Long-term changes of synaptic plasticity depend on protein synthesis and transcription. Ng (neurogranin) is a small protein concentrated at dendrites and spines of forebrain neurons, involved in synaptic plasticity through the regulation of CaM (calmodulin)-mediated signalling. Ng presents a central IQ motif that mediates its binding to CaM and PA (phosphatidic acid) and that can be phosphorylated by PKC (protein kinase C). In the present manuscript, we report that Ng displays a strong nuclear localization when expressed in cell lines and hippocampal neurons, either alone or fused to GFP (green fluorescent protein; GFP–Ng). Furthermore, using subcellular fractionation and immunocytochemical techniques, we were able to localize endogenous Ng in the nuclei of rat forebrain neurons. Nuclear localization of Ng depends on its IQ motif and is reduced by binding to cytoplasmic CaM. Also, PKC stimulation induces a transient nuclear translocation of Ng in acute hippocampal slices. A similar translocation is observed in the neurons of the cerebral cortex and hippocampus after the induction of generalized seizures in adult rats. In summary, the results of the present study show that a fraction of rat brain Ng is localized in the neuronal nuclei and that synaptic activity regulates its translocation from the cytoplasm. The possible involvement of Ng in the regulation of intranuclear Ca2+/CaM-dependent signalling and gene expression is discussed.
Long-term changes in synaptic connection strength depend on protein synthesis and transcription [1,2] and several signalling pathways leading to transcriptional activation have been implicated in learning and memory [3–5]. However, knowledge of synapse-to-nucleus signalling in neurons grows slowly and few molecules entering the nucleus upon stimulation of neuronal activity have been identified. Among them, there are transcription factors such as nuclear factor κB  and NFAT (nuclear factor of activated T-cells) , enzymes such as ERK2 (extracellular-signal-regulated kinase 2) , PYK2 (proline-rich tyrosine kinase 2)  and the class II histone deacetylases HDAC4 and HDAC5 , and other proteins such as CaM (calmodulin) , AIDA-1 , profilin  and, more recently, thymosin β15  and Jacob .
Ng (neurogranin), a small protein identified in brain [16–18], is a post-synaptic PKC (protein kinase C) substrate abundantly expressed in the cerebral cortex, hippocampus, amygdala and basal ganglia. Ng binds CaM in the absence, or at low levels, of Ca2+ and releases it at micromolar or higher concentrations. Ng phosphorylation at Ser36 by PKC prevents its binding to CaM [16,19]. More recently, it has been shown that Ng binds to PA (phosphatidic acid) and associates to cellular membranes . Ng interactions are mediated by a central 15–20 amino acid stretch, termed the IQ motif, that shapes an amphiphilic α-helix rich in basic and hydrophobic residues and shares strong homology with the IQ motif present in GAP-43, an axonal protein abundantly expressed in developing and regenerating neurons .
There is a relationship between Ng expression, synaptic plasticity and remodelling of neuronal connections. In the rat forebrain, Ng expression develops postnatally and peaks during the third week of life, coincident with a period of intense synaptogenesis [22,23]. In canaries and zebra finches, Ng expression is high in the song nucleus Area X during the juvenile song-learning period and decreases noticeably in the adults, who exhibit a robust stability of the song control circuit . Furthermore, Ng knock-out mice show severely reduced performance in the Morris water maze, deficits in high-frequency-induced LTP (long-term potentiation) in area CA1 and significantly decreased intracellular Ca2+ responses after strong tetanic stimulation . Actually, Ng-deficient mice have lower amounts of phosphorylated CaMKII (Ca2+/CaM-dependent protein kinase II) in their brains [26,27], suggesting an attenuated signalling through Ca2+/CaM-mediated pathways.
In neurons, Ng is localized in small aggregates distributed in cell bodies and dendrites, and concentrates at the perinuclear region and dendritic spines . Ng sorting into the somatodendritic compartment most likely depends on the dendritic targeting of its mRNA [29,30] and its local translation [31,32]. Earlier work has already described the presence of Ng in the nucleoplasm of cerebral cortex  and striatal  neurons. However, further studies using subcellular fractionation  or immunocytochemical  methods did not find Ng in the neuronal nucleus.
In the present study, we show that Ng is present at the nucleoplasm of neurons and moves from the cytoplasm to the nucleus in a synaptic activity-dependent manner. In the nuclear translocation mechanism described, the Ng IQ motif and its interaction with CaM seem to play an important role. Taken together these results open the possibility that Ng, tightly related to synaptic plasticity, may act as a nuclear regulator of Ca2+/CaM-dependent signalling, leading to fine tuning of transcriptional activity in neurons.
