In order to identify new binding partners for GluN2A, we performed a yeast two-hybrid screening using the C-terminal domain (aa 839–1461, without the aa 1462–1464 PDZ-binding sequence) as bait. We obtained a number of positive clones including RNF10 (Seki et al., 2000; Hoshikawa et al., 2008). In Schwann cells, RNF10 has a function in the transcriptional regulation of myelin formation (Hoshikawa et al., 2008). However, very little is known about the neuronal function of this protein (Seki et al., 2000; Lin et al., 2005; Hoshikawa et al., 2008; Malik et al., 2013). Using specific glial (GFAP) and neuronal (MAP2) markers, we confirmed that RNF10 is expressed in both glia (Figure 1A) and neurons (Figure 1B). Interestingly, in neurons RNF10 displayed a nuclear and somatodendritic distribution (Figure 1B). Moreover, transfected GFP-RNF10 displayed a prominent co-localization with PSD-95 and GluN2A at the dendritic spines of hippocampal neurons (Figure 1C). Similarly, analysis of endogenous RNF10 in DIV14 primary hippocampal neurons showed clustered RNF10 immunolabeling along dendrites and most puncta co-localized with GluN2A (Figure 1D, left panels) and PSD-95 (Figure 1D, right panels). PSDs from the rat hippocampus were purified to confirm the subcellular distribution of RNF10 by a biochemical approach. Subcellular fractionation demonstrated that RNF10 is associated with synaptic fractions and that it is prominently present in PSD fractions (Figure 1E). Finally, immunofluorescence analysis of the CA1 region of the adult rat hippocampus revealed the presence of an intense signal for RNF10 in the soma and nuclei together with a punctate staining along MAP2-positive dendrites (Figure 1F).

Figure 1

Download asset Open asset

Interestingly, GluN2A silencing in primary hippocampal neurons (Figure 1G) induced a significant decrease in RNF10 synaptic levels, as indicated by the reduction of RNF10 co-localization with PSD-95 (Figure 1H). Notably, the remaining dendritic RNF10 in shGluN2A neurons co-localized with the surface GluN2A pool unaffected by the knock down (Figure 1I). However, no modification of RNF10 nuclear level was observed following GluN2A silencing (data not shown; n=30; p=0.5491; shGluN2A vs scramble; unpaired Student’s t-test).

Different experimental approaches were used to substantiate the yeast two-hybrid data and to confirm the interaction between RNF10 and GluN2A. Co-immunoprecipitation (co-i.p.) studies performed from hippocampal P2 crude membrane fractions indicated a specific interaction of RNF10 with GluN2A but not with GluN2B subunit of the NMDARs (Figure 2A). No signal for GluN2A or GluN2B was obtained by using anti-synaptophysin as an irrelevant antibody or in the absence of antibody in the co-i.p. assay (Figure 2A). To further validate these findings we performed similar experiments with the GluN2B-associated synapse-to-nucleus messenger Jacob (Dieterich et al., 2008; Karpova et al., 2013). Indeed, the affinity-purified pan-Jacob antibody preferentially co-i.p. GluN2B from rat brain homogenate. Only a very faint band of GluN2A was detected in the complex with Jacob (Figure 2B), which might potentially represent synaptic tri-heteromeric GluN1/GluN2A/GluN2B NMDARs.

Figure 2

Download asset Open asset

In addition, cell lysates from COS-7 cells transfected with HA-GluN1 and GFP-GluN2A or GFP-GluN2B constructs were immunoprecipitated with anti-RNF10 and immunoblotted for GFP. The results confirmed that GluN2A interacts with RNF10, whereas GluN2B failed to associate with RNF10 (Figure 2C). To corroborate these results, we transfected HA-GluN1 and either GFP-GluN2A or GFP-GluN2B constructs in heterologous cells (COS-7) endogenously expressing RNF10. Immunofluorescence studies revealed the presence of RNF10 aggregates with a high co-localization degree with GluN2A but not with GluN2B (Figure 2D; 73.8% ± 2.9% GluN2A/RNF10 vs 31.3.% ± 1.7% GluN2B/RNF10; p<0.001, n=15; unpaired Student’s t-test). Finally, RNF10 clustering with GluN2A at synapses was confirmed by proximity ligation assay (PLA). As shown in Figure 2E, a large number of PLA signals were detected when the two antibodies RNF10 and GluN2A were used indicating these two proteins are in close proximity (<40 nm) to each other. As expected, PLA signals were distributed along MAP2-positive dendrites (Figure 2E, left panels) and at the top of GFP-positive dendritic spines (Figure 2E, right panels). For control experiments, only RNF10 primary antibody was used and no PLA signal was generated (Figure 2E).

Notably, transfection of COS-7 cells with GluN2A but not GluN2B induces a highly significant redistribution of endogenous RNF10 from the nucleus to the cytoplasm, thus suggesting that GluN2A traps RNF10 outside the nucleus (Figure 2F–G). Overall, these results demonstrate that RNF10 is a component of the excitatory PSD and that it is specifically associated with GluN2A-containing NMDARs.

We next mapped the binding interface of the interaction between RNF10 and GluN2A using truncation constructs in pull-down and co-i.p. experiments. The pull-down assay performed from adult rat brain tissue using glutathione S-transferase (GST)-GluN2A C-terminal domain fusion proteins (Figure 3A) revealed that RNF10 interacts with the GluN2A cytoplasmic tail (CT; aa 839–1464) but failed to bind the distal part of the GluN2A C-terminus (aa 1049–1464) (Figure 3B, upper panel). In addition, GluN2A(839–991) failed to interact with RNF10 (Figure 3B, lower panel), thus indicating GluN2A(991–1049) is needed for the formation of the GluN2A/RNF10 complex.

