Distinct sequence grammars of nucleosomal and linker DNA shape mutational and methylation landscapes in cancer

Nucleosomes, which regulate DNA accessibility and impact gene expression, DNA repair, and mutagenesis, have a higher-order sequence grammar underlying nucleosome organization. This grammar's relationship to cancer-associated mutational and epigenetic processes has been unclear.

Researchers have developed DeepND, an interpretable deep learning framework that can distinguish nucleosomal from inter-nucleosomal DNA and identify sequence features related to nucleosome architecture. This framework revealed 210 sequence motifs, 2-16 base pairs long, that were significantly enriched or depleted in nucleosomal DNA, expanding upon previously known nucleosomal motifs.

Among these motifs, 44 were associated with cancer-type-specific enrichment or depletion of somatic mutations, while six were linked to differential DNA methylation levels. In lung and endometrial cancers, mutations showed a notable ~10-base pair periodic enrichment pattern across various minor-groove orientations, suggesting differential sensitivity of mutational and DNA repair processes to nucleosomal rotational positioning.

Furthermore, sequence-specific methylation patterns were identified, indicating reduced accessibility of nucleosomal DNA to DNA methyltransferases. This included differential methylation at both inward- and outward-facing CpG motifs. In summary, this study provides a sequence-resolved characterization of nucleosome-level organization and its connection to cancer-specific mutational processes and epigenetic landscapes.