Antibodies and plasmids
Ng antibodies Ab205 and Ab756 have been described previously  and were used as clarified sera for immunoblot or affinity-purified against the relevant immunogens for IF (immunofluorescence). Rhodamine-labelled affinity-purified Ng Fab (Ng Rhod-Fab) was obtained from Ab756 IgG, by papain digestion, Fc fragment removal with Protein-A–Sepharose, labelling with 5-6-carboxy-tetramethyl-rhodamine-N-hydroxysuccinimide (Sigma) and affinity purification in a Ng-affigel column using standard protocols and labelling procedures . A monoclonal mouse antibody to CaM was provided by Dr J.M. McDonald (Department of Pathology, University of Alabama, Birmingham, AL, U.S.A.). The cDNAs for Ng, Ng-C3,4,9S, Ng-S36A and Ng-S36D were a gift from Dr Dan Gerendasy (Department of Molecular Biology, Scripps Institute, La Jolla, CA, U.S.A.). The cDNA for Ng-I33Q was donated by Dr Dan Storm (Department of Pharmacology, University of Washington, Seattle, WA, U.S.A.). Ng-IQless, dNg (double Ng) and 4Cys-Ng cDNAs were made by PCR cloning in pcDNA3. Ng-IQless lacks residues 31–45 of the Ng rat sequence. dNg consists of two complete Ng polypeptides assembled in tandem. 4Cys-Ng features an optimized biarsenical-binding tetracysteine motif (MDFLNCCPGCCMEPSAC..) (shown in bold) inserted at the N-terminal of Ng . cDNAs of Ng and mutants were subcloned (HindIII/BamHI) into pcDNA3 and psGFP2-C1 (where GFP is green fluorescent protein) vector, provided by Dr T.W. Gadella (Molecular Cytology, Swammerdam Institute for Life Sciences, University of Amsterdam, Amsterdam, The Netherlands) . The rat cDNA of CaM was cloned by PCR into pcDNA3 (pcDNA3-CaM).
Animals were treated according to the local and EEC rules. Animals were deeply anaesthetized by intraperitoneal (i.p.) administration of pentobarbital and transcardially perfused with ice-cold PB (0.12 M phosphate buffer, pH 7.3) for 5 min, and then cold 4% PFA (paraformaldehyde) in PB for 30 min. Brain pieces were post-fixed overnight in the same fixative and 100 μm sections were then cut on a vibratome. Endogenous peroxidase was blocked with 0.1% H2O2 in PB (20 min) and free-floating sections successively incubated in: (i) 5% heat-inactivated HS (horse serum) plus 0.2% (w/v) saponin in PB for 2 h; (ii) affinity-purified Ng antibody (Ab205, 1:500) in the same buffer overnight; (iii) biotinylated donkey anti-rabbit IgG (1:200) for 2 h; and (iv) ABC Elite reagent (Vector) for 1 h. Immunostaining was visualized using DAB (3,3′-diaminobenzidine) (Sigma) and H2O2 in PB. Non-specific staining was monitored in sections where no primary or no biotinylated secondary antibody was added. Sections were then either dehydrated and mounted with Eukitt (Merck) for wide-field light microscopy or enhanced with osmium tetroxide and cut in an ultramicrotome to obtain semi-thin sections (1 μm thick) that were observed at the optical microscope and further cut to obtain ultrathin sections (60–100 nm thick) that were analysed by EM (electron microscopy) (JEOL JEM1010) . Images were captured with a Photometrics Coolsnap FX digital camera attached to a Zeiss Axiovert200M microscope or with a Gatan BioScan camera for EM.
Cell culture and transfection
NIH 3T3 (mouse embryonic fibroblast), HEK-293 (human embryonic kidney 293) and HeLa (human cervical carcinoma) cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 1 mM glutamine and antibiotics (penicillin and streptomycin at 50 units/ml). For transfection, 1.5×105 cells were plated in P35 dishes containing (or not) four round coverslips and were transfected the following day with 1 μg of vector DNA and Lipofectamine™ 2000 (2:1 ratio to DNA) in OptiMem for 150 min. The cells were returned to their normal growth medium and used 24–48 h later. ReAsH-EDT2 (where EDT2 is 1,2-ethanedithiol) labelling of cells expressing 4Cys-Ng was performed as described in . Briefly, cells were rinsed twice with Hank's medium (137 mM NaCl, 5.3 mM KCl, 0.45 mM KH2PO4, 0.35 mM Na2HPO4, 1.25 mM CaCl2, 0.8 mM MgSO4, 1 mM NaHCO3, 1 mM pyruvate, 0.6% D-glucose and 10 mM Hepes, pH 7.3), incubated at 37 °C for 60 min in the same medium containing 1 μM ReAsH and 2 μM EDT2, washed 4×2 min with 250 μM BAL (2,3-dimercapto-1-propanol) in Hank's medium and returned to their original culture medium. 4Cys-Ng was recognized by affinity-purified Ng antibodies Ab205 and Ab756, either by immunoblot or IF. Specific in vivo labelling of tetracysteine motifs was assessed using cells expressing mCFP-tetraCys , a cyan fluorescent protein with an optimized tetracysteine motif at its C-terminus (provided by Dr J. Llopis, Lab. de Fisiología Experimental, Centro Regional de Investigaciones Biomédicas, Universidad de Castilla-La Mancha, C/Almansa, 14. 02006 Albacete, Spain). Cultures of hippocampal neurons were prepared from E18 (where E is embryonic day) rat embryos essentially as described in  and maintained in Neurobasal medium with B27 supplement and Glutamax. At DIV (days in vitro) 3, 2 μM Ara-C (cytosine arabinoside) was added to the culture medium and 50% of the medium volume was replaced once a week. For biochemical studies, neurons were seeded at 35000 cells/cm2 in P35 plates and at 15000 cells/cm2 for IF (18 mm coverslips in 12× multiwell plates). Transfection of cultured neurons was performed at DIV7 using Lipofectamine™ 2000.