Figure 3

Download asset Open asset

To strengthen these results, GFP-GluN2A (1–1049) truncation mutant (bearing a stop codon at aa 1049) was transfected in COS-7 cells to evaluate its capability to interact with endogenous RNF10 (Figure 3C,D). Immunofluorescence assay revealed a high GFP-GluN2A(1–1049)/RNF10 clustering leading also to RNF10 redistribution from the nucleus to the cytoplasm (Figure 3C; cytoplasm/nucleus ratio 0.98 ± 0.04 vs 0.26 ± 0.02 GFP-GluN2A(1–1049) vs untransfected, p<0.001, n=10; unpaired Student’s t-test), similarly to what has been described above for GFP-GluN2A construct (see Figure 2). Accordingly, RNF10 co-immunoprecipitated with both GFP-GluN2A and GFP-GluN2A (1–1049) from COS-7 lysates (Figure 3D), confirming that RNF10 interacts with the GluN2A juxta-membrane aa 839–1049 region. Interestingly, GluN2A(991–1029) domain previously shown to be responsible for a calcium-dependent binding of the NMDAR subunit with Calmodulin (CaM; Bajaj et al., 2014). Based on this consideration, we first evaluated whether calcium (Ca²⁺) could modulate GluN2A/RNF10 interaction. Using a pull-down assay we found that the presence of free Ca²⁺ (2 mM) in the buffer significantly reduced GluN2A binding to GST-RNF10 FL (Figure 3E; -37.8% ± 6.2%; p<0.01). Similarly presence of free Ca²⁺ (2 mM) in the buffer significantly reduced RNF10 binding to GST-GluN2A(839–1464) (Figure 3F; -65.4 ± 8.8%; p<0.05). As previously reported (Bajaj et al., 2014), Ca²⁺ induced also the binding of CaM to the GluN2A fusion protein (Figure 3F). In addition, co-incubation with exogenous CaM completely prevented the interaction between RNF10 and GluN2A in the pull-down assay (Figure 3G) only in the presence of Ca²⁺ in the buffer (Bajaj et al., 2014). Similarly, a co-i.p. assay from COS-7 cells transfected with GFP, GFP-GluN2A or GFP-GluN2B confirmed the capability of Ca²⁺/CaM to disrupt RNF10/GluN2A complex (Figure 3H).

The RNF10 protein contains a binding sequence for the transcription factor Mesenchyme Homeobox 2 (Meox2; Meox2 Binding Domain, MBD, aa 101–185; Lin et al., 2005), a Ringer Finger Domain (RFD, aa 225–270) and two putative nuclear localization sequences (NLS1, aa 591–599 and NLS2, aa 784–791). Several Myc-RNF10 mutants were prepared and co-transfected with GluN2A in COS-7 cells in order to identify the RNF10 domain responsible for the interaction with the NMDAR subunit. The RNF10(221–802) construct failed to interact with GluN2A as demonstrated by both co-i.p. (Figure 3I) and co-localization (Figure 3K,L) assays when compared to RNF10 full length (FL). Conversely, all RNF10 truncation mutants bearing the RNF10(1–221) domain co-immunoprecipitated with GFP-GluN2A (Figure 3J). Finally, pull-down assay performed from adult rat brain tissue using GST-RNF10 fusion proteins confirmed that the RNF10 N-terminal region 1–221 is crucial for the binding to GluN2A (Figure 3M).

RNF10 is a member of the Ring Finger Protein family, which has been generally implicated in development, transcriptional regulation, signal transduction, DNA repair and oncogenesis (Saurin et al., 1996). However, the neuronal function of RNF10 is still unknown (Lin et al., 2005; Malik et al., 2013). To understand the role of RNF10 in neurons, we silenced RNF10 expression by using a short hairpin (sh) RNF10 knock-down or a scramble sequence (as control) in primary hippocampal cultures. We tested three different sequences of RNF10 shRNA (shRNF10; see Materials and methods); shRNF10 that led to the highest level of RNF10 downregulation (>90%; TRCN0000041128) was selected and used in all experiments. Confocal imaging of hippocampal neurons transfected with shRNF10 or scramble plasmids demonstrated a significant effect of RNF10 knock-down on dendritic spine morphology (Figure 4A–E). In particular, RNF10 silencing produced a significant reduction in dendritic spine density (Figure 4B) without any effect on dendritic spine length (Figure 4C) or dendritic spine head width (Figure 4D). For a more detailed morphological analysis, dendritic spines were categorized according to their shape (mushroom, thin and stubby) using a validated classification method (Bourne and Harris, 2008). However, no effect of shRNF10 in the proportion of dendritic spine subtypes was observed (Figure 4E). Notably, the effect of RNF10 silencing on dendritic spine density was fully rescued by co-expressing a wild-type human variant of RNF10 resistant to shRNA (see Materials and methods; flag-RNF10; Figure 4A–E). Interestingly, overexpression of RNF10 per se had no effect on dendritic spine morphology (data not shown). We then verified whether the loss of spines following RNF10 protein knock-down was correlated with an altered expression of the main components of the excitatory synapse. To this end, we infected primary hippocampal neurons (Figure 4F, left panels) and organotypic hippocampal slices (Figure 4F, right panels) with pLKO-shRNF10 lentivirus or scramble sequence as a control. As expected, RNF10 silencing produced a significant reduction of RNF10 protein level compared to dissociated neurons and slices treated with scramble construct (Figure 4F). Most importantly, RNF10 silencing resulted in a significant decrease of GluN2A, PSD-95 and the GluA1 subunit of the AMPA receptor (AMPAR) protein levels in the total cell homogenate of pLKO-shRNF10-infected dissociated neurons and organotypic slices (Figure 4F).

Figure 4

Download asset Open asset

To learn more about the effect of RNF10 knock-down on gene expression, we performed a microarray analysis from organotypic hippocampal slices virally infected with pLKO-shRNF10 lentivirus in order to identify RNF10 target genes. Interestingly, heat map of differentially expressed genes (Figure 5A), gene ontology analysis (Figure 5B) and real-time PCR validation (Figure 4G) showed that RNF10 silencing modulated the expression of several genes involved in excitatory synaptic transmission and dendritic spine morphology (Vogt et al., 2007; Michaluk et al., 2011; Ramakers et al., 2012). Furthermore, we confirmed by WB analysis the effect of viral infection with pLKO-shRNF10 on protein levels of some of the RNF10 target genes, such as Ophn1, ArhGap4 and ArhGef6 (Figure 4H).

