Résumé
The unfolded protein response (UPR) is a stress-adaptation pathway and therapeutic target in cancer, yet its pro-survival versus pro-death outcome is difficult to predict because the three ER sensors, PERK, IRE1α, and ATF6, are highly interconnected. Transcriptomic analyses identified sensor-specific gene signatures associated with patient survival across malignancies, and indicated that low IRE1α activity (low XBP1 signature or higher expression of RIDD targets) correlates with improved outcome. We developed SNUPR (single nuclei analysis of the unfolded protein response), an accessible flow cytometry approach that profiles all three branches in nuclear suspensions. SNUPR reveals marked heterogeneity of UPR activation across cancer cell lines that cannot be inferred from sensor expression. This heterogeneity is derived from differences in the strength and duration of PERK-mediated translational inhibition, which gates downstream translation-dependent IRE1α and ATF6 transcriptional programs. Finally, in multiple myeloma, we show that bortezomib-tolerant cells depend on IRE1α activity for survival, linking UPR state to proteasome-inhibitor resistance and positioning SNUPR to guide branch-selective targeting.