La cellule de Purkinje (CP) est la seule sortie anatomique du cortex cérébelleux. Des études récentes ont montré que les récepteurs NMDA (NMDA-R) jouaient un rôle essentiel dans le Depression à long terme (DLT) à la synapse entre les fibres parallèles (FP) et les CPs. Les NMDA-Rs pourraient jouer un rôle prépondérant dans l’intégration des informations somato-sensorielles des FPs et ainsi contribuer au rôle du cervelet dans l'apprentissage moteur. Nous montrons que les NMDA-Rs sont fonctionnels et recrutés uniquement lors de patrons de décharges des FPs haute fréquences. Ces résultats étant potentiellement liés aux propriétés biophysiques des NMDA-Rs, nous avons démontré que la PLT dépend des NMDA-Rs comportant les sous unité GluN2A et que l'expression post synaptique de la plasticité s'effectuait à travers une diffusion anterograde du monoxyde d'azote (MA). De plus, nous avons confirmé et disséqué les propriétés de filtre passe haut des NMDA-Rs in vivo et in vitro.Nous avons montré que la PLT nécessitait des trains d'activité des FPs plus long que dans le cadre de la DLT, nous postulons que la quantité de MA produite est plus importante lors de l'induction de PLT. Utilisant nos données, nous avons implémenté un model mathématique de plasticité à la synapses FP-CP pouvant prédire le signe de plasticité synaptique selon les patrons d'activité rencontrés par cette synapse. / Synaptic plasticity is thought to be the cellular mechanism underlying learning and memory and has been the subject of intense experimental and theoretical research. The experimental work has led to detailed knowledge of the receptors and signalling pathways involved in the induction of different types of synaptic plasticity. In parallel, theoretical studies have built ’plasticity rules’, formal descriptions linking spike timings to changes in synaptic efficacy, such as the spike-timing-dependent plasticity (STDP) rule [Gerstner et al., 1996, Song et al., 2000]. However, these plasticity rules are generally quite abstract and their link to the underlying biophysical mechanisms is often unclear. The best known mechanisms in synaptic plasticity are linked to N-methyl-D-aspartate receptor (NMDA-R) function. NMDA-Rs are biophysical coincidence detectors of glutamate and membrane depolarization [Mayer et al., 1984, Nowak et al., 1984]. The activation of postsynaptic NMDA-Rs defines learning rules where the relative timing of pre- and post-synaptic activity is a key parameter [Debanne et al., 1994, Nevian and Sakmann, 2006, Sjostrom et al., 2003]. In the few cases where the participation of presynaptic NMDA-Rs has been proposed, these have invariably been involved in presynaptically-expressed LTD [Rodríguez-Moreno and Paulsen, 2008b, Sjostrom et al., 2003]. Cerebellar parallel fibre-Purkinje cell (PF–PC) synaptic plasticity follows non-Hebbian plasticity rules. We have previously reported that PF-PC LTD induction needs PF bursting activity (at least pairs of spikes) [Bidoret et al., 2009] and is linked to the presence of presynaptic NMDA-Rs [Casado et al., 2002b]. In this thesis, we set out to characterise the activity requirements for bidirectional synaptic plasticity in young and adult animals, and to investigate the signalling pathways involved. Surprisingly, we found that LTP induction shares many properties with LTD induction, including a similar frequency-dependence for presynaptic activity and an absolute requirement for NMDA-R activation and NO production. However, LTP requires a different source of post-synaptic calcium increase [Ly et al., 2013a]. In contrast with other synapses [Bender et al., 2006, Fino, 2010], our data indicate that both LTP and LTD share signalling mechanisms. These involve presynaptically produced NO and postsynaptic Ca rises. Supporting the notion that the frequency dependence of plasticity arises from the involvement of presynaptic NMDA-Rs, we provide the first direct evidence for Ca influx through presynaptic NMDA-Rs in PFs in young and adult animals, settling a long-lasting controversy [Bidoret et al., 2009, Casado et al., 2002a, Shin and Linden, 2005a, Wang et al., 2014a]. Based on our data, we propose a novel mechanistic plasticity rule. This deliberately parsimonious rule can be used to interpret and predict the plasticity arising from arbitrary patterns of PF and climbing fibre (CF) activity. Our results support the notion that bidirectional synaptic plasticity depends on multi-spike activity patterns in an intricate fashion [Bidoret et al., 2009, Froemke and Dan, 2002, Pfister and Gerstner, 2006, Sjöström et al., 2001].