Figure 5

Download asset Open asset

The synaptic and nuclear localization in neurons, the presence of two different NLS motifs and its interaction with GluN2A and Meox2 (Lin et al., 2005) suggest that RNF10 could be a novel synaptonuclear protein messenger. To validate this hypothesis, we assessed whether the modulation of neuronal activity affects the subcellular localization of RNF10 and its association with the interacting proteins. We first enhanced synaptic excitatory activity with the GABA-A receptor antagonist Bicuculline (50 μM) in the presence of the K⁺ channel blocker 4-AP (2.5 mM; Hardingham et al., 2002) ('Bic' treatment). Enhanced excitatory activity significantly decreased RNF10 immunoreactivity along dendrites as compared to untreated neurons, without affecting PSD-95 (Figure 6A) labeling and induced a significant increase in RNF10 nuclear staining (Figure 6D, left panels). This effect was not correlated to an alteration of RNF10 total protein amount as determined by WB analysis performed with neuronal lysates (data not shown; +7.8% ± 8.6%, cLTP vs. control; p>0.05, n=6). To further confirm that RNF10 accumulates in the nucleus following long-distance protein transport, we performed time-lapse confocal imaging of RNF10 fused to photoconvertible tdEOS (Figure 6B,C). We photoconverted RNF10-tdEOS in distal dendrites from green-to-red emission using 405 nm irradiation immediately after Bic treatment and tracked the red fluorescence signal over 90 min. Bic treatment induced a significant increase in RNF10-tdEOS photoconverted fluorescence intensity in the nucleus compared to untreated neurons accompanied by a decline in distal dendrites (Figure 6B–C).

Figure 6

Download asset Open asset

Interestingly, the co-incubation of Bic with the GluN2A-specific inhibitor NVP-AAM007 (NVP, 50 nM or 300 nM) led to a significant reduction of RNF10 nuclear accumulation compared to Bic treatment alone also at a concentration of 50 nM, which is known to preferentially block di-heteromeric GluN1/GluN2A and not tri-heteromeric GluN1/GluN2A/GluN2B NMDARs (Figure 6D; Foster et al., 2010). Similarly, Bic dependent increase in RNF10 nuclear staining was reduced by the co-incubation with the NMDAR blocker MK801 (data not shown). Moreover, hippocampal neurons (DIV14) were treated with Bicuculline (50 μM) and 4-AP (2.5 mM) in presence of the GluN2B antagonist ifenprodil (5 µM), corresponding to a well-validated protocol to obtain a specific activation of synaptic NMDARs (Syn; Figure 6E, left panel; Hardingham et al., 2002). Synaptic NMDARs stimulation induced a significant increase in RNF10 nuclear accumulation (Figure 6E, left panels). On the other hand, two different protocols known to activate extrasynaptic NMDARs (Hardingham et al., 2002; Karpova et al., 2013) failed to induce the translocation of RNF10 from dendrites to the nucleus (Figure 6E, right panel). In control experiments, we found that enhanced synaptic activity with Bicuculline (50 µm) and 4-AP (2.5 mM) resulted in an increase in nuclear Jacob as reported previously (Figure 6F; Behnisch et al., 2011; Karpova et al., 2013). This increase was abolished in the presence of the GluN2B antagonist ifenprodil (5 µM), but not with GluN2A antagonist NVP-AAM077 (50 nM) indicating that Jacob dissociates from the synapses following activation of GluN2B but not di-heteromeric GluN2A containing NMDAR (Figure 6F).

The import of proteins from the cytosol into the nucleus through the nuclear pore complex depends on the binding of importins to a specific nuclear localization sequence (NLS; Thompson et al., 2004; Jordan and Kreutz, 2009; Karpova et al., 2012). According to this scheme, importins function as adapter molecules by binding NLS-bearing proteins. Components of the classical nuclear import machinery are present at synapses in association with NMDARs and importins translocate to the nucleus in response to NMDAR stimulation (Thompson et al., 2004; Jeffrey et al., 2009; Marfori et al., 2011; Ch'ng et al., 2012). Analysis of the RNF10 sequence showed the presence of two putative NLS motifs (NLS1: aa 591–599; NLS2: aa 784–791; Seki et al., 2000; Marfori et al., 2011). We therefore investigated the neuronal localization of RNF10 by transfection of RNF10 FL and RNF10 truncated mutants lacking one [Myc-RNF10 (1–611)] or both NLS motifs [Myc-RNF10 (1–590)] in primary hippocampal neurons. Myc-RNF10 FL showed an intense nuclear accumulation (as the endogenous protein; Figure 7A). However, both truncated mutants failed to localize to the nucleus as indicated by a very low co-localization with the nuclear marker Dapi (Figure 7A). In particular, also the longest Myc-RNF10 (1–611) construct lacking only the C-terminal NLS2 domain accumulated in the cell soma but did not show a marked nuclear localization (Figure 7A), thus suggesting that the second NLS2 motif could be responsible for RNF10 nuclear translocation in neurons. Notably, the overexpression of Myc-RNF10 (1–611) construct produced a significant reduction in the density of PSD-95-positive puncta along dendrites (Figure 7B, upper panels) similar to that observed following RNF10 silencing (Figure 7B, lower panels), indicating that this protein fragment can act in a dominant negative manner.

Figure 7

Download asset Open asset

Based on these results, we next evaluated a possible RNF10 interaction with importins in hippocampal neurons. RNF10 displayed a prominent co-localization with importin α1 (KPNA2 gene product; see Zienkiewicz et al., 2013) and surface GluN2A along dendrites of DIV14 primary hippocampal neurons (Figure 7C). In agreement with previous data about other importins (Thompson et al., 2004; Jeffrey et al., 2009; Marfori et al., 2011; Ch'ng et al., 2012), analysis of importin α1 subcellular distribution showed its presence in purified PSDs from rat hippocampus (Figure 7D). Immunoprecipitation experiments performed revealed that RNF10 interacts with importin α1 in neuronal cells (Figure 7E, upper panel). Interestingly, RNF10 co-precipitation with importin α1 was detected both in PSD and homogenate fraction but it was much more prominent in PSDs (Figure 7E, lower panel). In order to verify the capability of RNF10 to interact also with other importins, we performed heterologous co-immunoprecipitation experiments of RNF10 fused to GFP with six importin α isoforms (KPNA1-KPNA6 gene products) fused to tag-RFP. As shown in Figure 7F, RNF10 can interact with majority of importin α isoforms except importin α3 (KPNA4 gene product).

Bic treatment significantly increased RNF10/importin α1 association (Figure 7G). Interestingly, also the RNF10 interaction with the transcription factor Meox2 was significantly increased by Bic treatment (Figure 7H). On the contrary, Bic treatment significantly decreased RNF10/GluN2A interaction compared to control neurons (Figure 7H), which is in agreement with the reduction of RNF10 labeling along dendrites (see Figure 6A).

Taken together these findings indicate that enhancing synaptic activity of di-heteromeric GluN2A-containing NMDARs results in the dissociation of RNF10 from the cytoplasmic tail of GluN2A at postsynaptic sites and a tighter association with neuronal importin and importin-dependent nuclear translocation, followed by an association of RNF10 with the transcription factor Meox2.

