P633: active chromatin regulatory landscape of stereotyped subsets in chronic lymphocytic leukemia reveals a distinctive signature in subset #8

Background: Chronic lymphocytic leukemia (CLL) patients with stereotyped B cell receptors have been characterized from the perspective of genetic and DNA methylation alterations. Recent studies have characterized the chromatin landscape of non-stereotyped CLLs and normal B cells (Beekman, et al. Nat Med. 2018). Aims: To characterize the active chromatin regulatory landscape of 4 major stereotyped subsets as compared to IGHV status-matched non-stereotyped CLLs. Methods: We used chromatin-immunoprecipitation followed by sequencing (ChIP-Seq) with an antibody for the H3K27ac (a bona fide histone mark for regulatory element activation) in sorted CLL cells from 15 normal B cell subpopulations and 44 CLL, including 18 cases from stereotyped subsets #1, #2, #4, and #8. To understand their functional impact, we integrated our data with 16 RNA-seq data considering the 3D chromatin configuration (Hi-C data: GM12878 and a representative U-CLL case) (Figure 1) Results: Unsupervised principal component analysis (PCA) of H3K27ac data revealed different profiles in CLL as a whole, regardless of stereotypy status, vs normal B cells (PC1, 25%). Of note, PC2 (14%) revealed a clear separation between subset #8, a highly aggressive CLL subtype with the highest risk of Richter’s transformation amongst all CLL, vs other CLLs or normal B cells, suggesting that subset #8 may have a specific acetylation pattern. Indeed, supervised differential acetylation analysis (FDR<0.01) uncovered a signature of 903 Differentially Acetylated Regions (DAR) in subset #8 vs U-CLL (Figure 1) and no differences for the other subsets. This signature was unrelated to +12, highly prevalent (~60%) in subset #8, or IgG expression, a hallmark of subset #8. Hierarchical clustering analysis of this signature revealed 6 clusters of regions with different chromatin dynamics highlighting the de novo active cluster1 (n=197 regions) in subset #8 vs other CLLs or normal B cells. These differentially acetylated regions (DARs) were enriched (FDR<0.05) in binding sites of several differentially expressed TFs in #8 vs U-CLL, e.g. members of SOX, GATA, and CEBP families as well as MEF2A and HNF1A, which showed the highest enrichment. Next, we used the topologically associating domains (TAD) resulting from Hi-C data and RNAseq data to identify the potential target genes of the 197 de novo DARs in subset #8. We identified the target genes in 78/197 (39.6%) DARs that were linked to 112 genes, as some active regulatory regions were associated with more than one gene. Among them, the DAR located in chr9:137085994-137086793 bp was related to the expression levels of 8 genes (LCNL1, CLIC3, FBXW5, AJM1, MAMDC4, CCDC183, TRAF2 and an un-annotated transcript) within the same TAD. This region also contained 6 additional H3K27ac peaks with higher signal in subset #8, though not reaching significance. This finding suggests deregulation of a broad region in chromosome 9 leading to the coordinated upregulation of 8 transcripts. Concerning the activated genes, TRAF2 is a remarkable example since TRAF family members (TRAF6, -4, -1) are found overexpressed in U-CLL and here we report an active region in subset #8 leading to TRAF2 overexpression. Image:Summary/Conclusion: We report a subset #8-specific chromatin acetylation signature containing de novo activated regions, linked to the upregulation of target genes, in some cases including blocks of adjacent genes. This study further supports the distinct biological nature of subset #8 and may help to understand the aggressive clinical behavior of this distrinctive CLL variant.