Neurogranin, a link between calcium/calmodulin and protein kinase C signaling in synaptic plasticity

Authors

  • F. Javier Díez-Guerra

    Corresponding author
    1. Departamento de Biología Molecular, Centro de Biología Molecular Severo Ochoa, Universidad Autónoma de Madrid –Consejo Superior de Investigaciones Científicas, Universidad Autónoma, 28049, Madrid, Spain
    • Centro de Biología Molecular Severo Ochoa, C/Nicolás Cabrera, 1, Campus de Cantoblanco, Universidad Autónoma de Madrid, Madrid 28049, Spain
    Search for more papers by this author
    • Tel: +34-911964612. Fax: +34-911964420


Abstract

Neurogranin (Ng) (also named RC3, p17 or BICKS) is a small protein originally identified in rat brain and abundantly expressed in several telencephalic areas, such as the cerebral cortex, hippocampus, amygdala, and striatum. In neurons, it is found concentrated at dendritic spines where it participates in synaptic signaling events through the regulation of calmodulin (CaM) availability. Ng features an IQ motif that mediates its interaction with CaM and phosphatidic acid (PA) and that is phosphorylated by protein kinase C (PKC) at serine 36 (Ser36). Ser36-phosphorylated Ng is unable to bind either CaM or PA. Ng knockout mice display an apparently normal phenotype; however, they show severe deficits in spatial and emotional learning and a decrease in LTP induction, mostly due to the attenuation of the signaling that depends on calcium/CaM kinase II (CaMKII), PKC, and protein kinase A (PKA) activation. The present review is an update on the most relevant information about Ng expression, localization, interactions, and modifications as well as on its role in synaptic plasticity. © 2010 IUBMB IUBMB Life, 62(8): 594–606, 2010.

Abbreviations

A2RE, A2 response element; AC, adenylate cyclases; Ca2+, calcium; CaM, calmodulin; CaMK, calmodulin kinase; CREB, cAMP response element binding protein; DGK, diacylglycerol kinase; DAG, diacylglycerol; DTT, dithiothreitol; SDS-PAGE, dodecyl sulphate polyacrylamide gel electrophoresis; EM, electron microscopy; ER, endoplasmic reticulum; GABA, γ-aminobutyric acid; hnRNP, heterogeneous nuclear ribonucleoprotein; LTD, long term depression; LTP, long term potentiation; Ng, neurogranin; NO, nitric oxide; NMDA, n-methyl-d-aspartate; HMG, non-histone high-mobility-group protein; NLS, nuclear localization signal; PA, phosphatidic acid; PLC, phospholipase C; PLD, phospholipase D; PKC, protein kinase C; PP, protein phosphatase; SNP, single nucleotide polymorphism; SD, sleep deprivation; TRE, thyroid hormone response element; UTR, untranslated region.

INTRODUCTION

The signaling pathways underlying the various forms of synaptic plasticity are commonly initiated by local changes in the amplitude and frequency of intracellular calcium (Ca2+) concentrations and transmitted by the ubiquitous and abundant calcium-binding protein calmodulin (CaM). CaM, acting as a signaling hub, feeds in and distributes Ca2+ signals through a variety of different effectors, as protein kinases and phosphatases, adenylate cyclases (AC) or nitric oxide (NO) synthase. There are several factors that affect the nature and complexity of the signaling derived from Ca2+/CaM; most importantly, the precise localization of the Ca2+ source and the local availability of CaM and its effectors. Other factors involved include the relative affinity of the effectors for Ca2+/CaM, the efficiency of the Ca2+ clearance mechanisms in the cytoplasm and the presence of Ca2+-sequestering proteins, such as parvalbumin, that confine the signaling events in space and time. Although CaM is believed to be in sufficient amounts to mediate Ca2+ signaling, the fact is that the overall CaM trapping capacity of its multiple effectors typically doubles its intracellular concentration, thus raising the possibility that CaM availability could constitute a limiting factor for intracellular calcium signaling (1, 2).

