Work

TFIIH protein imbalance does not predict exome mutation burden or cisplatin sensitivity across cancer models

Year: 2026
Type: preprint
Abstract: Background Protein abundance can diverge from transcript levels, potentially masking DNA-repair defects. We assessed whether transcription factor IIH (TFIIH) core-subunit imbalance provides a biomarker of nucleotide excision repair-associated phenotypes in cancer. Methods In nine CPTAC tumor cohorts, core-subunit protein residuals were estimated after adjustment for cognate RNA, copy number and tumor purity, and benchmarked against eight unrelated complexes. Seven cohorts (698 tumors) contributed exome single-nucleotide variant (SNV) burden. A parallel CCLE/DepMap analysis scored 361 cell lines; cisplatin-response data were available for 287 GDSC1 and 243 GDSC2 lines. Models adjusted for cell lineage, proliferation, mean TFIIH RNA and MLH1 expression. Published functional repair classifications were matched in an exploratory five-line pilot. Results TFIIH residual co-regulation was lower than the control-complex mean in all nine tumor cohorts (mean Spearman correlation 0.379 vs 0.487; paired p = 0.0039). Imbalance did not associate with SNV burden (pooled standardized β = −0.009, 95% CI − 0.085 to 0.067) or improve held-out prediction (ΔR² = −0.0064). Cisplatin association estimates were − 0.026 (95% CI − 0.149 to 0.097) in GDSC1 and + 0.007 (95% CI − 0.147 to 0.162) in GDSC2. No association survived false-discovery control across 606 exploratory drug tests; SLFN11 expression showed the expected cisplatin association. The functional pilot was too small for confirmatory validation. Conclusions Bulk TFIIH residual imbalance did not provide a reproducible biomarker signal for the indirect endpoints examined. Direct nucleotide excision repair assays are needed to establish whether other forms of TFIIH disruption track functional repair deficiency.
Language: English
Open Access status: green
OpenAlex ID: W7236077764
Cites: 19
Cited by:
Related to: 10