LTP is widely considered as one of the major cellular mechanisms that underlie learning and memory (Bliss and Collingridge, 1993) and requires de novo protein synthesis and gene transcription for its maintenance (Smolen et al., 2012). We next asked whether RNF10 translocates to the nucleus from distal dendrites following induction of LTP in mature hippocampal neuronal cultures. For induction of LTP we used two established protocols (Deisseroth et al., 1996; Otmakhov et al., 2004; Oh et al., 2006). First, we employed 18s@50Hz field stimulation, which is known to produce a stable synaptic potentiation in cultured hippocampal neurons (Deisseroth et al., 1996) and in parallel we performed time-lapse confocal imaging of RNF10 fused to photoconvertible tdEOS (Figure 8A–C). The rise in intracellular Ca²⁺ concentration upon LTP-inducing stimuli was verified by Fluo-4-AM fluorescence (data not shown). We photoconverted RNF10-tdEOS in distal dendrites (Figure 8A) and tracked the red fluorescence signal over time. We observed a significant increase in RNF10-tdEOS photoconverted green-to-red fluorescence intensity in the nucleus when compared to unstimulated neurons (Figure 8B,C) accompanied by a concomitant decline in distal dendrites upon stimulation (Figure 8C, lower panels). This emphasizes that the synaptodendritic but not the somatic pool of RNF10 largely contributes to its nuclear transport upon LTP-inducing stimuli. Thus, RNF10 accumulates in the nucleus following long-distance protein transport.

Figure 8

Download asset Open asset

We next used a validated chemical LTP (cLTP) protocol (Otmakhov et al., 2004; Oh et al., 2006), which is known to activate synaptic NMDARs in primary hippocampal neurons. As expected, induction of cLTP leads to an increase of CREB-Ser133 (P-CREB; Bito et al., 1996) and GluA1-Ser845 (P845-GluA1; Marcello et al., 2013) phosphorylation (data not shown). Under these experimental conditions, cLTP induction produced a significant decrease in endogenous RNF10 immunostaining along dendrites (Figure 8D) and a concomitant significant increase of RNF10 localization in the nucleus as revealed by western blotting in a crude nuclear fraction (Figure 8G) and co-localization assays with the nuclear marker Dapi (Figure 8F and Figure 9B). Consistent with the above-described results (see Figure 7H), cLTP detached RNF10 from GluN2A (Figure 8E). These effects were not correlated to an alteration of RNF10 total protein abundance as determined by WB analysis performed with neuronal lysates (data not shown; +9.5% ± 8.7%, cLTP vs. control; p>0.05, n=8). Finally, we used anisomycin to exclude the possibility that de novo RNF10 protein synthesis could account for the nuclear accumulation of RNF10 following cLTP (Figure 8F,G). Anisomycin treatment did not prevent nuclear accumulation of RNF10 in this assay (Figure 8F,G).

Figure 9

Download asset Open asset

The RNF10 (1–611) protein fragment, lacking the NLS2 motif, fails to localize to the nucleus (Figure 7A). Taking this into account, we induced cLTP in the presence of an exogenously added RNF10 NLS2 peptide, thus competing with endogenous RNF10/importin binding. RNF10 NLS2 peptide is mostly composed of basic residues having a very high similarity with the most common cell-permeable peptides (CPPs) such as the TAT peptide (Farkhani et al., 2014). However, before the evaluation of the effect of NL2 peptide on RNF10 subcellular distribution, we first verified the ability of this peptide to act as a CPP, that is, that it can cross the plasma membrane and enter neurons. As shown in Figure 9A, we detected TideFluor^TM-tagged TAT and NLS2 peptides into neurons to a similar extent, thus indicating the capability of the NLS2 peptide to work as a CPP. Notably, NLS2 peptide induced a significant reduction of RNF10/importin α1 interaction when compared to treatment with TAT (used as control peptide; Figure 9B). The NLS2 peptide completely prevented the LTP-dependent increase in nuclear RNF10, as indicated by immunofluorescence (Figure 9C,D) and by immunoblot analyses of RNF10 protein levels in a P1 crude nuclear fraction (Figure 9G,H). Importantly, treatment with NLS2 peptide also prevented the decrease in dendritic RNF10 labeling observed after induction of cLTP (Figure 9E,F). Overall, these results further confirm that NLS2-dependent binding to importin represents a key step for RNF10 trafficking to the nucleus.

Conversely, induction of chemical long-term depression (cLTD) did not modify RNF10 binding to GluN2A (Figure 9I; +0.9% ± 4.8%, cLTD vs. control; p>0.05) and the nuclear distribution of RNF10 in hippocampal neurons (Figure 9J).

We next asked whether RNF10 translocation from dendritic spines to the nucleus produced a modulation of the protein level of Meox2/RNF10 target genes. First, we analyzed the expression of p21^WAF1/cip1, a known Meox2 target gene (Lin et al., 2005; Malik et al., 2013). Bic treatment increased p21^WAF1/cip1 levels as evaluated by immunofluorescence (Figure 10A,B) and Western blot analysis (Figure 10C). Up-regulation of p21^WAF1/cip1 was significantly attenuated by co-treatment with the NMDAR blocker MK801 (Figure 10A–C). Similarly, stimulation of synaptic NMDARs induced an increase in p21^WAF1/cip1 that was blocked by bath application of MK801 (Figure 10D). Notably, the increase in p21^WAF1/cip1 protein levels following stimulation of synaptic NMDARs was completely prevented by RNF10 silencing (Figure 10E), indicating that modulation of p21^WAF1/cip1 expression is RNF10-dependent. In addition, the same effect on p21^WAF1/cip1 levels was observed following induction of cLTP. Again, the increased expression of p21^WAF1/cip1 was prevented by RNF10 silencing (Figure 10F). Finally, RNF10 silencing decreased the expression of the newly identified RNF10 target gene Ophn1 (see Figure 4G,H) also after synaptic stimulation (Figure 10G).

Figure 10

Download asset Open asset

Altogether these results indicate that stimulation of synaptic NMDARs or induction of LTP leads to RNF10 translocation to the nucleus, where it binds to Meox2 and stimulates the expression of RNF10/Meox2 target genes (Lin et al., 2005; Malik et al., 2013), such as p21^WAF1/cip1 and Ophn1.

