The cytoplasmic polyadenylation element-binding protein 3 (CPEB3), is an RNA-binding protein which in its soluble state is localized in membraneless neuronal RNA granules keeping target mRNAs in a repressed state. The stimulus-dependent aggregation of CPEB3 activates target mRNAs translation, a central event for the maintenance of long-term memory-related synaptic plasticity in mammals. To date, the molecular determinants that govern both connected events remain unclear. Here, to gain insight into these processes, the biophysical properties of the human CPEB3 (hCPEB3) are characterized. We found that hCPEB3 homotypic condensation is mainly driven by hydrophobic interactions and occurs under physiological conditions. Moreover, hCPEB3 biomolecular condensates are dynamic inside living cells, whose localization and stabilization are mediated by its RNA-recognition domains.In contrast, the hCPEB3 polar N-terminal region is crucial for hCPEB3 amyloid-like aggregation in vitro, which is disrupted by the polyglutamine binding peptide 1 (QBP1), Aβ42 seeds and Hsp70, highlighting the importance of the Q4RQ4 tract as well as the hydrophobic residues for hCPEB3 functional aggregation. Based on these findings, we postulate a model for hCPEB3's role in memory persistence that advances a rather sophisticated control for hCPEB3 condensate dissociation and amyloid-like formation to achieve its physiological function.hCPEB3's demixing and amyloidogenesis Highlights • hCPEB3 forms toxic intermediates that persist longer than in other functional amyloids.• RNA-recognition domains stabilize hCPEB3 granule formation and dynamics.• Different segments within hCPEB3 promote amyloidogenesis and liquid demixing.• hCPEB3 amyloid formation requires both hydrophobic and polyQ segments. Graphical AbstracthCPEB3's demixing and amyloidogenesis

Post-traumatic stress disorder (PTSD) is a mental health disorder triggered by the exposure to a traumatic event that manifests with anguish, intrusive memories and negative mood changes. So far, there is no efficient treatment for PTSD other than symptomatic palliative care. Based on the implication of the functional amyloid cytoplasmic polyadenylation element binding protein-3 (CPEB3) in the consolidation of memory, we propose its active amyloid state as a possible therapeutic target by blocking the consolidation of traumatic memories through polyglutamine binding peptide 1 (QBP1), an inhibitor of the amyloid oligomerization previously investigated in Drosophila.To test this idea in mammals, here we have developed a transgenic mouse that constitutively expresses QBP1 peptide. We first assessed the innocuousness of this peptide for the normal development of the animal, which also showed normal locomotor activity and anxiety. By performing a battery of standard memory paradigms, we then showed that hippocampaldependent and aversive memories were impaired in the QBP1 mice. Furthermore, protein expression in the hippocampi of experienced mice showed that QBP1 mice do not increase their levels of amyloid oligomerization, evincing the blockade of the CPEB3 protein in its inactive state.The ability of QBP1 to block aversive memories in mice represents the proof of concept of a novel pharmacological approach for prophylaxis and therapy of acute stress and post-traumatic stress disorders.