetylase inhibitors (HDACi) may reverse the bias toward transcriptionally repressed heterochromatin found in some tumors, restoring apoptotic and cell cycle regulation pathways (Figure). However, clinical results with HDACi to date have been relatively modest, and there is no reliable means to predict benefit.
The group led by Riccardo Dalla-Favera at Columbia University has performed whole-exome sequencing to look for recurrent patterns of somatic change in diffuse large B-cell lymphoma (DLBL), and from seven cases sequenced plus another 72 studied by single nucleotide polymorphism (SNP) arrays they identified frequent alterations in regions containing the acetyltransferase genes CREBBP and EP300. Both of these genes modify lysine residues on nuclear proteins and enhance transcription of many important genes through acetylation of chromatin itself, acetylation of transcriptional activators such as p53, and inactivation of some transcriptional repressors such as BCL6. This finding was a surprise, because almost no mutations in these genes are seen in epithelial tumors, so the group went back to sequence the coding regions of CREBBP in 134 DLBLs and found sequence variants in 30, all resulting in loss or truncation of the active histone acetyltransferase (HAT) domain. These findings were more common in germinal-center (GCB)-type DLBL than non-GCB (32% vs. 13%). In another nine cases, the SNP arrays revealed deletion of all or part of one allele of CREBBP, and four showed abnormalities in both alleles. A similar analysis of EP300 yielded 19 with sequence variants and seven with allelic deletions. Overall, 39 percent of DLBLs showed structural alteration of at least one of these genes.
The group then looked at follicular lymphoma, with similar results. Among 46 cases, 15 had CREBBP mutations, also clustered in the HAT domain and not found in the patients’ germline DNA, and four in EP300. Analysis of other types of low-grade B-cell lymphoma did not yield any similar abnormalities, suggesting they are particular to malignancy arising in the germinal center.
Protein expression in cases with CREBBP alterations showed the wild-type allele present but reduced protein levels, while EP300 alterations led to complete loss of expression, despite detectable mRNA. Reduced protein levels were also noted in some lines despite intact CREBBP and EP300, suggesting alternative mechanisms of inactivation in some lymphomas. The mutations in the two genes showed evidence of functional overlap, with a high proportion clustered in a region of high identity encoding the contact surface for coenzyme A, a cofactor in acetylation reactions.
Transient transfection experiments confirmed that CREBBP proteins with mutations in the HAT domain lost their ability to acetylate BCL6 and p53. Studies using reconstituted double-knockout mouse embryo fibroblasts showed that the mutant protein blunted cyclic-AMP-induced transcription and endogenous histone acetylation and also reduced cell growth.
The balance between histone acetylation and deacetylation determines the relative amounts of transcriptionally active euchromatin and repressed heterochromatin.
The balance between histone acetylation and deacetylation determines the relative amounts of transcriptionally active euchromatin and repressed heterochromatin.
In Brief
These observations suggest a previously unrecognized but surprisingly common pathogenic pathway in germinal center lymphomas, with a defect in one allele sufficient to produce important changes in the transcriptional profile of a cell despite the presence of a wild-type second copy. This suggests that HAT activity in B lymphocytes can be limiting, with small changes in effective gene dose having major consequences for their phenotype. The therapeutic implication is that attempts to treat using inhibitors of HDAC may be most effective in those cases in which HAT mutations have disturbed the normal balance of acetylation, although clearly other mechanisms will have a role in this. Why such a pathogenic mechanism should be particular to the germinal center remains a matter of speculation, but this paper highlights the power of whole-genome sequencing to produce entirely new information on the genetic changes that lead to malignancy.
