Epigenome erosion drives neural crest-like phenotypic mimicry in triple-negative breast cancer and other SOX10+ malignancies

AO_SCPLOWBSTRACTC_SCPLOWO_ST_ABSBackgroundC_ST_ABSIntratumoural heterogeneity is a poor prognostic feature in triple-negative breast cancer (TNBC) and other high-grade malignancies. It is caused by genomic instability and phenotypic plasticity, but how these features co-evolve during tumour development remains unclear. SOX10 is a transcription factor, neural crest stem cell (NCSC) specifier and candidate mediator of cancer-associated phenotypic plasticity. MethodsUsing immunophenotyping, we investigated the expression of SOX10 in normal human breast tissue and breast cancer (n=21 cosmetic breast reduction and 1,860 tumour samples with clinical annotation). We then defined the context and evolution of its expression in TNBC compared to 21 other malignancies using systems-level transcriptomics. ResultsSOX10 was detected in nuclei of normal mammary luminal progenitor cells, the histogenic origin of most TNBCs. In breast cancer, nuclear SOX10 predicted poor outcome amongst cross-sectional (log-rank p=0.0015, hazard ratio 2.02, n=224) and metaplastic (log-rank p=0.04, n=66) TNBCs. Systems-level transcriptional network analysis identified a core module in SOX10s normal mammary epithelial transcription program that is rewired to NCSC genes in TNBC. Reprogramming was proportional to DNA damage and genome-wide promoter hypomethylation, particularly at CpG island shores. Using a novel network analysis pipeline, we found that NCSC-like transcriptional reprogramming is also strongly associated with promoter hypomethylation in other SOX10+ malignancies: glioma and melanoma. ConclusionsWe propose that cancer-associated genome hypomethylation simulates the open chromatin landscape of more primitive cell states, and that on this relatively unrestricted background, SOX10 recreates its ancestral gene regulatory circuits by default. These findings provide new insights about the basis of intratumoural heterogeneity and resurrection of developmental phenotypes in cancer; and highlight the potential for therapeutics that limit chromatin remodelling.