GluN3A promotes dendritic spine pruning and destabilization during postnatal development

Synaptic rearrangements during critical periods of postnatal brain development rely on the correct formation, strengthening, and elimination of synapses and associated dendritic spines to form functional networks. The correct balance of these processes is thought to be regulated by synapse-specific changes in the subunit composition of NMDA-type glutamate receptors (NMDARs). Among these, the nonconventional NMDAR subunit GluN3A has been suggested to play a role as a molecular brake in synaptic maturation. We tested here this hypothesis using confocal time-lapse imaging in rat hippocampal organotypic slices and assessed the role of GluN3A-containing NMDARs on spine dynamics. We found that overexpressing GluN3A reduced spine density over time, increased spine elimination, and decreased spine stability. The effect of GluN3A overexpression could be further enhanced by using an endocytosis-deficient GluN3A mutant and reproduced by silencing the adaptor protein PACSIN1, which prevents the endocytosis of endogenous GluN3A. Conversely, silencing of GluN3A reduced spine elimination and favored spine stability. Moreover, reexpression of GluN3A in more mature tissue reinstated an increased spine pruning and a low spine stability. Mechanistically, the decreased stability in GluN3A overexpressing neurons could be linked to a failure of plasticity-inducing protocols to selectively stabilize spines and was dependent on the ability of GluN3A to bind the postsynaptic scaffold GIT1. Together, these data provide strong evidence that GluN3A prevents the activity-dependent stabilization of synapses thereby promoting spine pruning, and suggest that GluN3A expression operates as a molecular signal for controlling the extent and timing of synapse maturation.