The existence of proteins that sequester CaM, either in its Ca2+-free (apo-CaM) or Ca2+-bound (Ca2+/CaM) forms, is well documented (3, 4). These proteins do not have any known activity as CaM effectors and their role is to control the availability of free CaM and possibly to limit the cellular damage in situations of excessive intracellular Ca2+. Additionally, these proteins have tags that direct them to specific intracellular locations and, as a result, contribute to accumulate CaM at or close to Ca2+ hot spots to favor CaM exchange with the appropriately located effectors. There are two such proteins highly expressed in the brain, namely GAP-43 (B50, neuromodulin) which is localized in axons and growth cones and Neurogranin (Ng)—also known as RC3, p17 or BICKS—that is present in the somatodendritic compartment and concentrates at dendritic spines. Both have been related to synaptic plasticity early since their identification. In the present article, relevant information on Ng expression, localization, interactions, post-translational modifications and its role in synaptic plasticity has been organized and reviewed. For further insight, the reader is referred to earlier reviews (3–6).

EXPRESSION

Ng was initially identified in bovine brain as a perchloric acid-soluble and protein kinase C (PKC) substrate (p17) (7) and its cDNA cloned shortly afterwards by screening a rat brain cDNA library with a cortex-minus-cerebellum subtracted cDNA probe (8). Regional distribution studies showed that Ng is an abundant protein in several telencephalic areas such as the cerebral cortex, hippocampus, amygdala, and caudate-putamen, whereas it is practically absent in the thalamus, cerebellum, brain stem, and spinal cord (9). In the rat, Ng is detected as early as embryonic day 18 (E18) in the piriform cortex and the amygdala. However, more than 90% of Ng brain content builds up postnatally and follows two sequential waves of expression: an early stage during the first week of life, of low-intensity expression, where Ng is mostly found in neuronal cell bodies and a juvenile stage, during the second and third weeks, of high-intensity expression, where Ng staining becomes apparent in the neuropil, reflecting its intracellular spreading into distal parts of the dendrites and dendritic spines (10). Ng expression is restricted to neurons and is not present in glial cells. However, although generally considered a neuron-specific protein, Ng is expressed at low levels in the lung, spleen, and bone marrow (Human Protein Atlas). High levels of Ng expression have been found in platelets (11) and moderate levels in B-lymphocyte (12). In the neocortex and hippocampus, Ng is only expressed by principal excitatory neurons (13, 14). In the cerebellum, originally believed to contain no Ng, this protein is present in a subset of GABAergic Golgi cells (15, 16) and also transiently expressed in a subpopulation of Purkinje neurons during development (E15-P20) (17). More recently, Ng has been detected in the granule cell inhibitory interneurons of the mouse olfactory bulb, a continuously renewed cell type that also express CaM kinase IV (CaMKIV), but not the calcium binding protein calretinin (18).

The mammalian Ng gene (nrgn) spans ∼12.5 kbp and contains four exons and three introns, which predict a 78-amino-acid protein with five amino acids encoded by exon 1 and the remaining 73 amino acids encoded by exon 2. The third and fourth exons contain 3′ untranslated sequences (19–21). The gene lacks TATA, GC, and CCAAT boxes in the proximal upstream region of the transcriptional start site, located at about 250 bp upstream from the AUG initiation codon. However, this region contains putative binding sites for transcription factors such as Sp1 and AP2, which are likely responsible for conferring basal transcriptional activity (22). The region −1508/+256 of the rat gene is responsible for transcription stimulation by phorbol esthers, but not by cAMP, arachidonic acid, vitamin D, retinoic acid, or thyroid hormones. This region has four conserved AT-rich segments of great homology to those present in the PKCγ gene, which shows a similar expression pattern in the brain during development. The chromosomal non-histone high-mobility-group protein I (HMG-I) has been shown to bind the AT-rich sequences within the promoters of PKCγ and Ng genes (23).

The Ng gene was the first neuronal gene whose expression was reported to be significantly altered by thyroid hormone deprivation (24). Thyroid hormone does not affect the timing of Ng expression, but is needed for full expression in the brain during the second stage of postnatal development and also in adult animals (25). In vivo, several neuronal populations are sensitive but others are insensitive to thyroid hormone (26). Thus, layer VI of the cerebral cortex, layers 2–3 of the retrosplenial cortex, the dentate gyrus and the caudate were affected by hypothyroidism, whereas upper layers of the cerebral cortex, the pyramidal layer of the hippocampus, and the amygdala were not (27). No responsive element for thyroid hormone (TRE) was found upstream of the transcription origin (28). However a receptor-binding site was localized in the first intron, with a sequence (GGATTAAATGAGGTAA) that is closely related to the consensus T3-responsive element of the direct repeat (DR4) type that binds heterodimers of T3 receptors (T3R) and 9-cis-retinoic acid receptors, but not T3R monomers or homodimers (29). These data substantiate the finding that Ng brain expression is reduced in aged and vitamin A-deprived animals and is responsive to retinoids (30–32). A second sequence (TTCCAAAATGG) located adjacent to the TRE in the first intron, was shown to bind a 121 kDa developmentally regulated protein and to interfere with T3 transactivation (33). In summary, the regulation of Ng expression by thyroid hormones and retinoids is complex and still needs further evidence for its complete understanding.

