Translatome-based classification reveals a dual metabolic dependency of a new tumor subtype of pancreatic cancer

ABSTRACT Molecular profiling of Pancreatic Ductal Adenocarcinoma (PDA), based on transcriptomic analyses, identifies two main prognostic subtypes (basal-like and classical), but does not allow personalized first-line treatment. To date, tumors have not been profiled based on protein synthesis rates, yet the step of mRNA translation is highly deregulated in both PDA cancer cells and their microenvironment. Using a collection of twenty-seven pancreatic Patient-Derived Xenografts (PDX), we performed genome-wide analysis of translated mRNA (translatome). Unsupervised bioinformatics analysis revealed a new tumor subtype harboring a low protein synthesis rate, but associated with a robust translation of mRNAs encoding effectors of the integrated stress response (ISR), including the transcription factor ATF4. Functional characterization of the “ISR-activated” human cancer cells revealed a high resistance to drugs, low autophagic capacities, and importantly, metabolic impairments in the serine synthesis and transsulfuration pathways. Overall, our study highlights the strength of translatomic profiling on PDA, which here revealed an unforeseen drug-resistant cancer cell phenotype, whose auxotrophy to both serine and cysteine may be amenable to targeted therapy.