RNP granule remodelling in response to neuronal activity

One of the most fascinating – and still open – questions in neuroscience is how neuronal cells can form, store and then recall memories. Previous work has shown that Long-term memory (LTM) formation requires de novo protein synthesis, involving not only translation of newly transcribed RNAs, but also local, experience-induced translation of quiescent mRNAs carried and stored at synapses. For their transport and translational control, mRNAs are packaged with regulatory RNA binding proteins (RBPs), mainly translational repressors, into ribonucleoprotein (RNP) granules. To date, how neuronal RNP granules are remodelled in response to neuronal activity to relieve translation repression of mRNAs is unclear. Furthermore, the functional impact of such a remodelling in the establishment of long-term memories remains to be demonstrated in vivo. The objective of my PhD was to 1) investigate the in vivo mechanisms underlying activity-dependent remodelling of neuronal RNP granules; 2) test the hypothesis that RNPs could be involved in LTM-underlying mechanisms by regulating gene expression. To this end, I used as paradigm RNPs containing the conserved RBP Imp in Drosophila. First, I studied the impact of neuronal activity on Imp RNP properties by treating Drosophila brain explants with either KCl or the tyramine neuropeptide. In both cases, a disassembly of Imp RNPs was observed, characterized by a loss of both Imp and other RNP-component granular patterns, and a de-clustering of RNP-associated mRNA molecules. RNP disassembly could be reverted upon Tyramine withdrawal and was not observed in hyperpolarized neurons. Furthermore, my data suggest that RNP-disassembly is linked to increased translation of associated mRNAs, consistent with a model in which activity-induced RNP remodelling would lead to translational de-repression. Second, I investigated the mechanisms controlling RNP remodelling. A candidate regulator was CamkII, a conserved Ca2+ -activated kinase identified as a partner of Imp in an IP-Mass Spectrometry analysis. During my PhD, I could validate the Imp-CamkII interaction and showed that it is not mediated by RNA but depends on CamkII activity. Furthermore, I showed that inactivating CamkII function prevents the disassembly of Imp RNPs observed upon neuronal activation of brain explants, suggesting that CamkII may be involved in the activity-dependent remodelling of Imp RNP granules. These results are particularly interesting in the context of establishment of LTM, as CamkII has long been recognized as essential for LTM. Moreover, we recently showed in Drosophila that interfering with Imp function in a population of CNS neurons involved in learning and memory – the Mushroom Body γ neurons -, dramatically impairs LTM and that this effect relies on Imp C-terminal Prion-like domain, a domain known to be involved in RNP homeostasis. Altogether, my thesis work suggests a model where CamkII-dependent remodelling of Imp RNPs in response to neuronal activation might underlie LTM formation in vivo.