Data from Transcription Factors Involved in Tumorigenesis Are Over-Represented in Mutated Active DNA-Binding Sites in Neuroblastoma

The contribution of coding mutations to oncogenesis has been largely clarified, whereas little is known about somatic mutations in noncoding DNA and their role in driving tumors remains controversial. Here, we used an alternative approach to interpret the functional significance of noncoding somatic mutations in promoting tumorigenesis. Noncoding somatic mutations of 151 neuroblastomas were integrated with ENCODE data to locate somatic mutations in regulatory elements specifically active in neuroblastoma cells, nonspecifically active in neuroblastoma cells, and nonactive. Within these types of elements, transcription factors (TF) were identified whose binding sites were enriched or depleted in mutations. For these TFs, a gene expression signature was built to assess their implication in neuroblastoma. DNA- and RNA-sequencing data were integrated to assess the effects of those mutations on mRNA levels. The pathogenicity of mutations was significantly higher in transcription factor binding site (TFBS) of regulatory elements specifically active in neuroblastoma cells, as compared with the others. Within these elements, there were 18 over-represented TFs involved mainly in cell-cycle phase transitions and 15 under-represented TFs primarily regulating cell differentiation. A gene expression signature based on over-represented TFs correlated with poor survival and unfavorable prognostic markers. Moreover, recurrent mutations in TFBS of over-represented TFs such as EZH2 affected MCF2L and ADP-ribosylhydrolase like 1 expression, among the others. We propose a novel approach to study the involvement of regulatory variants in neuroblastoma that could be extended to other cancers and provide further evidence that alterations of gene expression may have relevant effects in neuroblastoma development.

These findings propose a novel approach to study regulatory variants in neuroblastoma and suggest that noncoding somatic mutations have relevant implications in neuroblastoma development.