The Ng mRNA was also independently identified in a search for transcripts which were affected by sleep deprivation (SD) in the rat brain (34). Interestingly, a 40% decrease in Ng mRNA levels was observed in subcortical areas (basal ganglia), but not in the cerebral cortex or the hippocampus, after 24 H of SD (35). When the levels of protein were analyzed, a 37% decrease was found in the cerebral cortex, but not in other areas, therefore suggesting a differential regulation of protein and mRNA expression across brain regions. Although in the human brain there is only a single transcript of 1.3 kbp (36–38), two mRNA isoforms of about 1.4 and 0.9 kbp have been reported in rat and mouse species (8, 34). The short mRNA may be generated by premature transcription termination and polyadenylation within exon 4. On the other hand, the extended 3′ untranslated region (UTR) of the long transcript contains potential target sites of miRNA regulation (39). In situ hybridization studies have shown that the Ng mRNA is selectively translocated to dendrites (40), and that such translocation is impaired in the cortex of Alzheimer disease patients (36). Two reports have described cis-acting elements in the 3′-UTR of Ng mRNA involved in its transport into dendrites (41, 42). One of them identifies an A2 response element (A2RE)-like sequence (CCCUGAGAGCA) in the longer Ng transcript, at positions 1169–1180. This sequence is similar to the A2RE sequence in the mRNA of CaMKIIα that mediates its binding to heterogeneous nuclear ribonucleoprotein (hnRNP) A2, an essential component of the A2 pathway for dendritic targeting of mRNAs (43). This process is associated to translational silencing and in situ regulation by synaptic activity (44). The current data strongly point to a local regulation of Ng translation in dendrites by synaptic activity. However, more work is necessary to characterize the mechanisms that are physiologically relevant for the dendritic transport and translational control of the Ng mRNA.

PHYSICOCHEMICAL PROPERTIES AND INTERACTIONS

Ng is an acidic (pI = 5.6) 78 amino acid polypeptide whose sequence is highly conserved among mammalian species (+96% homology) (45–47). On sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE), Ng monomers migrate with an apparent mass of 15–19 kDa, far from its predicted mass of 7.5 kDa. Ng is a heat and acid-stable protein. Structural studies indicate that it exists in an unfolded or random form and that only a central region (G25–A42) shows motional restrictions compatible with the presence of a helical structure (48). This region, which contains basic and hydrophobic amino acids, shows high propensity to form an amphiphilic α-helix (49) and shares high homology to other IQ motif-containing proteins, such as GAP-43.

IQ motifs, defined by a loose consensus sequence (IQXXXRGXXXR), were first identified as the light chain binding sites in conventional myosins and subsequently recognized as CaM binding sequences in other proteins (50). IQ motifs display diverse modes of interaction with CaM that include Ca2+-dependent and independent binding (51). The IQ motif of Ng, as its homologous in GAP-43, contains a phosphorylation site for protein kinase C (PKC) (46, 47) and mediates an interaction with CaM that has ionic, hydrophobic, and structural components and displays two different affinity states, a high affinity state in the absence of Ca2+ and a low affinity state at micromolar or higher Ca2+ concentrations (52). Although the Ca2+ dependence of the interaction can be reduced under several circumstances (Vg. at low salt), the phosphorylation at the IQ motif completely prevents it.