Coverslips used for IF were cleaned in 65% nitric acid overnight, extensively washed with distilled water and heat-sterilized at 120 °C in an oven. Unless stated, the IF procedure was as follows (normal IF): after a quick rinse with Hank's medium, cells were fixed with 2% PFA in Hank's medium for 20 min, washed three times with PBS and incubated in 0.1 M glycine in PBS to inactivate free aldehydes (15 min). Blocking non-specific binding and permeabilization was achieved by incubating in blocking buffer [5% heat-inactivated HS, 0.2% (w/v) Triton X-100 in PBS] for 30 min. Primary and secondary antibody incubations were performed in blocking buffer for 90 and 45 min respectively, followed by three 5 min washing steps in PBS. To visualize cell nuclei, an additional incubation with To-Pro3 (1:500, 30 min) was made, followed by extensive washing. Finally, coverslips were rinsed in distilled water and mounted in Mowiol. Negative controls were made by omitting the primary antibody. Visual inspection and image acquisition was mostly performed using 63× PlanApo 1.4 numerical aperture and 100× PlanApo 1.4 numerical aperture objectives in a Zeiss Axiovert200M microscope equipped with a Photometrics Coolsnap FX monochrome camera and MetaMorph 6.1 software (Universal Imaging). In addition, confocal images were acquired with a Zeiss LSM 510 confocal microscope using multitrack scanning mode, at 1024×1024 (no zoom and 2× averaging) using the following laser lines: 488 nm (30 mW Argon, 8–10%), 543 nm (1.2 mW HeNe, 30%) and 633 nm (5 mW HeNe, 50%). Digital image processing (averaging, background subtraction, shading correction) and montage were performed using ImageJ software (http://rsb.info.nih.gov/ij/). To quantify colocalization and Pearson's coefficients, images were registered using the TurboReg plugin (http://bigwww.epfl.ch/thevenaz/turboreg/) and then analysed with the intensity correlation analysis and colocalization threshold plugins stored at the WCIF ImageJ collection of plugins (http://www.uhnresearch.ca/facilities/wcif/imagej/).
Determination of Ng Nuc/Cyto (nuclear/cytoplasm) ratios by image analysis
Cells expressing GFP, GFP–Ng or GFP–Ng mutants were fixed and their nuclei counterstained with To-Pro3. Cells exhibiting an unusual size, morphology or expression levels were discarded. Using the confocal microscope and the 63× lens, a single optical slice was chosen from each imaged cell at the height where the nucleus exhibited the longest diameter (see Figure 4A). The images obtained using 488 nm (GFP) and 633 nm (ToPro3) laser excitation were background subtracted and the above-background areas were selected and used to outline the cellular and nuclear perimeters respectively. The Nuc region was subtracted from the total cell area to obtain the Cyto region. Finally, the mean fluorescence intensity elicited by 488 nm excitation was determined for both regions and the Nuc/Cyto ratio was calculated.
Hippocampal slices, tissue and cell culture extracts
Adult male rats were killed and their brains quickly removed, cooled and cleaned of meninges and blood clots. Both hippocampi were carefully exposed, unrolled along the hippocampal fissure, dissected out and then cut into 350 μm slices using a McIlwain tissue chopper. The slices were equilibrated for 60 min in air bubbling Hank's medium and transferred to flat-bottom tubes (3 slices per tube), containing 0.5 ml of fresh oxygenated Hank's medium. For treatments, drugs were added directly to the medium and the slices were incubated at 30 °C with shaking. Total extracts of hippocampal slices, fresh tissue or cell cultures were obtained by homogenization in 2% SDS, 5 mM Tris/HCl, pH 6.8, 1 mM EDTA and 10 mM 2-mercaptoethanol, heated at 90 °C (2 min) and centrifuged.
