The control of genomic imprinting is intricately linked with DNA methylation (Li et al., 1993). Germline methylation at specific loci set up regulatory networks for the monoallelic expression of imprinted genes. The methyltransferases DNMT3a and DNMT3L work in concert to establish these methylation patterns, and the removal of either from mouse embryos results in the abolition of imprinting in these animals (Bourc'his et al., 2001; Kaneda et al., 2004).
Differentially methylated regions (DMRs) are methylated on one allele but not the other. The presence of DNA methylation usually corresponds with a lack of transcription from that region due to the recruitment of chromatin condensation complexes by methyl binding proteins (Jones et al., 1998; Nan et al., 1998). The regulation of imprinted clusters is often quite complex, and methylation does not necessarily correspond with suppression. Where a DMR corresponds with a promoter of one of the cluster transcripts, that gene will be repressed on the methylated allele.
At some imprinted domains, however, such as the KCNQ1 region on hChr11, the expression of a ncRNA molecule from the unmethylated allele results in repression of other genes in cis, so in fact the majority of imprinted transcripts from that cluster are expressed from the methylated allele (Fitzpatrick et al., 2002). As previously decribed (Table 1.3 in Chapter 1, Introduction) The creation of transgenic mice
lacking known DMRs has identified several of them as ‘imprint control regions’
(ICRs).
In the human, the disruption of allelic ICR methylation, which may involve either a loss or a gain of overall methylation, can result in several different imprinting syndromes. Ten percent of the fetal overgrowth disorder, Beckwith Wiedemann Syndrome (BWS) cases are caused by methylation changes at IGF2/H19 (ICR1), and 40-50 % by loss of methylation at the KvDMR (ICR2) (Elliott and Maher, 1994). As many as half of all Silver Russell syndrome (SRS) cases are due to methylation defects at ICR1 (Abu-Amero et al., 2008). Mutation of the ICR sequence can result in the inappropriate setting of methylation during gametogenesis. Two percent of Prader-Willi syndrome (PWS) and 8 % Angelman syndrome (AS) cases are caused by mutation of the paternal (unmethylated) and maternal (methylated) ICRs respectively (Buiting et al., 1995; Saitoh et al., 1996). The genome-wide disruption of methylation during cancer development often results in the biallelic expression, or repression, of imprinted genes notably IGF2, a potent growth enhancer (Cui et al., 2003). Children affected by BWS caused by loss of methylation at IGF2/H19 are also highly likely to develop Wilm’s tumour (Rump et al., 2005; Weksberg et al., 2001)
Cells growing in culture often display changes from the cell population from which they were derived, commonly involving methylation (Bibikova et al., 2006).
Prolonged culture growth, and stress put on the cells during cryopreservation or passaging can result in the accumulation of induced methylation changes (Maitra et al., 2005). Where methylation changes correspond with ICR methylation, concurrent loss of imprinting can occur. Murine ES cells display widespread loss of imprinting, which is correlated with a loss of appropriate allelic methylation (Dean et al., 1998;
Humpherys et al., 2001). Changes in methylation are often locus specific, indicating that there are regions prone to disruption, but others which are more robust.
Disruption of imprinting of IGF2/H19 of primate ES cells was found to correlate with hypermethylation of the CTCF binding site, whereas maintained imprinting of SNRPN was consistent with the maintenance of the DMR (Mitalipov et al., 2007).
Data on human ES cells reveals one set of genes whose allelic expression can be correlated with ICR methylation status, and a second for which both expression and methylation vary between different cell lines independently (Kim et al., 2007b).
Methylation status of KvDMR, PEG3, NESP55 and SNRPN (PWS/AS-IC) DMRs
correlated with allelic expression in hES cell lines, whereas status of the MEST DMR and IGF2 DMR2 did not (Kim et al., 2007b). In another study, biallelic expression of H19 was observed following prolonged passage, but was associated with DMD hypermethylation, the opposite of what would be expected normally (Rugg-Gunn et al., 2005).
In the present study, a gene specific loss of imprinting has been observed.
This loss of imprinting could be due to changes in methylation at differentially methylated ICRs. To investigate whether this was the case, methylation at DMRs of the MEST, PEG10, IGF2, KCNQ1, SNRPN, PEG3, and GNAS imprinted gene clusters was analysed. Combined bisulphite and restriction analyses (COBRA) were used to quantify allelic methylation, followed by bisulphite sequencing to measure bisulphite conversion and the methylation status of each CpG in the region analysed.
Figure 6.1 Combined bisulphite and restriction analysis (COBRA)
Combined bisulphite and restriction analysis (COBRA) uses bisulphite treatment and endonuclease recognition to distinguish methylated and unmethylated DNA.
Genomic DNA is incubated with sodium bisulphite, resulting in the conversion of unmethylated cytosines to thymines, and then amplified using bisulphite-specific primers for the region of interest. The inclusion of a CpG dinucleotide containing-restriction enzyme recognition site (e.g. TCGA) into the amplicon, allows the semi-quantitative resolution of methylated and unmethylated alleles by size following digestion by the restriction enzyme. (i) PCR product prior to incubation with enzyme, then (ii), (iii) and (iv) after incubation: (ii) completely unmethylated region; (iii) fully
methylated region; (iv) region containing equal amounts of methylated and unmethylated DNA – a candidate DMR.
The COBRA technique allows estimation of the ratio of methylated to unmethylated alleles (Xiong and Laird, 1997). It only allows analysis of one or two CpG dinucleotides per assay, however, so to measure the extent of methylation throughout the whole region, the PCR amplicon may be cloned and sequenced.
Bisulphite sequencing additionally permits estimation of the efficiency of bisulphite conversion. However, bias during ligation and colony growth for PCR products representing methylated or unmethylated DNA cannot be controlled for, so sequencing does not necessarily give an accurate ratio of each type of DNA.