We finally analyzed whether RNF10 silencing leads not only to a decrease of dendritic spine density (Figure 4A–E) but also to an alteration of the plastic properties of synapses. To this end, we recorded miniature EPSCs (mEPSCs) in hippocampal neurons before and for 2 hr after cLTP induction. In basal conditions, down-regulation of RNF10 led to a significant decrease of the frequency (Figure 11A,C) but not the amplitude (Figure 11A,B) of mEPSCs. Notably, no alteration of mEPSCs frequency was observed by co-expressing shRNF10 with the wild-type human variant of RNF10 resistant to shRNA (FlagRNF10; Figure 11A,C). Furthermore, while in control neurons cLTP induction resulted in a long-lasting increase of the amplitude and frequency of mEPSCs, this effect was totally prevented in neurons transfected with shRNF10 (Figure 11D,E). Again, co-transfection of FlagRNF10 fully rescued a physiological long-lasting increase of the amplitude (Figure 11D) and frequency (Figure 11E) of mEPSCs.

Figure 11

Download asset Open asset

We then tested whether different levels of RNF10 expression could affect not only spine density (see Figure 4A–E) but also interfere LTP-dependent modulation of dendritic spine size (Bosch and Hayashi, 2012). To this end, we overexpressed or silenced RNF10 in hippocampal neurons (DIV7) followed by induction of cLTP (DIV14). As expected, cLTP increased significantly PSD-95 cluster width in control neurons (Figure 11F,G). RNF10 overexpression produced an increase in PSD-95 clusters width also in absence of stimulation and occluding any further increase of PSD-95 clusters width following induction of cLTP (Figure 11F). On the other hand, RNF10 silencing blocked any cLTP-dependent modification of PSD-95 clusters (Figure 11G). Evaluation of GluA1 synaptic cluster size confirmed that RNF10 silencing precludes any increase in GluA1 clusters width induced by cLTP (Figure 11I).

Post-synaptic densities (PSDs) from rat hippocampus were isolated as previously reported (Gardoni et al., 1998). To purify the postsynaptic Triton-insoluble fraction (TIF), cell lysates were centrifuged at 13,000 g for 15 min at 4°C. The resulting pellet was resuspended in 150 mM KCl, 0.5% Triton and spun at 100000 g for 1 hr at 4°C. The final pellet (TIF) was homogenized with a glass-glass potter in 20 mM Hepes buffer containing Complete. All purifications were performed in the presence of complete sets of protease and phosphatase inhibitors (Roche Diagnostics, Monza, Italy) Protein content of the samples has been quantified by using Bio-Rad (Hercules, CA, USA) protein assay. After measuring protein concentration, all samples have been standardized at 1 μg/μl concentration and the same protein amount loaded in each lane for western blot analysis.

Organotypic hippocampal slice cultures were prepared as previously described (Pellegrini Giampietro et al., 1999). Hippocampal neuronal primary cultures were prepared from embryonic day 18–19 (E18-E19) rat hippocampi as previously described (Piccoli et al., 2007). Neurons were transfected at DIV7 using calcium-phosphate method. For induction of 'Bic' treatment: hippocampal neurons (DIV14) or organotypic hippocampal slices (DIV14) were incubated with Bicuculline (50 μM; Tocris) and 2.5 mM 4-Aminopyridine (4-AP;Tocris) in Neurobasal medium supplemented with B27; 'Syn' treatment: stimulation of synaptic NMDA receptors was obtained by treating hippocampal neurons at DIV14 with 50 μM Bicuculline (Tocris), 2.5 mM 4-AP and 5 μM ifenprodil in Neurobasal medium supplemented with B27; 'Extrasyn#1' treatment: stimulation of extrasynaptic NMDA receptors was obtained by pretreating hippocampal neurons at DIV14 with 50 μM Bicuculline, 2.5 mM 4-AP and 10 μM MK801 for 30 min, washing them with Neurobasal medium supplemented with B27, incubating with 50mM KCl (Carlo Erba) for 10 min and then with 40 μM Glutamate (Sigma); 'ExtraSyn#2' treatment was performed as previously described by Karpova and co-workers (2013). cLTP (Otmakhov et al., 2004; Oh et al., 2006): Primary neuronal cultures at DIV14 were first incubated in ACSF (125 mM NaCl, 2.5 mM KCl, 1 mM MgCl₂, 2 mM CaCl₂, 33 mM D-glucose and 25 mM HEPES, pH 7.3) for 30 min, followed by 16 min of stimulation with 0.05 mM Forskolin (Sigma Aldrich), 0.1 mM Picrotoxin (Tocris) and 100 nM Rolipram (Calbiochem) in ACSF (no MgCl₂) to induce NMDAR-dependent cLTP. Control was incubated in ACSF. After stimulation, neurons were replaced in regular ACSF for 2 hr. cLTD (Marcello et al., 2013): primary neuronal cultures at DIV14 were first incubated in ACSF, followed by incubation in ACSF (in presence of MgCl₂) and 50 μM NMDA (Sigma Aldrich) for 10 min. Neurons were then replaced in regular ACSF for 1 hr to induce cLTD. For treatment with 'NLS2 peptide' hippocampal neurons (DIV14) were treated with active (NLS2: RKRKRQKQK) or control (TAT: GRKKRRQRRRPQ) peptide for 24 hr and then treated to induce cLTP the presence of the same peptides.

COS-7 cells were maintained in DMEM + Glutamax medium (GIBCO-BRL) supplemented with 10% Fetal Bovine Serum (FBS) (GIBCO-BRL) and Pen/Strep (GIBCO-BRL). The day before transfection, COS-7 cells were placed in a 6 wells multiwell (for cells lysis) or 12 wells multiwell (for cells immunostaining), then cells were transfected using lipofectamine method (Invitrogen). After 36 hr, COS-7 cells were lysed for co-immunoprecipitation and western blotting or fixed for immunostaining.

RNF10-tagged with GFP was co-expressed with importin α isoforms (KPNA1-KPNA6) tagged with tagRFP in HEK293T cells. Thirty-six hours post-transfection proteins were extracted with RIPA buffer containing phosStop (50 mM Tris pH=8, 1% NP-40, 0.5% NO deoxycholate, 0.1% SDS, 150 mM NaCl, 1% Triton X-100, protease inhibitor). RNF10-GFP was immunoprecipitated from cell extract using anti-GFP MicroBeads (MACSMolecular). Co-immunoprecipitated importinerase-α isoforms were detected in complex with RNF10 using anti-tagRFP antibodies (Evrogen).