Aliquots from extracts were mixed with electrophoresis sample buffer, heated at 90 °C for 1 min, separated by SDS/PAGE and processed for immunoblot as described previously . Primary antibodies used were: Ng Ab756 (rabbit, 1:40000), phospho-Ng(Ser36) (rabbit, 1:5000, Upstate), anti-GFP (mouse, 1:10000, Roche) and anti-CaM (mouse, 1:20000). Immunoreactive bands were visualized using biotinylated secondary antibodies (1:5000, Jackson), Vectastain ABC Elite (Vector) and ECL® (enhanced chemiluminescence) detection (ECL® kit; Amersham) or DAB/H2O2.
The Ne-Per reagents (Pierce) were used to prepare nuclear and cytoplasmic extracts from rat brain tissue and cell cultures. We checked other methods to prepare nuclei-enriched fractions, including isolation of nuclei using a sucrose-density barrier . With all of them, highly purified nuclear fractions were obtained, as demonstrated by the distribution of nuclear and cytoplasmic markers. In general, a compromise between the purity of isolated nuclei and the time of fractionation needs to be reached, especially for small size components that may leak from nuclei during the procedure. In our hands, the Ne-Per reagents consistently gave the lowest values of nuclear Ng in brain tissue and good separation of nuclear and cytoplasmic markers (results not shown).
PTZ (pentylenetetrazole) treatment
Adult rats (3 months old) were injected with PTZ (Sigma) in saline at 45 mg/kg (i.p.) of body weight. Control animals received the same volume of saline. Treated animals developed strong convulsions, most of them at 2–3 min after PTZ injection, that were rated of degree 5, according to the scale of McIntyre and Racine . In the following 10–15 min, some animals had more seizures, coming in short periods and declining in strength, and then gradually entered into a period of rigidity. Animals that suffered severe and prolonged seizures were administered a single dose of thiopental (25 mg/kg of body weight, i.p.) and separated from the experimental group. Rats were killed at different time points after PTZ injection, their brains removed and the hippocampus and cerebral cortex quickly dissected, weighed and frozen on dry ice. Nuc and Cyto extracts were separated from tissue samples using the Ne-Per kit (Pierce) and analysed by immunoblot.
Restriction enzyme digestions, DNA ligations, site-directed mutagenesis and other recombinant DNA procedures were performed using standard protocols. All DNA constructs were verified by DNA sequencing.
Experiments were normally repeated independently at least three times and typical results from IF and immunoblots are shown. To evaluate the differences between experimental groups, unpaired t test (two-tailed) was used. Unless stated, means±S.E.M. are represented. The differences were considered statistically significant when P<0.05 (*P<0.05; **P<0.01; ***P<0.001).
Ng accumulates at the cell nucleus
A number of GFP–Ng constructs were prepared to study the dynamics of Ng subcellular localization. Following GFP–Ng expression, a strong nuclear localization was noticed, which contrasted with the description of Ng as a predominantly cytoplasmic protein . To investigate this, the intracellular distribution obtained from direct GFP–Ng fluorescence was compared with that derived of indirect IF with Ng antibodies. As shown in Figure 1(A, top row), Ng immunoreactivity in GFP–Ng-expressing cells is observed at the cytoplasm but not in the nucleus, whereas GFP–Ng fluorescence is strong in the nucleus and less evident at the cytoplasm. This puzzling result was also obtained with different cell lines (HEK-293, HeLa, NIH 3T3), cultured hippocampal neurons (Supplementary Figure S1 at http://www.BiochemJ.org/bj/424/bj4240419add.htm), different Ng antibodies (Ab756 and Ab205) and constructs of Ng fused to several fluorescent proteins (results not shown). In addition, proteolysis of GFP–Ng was discarded by immunoblot analysis of total cell extracts with GFP and Ng antibodies (see Supplementary Figure S1).
Nuclear localization of GFP–Ng
Since GFP–Ng remains intact, it could be that Ng antibodies were not completely detecting GFP–Ng due to either poor antibody penetration or to epitope masking. To improve Ng titration, we used two strategies: (i) extend primary antibody incubations to 18 h or longer (long IF); and (ii) incubate live cells with 0.05% saponin for 30 s before fixation (sap+long IF), to extract cytoplasmic material and facilitate antibody access. As shown in Figure 1(A, middle row), longer primary antibody incubations allowed us to see some nuclear Ng immunostaining. Further, saponin permeabilization led to a nearly complete loss of cytoplasmic immunoreactivity, but also to enhanced detection in the nucleus (Figure 1A, bottom row). These results suggest that intact GFP–Ng is actually present in the nucleoplasm. However, they do not demonstrate the nuclear localization of Ng, since Ng localization could be altered by GFP tagging. To investigate this, we used the long IF procedure with Ng-expressing cells and found Ng immunoreactivity in the nuclei (Figure 1B, middle row). More clear evidence was obtained after saponin permeabilization, since Ng-expressing cells exhibited strongly fluorescent nuclei, although with most of the immunoreactivity washed out from the cytoplasm (Figure 1B, bottom row). We also found that poor detection of nuclear Ng in normal IF conditions was possibly due to limited antibody penetration and not epitope masking, since similar results were obtained using GFP-expressing cells and GFP antibodies (see Supplementary Figure S2 at http://www.BiochemJ.org/bj/424/bj4240419add.htm).