PKC phosphorylates Ng at a single site (Ser36) within the IQ motif (46, 47). It is widely accepted that PKC is the only kinase responsible for Ng phosphorylation in vivo, although phosphorylase kinase is also able to phosphorylate Ng at Ser36 in vitro, but with a much lower efficiency (53). A synthetic peptide containing residues 28–43 of Ng is commonly used as a substrate in assays of PKC activity in tissue homogenates, further supporting the selective phosphorylation of Ng by PKC (54, 55). In vivo Ng phosphorylation in response to synaptic activation stimuli has been shown in several preparations such as brain slices (56, 57) and cultured hippocampal neurons (58). Using PKCγ knockout mice, it could be demonstrated that this PKC isoform is the only kinase involved in Ng phosphorylation upon high K+ depolarization or glutamate receptor activation (59). Interestingly, Ng and PKCγ show similar expression patterns during development and also share a highly coincident localization in the brain, both at the regional and the subcellular levels. Several studies have analyzed the in vivo phosphorylation status of rat brain Ng using mass spectrometry. While initial work proposed that phospho-Ng is the major isoform in vivo (60), later studies questioned these data and showed that the N-terminus was acetylated in vivo (61). Our results, using phospho-Ng specific antibodies, suggest the presence of low amounts of phosphorylated Ng in freshly extracted rat brain. Therefore, it seems reasonable that, if Ng plays a role as a CaM reservoir, most of it should be dephosphorylated to accomplish this task. Phospho-Ng can be efficiently dephosphorylated in vitro by Ca2+/CaM-activated protein phosphatase 2B (PP2B or calcineurin) and also by PP1 and PP2A (62). Our data suggest that the rate of in vivo Ng dephosphorylation should be high since incubation of hippocampal slices with protein phosphatase inhibitors such as okadaic acid or calyculin increase Ng phosphorylation levels even more than PKC activation by phorbol esters, therefore indicating that the lifetime of phospho-Ng after PKC activation is short. There is abundant evidence showing that Ng phosphorylation increases after LTP induction (56, 57, 63) and decreases after LTD induction (64, 65) and that, in both paradigms, the phosphorylation changes are transient.

Rat brain Ng contains four cysteines that are readily oxidized by many oxidants in vitro (66) and are responsible for the marked tendency of purified Ng to form oligomers in the absence of reducing agents. The rate of oxidation of Ng by nitric oxide (NO) is considerably faster than that of serum albumin, glutathione, or even DTT. Oxidized Ng displays an increased mobility on SDS-PAGE, reduced binding to CaM and significantly less phosphorylation by PKC. Individual mutation of Ng cysteines indicates that Cys51 preferentially forms an intramolecular disulfide bond with Cys9 and, with a lower probability, with Cys3 or Cys4 (67). This suggests that an unconstrained or relaxed conformation of Ng is necessary for a proper interaction with CaM and an efficient phosphorylation. Also, the increased mobility of oxidized Ng on SDS-PAGE agrees with the notion that Ng has a strong tendency to stabilize a rigid α-helix in hydrophobic and negatively charged environments. This occurs when bound to CaM and also in the presence of negatively charged detergents such as SDS, what explains its anomalous electrophoretic mobility. Using non-reducing SDS-PAGE to measure both oxidized and reduced Ng, it has been shown that N-methyl-D-aspartate (NMDA) receptor (NMDA-R) activation in brain slices leads not only to a transient increase in Ng phosphorylation but also to a transient enhancement of Ng oxidation which depends on NO synthase activity (68). Studies using hippocampal slices (69) and N2A neuroblastoma cells (70) have correlated Ng oxidation with increased Ca2+/CaM signaling. However, these data should be interpreted with caution since Cys3, 4, and 9 are well conserved among species, but Cys51 is not. Thus, in humans, intramolecular sulfide bonds could only be formed between Cys9 and either Cys3 or Cys4 and it is not clear that such bonds could affect the ability of Ng to bind CaM or to be phosphorylated by PKC, since the IQ motif and surrounding regions would not be affected by similar conformational restrictions.

SUBCELLULAR LOCALIZATION

As gene expression or post-translational modification, intracellular localization is a well established determinant of protein function that acquires a special relevance in nerve cells, which display a high degree of intracellular compartmentalization. Ng was initially tagged as a “soluble” protein. The term “soluble” is sometimes confused with “cytoplasmic or cytosolic,” although it merely reflects the protein behavior during extraction and not its adscription to any intracellular location. Using specific antibodies, it was soon realized that Ng localizes in the neuronal soma and dendrites (9, 71) and is very rarely observed in axons (71, 72). Considering its small size, soluble character, and high intracellular levels, it seems reasonable to think that additional mechanisms would be involved in maintaining its somatodendritic confinement. One such mechanism could be in situ translation of Ng transcripts in dendrites that, if coupled to high turnover ratios, would lead to a polarized accumulation in dendrites. Additionally, intracellular sorting of Ng could be determined by interactions with intracellular components that are resident and promote its retention in the somatodendritic compartment.

