Acinar cells synthesize and secrete large quantities of proteins, mainly digestive enzymes such as trypsin and amylase, and therefore have an exceptionally high level of translational capacity. Under pathological conditions, it has been established, using mouse models, that disruption of acinar cells leads to the development of pancreatic tumors via acino-canal metaplasia (ADM), resulting in the formation of neoplastic lesions known as pancreatic intraepithelial neoplasia (PanIN). These lesions are one of the precursors of pancreatic ductal adenocarcinoma (PDAC). While numerous transcriptional control mechanisms have been identified during pancreatic tumorigenesis, the mechanisms regulating translation are still largely unknown. In my thesis work, we focused on characterizing these mechanisms using two different approaches. My first study sought to understand how translation is controlled when tumorigenesis is initiated from pancreatic acinar cells, given that this cell type shows an exceptionally high capacity for protein synthesis, and that cancer cells generally show higher levels of protein synthesis than their non-cancerous counterparts. It is known that cancer initiation is associated in many cell types with an increase in different pathways, namely translation levels, ribosomal activity, expression of translation initiation factors (eIFs) and endoplasmic reticulum (ER) stress. In this context, the question arises as to how acinar cells, which have high levels of ER stress linked to their high levels of protein synthesis, will adapt these parameters to a context of tumor initiation. Using a mouse model that mimics the initiation of pancreatic cancer, we found that during this initiation, acinar cells can still increase their level of protein synthesis. This increase is based on an increase in ribosome biogenesis and the expression of numerous eIFs, as has been observed in other cell types. In contrast, we observed a decrease in ER stress. Given that ER stress is elevated in the presence of high activity of protein N-glycosylation enzymes, a process essential for secretion, our results suggest that this reduction is associated with the reorganization of the genetic programme present during acino-canal metaplasia, which leads to a reduction in N-glycosylation enzymes. Thus, this work provides evidence for pancreas-specific adaptation of translation regulation during tumorigenesis, which is characterized by the unexpected combination of an increase in the rate of protein synthesis and a decrease in ER stress. This highlights the diversity and adaptive capacities of protein translation mechanisms in transforming tissues. My second study focused on the roles of two RNA-binding proteins (RBP) involved in the control of translation during pancreatic tumorigenesis. We initially discovered an increase in the expression of two RBP, named G3BP1 and G3BP2, in PanIN cells compared with acinar cells. Intriguingly, this increase was accompanied by a shift in cellular localization, with G3BP1 and G3BP2 moving from a predominantly perinuclear localization in acinar cells, suggesting a potential role in nuclear transport, to a cytoplasmic localization in PanIN. To explore their role in pancreatic cancer, we genetically inactivated their expression in a pancreatic tumor initiation model using CRISPR/Cas9. Inactivation of either isoform in acinar cells does not interfere with tumor initiation. In contrast, the combined absence of G3BP1 and G3BP2 greatly reduces the formation of PanIN precancerous lesions. This reduction could be attributed to regulation of the translation of certain mRNAs. This study shows the functional redundancy of G3BP1 and G3BP2, and supports a pro-oncogenic role for both proteins in pancreatic tumorigenesis. In an independent study, we also showed that this role did not depend on the presence of G3BP1 and G3BP2 in stress granules, cytoplasmic organelles potentially involved in tumor progression.