Since previous reports had been somehow contradictory [23,28,33,34], we looked for additional evidence to support the nuclear localization of Ng and, at the same time, to validate GFP–Ng as a Ng reporter. We prepared a construct for 4CysNg, a Ng mutant that carries an optimized tetracysteine tag at its N-terminus, which minimally disrupts its sequence and size . 4Cys-Ng, which can be visualized after ReAsH labelling, showed a strong nuclear accumulation that could not be detected by normal IF, as expected (Figure 1C). A different approach was to use smaller probes to titrate Ng, since they should have better intracellular penetration. Monovalent Fab fragments from immune sera retain the ability to bind to their original epitope and are three times smaller than the parent immunoglobulin molecules. Thus, an affinity purified rhodamine-labelled Fab fragment, Ng Rhod-Fab, was prepared from Ab756 and used to localize Ng in GFP–Ng-expressing cells. As shown in Figure 1(D), the labelling profile of Ng Rhod-Fab was practically identical with that derived from the fluorescence of GFP–Ng (the Pearson's coefficient for the colocalization was 0.996), indicating that the Fab fragment achieves higher titration levels than the parent IgG. This additional evidence further supports the view that Ng accumulates in cell nuclei and that GFP–Ng-derived fluorescence accurately reports the intracellular localization of Ng. Finally, when Ng Rhod-Fabs were used with Ng-expressing cells, Ng accumulation was also observed in the nucleus (see Supplementary Figure S2).
Next we investigated whether endogenous Ng also exhibits nuclear localization. For this, we measured Ng levels in purified Nuc and Cyto fractions from striatum, hippocampus and cerebral cortex. We found that Ng is present in the nuclear fractions of the three regions analysed, and that its levels gradually increase from postnatal day 7 to 32 and then decline slightly in the adult (Figure 2). Intracellular Ng localization was also studied by immunocytochemistry followed by optical or electron microscopy imaging. Ng expression is high in the forebrain  and, in particular, in the hippocampus and in layers II–IV of the cerebral cortex (Figure 3A). Ng could be clearly detected in nuclei of hippocampal (Figure 3E, thick sections) and cerebral cortex neurons (Figures 3B–3D, semithin sections). Interestingly, Ng concentration varied markedly in the nuclei of adjacent neurons (Figures 3C and 3D). This was also observed in CA2 neurons by EM imaging (Figures 3F and 3G), confirming that the nuclear levels of Ng can be differentially regulated in neighbouring neurons.
Nuclear localization of Ng in several brain regions during postnatal development
Immunocytochemical localization of Ng in the rat brain
Nuclear accumulation of Ng requires an intact IQ motif
We then studied which part of the Ng sequence is more relevant for its nuclear localization. We first analysed Ng levels in nuclear and cytoplasmic fractions from cells expressing several Ng mutants. However, the comparison was difficult due to differences in transfection efficiency and expression levels. On the other hand, observation at the fluorescence microscope suggested marked differences of nuclear localization among the GFP–Ng mutants. To quantify them, we used a simple procedure to measure the mean intensity values of fluorescence in the Nuc and Cyto regions of cells expressing GFP or GFP–Ng and mutants (Figure 4B). These values were used to calculate the Nuc/Cyto mean intensity ratio (see the Experimental section and Figure 4A). As shown in Figure 4(C), GFP–Ng-expressing cells showed a Nuc/Cyto ratio 2.5-fold greater than that of GFP, suggesting that Ng carries signals that promote its nuclear accumulation. The fact that GFP–dNg, which features two Ng molecules assembled in tandem, displayed a Nuc/Cyto ratio 3.5-fold greater than that of GFP further supported this view. On the other hand, the deletion of the IQ motif (GFP–Ng-IQless) resulted in lower Nuc/Cyto ratio values, significantly lower than those of GFP–Ng and similar to those of GFP, suggesting that the IQ motif carries the relevant signal needed for nuclear accumulation. It should be noted that the C-terminus of the Ng IQ motif (R43KKIK) shows a striking similarity with previously described NLSs (nuclear localization signals) . Addition of leptomycin B (25 nM, up to 6 h), a potent and specific nuclear export inhibitor, did not modify the Nuc/Cyto ratios of GFP, GFP–Ng, GFP–dNg and GFP–Ng-IQless (results not shown), suggesting that nuclear export mechanisms were not involved. Other GFP–Ng point mutants studied showed ratios similar to that of GFP–Ng, whereas N-terminal and C-terminal deletions progressively decreased the ability of GFP–Ng to show nuclear accumulation.