Under the electron microscope (EM), Ng displays a typical granular distribution both in the cytoplasm and nucleoplasm of striatal (71), cortical, and hippocampal neurons (73). In dendrites, the strongest immunolabeling is found in spine heads, whereas in the cell body, Ng is most abundant in the perikaryal region. Dispersed labeling is often interrupted by highly electron-dense deposits that confer the granular appearance to the staining and suggest the presence of aggregated Ng. These granules, which gave to Ng its name, are often attached or in close proximity to mitochondria, the endoplasmic reticulum (ER), and the trans-Golgi network (73). The association of Ng to membranes was first suggested by in vitro experiments showing Ng binding to acidic phospholipids (74) and, later, by the identification of Ng in microsomal and synaptosomal fractions of the rat brain (75). More recently, Ng association with cellular membranes has been proposed to be mediated by its direct interaction with phosphatidic acid (PA) (76). PA is a minoritary cellular phospholipid that, in addition to its role as a precursor in the synthesis of other phospholipids, functions as a signaling molecule in several processes (77). PA generated from phospholipase D (PLD) or diacylglycerol (DAG) and diacylglycerol kinase (DGK) has been proposed to function as a local and transient membrane signal that recruits appropriate proteins to participate in signaling cascades (78). In this sense, PA could play a relevant role directing Ng accumulation into selected intraneuronal compartments, such as dendritic spines. Interestingly, the interaction between Ng and PA is inhibited by CaM (at low Ca2+ levels) and by phosphorylation at Ser36 (76), so both mechanisms could intervene in the regulation of the proposed PA-dependent recruitment of Ng to membranes. Recent work has shown that Ng accumulates near the extrasynaptic membrane in dendritic spines of CA1 neurons, thus supporting the proposal that Ng membrane targeting by PA could determine its intracellular localization (79).

Although some studies had reported the presence of Ng in neuronal nuclei (9, 71), other studies were not able to find nuclear Ng (73, 75). Recent work reconciles these results showing that detection of nucleoplasmic Ng is highly dependent on the experimental procedure employed (80). It also shows that Ng accumulates in nuclei when not retained by cytoplasmic CaM or PA and that the nuclear import depends on a cluster of basic amino acids located at the C-terminal of the IQ motif acting as a nuclear localization signal (NLS). Further, the study shows that the nuclear translocation of Ng is regulated by synaptic activity and postulates a role for Ng in transcriptional regulation in neurons acting on intranuclear CaM and PA-dependent signaling pathways (81, 82).

MECHANISMS AND FUNCTION

There is a clear correlation between Ng levels and the cognitive function. For example, ageing and hypothyroidism are conditions that are associated with cognitive deficits and also with decreased levels of Ng in pyramidal neurons (83, 84). Mice deficient in Ng grow normally and present apparently normal phenotypes (85, 86). Nevertheless, they show severe deficits in visual-spatial learning and a marked tendency towards stress and anxiety. In the Morris's test, Ng−/− mice are unable to establish the search criterion to reach the submerged platform and they swim without showing clear signs of orientation (87). Also, these mice present alterations in the induction of LTP and reduced levels of phospho-CaMKIIα (88, 89).

It has been clearly demonstrated that Ng role in synaptic plasticity depends on its ability to bind and sequester CaM at dendritic spines (Figure 1). On the other hand, postsynaptic levels of free Ca2+/CaM have been shown to be a limiting factor during the induction of LTP (2). Therefore, it seems obvious that Ng should affect synaptic transmission by regulating the local availability of CaM and controlling the spatiotemporal patterns of postsynaptic Ca2+/CaM dependent signaling (90). Experimental evidence supporting this scheme is multiple. For example, injection of antibodies raised against Ng into hippocampal CA1 pyramidal cells prevents the induction of LTP without affecting post-tetanic potentiation (91). Also, knockout of Ng leads to a large decrease in the LTP induced by a single 100 Hz, 1 second tetanus, whereas LTD is slightly enhanced (85, 89). More recently, it has been shown that Ng overexpression in CA1 neurons enhances postsynaptic sensitivity and potentiates synaptic transmission, whereas in situ silencing of endogenous Ng expression blocked LTP induction (79). Interestingly, the synaptic enhancement induced by Ng occluded LTP, in the same way as postsynaptic microinjection of Ca2+/CaM does (2), therefore reinforcing the idea that enhanced synaptic strength induced by Ng is due to increased CaM recruitment to the synapse (92).