Real-time quantitative PCR (qPCR) was performed as previously described (De Fabiani et al., 2003). The following gene expression assay kits from Life Technologies have been used: Rn01409258_g1 for S100a11; Rn01504461_g1 for Rac2; Rn01526492_m1 for Apbb1ip; Rn01422083_m1 for Il18; Rn01430875_g1 for Timp1; Mm00623991_m1 for Itgb8; Rn01513693_m1 for Thbs1; Rn00580728_m1 for Cd36; Rn00579162_m1 for Mmp9. For microarray experiments, RNA was analyzed by the Genopolis Consortium using an Affimetrix platform (GEO accession number GSE69267). Data was primarily analyzed using the TAC and Partek Genomic suite softwares. The Robust Multichip Average (RMA) method was employed to calculate probe set intensity (Irizarry et al., 2003). Differentially-expressed genes whose fold change was higher or equal than 1.75 (up-regulated genes) and lower or equal to 0.65 (down-regulated genes) with a p value lower than 0.05 were selected as significantly modulated. The expression value of each probe is divided by the mean value of the scramble controls and the log2 of this ratio has been reported. Gene Ontology biological process and enriched pathways analyses were performed using Panther software (http://www.pantherdb.org/).

Hippocampi from adult rats were homogenized at 4°C in an ice-cold buffer containing 0.32 M Sucrose, 1 mM Hepes, 1 mM NaF, 0.1 mM PMSF, 1 mM MgCl in presence of protease inhibitors (Complete, GE Healthcare, Mannheim, Germany) and phosphatase inhibitors (PhosSTOP, Roche Diagnostics GmbH, Mannheim, Germany), using a glass-Teflon homogenizer. Homogenates were then centrifuged at 1000 g for 5 min at 4°C, to remove nuclear contamination and white matter. The supernatant was collected and centrifuged at 13,000 g for 15 min at 4°C. The resulting pellet was resuspended in hypotonic buffer (1 mM Hepes with Complete) and referred as P2 crude membrane fraction. Aliquots of proteins from homogenate, P2 crude membrane fractions or PSDs were incubated overnight at 4°C in RIPA buffer containing 50mM Tris HCl (pH 7.2), 150 mM NaCl, 1% NP-40, 0.5% deoxycholic acid, 0.1% sodium dodecyl sulphate (SDS) in a final volume of 150 µl with the antibody. Pre-incubation with 1% SDS was used for co-i.p. experiments performed in PSDs (Gardoni et al., 2001). As a control, one sample was incubated in the absence of the antibody (no IgG lane) and another sample in presence of an irrelevant antibody. Protein A-sepharose beads (Sigma-Aldrich) were added and incubation was continued for 2 hr, at room temperature, with shaking. Beads were collected by centrifugation and washed three times with RIPA buffer before adding sample buffer for SDS-PAGE and boiling for 5 min. Beads were collected by centrifugation, all supernatants were applied onto 7%–12% SDS-PAGE and revealed by western blotting.

PLA was performed as previously described (Söderberg et al., 2006; Augusto et al., 2013) in primary hippocampal neurons. Cells were fixed with 4% PFA - 4% Sucrose for 5 min at 4ºC. Cells were then rinsed in PBS, permeabilized with 0.1% Triton X-100 in PBS for 15 min at room temperature and blocked with 5% BSA in PBS for 30 min at room temperature. Subsequently, cells were incubated with goat polyclonal anti-RNF10 antibody (1:200; Santa-Cruz) and rabbit polyclonal anti-GluN2A antibody (1:200; Lifetechnologies) overnight at 4ºC. After washing in PBS, the cells were incubated for 1 hr at 37ºC with the PLA secondary probes anti-goat Plus and anti-rabbit Minus (Olink Bioscience). Afterward, the cells were washed twice with Duolink II Wash Buffer A (Olink Bioscience) and incubated with the ligation-ligase solution (Olink Bioscience) for 30 min at 37ºC. After a new rinse, the cells were incubated with DNA polymerase (1:80; Olink Bioscience) in the amplification solution (Olink Bioscience) for 100 min at 37°C. After washes in consecutive decreasing concentrations with Duolink II Wash Buffer B (Olink Bioscience), the cells were then incubated with chicken polyclonal anti-GFP/rabbit polyclonal anti-MAP2 (1:300/1:400; Milipore/Milipore) overnight at 4ºC. After washing in PBS, the cells were incubated for 1 hr at room temperature with secondary goat anti-chicken-AlexaFluor 488/goat anti-rabbit-AlexaFluor 647 (Lifetechnologies). The cells were washed with PBS and mounted on slides with Fluoroshield mounting medium (Sigma).

The C-terminal tail of the GluN2A subunit was subcloned downstream of glutathione S-transferase (GST) in the BaMHI and HindIII sites of the expression plasmid pGEX-KG by PCR using the Pfu polymerase (Stratagene) on a GluN2A cDNA template (kind gift from S. Nakanishi). GST-GluN2A fusion proteins containing the cytoplasmic domain of GluN2A (839–1464 or 1049–1464 or 1244–1461 or 1244–1464) were expressed in Escherichia coli and purified on glutathione agarose beads (Sigma Aldrich, St. Louis, MO) as previously described (Gardoni et al., 2001).

Three hundred micrograms of proteins were incubated with GST to a final volume of 1 ml with Tris Buffered Saline solution for 45 min with shaking. Samples were centrifuged at 1000 rpm for 5 min and the pellets were discarded. Supernatants were incubated for 2 hr with GST fusion proteins or GST alone. Beads were washed four times with TBS and 0.1% Triton X-100. Bound proteins were resolved by SDS-PAGE and subjected to immunoblot analysis.

