Neuronal ELAV proteins enhance mRNA stability by a PKCα-dependent pathway

Abstract

More than 1 in 20 human genes bear in the mRNA 3′ UTR a specific motif called the adenine- and uridine-rich element (ARE), which posttranscriptionally determines its expression in response to cell environmental signals. ELAV (embryonic lethal abnormal vision) proteins are the only known ARE-binding factors that are able to stabilize the bound mRNAs, thereby positively controlling gene expression. Here, we show that in human neuroblastoma SH-SY5Y cells, neuron-specific ELAV (nELAV) proteins (HuB, HuC, and HuD) are up-regulated and redistributed by 15 min of treatment with the activators of PKC phorbol esters and bryostatin-1. PKC stimulation also induces nELAV proteins to colocalize with the translocated PKCα isozyme preferentially on the cytoskeleton, with a concomitant increase of nELAV threonine phosphorylation. The same treatment promotes stabilization of growth-associated protein 43 (GAP-43) mRNA, a well known nELAV target, and induces an early increase in GAP-43 protein concentration, again only in the cytoskeletal cell fraction. Genetic or pharmacological inactivation of PKCα abolishes nELAV protein cytoskeletal up-regulation, GAP-43 mRNA stabilization, and GAP-43 protein increase, demonstrating the primary role of this specific PKC isozyme in the cascade of nELAV recruitment. Finally, in vivo PKC activation is associated with an up-regulation of nELAV proteins in the hippocampal rat brain. These findings suggest a model for gene expression regulation by nELAV proteins through a PKCα-dependent pathway that is relevant for the cellular programs in which ARE-mediated control plays a pivotal role.

Long-lasting changes in cellular functions require reprogramming of protein synthesis as a result of cell signaling events that influence nuclear transcription and/or the fate of the transcribed mRNAs, ultimately leading to changed mRNA availability to the ribosome. Posttranscriptional mechanisms are emerging as key controllers of gene expression (reviewed in refs. 1 and 2) and are postulated to be critical for the localized changes in protein levels involved in cell differentiation and in the maintenance of the differentiated phenotype, especially in polarized cells such as neurons (3). Modulation of mRNA decay appears to be an efficient posttranscriptional way of controlling expression, because small changes in mRNA half-life can radically alter the abundance of a given mRNA and the amount of the relevant protein (4). Indeed, the decay rates of many mRNAs are governed by defined sequence determinants and by RNA-binding proteins (RBPs) acting on these determinants. The best-characterized regulative cis motifs in mammalian mRNAs are the AREs (adenine- and uridine-rich elements), which are found in the 3′ UTRs of mRNAs endowed with a rapid response to cell environmental stimuli, as in many cytokines and oncogenes (reviewed in ref. 5).

In the human genome, a general ARE consensus is present in 5–8% of expressed genes (6), and it represents a docking site for RBPs controlling mRNA stability, probably by modulation of exosome activity (7). ARE-dependent mRNA decay has been shown to be a target of at least two signaling cascades. The first is the p38 mitogen-activated protein kinase (MAPK)–MAPKAPK2 pathway, which, when activated, stabilizes ARE-bearing interleukin mRNAs (8–10), possibly through inactivation of the ARE-binding, mRNA-destabilizing RBP tristetraprolin (11, 12). The second pathway, which has been less investigated, is triggered by phorbol esters (phorbol 12-myristate 13-acetate, PMA) and calcium ionophore administration to culture cells, leading again to stabilization of ARE-bearing mRNAs (13–19). For its features, this pathway could involve the calcium- and diacylglycerol-regulated PKC isozymes, possibly resulting in the activation of a downstream function able to induce stabilization of ARE-bearing mRNAs. Fifteen years ago, Malter and coworkers (20, 21) identified a factor of ≈32 kDa, which they called AUBF for AU-rich binding factor, that was induced to bind ARE sequences after brief PMA treatment or calcium influx and was inactivated by dephosphorylation in peripheral blood mononuclear cells. AUBF was shown to be almost entirely located on polysomes when stimulated (22).

