On-target WRN mutations drive resistance to WRN inhibitors in TA-repeat-expanded MSI cancers

Werner helicase inhibitors (WRNi) are in clinical development for microsatellite-unstable (MSI) tumours with defective DNA mismatch repair. A key genetic feature underlying this vulnerability is the expansion of TA dinucleotide repeats, which renders cells selectively reliant on WRN function. Despite this clear synthetic lethality, routes of tumour cell evolution under the selective pressure of WRN genetic loss or inhibition remain entirely unknown. Here, we investigate how cancer cell evolution shapes response to WRN inhibition and informs resistance mechanisms. Genome-wide CRISPR screens combined with complete WRN knockout revealed no bypass pathways, underscoring WRN's essential and non-redundant function in MSI cells. Pharmacogenomic profiling identified modulators of WRNi sensitivity, including the MRN complex, ATM, and SMARCAL1, but none rescued viability, confirming the robustness of the WRN-MSI synthetic-lethal interaction. WRN-MSI synthetic lethality remains robust across diverse models, including patient-derived organoids and immunotherapy-refractory tumours. Using semi-saturation mutagenesis, prolonged drug exposure across cell lines from diverse tissue lineages, and in vivo validation, we identified a spectrum of recurrent on-target WRN mutations driving acquired resistance to multiple WRNi. This was supported by in vivo evidence of acquired WRN mutations in both cell line-derived and patient-derived xenografts treated with HRO761, providing direct preclinical validation of this resistance mechanism. TA repeats, inferred by short-read sequencing and measured by long-read sequencing, remained stable under WRNi treatment and were unlinked to resistance. Resistant clones showed no evidence of MMR restoration or other bypass mechanisms. Some WRN mutations conferred broad cross-resistance, whereas others preserved sensitivity to alternative WRNi; for example, I852F retained sensitivity to VVD-133214 but not to HRO761, whereas F730L conferred pan-resistance to both yet remained vulnerable to GSK4418959. All three compounds are clinically advanced WRN inhibitors currently progressing through Phase I trials. Some differential sensitivity and resistance patterns were also validated in patient-derived organoids that acquired secondary resistance to HRO761 in vitro. Finally, resistant clones remained vulnerable to rational strategies: combining WRNi with irinotecan suppressed resistant outgrowth, while ATR inhibitors and orthogonal WRNi offer additional routes to extend response. These findings establish on-target WRN mutation as the dominant mechanism of resistance to WRN inhibitors and define a framework for resistance-informed clinical trial design. They also outline actionable strategies to detect and overcome resistance, including ctDNA-based molecular monitoring and rational combination therapies to extend clinical benefit.