For co-localization and morphological studies, treated hippocampal neurons were fixed 7 min in 4% paraformaldehyde plus 4% sucrose in PBS at room temperature; then extensively washed with PBS supplemented with CaCl₂ and MgCl₂ (PBS-C.M), permeabilized with 0.2% Triton X100 in PSB-C.M and blocked for 2 hr at room temperature with 5% BSA in PBS-C.M. Primary and secondary antibodies were applied in 5% BSA in PBS-C.M. Cells were chosen randomly for quantification from different coverslips. Fluorescence images were acquired by using Zeiss Confocal LSM510 Meta system with a sequential acquisition setting at 1024x1024 pixels resolution; for each image two up to four 0.5μm sections were acquired and a z projection was obtained (Malinverno et al., 2010). All images were acquired standing the signals within the linear range in order to perform a reliable quantification (avoiding the presence of any saturated pixel) and to compare appropriately all experimental conditions. For analysis of dendritic spine morphology, primary hippocampal neurons were transfected with GFP-containing constructs at DIV7. Cells were fixed and immunolabeled for GFP at DIV14. The SP5 CLSM system (Leica-Microsystems, Mannheim, Germany) equipped with Diode (405 nm), Argon (488 nm) and Diode Pumped Solid State (561 nm) lasers was used for time-lapse imaging of RNF10 fused to tdEOS. Images were taken with HCX APO L20x/1.00W objective (Leica, Germany). Along the z-axis at list 10 optical sections with focus depth of 300–400 nm were taken in order to cover the complete volume of imaged neurons. Quick change 18 mm Chamber with Field Stimulation (RC-49MFS, Warner Instruments) was used for mounting coverslips with living overexpressing RFN10-tdEOS neurons in Neurobasal medium (Gibco) on the microscope stage. Master-8/U 8-channel programmable pulse generator (Science Products) was used to generate timing and two channels were synchronized to produce complex patterns of pulses (1 ms pulse duration) with inter pulse interval of 20 ms. To eliminate the distortion and thereby to achieve a rectangular pulse permitting accurate measurement, the constant-current stimulus isolation unit (SIU-102, Warner Instruments) was used in the output of a stimulus generator. The LTP was induced by applying current pulses 18s@50 Hz via field stimulation electrode. Response of cultured hippocampal neurons on field synaptic stimulation was verified by Ca²⁺-imaging with Fluo-4-AM (1 µM) (data not shown).

Miniature EPSCs were recorded on DIV14 hippocampal neurons at room temperature, at a holding potential of -60 mV. The recording pipettes had resistance of around 5 MΩ when filled with the following medium (in mM): 140 CsCl, 0.5 CaCl₂, 20 EGTA, 10 HEPES, 10 D-glucose, pH 7.2 and osmolarity of 300 mOsm. The high concentration of EGTA avoided slow Ca²⁺-dependent desensitization of NMDARs. Neurons were perfused continuously with the following external medium (in mM): 140 NaCl, 2 CaCl₂, 3 KCl, 10 HEPES, 10 D-glucose, 0.01 glycine, 0.01 bicucullin, 0.0003 tetrodotoxin, pH 7.4 and osmolarity of 330 mOsm. Currents were recorded through an Axopatch 200B amplifier, filtered at 1 kHz, digitized at 3 kHz and analyzed using the pClamp 10.0 software of Axon Instrument (Molecular Devices, Sunnyvale, CA).

Acquisition and quantification of western blotting was performed by means of computer-assisted imaging (ChemiDoc system and Image lab 4.0 software; Bio-Rad). Western blotting experiments have been normalized on tubulin or H3 levels, depending if they have been performed from total cell lysate (homogenate) or P1 nuclear fraction, respectively. Data obtained by pull-down assays were normalized on the amount of GST fusion protein. The data from all quantifications were analyzed using Graphpad Prism software and levels and values were expressed as mean ± standard error of the mean (SEM). Co-localization analysis was performed using Zeiss AIM4.2 software. Statistical evaluations were performed by Student t test or Pearson correlations or, as appropriate, by one-way ANOVA followed by Bonferroni’s or Tukey's as a post-hoc test. When appropriate, experiments have been performed in blind conditions.

The following antibodies were used: monoclonal antibody (mAb) anti-α-calcium/calmodulin-dependent kinase II (αCaMKII), polyclonal antibody (pAb) anti-GluN2A, pAb anti-CREB, mAb anti-Myc, pAb anti-GluA1 and pAb anti-p-CREB (Ser-133) were purchased from Millipore (Billenca, MA,USA); mAb anti-Meox2 and pAb anti-Synaptotagmin were purchased from Abcam (Cambridge, MA, USA); mAb anti-GluR2, mAb anti-GFP, mAb anti-GST and anti-PSD-95 were purchased from NeuroMab (Davis, CA); mAb anti-α-Tubulin and mAb anti-Flag were purchased from Sigma-Aldrich (St. Louis, MO); mAb anti-p21 and mAb anti-Rch1 (importin α1, KPNA2 gene product) were purchased from BD Biosciences (Franklin Lakes, NJ); pAb anti-histone H3 was purchased from Proteintech (Chicago, IL); pAb anti-RNF10 were purchased from ProteinTech and Santacruz; mAb anti-GluN2B and p-GluA1(845) were purchased from Invitrogen (Carlsbad, CA); pAb anti-BDNF was purchased from Genetex (Irvine, CA); mAb anti-Map2 was purchased from Immunological Sciences (Roma, Italy); mAb anti-JL8 was purchased from Clontech (Mountain View, CA); pAb anti-ArhGap4 was purchased from Novus Biologicals (Cambridge, UK); mAb anti-GFAP, pAb anti-Ophn1 and pAb anti-ArhGef6 were purchased from Cell Signaling (Danvers, MA). Peroxidase-conjugated secondary anti-mouse Ab was purchased from Pierce (Rockford, IL) while peroxidase-conjugated secondary anti-rabbit Ab was purchased from Bio-Rad (Hercules, CA). AlexaFluor secondary Abs were purchased from Invitrogen (Carlsbad, CA).

The specificity of the RNF10 antibody (ProteinTech) in recognizing RNF10 in western blotting and immunocytochemistry was carefully verified by RNF10 overexpression and silencing in primary hippocampal neurons (data not shown). The specificity of the importin α1 antibody (BD Biosciences) in recognizing KPNA2 gene product in western blotting was carefully verified by KPNA1-6 overexpression in HEK293T cells (data not shown).

The RNAi Consortium (TRC) shRNA target gene set in pLKO.1 lentiviral vector for rat RNF10 (accession number NM_001011904.1) along with scrambled vector were purchased from Open Biosystems (Thermo Scientific, Milano, Italy). The set contained three vectors (the TRC IDs: TRCN0000041128, mature antisense TTCAGGTTGATCTTCTTAGGG; TRCN0000041131, mature antisense TTTCTGTGAATTGGAGCGACG; and TRCN0000041132, mature antisense ATACCCAGCAATCTCTACCAC), all of them were validated by real-time PCR on rat hippocampal primary cultures. Vector TRCN0000041128 showed the highest level of RNF10 downregulation (>90%, data not shown) and was used in all further experiments (designated as shRNF10). To knock down rat GluN2A gene expression, a set of 4 short hairpin RNA (shRNA) expression vectors in pGFP-V-RS was purchased from Origene (Rockville, MD). The sequences were as follows: A-gacagcactcctatgataacattctggac; B-tctaccagcaggactggtcacagaacaac; C-tgcgagccagatgacaaccacctcagcat; D-agtagaggtggctgtcagcactgaatcca. A scrambled control from Origene (Rockville, MD) was also used as a negative control. Sequence called B showed the best knockdown and was therefore used for further experiments.

