The spine coverage rate was higher in immature mice at P10 (Figures 5I–5K). Because the total perimeter of the spines was not different between adult and immature mice (Figure 5L), higher spine coverage rates were most probably caused by the structural differences in the PF terminals. Taken together, our results show that PFs extend axonal protrusions that cover the surface of PC spines in the immature cerebellum in vivo during the peak of PF-PC synapse formation. To examine whether PF protrusions require Cbln1-GluD2 interaction in vivo, we introduced GFP into EGL by

electroporation at P7 in cbln1-null, glud2-null, and wild-type cerebella and analyzed PFs at later postnatal days ( Figure 6A). We found that PF protrusions were reduced in cbln1-null and glud2-null mice both at P18 and P25 ( Figure 6B). Similarly, Autophagy Compound Library modest but statistically significant reduction in PF boutons was observed Pfizer Licensed Compound Library concentration in cbln1-null mice at P18 and P25 and glud2-null mice at P25 ( Figure 6C). We have previously shown by electron

microscopy that the density of PF-PC synapses is reduced by as much as 80% in adult cbln1-null mice ( Hirai et al., 2005). Thus, in the present analysis, we may have overestimated PF boutons by including boutons that belong to PF-interneuron synapses and bouton-like axonal swellings lacking postsynaptic contacts. Nevertheless, these results suggest that morphological changes in PFs require Cbln1 and GluD2 in vivo. We have

previously shown that single injection of recombinant WT-Cbln1 into adult cbln1-null mice in vivo induces significant increase in PF-PC synapses ( Ito-Ishida et al., 2008). Therefore, we next examined whether complementation of cbln1-null mice with recombinant WT-Cbln1 could also restore PF protrusions during development. Indeed, injection of WT-Cbln1 into cbln1-null mice at P14 increased the density of PF protrusions ( Figures 6D and 6E) and PF boutons ( Figures 6D and 6F) at P15. Such changes were not induced by injection of CS-Cbln1 ( Figures 6D–6F). These results indicate that PF protrusions depend on the Cbln1-GluD2 only interaction in vivo. Our results from the coculture assay suggested that interaction between Nrx and Cbln1 is required for protrusion formation (Figures 4H and 4I). To clarify the roles of Nrx in vivo, we examined the effect of altering Nrx levels on PF structure in the developing cerebellum. Overexpression of Flag tagged Nrx1β (+S4), a splice variant which binds to Cbln1, specifically increased the density of PF protrusions, while no change was observed in the density of boutons (Figures 7A–7C). The result suggests that endogenous Nrx level is not saturated and protrusive changes can be triggered by increasing Nrx. The bouton density, which should be determined by the number of PF-PC contacts, may be already too high in endogenous condition to induce any additional changes.