Aberrant activation of the Wnt/β-catenin pathway in primary CLL is known to contribute to the defect in apoptosis inherent to this malignancy
[3, 6, 7, 17]. Different molecular mechanisms for this were suggested, including the autocrine activation by the overexpressed Wnt ligands
, silencing of E-cadherin due to aberrant splicing
 or epigenetic down-regulation of pathway inhibitors by promoter hypermethylation
. Stimulated by the findings of aberrantly methylated CDH1 and SFRPs
 as well as our own evidence on hypermethylation of DKK2 and DKK3 in primary CLL, we quantitatively characterised the methylation status and the expression of the Wnt/β-catenin pathway inhibitors. Ten of the twelve genes exhibited high DNA methylation in at least one of the CLL cell lines and also in patient samples compared to non-tumour material. The absolute level of methylation, however, varied strongly between the genes.
By far the biggest difference in methylation between tumour and normal could be seen for SFRP1, which was hypermethylated in all tumour samples. Surprisingly, however, the apparent change in transcript level upon DNA demethylation with 2 μM 5-aza-dC was insignificant. In contrast, methylation in the promoter regions of DACT1 and DKK4 did not differ between tumour and normal. However, their transcription was affected by changes to the degree of DNA methylation, although for DACT1 this happened clearly indirectly via a yet unknown factor. The fact that DNA methylation was found to be involved in the regulation of the antagonist genes’ expression in the cell lines and the close similarity of the methylation patterns to those of primary CLL B cells suggests the existence of an epigenetic silencing mechanism in this haematological malignancy.
Owing to the absence of the CpG island within its 5´ region, SFRP3 was the only member of the SFRP class, whose methylation status had not been analysed in most of reports published earlier, irrespective of the studied tumour type. However, we could demonstrate that hypermethylation of CpG sites in the first exon was associated with an apparent transcriptional down-regulation to an extent that was well beyond that seen for other genes. Also overall, the variation of methylation upon addition of the demethylation agent differed between genes. There might be a correlation between the intensity of the effect observed and the mere number of CpG sites or their frequency in a given stretch of DNA sequence. However, the data set from this study was too small for any significant evaluation and more quantitative analyses are required to proof an actual relationship.
Also, the degree of demethylation induced by 5-aza-dC had different apparent effects on the transcription levels. The extent of variation may be controlled by the sequences next to the CpG sites. Such an effect is known for the formation of left-helical Z-DNA structures, which are most likely to occur in methylated d(CG) sequences
. A conformational twist from right- to left-helical secondary structure can occur from one base to the next and back, resulting in a net structural variation that only disturbs or relaxes the right-turning helix rather than inducing a strong conformational change. This variation in the structural components of a sequence could explain an associated variation in gene activity and could topologically affect also DNA stretches that have some distance from the actual CpG site.
Because epigenetic down-regulation of the Wnt/β-catenin pathway inhibitors may be instrumental in the constitutive Wnt/β-catenin signalling in primary CLL
[3, 6, 7, 17], we wondered if the pathway is active in the CLL cell lines and can be modulated upon DNA hypomethylation. However, no constitutive Wnt/β-catenin signalling could be detected in either of the two cell lines by measuring the level of the transcriptionally active β-catenin fraction. This obvious discrepancy with primary CLL
[3, 7] might reflect a secondary loss of the constitutively active pathway in the established cell lines. This observation is in agreement with a recent report
, which has documented much less β-catenin and lymphoid enhancer-binding factor 1 (LEF-1, a direct target of the pathway
) in the CLL cell lines JVM-1 and MEC-1 in comparison with the patient samples. Given a dormant Wnt/β-catenin pathway in the CLL cell line models, it was therefore not possible to ascertain, if pharmacological restoration of all the pathway antagonists can affect its activity. Nevertheless, the role of epigenetic silencing of one of the inhibitors, E-cadherin, could be demonstrated. We showed that its expression is regulated by promoter methylation both on transcriptional and protein levels. Significantly up-regulated upon 5-aza-dC treatment, E-cadherin binds β-catenin thereby capturing it on the cellular membrane. We speculate that epigenetic restoration of E-cadherin expression in the cells with aberrantly active Wnt/β-catenin signalling might suppress it in a similar way as the enforced E-cadherin expression in the primary CLL B cells lead to down-regulation of the pathway
. Thus, this finding may be of therapeutic interest.
Finally, observations from this study draw attention to two aspects, which warrant further clarification. First, the concurrent hypermethylation and silencing was shown for the genes of a single pathway, which are located on different chromosomes. This finding may support a model of carcinogenesis suggested earlier, in which epigenetic silencing is not completely stochastic but might reflect the existence of a directed program, by which functionally related groups of genes important for development of tumours are silenced through promoter methylation
. Our data may add further evidence for such a guiding role of Polycomb-group repressive complexes (PRC) in patterning aberrant DNA hypermethylation in cancer. These epigenetic regulatory proteins induce repressive chromatin states by covalently modifying histones (H3K27me3) within promoters of many developmentally regulated genes in embryonic and adult stem cells
[49, 50]. Also in B cells, a PRC2 component EZH2 was shown to contribute to epigenetic reprogramming of naïve B cells as they transit through germinal centre, which may facilitate proliferation and lymphomagenesis
. Given the facts that (1) promoter regions of the Wnt pathway antagonists are repressively marked with H3K27me3 in either naïve B cells, centroblasts or embryonic stem cells
[51, 52], (2) promoters of DKK1, DKK2, SFRP1 and SFRP2 are occupied with the PRC2 component EZH2 in these cells, (3) recruitment of DNA methyltransferases by EZH2
 and (4) age dependent aberrant hypermethylation of PRC targets
 have been reported, it is plausible to assume that concurrent hypermethylation of specific groups of genes is an interrelated part of general epigenome reprogramming orchestrated by PRC.
Second, despite the fact that cancer-associated CpG hypermethylation has been shown early in development of solid tumours
, nothing is known about epigenetic alterations in a CLL precursor state monoclonal B cell lymphocytosis (MBL)
. However, up-regulation of LEF-1, a direct target of the Wnt pathway and a pro-survival factor, has recently been reported in MBL patients, who are known to be at risk for progression to CLL
. Therefore, deregulation of the Wnt/β-catenin pathway may have a role in CLL leukaemogenesis and further methylation analysis of the antagonists in MBL is desirable in view of possible benefits for CLL prevention by DNA demethylating drugs, if epigenetic aberration of these genes is detected at an early stage of MBL.