For the testing and validation of sh-resistant form of RNF10, rat primary hippocampal cultures were transfected with GFP (Ctrl), rat-shRNF10, RFP-humRNF10, and rat-shRNF10 + TRP-humRNF10. Total RNA was extracted with Tryzol 48 hr after transfection, retrotranscribed (Improm II, Promega) and analyzed by quantitative PCR using primers that specifically amplify rat RNF10 or human RNF10. Transfection with rat-shRNF10 resulted in three fold decrease of endogenous RNF10 mRNA, whereas transfection with human RFP tagged RNF10 resulted in 45-fold increase in mRNA for human RNF10, the level which was not affected by rat-shRNF10 co-transfection. Interestingly, the overexpression of human RNF10 in rat hippocampal cells abolished effect of shRNF10 on levels of endogenous RNF10 mRNA. The specificity of rat-shRNF10-mediated knock down was further tested using human neuroblastoma line SY-S5Y5. No effect on both endogenous and everexpressed RNF10 was observed after transfection with rat-shRNF10.

Cloning of pLV-RNF10-EGFP. The cloning of this construct was accomplished in two steps. First, EGFP was amplified from the pRRLSIN.cPPT.hCMV-EGFP.WPRE (pLV-EGFP) plasmid to create in frame sequence with XbaI-SalI sites at 5’ of EGFP and XhoI sites at 3’ of EGFP using primers: forward 5’-aagcttTCTAGAgaattcGTCGACATGGTGAGCAAGGGCGAGGA; reverse 5’- aagcttCTCGAGTTACTTGTACAGCTCGTCCAT. Amplificate was digested with XbaI and XhoI enzymes and ligated in XbaI-SalI digested pLV vector creating the intermediate pLV-(XS)EGFP bearing XbaI-SalI cassette at 5’ of EGFP. After the sequence was confirmed by sequencing, human RNF10 was excised from pCMV-Tag4-RNF10 plasmid by XbaI and SalI and ligated in XbaI/SalI digested pLV-(XS)EGFP yielding pLV-RNF10-EGFP plasmid.

For production of lentiviral particles 2 mln of HEK293T producing cells were plated on 100 mm plates (Costar, Milan, Italy) in complete medium (DMEM, Cat. D5671; 2 mM Glutamine; 100 units/ml penicillin and 100 μg/ml streptomycin (Sigma); 10% Foetal Bovine Serum (FBS, Immunological Sciences, Milano, Italy). For overexpression of RNF10, the day after plating the cells were transfected with 3μg pL-shRNF10 or pGIPZ-scr and with 2.5 μg psPAX2 and 0.75 μg pMD2.G retroviral packaging constructs using Lipofectamine 2000 (Life Technologies, Milan, Italy) in 7 ml of complete medium. For RNF10 silencing, the cells were transfected with 3 μg pLV-RNF10-EGFP, and with 2 μg pMDLg/pRRE, 0.75 μg pRSV-REV, 11 μg pMD2.G lentiviral packaging constructs. 48 hr after transfection culture medium was collected, filtered through 0.45 μm PVDF filter and precipitated with polyethylene glycol (PEG; 5x PEG solution contained: 200 g PEG, Sigma Cat. 89510, 12 g NaCl in 2 mM Tris pH 7.5) overnight. Precipitates were then collected by centrifugation at 1500 xg for 30 min at 4°C, resuspended in cold HEPES balanced salt solution (HBSS, Sigma), aliquoted and snap frozen. Aliquots were kept at -80°C. Each viral batch was tittered by serial dilution infections of rat hippocampal primary cells by means of real-time PCR (for RNF10 silencing) using primers for rat RNF10 (to evaluate efficacy of downregulation): forward 5’-GAACAAGAGACAGGGCAGGT, reverse 5’-GACAGAGGGGTCAGCAGAAA. For experiments, a viral titer was used yielding > 80% of RNF10 downregulation. Virus for overexpression was tittered by FACS, and a titer yielding > 80% EGFP positive cells was used.

Y2H was conducted according to the manufacture procedure guidelines using the Mate & Plate™ Library - Mouse Brain (Normalized) (cat # 630488, TakaraBio/Clontech Europe, France). Briefly, the GluN2A(839–1461) C-tail (bait) was cloned in the pGBKT7 plasmid and transformed in the AH109 haploid yeast strain (MATa). This was mated overnight with the Mouse Brain normalized library (prays) cloned in pGADT7 transformed in Y187 haploid yeast strain (MATα). Yeast were plated after 24 on selective plate, allowing only the growth of diploid where a protein interaction between the bait and pray protein occurred (absence of Leucine, Adenine, Tryptophan and Histidine). Then the diploid where tested by a colorimetric assay (α-gal) to avoid the presence of false positive. Plasmids (21 positive clones) were extracted from the yeast and sequenced. Five genes were identified following sequencing, one of those being RNF10.

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

This work was supported by: FP7-PEOPLE-ITN-2008 #238608 (SYMBAD) to MDL; Fondazione CARIPLO (grant number 0795-2013) to MDL and AAG; FIRB Accordi di Programma RBAP11HSZS to MDL; Agence Nationale de la Recherche [ANR-13-JSV4-0005-01, SYNcity], Région Languedoc-Roussillon (Chercheur d’Avenir) to JP; PRIN2010-2011 2010AHHP5H to FaG; PRIN2010-2011 2010PWNJXK to MDL and PLC; DFG (SFB779 TPB8 / Kr1879/5-1/6-1) and FP7 MC-ITN NPlast to MRK; CBBS to AK and MRK.

We thank Dr. Richard Huganir (Johns Hopkins University School of Medicine, Baltimore, MD) for the GFP-GluN2A construct and Dr. Toru Ogata (The University of Tokyo, Bunkyo-ku, Tokyo, Japan) for the rat RNF10 cDNA. We want to thank Dr. J Wiedermann and Dr. U Thomas for providing tdEOS construct and Dr. M Mikhaylova for help with Jacob-GluN2B co-i.p. experiments.

Animal experimentation: All procedures were performed in accordance with the current European Law and were approved by the Italian Ministry of Health (as indicated in Dlgs N. 295/2012-A).

© 2016, Dinamarca et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.