Nuclear localization of Ng depends on its IQ motif
Cytoplasmic CaM regulates Ng nuclear translocation in hippocampal neurons
Cultured hippocampal neurons express very small amounts of Ng  and this precludes its localization by IF. Thus hippocampal neurons were transfected at DIV7 and analysed at DIV15–18, a stage where neuronal maturation is mostly completed . As noticed above in the immunocytochemical study, the nuclear localization of GFP–Ng varied markedly among neighbouring neurons. Furthermore, by simple visual inspection, we noticed that GFP–Ng-I33Q and GFP–Ng-S36D showed strong nuclear localization, whereas nuclear concentration of GFP–Ng-S36A was low. In view of such heterogeneity and to establish a quantitative comparison among mutants, we decided to categorize neurons into three groups defined on the basis of their relative fluorescence intensity in the nucleus and cytoplasm: (i) the Nuc>Cyto group, showing a clear nuclear accumulation; (ii) the Nuc<Cyto group, with less fluorescence in the nucleus than the surrounding cytoplasm; and (iii) the Nuc≈Cyto group, gathering neurons that could not be clearly assigned to the other two groups (Figure 5, top panel). Consistent with our previous impression, we found that 23.81±2.07% of GFP–Ng neurons fell into the Nuc>Cyto group, and that this decreased to 10.02±3.25 for GFP–Ng-S36A and increased to 81.70±1.99 and 88.50±2.42 for GFP–Ng-S36D and GFP–Ng-I33Q respectively. Furthermore, 33.73±3.35% of the GFP–Ng neurons were sorted into the Nuc<Cyto group and this increased to 61.59±5.12 for GFP–Ng-S36A and decreased to 2.85±1.27 and 1.07±0.71 for GFP–Ng-S36D and GFP–Ng-I33Q respectively (Figure 5, lower panel). Finally, GFP-expressing neurons, included as a reference, were mostly assigned to the Nuc≈Cyto group, with values (68.98±5.93%) similar to those obtained for GFP–Ng-IQless (69.81±6.10%). To summarize, Ng mutants that do not bind CaM, such as Ng-I33Q and Ng-S36D, show strong nuclear accumulation, whereas Ng-S36A, whose binding to CaM cannot be inhibited by PKC phosphorylation, shows a weak nuclear localization.
Distinct nuclear localization of Ng mutants in hippocampal neurons
With these results, we hypothesized that CaM binding to cytoplasmic Ng could mask a putative NLS in the Ng sequence and diminish its nuclear translocation. In support of this view, we found that when CaM was overexpressed in NIH 3T3 cells that also expressed Ng, the levels of Ng present in purified nuclear fractions diminished (Figure 6A). Also, in hippocampal neurons, CaM overexpression significantly reduced the percentage of GFP–Ng and GFP–Ng-S36A neurons in the Nuc>Cyto group, therefore lowering the nuclear accumulation of the Ng forms that bind CaM (Figure 6B). However, it did not significantly alter the nuclear localization of the Ng forms that do not bind CaM, most notably GFP–Ng-I33Q and GFP–Ng-IQless. In other words, CaM overexpression only reduces the nuclear levels of Ng when it is able to bind Ng Since the Ng–CaM interaction is prevented when Ng is phosphorylated , we assayed the effect of PKC stimulation on the nuclear localization of GFP–Ng and mutants. Thus hippocampal neurons were treated with PMA or Bis-I (bisyndolil-maleimide-I), an activator and an inhibitor of PKC activity respectively (Figure 6C). As expected, those mutants that cannot be phosphorylated by PKC (GFP–Ng-S36A and GFP–Ng-S36D) did not alter their nuclear localization levels. Neither did neurons expressing GFP–Ng-I33Q, which can be phosphorylated by PKC but does not bind CaM, alter its nuclear localization. However, the nuclear localization of GFP–Ng significantly increased when PKC stimulation was coupled to phosphatase inhibition with OA (okadaic acid), suggesting that inhibition of dephosphorylation after PKC phosphorylation reduces Ng–CaM reassociation and leaves more Ng free to move into the nucleus.
