Loss of CYLD was linked to more unstable DNA and a shift away from homologous recombination in cancer-cell experiments, and CYLD-deficient cells showed greater sensitivity to the PARP inhibitor Olaparib. The finding brings together two strands of evidence in the study: analyses tied chromosome 16q loss to lower CYLD messenger RNA and higher tumor mutation burden, wGII and HRD scores, while cell experiments found more chromosomal breaks, more ionizing-radiation-induced DNA damage and persistent γ-H2AX, a laboratory signal of DNA damage.
The paper's central question was whether CYLD loss associated with chromosome 16q deletion alters TIRR-53BP1-dependent repair of DNA double-strand breaks and sensitivity to Olaparib. TIRR and 53BP1 became the focus because the proposed model placed them between CYLD loss and the cell's choice of repair route. That model remains a laboratory proposal, not a clinical finding.
The clue was TIRR
Researchers tested the mechanism in PC-3 prostate cancer cells and UM-UC-3 bladder cancer cells, with HEK293T cells used for molecular assays. They depleted or knocked out CYLD, reintroduced wild-type CYLD or catalytically inactive mutants, and manipulated TIRR. These comparisons let the team ask whether the enzyme activity of CYLD, rather than its presence alone, tracked with TIRR stability and DNA-repair behavior.
CYLD interacted with TIRR, and CYLD knockdown was accompanied by lower TIRR protein without a change in TIRR messenger RNA. The reported TIRR protein half-life was also shorter after CYLD loss, suggesting that the difference lay in protein stability rather than RNA production. Half-life here refers to how long the protein remained stable in the assay.
Biochemical assays supplied a possible mechanism. They identified predominantly K48-linked TIRR polyubiquitin, meaning chains of the ubiquitin tag attached to TIRR, and found that wild-type CYLD reduced that ubiquitination. The M2 domain mediated CYLD-TIRR binding, while inactive C601A and C601S mutants did not reduce TIRR ubiquitination. Together, those results implicated CYLD's catalytic activity in the TIRR-stability finding.
A different repair route
When CYLD or TIRR was depleted, 53BP1 accumulated more strongly and stayed longer at DNA damage sites. Researchers measured this as 53BP1 foci, or visible spots of the protein in the cell. Restoring TIRR or adding back wild-type CYLD reduced the excess signal, but the catalytically inactive C601S form did not.
The repair consequences appeared in reporter assays that separate HR from NHEJ, two routes tested by the study. PC-3 cells received the reporter constructs and the I-SceI enzyme, and the signal was read by flow cytometry 48 hours after transfection, normalized against a positive-control GFP signal. CYLD-depleted cells showed less HR and more NHEJ, as well as reduced BRCA1, RAD51 and RPA2 foci. Wild-type CYLD rescued the reporter defect, whereas C601S did not.
Drug response in cells and mice
The shift in repair was accompanied by a drug response. CYLD-depleted PC-3 and UM-UC-3 cells had lower Olaparib IC50 values, a laboratory measure of the drug concentration needed for a midpoint inhibitory effect, and formed fewer or smaller colonies. Reconstituting wild-type CYLD or TIRR attenuated the cellular response to Olaparib, while C601S did not. The result identifies a vulnerability in these models, not evidence that patients would respond the same way.
The animal experiment extended the comparison to PC-3 xenografts in male SCID mice. It compared control and CYLD-knockdown tumors under vehicle or Olaparib treatment, with nine mice reported in each group. CYLD-deficient xenografts showed stronger Olaparib-associated growth inhibition, and the reported treatment did not affect mouse body weight.
Why the finding remains preliminary
Human tissue findings offered a limited check on the laboratory pattern. In one immunohistochemistry analysis, which measures protein signals in tissue, the study examined 28 prostate cancer specimens and 27 bladder cancer specimens. CYLD and γ-H2AX showed an inverse relationship, meaning higher CYLD was reported alongside less of the DNA-damage signal. Because this was an association in tissue samples, it cannot establish that CYLD levels caused the difference.
Public-dataset analyses also produced signals related to radiation response. In the TCGA HNSC cohort, CYLD expression positively correlated with a radiosensitivity index, with a Pearson correlation of 0.26 and p < 0.001. In a radiotherapy-treated LGG cohort, the survival comparison between CYLD-low and CYLD-high groups was reported as significant, with a two-sided log-rank p < 0.05. These were retrospective subgroup associations, not randomized treatment effects, and they do not establish that CYLD predicts radiotherapy benefit.
The evidence therefore remains preclinical. The mechanistic work centered on the tested cell lines, and the xenograft work used one PC-3 model in immunodeficient male SCID mice. The study did not include a clinical dataset pairing CYLD or chromosome 16q status with PARP-inhibitor response, and it does not provide exact IC50 values, tumor-volume values or survival estimates. Study data are available from the corresponding author on request but are not publicly available because of privacy or ethical restrictions.
The authors propose the CYLD-TIRR-53BP1 axis as a biomarker framework for HR deficiency and PARP-inhibitor response. The experiments support mechanistic relationships in the tested models, but they do not show that CYLD loss predicts PARP-inhibitor benefit in patients, that CYLD is the only relevant gene within chromosome 16q, or that the mechanism extends across tumor types. Clinical datasets with paired CYLD or 16q status and PARP-inhibitor outcomes will be needed to test whether the proposed biomarker holds up.
Paper data and sources
Original title: Loss of CYLD on Chromosome 16q Impairs Homologous Recombination and Genomic Stability Through TIRR Degradation.
Authors: Mingming Lu, Jialu Kang, Qi Ye et al.
Journal/Repository: Advanced science (Weinheim, Baden-Wurttemberg, Germany)
Status: Peer-reviewed
First online: 2026-08-21
DOI: 10.1002/advs.77298
Original paper