Neurofibromatosis type 1 (NF1) is caused by NF1 gene mutations. The phenotype is highly variable, with 'modifiers' being discussed as potential determinants. Mismatch repair deficiency was shown to cause NF1 mutations, but constitutional mutation of mismatch repair genes was identified only once in a NF1 patient. We aimed to analyze whether DNA methylation of mismatch repair gene promoters, known to lead to transcriptional silencing, is associated with increased tumor load in NF1 defined by the number of cutaneous neurofibromas. Leukocyte DNA of 79 controls and 79 NF1 patients was investigated for methylation of mismatch repair genes MLH1, MSH2, MSH6, and PMS2 by methylation-specific PCR and pyrosequencing. MLH1, MSH6, and PMS2 promoters were not methylated. By contrast, we found promoter methylation of MSH2 with a higher rate of methylation in NF1 patients compared with controls. Furthermore, when comparing NF1 patients with a low vs those with a high number of cutaneous neurofibromas, MSH2 promoter methylation was significantly different. In patients with a high tumor burden, methylation of two (out of six) CpGs was enhanced. This finding was not confounded by age. In conclusion, enhanced methylation involving transcription start points of mismatch repair genes, such as MSH2 in NF1, has not been described so far. Methylation-induced variability of MSH2 gene expression may lead to variable mismatch repair capacity. Our results may point toward a role of MSH2 as a modifier for NF1, although the amount of DNA methylation and subsequent gene expression in other cell types of NF1 patients needs to be elucidated.

Neurofibromatosis type 1 (NF1) is a common autosomal dominant disorder caused by heterozygotic inactivation of the NF1 tumor suppressor gene at 17q11.2. The associated phenotypes are highly variable, and modifying genes have been proposed to explain at least in part the intriguing expressivity. Given that haploinsufficiency of the NF1 gene product neurofibromin is responsible for some of the clinical manifestations, variations in expression of the wildtype NF1 allele might modify the phenotype. We therefore investigated epigenetic molecular modifications that could result in variable expression of the normal NF1 allele. To exclude confounding by DNA sequence variations, we analyzed monozygotic twin pairs with NF1 who presented with several discordant features. We fine-mapped the methylation pattern of a nearly 1 kb NF1 promoter region in lymphocytes of 8 twin pairs. All twin pairs showed significant intra-pair differences in methylation, especially of specific promoter subregions such as 5'UTR, exon 1 and intron 1 (+7 to +622), transcription factor binding sites and promoter elements like NF1HCS. Furthermore, we detected significant intra-pair differences in cytosine methylation for the region from -249 to -234 with regard to discordance for optic glioma with a higher grade of methylation in glioma cases. In conclusion, our findings of epigenetic differences of the NF1 promoter in leukocytes within mono zygotic twin pairs may serve as a proof of principle for other tissues. The results point towards a role of methylation patterns of the normal NF1 allele for expression differences and for modification of the NF1 phenotype.

-10% >10-20% >20 -30% >30% methylation 0-10% >10-20% >20 -30% >30% Figure 3