Back

HOXDeRNA activates a cancerous transcription program and super-enhancers genome-wide

Deforzh, E.; Kharel, P.; Karelin, A.; Ivanov, P.; Krichevsky, A. M.

2023-06-30 molecular biology
10.1101/2023.06.30.547275 bioRxiv
Show abstract

BackgroundThe origin and genesis of highly malignant and heterogenous glioblastoma brain tumors remain unknown. We previously identified an enhancer-associated long non-coding RNA, LINC01116 (named HOXDeRNA here), that is absent in the normal brain but is commonly expressed in malignant glioma. HOXDeRNA has a unique capacity to transform human astrocytes into glioma-like cells. This work aimed to investigate molecular events underlying the genome-wide function of this lncRNA in glial cell fate and transformation. ResultsUsing a combination of RNA-Seq, ChIRP-Seq, and ChIP-Seq, we now demonstrate that HOXDeRNA binds in trans to the promoters of genes encoding 44 glioma-specific transcription factors distributed throughout the genome and derepresses them by removing the Polycomb repressive complex 2 (PRC2). Among the activated transcription factors are the core neurodevelopmental regulators SOX2, OLIG2, POU3F2, and SALL2. This process requires an RNA quadruplex structure of HOXDeRNA that interacts with EZH2. Moreover, HOXDeRNA-induced astrocyte transformation is accompanied by the activation of multiple oncogenes such as EGFR, PDGFR, BRAF, and miR-21, and glioma-specific super-enhancers enriched for binding sites of glioma master transcription factors SOX2 and OLIG2. ConclusionsOur results demonstrate that HOXDeRNA overrides PRC2 repression of glioma core regulatory circuitry with RNA quadruplex structure. These findings help reconstruct the sequence of events underlying the process of astrocyte transformation and suggest a driving role for HOXDeRNA and a unifying RNA-dependent mechanism of gliomagenesis.

Matching journals

The top 12 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.