Figure 1.

Schematic diagram illustrating Ng involvement in postsynaptic signaling mechanisms.

CaM diffusion from dendrite shafts to spines is slow and most likely not relevant for LTP induction. Then, the proposed mechanism assumes that Ng, which is already present in dendritic spines, concentrates CaM to participate in postsynaptic signaling processes. In this way, dendritic spines containing more Ng would trap more CaM that, upon stimulation, would be locally released to saturate signaling pathways, such as CaMKII, leading to synaptic potentiation. Conversely, spines with less Ng would release less CaM, which would only be sufficient to activate the signaling pathways associated with synaptic depression, such as calcineurin, whose requirements for Ca2+/CaM are much lower. A variety of experimental data provide support to this mechanism. Most notably, the realization that the Ng knockout mouse does not generate normal amounts of autonomously active CaMKII (85, 86). Other studies with knockout mice have shown a generalized attenuation of synaptic signaling affecting PKC and protein kinase A (PKA) downstream targets such as mitogen-activated protein kinase (p42), 90-kDa ribosomal S6 kinase, and the cAMP response element binding protein (CREB) (69, 88). Furthermore, knockout of Ng greatly reduces the typical increase of intracellular Ca2+ observed after tetanic stimulation in hippocampal slices (89). Similarly, in Ng-deficient cultured cells, Ca2+ oscillations and responses to several stimuli were shown to be reduced (70, 93). Ng effects on intracellular Ca2+ have being explained using mathematical models that take into account the Ca2+ buffering capacity of CaM (94) or simply by the homeostatic action of Ca2+/CaM on the various mechanisms controlling intracellular Ca2+ dynamics, namely, Ca2+ entry, extrusion, and intracellular buffering (93, 95).

Despite its attractive simplicity and support, this functional model leaves several questions open. For example, what is the role of Ng phosphorylation? Several studies have shown that Ng phosphorylation increases during LTP induction (56, 57, 63) and decreases with LTD induction (64, 65). Experimentally, it is not easy to separate Ng phosphorylation and CaM binding, because mutations at Ser36 also affect Ng association to CaM. Since phosphorylated Ng does not bind CaM or PA in any circumstance, it could be speculated that the role of phosphorylation is to prevent the recapture of CaM when intracellular Ca2+ returns to basal levels. In a way, this role could be considered as a “memory effect,” as occurs with CaMKII autophosphorylation, since it would prolong CaM availability and signaling in the postsynaptic space beyond the time window of the original stimulus. PKCγ has been shown to be the isoform involved in the in vivo phosphorylation of Ng (59). Since PKCγ needs Ca2+ and DAG for activation, only those stimuli capable to activate ionotropic and metabotropic receptors simultaneously could significantly increase Ng phosphorylation. Thus, Ng phosphorylation, progressively increased by higher frequency stimuli, would promote sustained activation of synaptic transmission. In this scenario, Ng phosphorylation would act as a “slider” that shifts the synaptic frequency-response curve to the left (enhanced postsynaptic sensitivity) or to the right (decreased postsynaptic sensitivity) depending on the level of phosphorylation, as proposed earlier (86). Finally, there are several studies suggesting that phospho-Ng has a signaling role “per se” based on the evidence obtained in Xenopus oocytes injected with Ng mRNA (96, 97). These reports showed that Ng is able to directly stimulate G-protein coupled second messenger pathways only when phosphorylated by PKC, thus assigning a protagonist role to phospho-Ng as a signaling enhancer in addition to its role as intracellular CaM buffer.

Another unanswered question is how Ng localizes at or near the postsynaptic membrane. If Ng is responsible for recruiting CaM to dendritic spines, then specific mechanisms regulating Ng intracellular targeting should exist. One possibility is that N-terminal cysteines were palmitoylated and involved in membrane targeting, as shown with the homologous cysteines in GAP43. However, to my knowledge, in vivo acylation of Ng cysteines has not been reported to date. The mass-spectrometry data obtained in rat brain Ng (60, 61) and the high Ng solubility when extracted from brain homogenates do not support the existence of such modifications in vivo. In a previous study, we showed that Ng binds to PA and that the intracellular distribution of Ng is affected by overexpression of PLD, an enzyme that catalyzes the production of PA from phosphatidylcholine (PLC) (76). So, it seems feasible that PA generated at the postsynaptic membrane could be the signal that attracts Ng to active synapses, as happens with several members of the Ras signaling pathway (98, 99). In fact, DAG generated downstream of metabotropic receptor activation could be transformed to PA by diacylglycerol kinase ζ (DGK ζ), a DGK isoform that binds PSD-95 and is abundant at dendritic spines (100). In this way, stimulation of metabotropic receptors at the synapse would ultimately be responsible for Ng accumulation and the consequent increase of CaM availability and signaling.

