Benign peripheral nerve sheath tumors are the clinical hallmark of Neurofibromatosis Type 1. They account for substantial morbidity in NF1 and are difficult to manage. Cutaneous (CNF) and plexiform neurofibromas (PNF) share identical histology, but maintain different growth rates and risk of malignant conversion. The reasons for their disparate clinical behavior are not well explained on the basis of recent genome or transcriptome profiling studies. We hypothesized that CNFs and PNFs are epigenetically distinct tumor types that exhibit differential signaling due to genome-wide and site-specific methylation events. We interrogated the methylation profiles of 45 CNFs and 17 PNFs (Illumina EPIC 850K) using normal tissue controls from NF1 subjects. Based on these profiles, we confirm that CNFs and PNFs are epigenetically distinct tumors with broad differences in higher order chromatin states, and specific methylation events altering genes involved in key biological and cellular processes such as inflammatory mediator regulation of TRP channels, RAS/MAPK signaling, actin cytoskeleton rearrangement, and oxytocin signaling.Based our identification of 2 separate DMRs associated with alternative leading exons in MAP2K3, we demonstrate differential RAS/MKK3/P38 signaling between CNFs and PNFs. Epigenetic reinforcement of RAS/MKK/P38 was a defining characteristic of CNFs leading to pro-inflammatory signaling and chromatin conformational changes, whereas PNFs signaled predominantly through RAS/ERK. Tumor size also correlated with specific CpG methylation events. Taken together, these findings confirm that epigenetic regulation of RAS signaling fates accounts for observed differences in CNF and PNF clinical behavior. CNFs may also respond differently than PNFs to RAS-targeted therapeutics raising the possibility of targeting P38mediated inflammation for CNF treatment.Copy number estimation from EPIC methylation arrays Copy number alterations were computed using SeSAMe (v1.3.2) with minor modifications for plotting functionality. Briefly, data were processed using the "open SeSAMe" procedure, producing a signal set object. Next, samples were segmented and called for copy number differences, comparing CNFs to normal skin and PNFs to normal nerve. The genome annotation used was hg19.
RNA-sequencing data analysis