This study represents the most comprehensive epigenetic analysis of CNFs and PNFs to date. Greater than 99% of NF1 patients exhibit both CNFs and PNFs over the course of their lifetime accounting for a substantial negative impact on quality of life4. Pain is a constant feature of both neurofibroma subtypes, yet how pain signaling occurs in peripheral nerve tumors is poorly understood. Despite the recent demonstration of MEK inhibitor effectiveness in PNF treatment34, it is unclear whether CNFs respond with equal efficacy. More therapies are needed to treat neurofibroma tumor progression and symptoms, such as pain and itching, as all of these clinical features contribute significantly to morbidity in NF1. Our findings address an unmet clinical need for neurofibroma treatment and offer significant mechanistic insight into how benign nerve tumors initiate, progress, and generate symptoms through epigenetic means. It is now conceivable that NF1-deficient neurofibromas can be treated by targeting epigenetic mechanisms that reinforce RAS signaling.
In the absence of distinctive transcriptomic or genomic alterations in CNFs and PNFs apart from putative NF1 deficiency, our work confirms that methylation events are key molecular determinants of nerve tumor initiation, growth, and pain generation. How epigenetic regulation of kinase signaling affects cancer predisposition in these tumors remains unclear; however, recent data confirms that accumulating epigenetic alterations in a single field or region are associated with elevated cancer risk35–39. Thus, our data lay the groundwork for future studies examining how epigenetic alterations affect PNF conversion into MPNSTs, as well as protective mechanisms that spare CNFs from cancerous progression or unchecked tumor growth. The identification of robust differences in the methylation profiles of CNFs versus PNFs also confirms their distinct biology, as well as laying the groundwork for future development of clinical biomarkers. Moreover, the
Chromatin conformational states differed significantly between CNFs and PNFs and were strongly linked to both site-specific and geographic-specific methylation events. These findings suggest that chromatin accessibility broadly affects gene expression in CNFs and PNFs. More work is needed to determine how epigenetic alterations affect regulatory genes that are known to contribute to tumor size and, ultimately, cancer predisposition. Based on our probe-based analysis of tumor tissue, we identified 34 CpG methylation sites that were statistically correlated with CNF size. Unfortunately, the genes corresponding to the individual methylation probe sites could not be identified with statistical confidence, nor could we link these methylation events with specific biological processes or signaling pathways. Regardless, these data confirm that CpG methylation influences CNF tumor size, possibly through a novel mechanism. More work is needed in this area.
Our data confirms that CNFs and PNFs strongly exhibit differential methylation at two established DMRs (i.e. DMR1 and DMR2) that are situated immediately upstream of the MAP2K3 transcriptional start site. This pattern of differential methylation resulted in upregulated expression of MKK3 and p38 in CNFs, whereas in PNFs the reciprocal effect was observed with downregulated expression (Fig. 5). This effect was consistent within and across tumor types despite expected signaling heterogeneity from analyzing whole tumor tissue from multiple subjects. The cell types that contributed to the observed differences in methylation profiles could not be determined. Unfortunately, deconvolution analysis is dependent on cell type-specific profiles which are lacking for NF1-deficient neurofibromas.
MAP2K3 was previously identified as a candidate imprinted gene in the context of NF1 deficiency40, but the roles of its upstream DMRs are not well characterized. DMR1 is generally thought to regulate expression of an alternative coding region with sequence homology to exon 1, whereas DMR 2 regulates exon 1 directly. The importance of alternative exon expression in cancer is increasingly being recognized as it has been used to identify breast cancer subtypes using RNAseq data from the The Cancer Genome Atlas (TCGA) Breast Invasive Carcinoma (BRCA) cohort40. DNA methylation status was also shown to affect expression of alternative exons in the sphingosine 1-phosphate (SPHK1) gene in gastroesophageal cancer41,42. Apart from these studies, the impact of alternative exon expression on tumorigenesis has not been well described, nor has the role of methylation in defining which MAP2K3 exon is preferentially expressed.
Our work extends these important findings by identifying alternative exon utilization as a potential regulatory mechanism for the MKK3/ p38 signaling axis. MKK3/p38 is a critical pathway that couples RAS-mediated growth and proliferation with inflammation (e.g. EGR1) and chromatin remodeling (e.g. SWI-SNF). More broadly, these data strongly point towards epigenetic control of RAS signaling fates downstream of NF1. We propose a schema where p38 activation in response to cellular stress and cytokine signaling inputs is reinforced in CNFs, whereas PNFs appear to signal predominantly through RAS/MEK/ERK leading to growth and proliferation (Fig. 6-schematic). Future studies are needed to better define the implications of p38 activation in neurofibromas and their various cellular constituents. It is important to note, however, that crosstalk between the MKK3/p38 and RAS/MAPK signaling pathways has not been extensively studied. Prior work suggests a potential inhibitory role for RAS/ERK in mitigating p38-mediated inflammation43. Interestingly, p38 is not typically activated in response to mitogenic stimuli, but we observed a high degree of correlation between pp38 and pERK expression. These results suggest that differential methylation may enhance crosstalk between MKK3/p38 and RAS/ERK leading to mixed signaling effects in the context of NF1 deficiency.
Proof of this concept comes from our observation that upregulated MKK3 expression, in turn, correlated with both p38 expression (p38) and activation (pp38) indicating strong epigenetic reinforcement of the MKK3/ p38 axis in CNFs (Fig. 4). Two expected results of p38 activation are activation of the MKK3/p38/EGR1 inflammatory cascade44 and changes in chromatin conformation mediated through the SWI-SNF complex family45. Relevant to the role of EGR1 in the pro-inflammatory response, it is intriguing that in our unbiased gene set enrichment analysis we identified inflammatory mediator regulation of TRP channels and phospholipase D signaling as the most significant altered signaling pathways related to DMRs (Fig. 4a), granted EGR1, itself, was not found to be differentially methylated (data not shown). Pain is a constant feature of CNFs and PNFs leading to significant morbidity. Pain signaling in nerve tumors is not well understood and difficult to manage, clinically. These data identify a potentially novel mechanism for epigenetic regulation of pain signaling in nerve tumors and a targetable signaling axis in MKK3/p38.
p38 is involved in the direct recruitment of SWI/SNF complexes to gene promoters resulting in chromatin modification and enhanced expression 46. Although the methylation states of SWI/SNF complex family member DMRs were not discordant between CNFs and PNFs, it is plausible that reinforced MKK3/p38 signaling would exert its effect through SWI/SNF leading to the observed conformational changes. Further studies are needed to determine how SWI/SNF affects expression of genes involved in growth, proliferation, and inflammation. Moreover, the effects of targeting p38 may be amplified by expected loss of recruitment of SWI/SNF complexes to target genes.