More recent findings have contributed to strengthen the role of Ng in learning and memory processes. For example, songbirds learn to sing during the juvenile stage, generally imitating the singing of an adult. Once learned, their typical singing does not change throughout life. In the adult zebra finch, Ng expression is high in most telencephalic areas except in the song areas, where expression is very low. However, during the juvenile period, Ng expression in these areas is high but diminishes dramatically when the song learning period finishes (101). This finding suggests that Ng expression is important for the learning process, and also, that a drastically reduced Ng expression is necessary to stabilize synapses and circuits responsible for the invariant reproduction of the singing. Consistent with this view, Ng expression has been shown to be downregulated during the period of memory consolidation after Pavlovian fear conditioning (102). It is well known that environmental enrichment enhances hippocampal neurogenesis and cognitive functions (103). Recent work has shown that short-term (3 weeks) enrichment increased the expression of hippocampal Ng in wild type (Ng+/+) and heterozygous (Ng+/−) mice and that this increase positively correlates with improvements in LTP and performances in spatial and emotional learning tasks, whereas homozygous (Ng−/−) mice were only minimally affected (104). On the other hand, long-term (+10 weeks) exposure to enriched environments resulted beneficial to all three genotypes in preventing age-related cognitive decline. However, hippocampal slices of the enriched Ng−/− mice, unlike those of Ng+/+ and Ng+/− mice, did not show enhancements in the induction of LTP in the CA1 region (105), thus suggesting that cognitive function can be improved through LTP and Ng-independent pathways.

PATHOLOGIES AND FUTURE PERSPECTIVES

Ng has been associated to several pathologies. For example, it has been related to drug addiction based on results showing that inhibition of Ng expression by antisense nucleotides attenuates opioid tolerance (106) and disrupts acute opioid dependence in mice (107). However, the more compelling evidence relating Ng with disease has come from genetic studies. Thus, Ng has been associated to decrements in the cognitive function in the Jacobsen syndrome, a disease caused by the deletion of the end of the long arm of chromosome 11, whose patients exhibit variable degrees of mental retardation (108). More recently, evidence has accumulated correlating Ng with schizophrenia. Profound reductions in Ng have been observed in layers III and V of areas 9 and 32 of the prefrontal cortex, an important region involved in higher cognitive functions (109). Also, genotype distribution studies and genome wide scans have identified at least two single nucleotide polymorphisms (SNPs) that associate Ng to schizophrenia. None of them maps to the coding region of the gene. The SNP rs7113041 is located near the TRE localized at the first intron and was found in a male Portuguese population (110). On the other hand, the SNP rs12807809 is located 3,457 bases upstream of the translation initiation and was identified more recently in a study which combined SNP data from several large genome-wide scans (111).

Future applied research on Ng will surely focus on the development of strategies to effectively alleviate mental retardation and to improve memory and cognitive functions in general. However, to reach this point, more basic research is needed in all the areas covered by this review. Thus, it will be necessary to improve our current understanding of the mechanisms controlling Ng expression, both at the transcriptional and translational levels. Good hints in this regard might be obtained from developmental studies where the spatiotemporal patterns of Ng expression could be associated with concurrent events. Further structural and functional studies are also needed to unveil the versatility of IQ motifs and their multiple interactions. In this respect, it would be interesting to compare the activity of other IQ motifs that occur naturally in other proteins, such as PEP-19 (112), in different synaptic plasticity paradigms. Finally, as stated above, more evidence is necessary to characterize the signaling processes affecting Ng phosphorylation and intraneuronal localization and to better delineate the crosstalk points between the Ca2+/CaM and PKC signaling pathways. I hope that the present compendium on Ng would contribute to prompt new ideas and projects that promote the advance of our current understanding of Ng molecular physiology.

Acknowledgements

This work was supported by a grant from the Spanish Ministry of Science and Technology [grant number BFI2002–01581]. The author thanks Fundación Ramón Areces for institutional support.

Ancillary