Back

SHLD2 loss is a synthetic vulnerability to Polθ inhibition combined with radiotherapy

Rodriguez-Berriguete, G.; Thambiayah, P.; Cicconi, A.; Machado, N.; Gotorbe, C.; Nderitu, D.; Cheng, W.-C.; Boursier, M. L.; Cerutti, A.; Grinkevich, V.; Hill, B. R.; Koler, K.; Langdon, S. A.; Majithiya, J. B.; Menon, S.; Moore, S.; Neves, J.; Palmer-Deverill, N. M.; Rajendra, E.; Roy-Luzarraga, M.; Thapa, A.; Heald, R. A.; Smith, G. C. M.; Robinson, H. M. R.; Ranzani, M.; Higgins, G. S.

2025-07-07 cancer biology
10.1101/2025.07.04.662969 bioRxiv
Show abstract

DNA polymerase theta (Pol{theta}) plays a crucial role in the repair of DNA double-strand breaks (DSBs) by microhomology-mediated end joining (MMEJ). We previously demonstrated that Pol{theta} inhibition (Pol{theta}i) is an effective and well-tolerated approach to sensitise tumours to radiotherapy (RT). Here, we profiled 54 cancer cell lines and found that Pol{theta}i induces significant radiosensitisation in most models, though with marked variability not explained by indicators of Pol{theta} activity. To pinpoint molecular determinants of radiosensitisation by Pol{theta}i, we performed a CRISPR knockout screen which revealed loss of the TP53BP1/Shieldin pathway component SHLD2 (FAM35A) as a vulnerability to Pol{theta}i combined with RT. We demonstrated that SHLD2 loss not only increases sensitivity to RT alone, but also enhances the radiosensitising effect of Pol{theta}i, both in vitro and in vivo. Importantly, we found that SHLD2 is deleted in a subset of human prostate cancers, often co-occurring with PTEN loss, an adverse prognostic factor. Furthermore, we show that SHLD2-deficient cancer cells are more reliant on Pol{theta} to prevent DSB accumulation and chromosomal instability. In summary, we discovered SHLD2 loss as a novel collateral vulnerability that can be exploited through combined treatment with Pol{theta}i and RT.

Published in Science Advances (predicted rank #6) · training set

Matching journals

The top 3 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.