, 1995). Spine density did not differ significantly between scrambled and Cyfip1 shRNA neurons (not shown). However, Cyfip1 knockdown robustly affected spine morphology: spines with mature phenotype (i.e., “stubby” and “mushroom”) were significantly reduced in Cyfip1-silenced neurons compared to control, whereas elongated, immature-looking spines increased in

number ( Figures 5D, 5E, S5E, and S5F). Mean head width was unchanged (not shown), but mean spine length was increased as a consequence of Cyfip1 silencing; cumulative Proteases inhibitor probability plots corroborated these results ( Figure 5F). To exclude the possibility that the phenotype might be due to off-target effects, we performed a rescue experiment by cotransfecting the sh315 (against Cyfip1 3′UTR) and the Cyfip1 Fulvestrant manufacturer WT coding sequence. The construct was able to restore normal CYFIP1 levels ( Figure S5E), and

consequently proper spine distribution ( Figures 5D, 5E, and S5F) and mean spine length ( Figure 5F). Finally, we aimed at investigating the contribution of CYFIP1-eIF4E and WRC to spine formation. Therefore, we cotransfected the CYFIP1 mutants validated above with Cyfip1 sh315 to knockdown the endogenous protein ( Figure S5E), and analyzed dendritic spine morphology. All mutants failed to restore the normal spine distribution and spine length ( Figures 5D, 5F, and S5F), indicating that both CYFIP1 complexes are equally important for proper spine formation. In conclusion, CYFIP1 deficiency alters the proper functioning of two complexes modulating critical synaptic processes, i.e., protein synthesis and actin cytoskeleton remodeling, both of these ultimately leading to defects in spine morphology. To further expand the knowledge of CYFIP1 in the brain, we studied its interactome in mouse cerebral cortex through immunoprecipitation with a specific anti-CYFIP1 antibody and tandem mass spectrometry (MS). In whole cortical lysates, we identified a total of 27 CYFIP1-associated proteins, of which 74% are

RNA-binding proteins (RBPs) (Figure 6A), comprising either known (FMRP and PABP-1) (Napoli et al., 2008 and Schenck et al., 2003) or novel (ELAV-like proteins, Caprin1 and hnRNPQ/SYNCRIP) partners; these are listed in Tables S2 and S3. The association mafosfamide of some interactors (ELAVL4, PABP1, Caprin1, SYNCRIP, FMRP, eIF4E, and DCTN1) was validated by reverse immunoprecipitation (Figures 6B and S6). To investigate whether these interactions depend upon RNA, CYFIP1 was immunoprecipitated from RNase-treated cortical lysates. Whereas the binding of CYFIP1 to PABP1, DCTN1 and eIF4E was not compromised by RNA degradation, the interaction with SYNCRIP, ELAVL4, and ELAVL1 was no longer detected (Figure 6C) implying that the CYFIP1 complexes contain both protein and RNA molecules. The association of FMRP with CYFIP1 was slightly reduced by treatment with RNase, confirming previous indications that RNAs (e.g., BC1) can strengthen this interaction (Napoli et al.