Epigenome-wide analysis of T-cell large granular lymphocytic leukemia identifies BCL11B as a potential biomarker

Abstract

Background: The molecular pathogenesis of T-cell large granular lymphocytic leukemia (T-LGLL), a mature T-cell leukemia arising commonly from T-cell receptor αβ-positive CD8⁺ memory cytotoxic T cells, is only partly understood. The role of deregulated methylation in T-LGLL is not well known. We analyzed the epigenetic profile of T-LGLL cells of 11 patients compared to their normal counterparts by array-based DNA methylation profiling. For identification of molecular events driving the pathogenesis of T-LGLL, we compared the differentially methylated loci between the T-LGLL cases and normal T cells with chromatin segmentation data of benign T cells from the BLUEPRINT project. Moreover, we analyzed gene expression data of T-LGLL and benign T cells and validated the results by pyrosequencing in an extended cohort of 17 patients, including five patients with sequential samples.

Results: We identified dysregulation of DNA methylation associated with altered gene expression in T-LGLL. Since T-LGLL is a rare disease, the samples size is low. But as confirmed for each sample, hypermethylation of T-LGLL cells at various CpG sites located at enhancer regions is a hallmark of this disease. The interaction of BLC11B and C14orf64 as suggested by in silico data analysis could provide a novel pathogenetic mechanism that needs further experimental investigation.

Conclusions: DNA methylation is altered in T-LGLL cells compared to benign T cells. In particular, BCL11B is highly significant differentially methylated in T-LGLL cells. Although our results have to be validated in a larger patient cohort, BCL11B could be considered as a potential biomarker for this leukemia. In addition, altered gene expression and hypermethylation of enhancer regions could serve as potential mechanisms for treatment of this disease. Gene interactions of dysregulated genes, like BLC11B and C14orf64, may play an important role in pathogenic mechanisms and should be further analyzed.

Keywords: BCL11B; DNA methylation; Large granular lymphocytic leukemia; Pyrosequencing; T-LGLL.

Fig. 1

Genome-wide methylation signature of T-LGLL compared to normal T-cell subsets. Unsupervised Principal Component Analysis (PCA) of T-LGLL samples (LGL = purple) and benign T-cell subsets (CD4⁺ central memory = yellow; CD4⁺ effector memory = rose; CD4⁺ naïve = red; CD8⁺ central memory = green; CD8⁺ effector memory = blue; CD8⁺ naïve = gray) of the top two components. Beta-values of 39,930 CpG loci with the highest variation (sd/sd_max> = 0.4) are visualized

Fig. 2

Genome-wide methylation signature of T-LGLL and CD8⁺ memory cells. Heatmap of supervised cluster analysis of T-LGLL samples (purple) and CD8 positive memory cells (CD8⁺ central memory = green; CD8⁺ effector memory = blue), using adjusted p value < 0.005 and delta Beta > 0.25 lead to visualization in this heatmap of 2216 highly differentially methylated CpG loci

Fig. 3

Annotation of differentially methylated CpG loci in T-LGLL with chromatin states of CD8⁺ memory cells. A Differentially hyper- or hypomethylated CpG loci between T-LGLL and normal T-cell subsets were annotated by the chromatin segmentation data of CD8⁺ memory cells from the BLUEPRINT project (chromatin segmentations E1–E11 are depicted in different colors in %, color code explained in Additional file 1: Table S17). The first “background” bar represents the loci of the array. From left to right the following CpG subsets are shown: (1) All 450 k CpGs, annotated to chromatin states (2) Significant hypermethylated CpGs in T-LGLL samples vs non-neoplastic T cells, (3) significant. hypomethylated CpGs in T-LGLL samples vs non-neoplastic T cells, (4) significant hypermethylated CpGs located in promoters or enhancers (as defined by the Illumina 450 k annotation, see Additional file 11) in T-LGLL samples vs non-neoplastic T cells, (5) significant hypomethylated CpGs located in promoters or enhancers (as defined by the Illumina 450 k annotation, see methods) in T-LGLL samples vs non-neoplastic T cells. B Chromatin state enrichment of CD8-positive memory genome regions containing significantly differentially methylated CpG loci (hyper and hypomethylated, bar 2 and 3, Fig. 3A, respectively) compared to 450 k background CpG loci (bar 1, Fig. 3A). (*) Significant enrichments (hypergeometric test, p value < 0.01). Chromatin segmentation E1–E11 is described in Additional file 1: Table S17

Fig. 4

Expression analysis of T-LGLL samples compared to normal T-cell subsets. Expression analysis (qPCR) of T-LGLL samples of indicated genes identified as hyper- or hypomethylated compared to healthy donor-derived samples (CD8⁺ T-cell subsets: T naïve = CD8⁺ naïve cells; T CM = CD8⁺ central memory cells; T EM = CD8⁺ effector memory cells; T-LGLL = T-LGLL samples; CD8 HD = healthy donor bulk CD8⁺ cells). High average delta CT values indicate low expression

Fig. 5

Interaction of the gene BCL11B with other genes in CD8⁺ cells. BCL11B long-range interactions from Promoter Capture Hi-C for total CD8-positive T cells (red: BCL11B–C14orf64 (LINC01550) interactions). Genomic regions are depicted in blue circles (promoter region) or blue square (enhancer region)

Fig. 6

Methylation values over time for ten CpG sites in “patient 2” determined by bisulfite pyrosequencing. Response to therapy is indicated by descending methylation values. Sampling was not performed in a linear manner