CaM regulates Ng nuclear translocation
Ng translocation to neuronal nuclei is stimulated by synaptic activity
Once the nuclear localization of Ng had been demonstrated, we wanted to investigate its regulation in vivo. Using acute hippocampal slices, we found that PKC stimulation promotes a transient nuclear accumulation of Ng that peaked 15–20 min after the addition of PMA and slowly declined thereafter (Figure 7A). Also, in situations of increased excitatory synaptic activity, such as the incubation with 50 μM bicuculline, higher levels of nuclear accumulation were also observed (Figure 7B). These results suggest that increased intracellular Ca2+ levels and PKC activity elicited by synaptic activity would lead to a decrease in Ng-CaM levels and would leave more Ng free to move into the nucleus. To investigate the role of Ng phosphorylation in the mechanism of nuclear translocation, we analysed the specific activity of phosphorylated Ng in the nucleus and cytoplasm, using a specific phospho-Ser36-Ng antibody. As expected, Ng phosphorylation at Ser36 increases after PKC stimulation, and this is followed by an increased nuclear accumulation. In addition, phosphatase inhibition by OA not only led to higher levels of Ng phosphorylation, but also to a further increase in nuclear accumulation (Figure 7C, left panel). However, when the specific activity of Ng phosphorylation was measured in the nuclear and cytoplasmic fractions, no significant differences were found (Figure 7C, right panel). These results suggest that Ng phosphorylation itself does not signal or is not required for nuclear accumulation. Therefore, since PKC-mediated Ng phosphorylation is followed by nuclear accumulation, we propose that the role of Ng phosphorylation is to prevent the formation of Ng–CaM complexes and, consequently, to increase the pool of free Ng ready for nuclear translocation. This proposal, as suggested by the results obtained after phosphatase inhibition (Figures 6C and 7C), assumes that Ng dephosphorylation is very active and that phosphatase activity could play a role both in the magnitude and the duration of Ng nuclear transients.
Ng nuclear translocation is regulated by synaptic activity
Since the evidence suggested that excitatory synaptic activity favours the nuclear translocation of Ng, we investigated this hypothesis in live animals. For this, we treated adult rats with PTZ, a convulsant that elicits high levels of excitatory synaptic activity in the hippocampus and cerebral cortex , and brings about the rapid appearance of generalized seizures. The animals were then killed at several times after injection and the nuclear and cytoplasmic fractions, purified from the cerebral cortex and hippocampus, analysed by immunoblot. As shown in Figure 7(D), the Ng Nuc/Cyto ratio rapidly increases after PTZ administration in the hippocampus, reaching maximal levels at 30 min and then declining slowly during a period of 4 h. In the cerebral cortex (Figure 7E), two peaks of nuclear translocation were found, one at 15 min and the other at 120 min after PTZ. These results strongly support the hypothesis that excitatory synaptic activity stimulates Ng nuclear accumulation.
In the present study, we have shown that Ng is present in neuronal nuclei and actively translocates from the cytoplasm in response to excitatory synaptic activity. The nuclear localization of Ng was not clearly established in the literature. Represa et al.  observed occasional nuclear staining in some cerebral cortex pyramidal neurons and Watson et al.  reported nucleoplasmic immunolabelling in striatal neurons. However, other studies were not able to detect Ng in neuronal nuclei Ng [33,34]. More recently, the presence of Ng in a nuclear-enriched fraction has been described . We think that these discrepancies may be related to the difficulties that we found immunostaining intranuclear Ng. The inability of several Ng antibodies to recognize intranuclear Ng was somehow surprising. However, the results obtained after saponin permeabilization or using Ng Rhod-Fabs confirmed it. The strong nuclear staining observed for ReAsH-labelled 4Cys-Ng also supported this view, and suggested that GFP-tagging does not affect Ng subcellular distribution, therefore promoting GFP–Ng as a faithful reporter of intracellular Ng, as happens with other GFP-tagged proteins.
It is known that nuclear pore complexes allow free diffusion of molecules smaller than 40 kDa . Given its small size, Ng diffusion from the cytoplasm could well account for its nuclear localization. However, as shown in the present study, Ng is not merely present but substantially concentrated within the nucleus. We have shown that GFP–Ng is 2.5-fold more concentrated at the nucleus than GFP, and GFP–dNg even more concentrated (3.5-fold). Assuming that GFP distributes evenly in the cytoplasm and nucleoplasm, our results indicate the existence of active nuclear import mechanisms, intranuclear retention mechanisms, or both. Indeed, the C-terminal portion of the Ng IQ motif features a positively charged sequence (R43KKIK), which is conserved from birds to humans and could act as a NLS. The low nuclear presence of GFP–Ng-IQless, a mutant that lacks the putative NLS, further supports this view and strongly suggests a role of the IQ motif in Ng nuclear translocation. Once in the nucleus, Ng would quickly diffuse back to the cytoplasm due to its small size, unless it interacts with intranuclear components. This view is supported by the lack of effect observed for leptomycin B on nuclear accumulation. A rapid nucleocytoplasmic exchange of Ng could lead to the appearance of nuclear transients, as described in yeast for several transcription factors .
