Back

SON-mediated circular RNA suppression promotes PKR signaling and radiation resistance in glioblastoma

Harvey, C. A.; Dunlap, S. K.; Mohamed, M.; Kudaravalli, E.; Elkholy, A.; Gehrs, L.; Zakakhosravi, S.; Park, B.; Min, K. W.; Murphy, J. M.; Iwaloye, O.; Osuka, S.; Lim, S.-T. S.; Willey, C. D.; Flemington, E. K.; Ahn, E.-Y. E.

2026-09-29 cancer biology
10.64898/2026.09.28.755153 bioRxiv
Show abstract

Glioblastoma (GBM) exhibits extensive loss of circular RNAs (circRNAs), yet the mechanisms driving this depletion and its functional consequences remain unclear. Here, we identify the splicing co-factor SON as a previously unrecognized suppressor of circRNA biogenesis in GBM. CircRNA sequencing revealed widespread increases in circRNA abundance following SON knockdown. Conversely, elevated SON expression in GBM, with further increases in radiation-resistant models, was associated with widespread circRNA loss, including reduced expression of circUSP1 and circSUCO. SON knockdown increased circRNA abundance and attenuated activation of the dsRNA-responsive kinase PKR and downstream NF{kappa}B signaling, whereas radiation-resistant GBM cells exhibited enhanced PKR/NF{kappa}B activation that was attenuated by SON depletion. Importantly, overexpression of the SON-suppressed circRNAs circUSP1 and circSUCO reduced PKR activation and selectively decreased clonogenic survival following irradiation in radiation-resistant GBM cells, with little effect on parental cells. Together, these findings identify a previously unrecognized SON/circRNA/PKR regulatory pathway linking RNA processing to acquired radiation resistance in GBM and highlight specific PKR-inhibitory circRNAs as potential therapeutic tools for resensitizing radiation-resistant GBM to radiotherapy.

Matching journals

The top 10 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.