Abstract
Deletions in chromosomal band 13q14.3 occur in a variety of human neoplasms like chronic lymphocytic leukaemia (CLL), indicating a tumor suppressor mechanism (TSM) in this region. Intriguingly, several characteristics of the region of interest point to an epigenetic pathomechanism:
candidate protein-coding genes and non-coding RNA genes including miR15a and miR16-1 lack point mutations in the majority of patients, yet these genes are significantly downregulated in almost all CLL patients the presence of large non-coding RNA genes in 13q14.3 is reminiscent of imprinted regions where only one gene copy is active.
We have recently shown that already in healthy tissue only one gene copy of 13q14.3 is active while one gene copy is randomly chosen for silencing. Thus, loss of the single active copy is sufficient for complete loss of gene function in tumor cells.
In order to elucidate the epigenetic regulatory mechanism, we analysed DNA- and Histone-methylation of all CpG islands in the region in non-malignant B-cells and CLL cells. Using aPRIMES and ChIP-qPCR as screening tools, BioCOBRA as a quantitative high-throughput method and bisulfite sequencing for validation, we could identify two candidate regulatory elements with abnormal chromatin in CLL patients (n=80, median 57% DNA-methylation, range 0–100%) as compared to healthy probands (n=20, median 88% DNA-methylation, range 74–100%, p<0.003). Interestingly, this epimutation can be found in all cytogenetic subgroups of CLL patients and is independent of IgV(H) mutation status, making it a prime candidate for an underlying epigenetic defect in CLL. Pilot studies suggest that this epimutation regulates gene expression of the critical region via large non-coding RNA genes.
In order to find out how loss of function of the 13q14 genes could result in the pathophenotype of CLL cells, we overexpressed and knocked-down RFP2, C13ORF1, KPNA3 and the largen non-coding RNA gene Dleu2 in two different cell lines and used custom oligonucleotide microarrays and timecourse experiments (n=68 array hybridizations) to identify genes that were subsequently deregulated and thus potential target genes. Less than 1% of genes represented on the arrays were significantly deregulated (median 211/25100 genes, range 44–370), showing the high specificity of the procedure. Using ingenuity pathway analyses, we found that modulation of the expression of 13q14.3 candidate genes deregulates most significantly NFkB target genes and components of the NFkB pathway itself. For a detailed validation analysis we focused on RFP2 and could show that it robustly and quickly induces NFkB activity in fibroblasts (HeLa), kidney cells (HEK-293) and CLL cell lines (Granta-591). However, analyses by oligonucleotide ELISA, Western Blot and EMSA-Band-Shift assays suggest that activation of NFkB occurs not via modulation of components of the canonical or non-canonical NFkB signalling pathways.
Therefore, we propose a model for the TSM in 13q14.3 where
in healthy B-cells, only one gene copy is active while the second is epigenetically silenced expression of candidate genes is deregulated in CLL cells by epimutation that is present in all cytogenetic subgroups and that this loss of function of 13q14 candidate genes results in deregulation of the NFkB signalling pathway which will change the activation level of CLL cells and their sensitivity to induction of apoptosis.
PSoL: a positive sample only learning algorithm for finding non-coding RNA genes