How do neurons regulate Ng nuclear translocation? We propose that the amount of Ng effectively transported into the nucleus is regulated by its interaction with cytoplasmic components. Ng association with membranes  or interaction with cytoplasmic CaM could mask its NLS and prevent its nuclear translocation. Similar mechanisms have been described for proteins that shuttle between the cytoplasm and the nucleus . For example, the nuclear localization of RGK proteins is modulated by their interaction with CaM or 14-3-3 proteins , in the same way as proposed in the present study for Ng. The evidence obtained in the present study to support this view is summarized as follows. First, neurons with high levels of Ng display a notorious nuclear accumulation, suggesting that the excess of Ng not retained in the cytoplasm is free for nuclear import. Secondly, since at least some Ng could be localized in nuclei due to passive diffusion through the nuclear pores, the finding of neurons with no intranuclear Ng (Figures 3B–3D) suggests the existence of retention mechanisms in the cytoplasm. Thirdly, Ng mutants that do not bind CaM, such as Ng-S36D and Ng-I33Q, are highly concentrated in nuclei of hippocampal neurons, whereas Ng-S36A, whose binding to CaM is not affected by phosphorylation, displays a weak nuclear localization. These results suggest that Ng binding to cytoplasmic CaM prevents its nuclear import. The nuclear localization differences observed between hippocampal neurons and NIH 3T3 cells for these mutants are most likely due to the high expression attained shortly after transfection in NIH 3T3 cells, which would saturate and overflow the cytoplasmic retention mechanisms. Fourthly, CaM overexpression in neurons and cell lines leads to a reduction of nuclear Ng, consistent with its role as a retention mechanism. Finally, stimulation of Ng phosphorylation in cultured neurons and acute hippocampal slices, which reduces the levels of CaM–Ng interaction, leads to increased nuclear accumulation of Ng.
Although more data are needed, the present study provides sufficient evidence to suggest that nuclear Ng has physiological relevance. Nuclear localization of Ng is developmentally regulated in the rat brain; it has very low levels in the first week of life that increase dramatically during the second and third weeks, a period of intense synaptogenesis. Also, intranuclear Ng levels are noticeably heterogeneous among neighbouring neurons, suggesting a relationship between nuclear accumulation and synaptic activity. Finally, the induction of a generalized increase in excitatory synaptic activity after PTZ administration led to a transient translocation of Ng to neuronal nuclei in the cerebral cortex and hippocampus. The next question arising is: what is the role of Ng in the neuronal nucleus? At the moment, we have no data to answer that question. Detailed intranuclear localization studies are needed to reveal specific association of Ng with intranuclear components. So far, Ng functionality has been related to Ca2+/CaM signal transduction [25,54], and there are several Ca2+/CaM-regulated processes within neuronal nuclei that affect long-term synaptic plasticity. Nuclear CaM is needed for activity-induced phosphorylation of CREB (cAMP-response element-binding protein) and the expression of activity-induced genes . A candidate to transduce nuclear Ca2+/CaM signalling in long-term plasticity is CaMKIV . On the other hand, PA is a prominent product of endogenous neuronal nuclear lipid phosphorylation , probably generated by nuclear diacylglycerol kinases . Since Ng binds to PA , Ng could have a role in intranuclear PA signalling.
Ca2+/CaM-dependent protein kinase
- Cyto region
days in vitro
green fluorescent protein
human embryonic kidney 293
- Ng Rhod-Fab
affinity-purified rhodamine-labelled Fab fragment for Ng
nuclear localization signal
- Nuc region
protein kinase C
Alberto Garrido-García was responsible for the main part of the experimental work, including cultures of cell lines and hippocampal neurons, immunoblots and immunofluorescence experiments. Beatriz Andrés-Pans was actively involved in the analysis of the nuclear localization of Ng and a number of mutants overexpressed in NIH 3T3 cells, using immunoblot and image analysis. Lara Durán-Trío was focused on the demonstration of the nuclear translocation of Ng using cultured hippocampal neurons and acute hippocampal slices, and also participated in the development of the 4Cys-Ng construct. F. Javier Díez-Guerra was involved in the immunocytochemistry experiments, the preparation of constructs and the rhodamine-labelled Fab probe and the fluorescence microscopy analysis. He also led the project and wrote the manuscript.
We thank Ms Beatriz Domingo and Dr Juan Llopis for their collaboration in the ReAsH labelling experiments. We thank Dr Noa Beatriz Martín-Cófreces for helpful revision of the manuscript.
This work was supported by a grant from the Spanish Ministry of Science and Technology [grant number BFI2002–01581]. We thank Fundación Ramón Areces for institutional support.
Present address: Depto Neurobiología Molecular, Celular y del Desarrollo, Instituto Cajal (CSIC), 37 Av. Doctor Arce, Madrid 28002, Spain