ELAV (embryonic lethal abnormal vision) proteins, or Hu antigens, represent the best-studied ARE-binding RBPs and are known from a substantial body of evidence to stabilize target mRNAs in the cytoplasm (reviewed in refs. 23 and 24). In vertebrates, HuB, HuC, and HuD are neuron-specific members of the ELAV family (nELAV proteins), whereas HuR is ubiquitously expressed; all four proteins are highly homologous in sequence, are ≈40 kDa in size, and contain three ≈90-aa-long RNA recognition motif-type RNA-binding domains (25). ELAV proteins shuttle between the nucleus and the cytoplasm (26–29) and, in the cytoplasm, can colocalize with polysomes and with the cytoskeletal apparatus (30–32). In PC12 cells, genetic knock-down by antisense RNA expression of one of the nELAV proteins, HuD, blocks the neurite outgrowth phenotype induced by PMA in this neuronal differentiation model and prevents PKC-mediated stabilization of the ARE-bearing, HuD-bound growth-associated protein 43 (GAP-43) mRNA (33).

In view of these and other features of ELAV RBPs, we hypothesized that these proteins represent a final target of the signaling cascade involving PKC and resulting in stabilization of ARE-bearing mRNAs. Employing a human neural cell model, SH-SY5Y neuroblastoma cells, and focusing on the well described activity of the nELAV proteins on the GAP-43 mRNA, we found that activation of PKCs by diacylglycerol analogues is indeed able to promote nELAV nuclear export, nELAV up-regulation, and nELAV cytoskeletal colocalization with the PKCα isozyme as early events, resulting in their increased phosphorylation. These changes were associated with the already described stabilization of GAP-43 mRNA. Unexpectedly, we also found that nELAV proteins activated in this way promote a compartmentalized increase of the GAP-43 protein. These results allowed us to define a previously unrecognized pathway, which we showed is also susceptible to in vivo pharmacological modulation, for the activity of nELAV proteins on their target mRNAs as positive modulators of gene expression in neural cells.

Materials and Methods

Cell Cultures. SH-SY5Y human neuroblastoma cells were grown in MEM (Eagle's minimal essential medium) supplemented with 10% FCS/penicillin/streptomycin/nonessential amino acids/sodium pyruvate (1 mM) at 37°C in an atmosphere of 5% CO₂ and 95% humidity. The PKCα-depleted, stably transfected SH-SY5Y-derived cell line (KD SH-SY5Y), was provided by Thomas B. Shea (McLean Hospital, Boston) and was obtained by transfection of a PKCα sequence in the antisense orientation (34). The KD SH-SY5Y cell line was maintained in the same medium supplemented with the selecting agent G418 (400 μg/ml, GIBCO/Life Technologies).

Treatments. In vitro. SH-SY5Y human neuroblastoma cells were exposed to the solvent (DMSO) or to 100 nM PMA (Sigma) for 15 min, as indicated in Results. The Ca²⁺-dependent PKCα inhibitor Gö6976 (Calbiochem) was used at 2 μM concentration. 5,6-Dichlorobenzamidazole riboside (DRB) (Sigma) was used at a concentration of 50 μM. Treatments were performed in Krebs–Ringer saline solution (pH 7.4) containing 20 mM Hepes, 125 mM NaCl, 5 mM KCl, 1 mM Na₂HPO₄, 1 mM MgSO₄, 1 mM CaCl₂, and 5.5 mM glucose. When the cells were prepared for SDS/PAGE, the treatments were stopped with buffer A, which contained 20 mM Tris (pH 7.4), 2 mM EDTA, 0.5 mM EGTA, 50 mM mercaptoethanol, 0.32 mM sucrose, and a protease inhibitor mixture (Roche Diagnostics) at the dilution suggested by the manufacturer. In vivo. Eight-week-old male Wistar rats (250 g) received intracerebroventricular injections of DMSO (vehicle) or 2 μM bryostatin-1 dissolved in DMSO. Animals were killed 24 h after treatments.

Preparation of the Cellular Fractions and Western Blotting. Subcellular fractioning (cytosol, membrane, and cytoskeleton) and Western blots were performed as described in ref. 35, with slight modifications. See Supporting Materials and Methods, which is published as supporting information on the PNAS web site, for Western blotting conditions.

Immunoprecipitation. SH-SY5Y human neuroblastoma cells were incubated at room temperature in Krebs–Ringer saline solution with or without 100 nM PMA at different times. Immunoprecipitation was performed according to a previously published protocol (36), with minor modifications. For more information, see Supporting Materials and Methods.

Immunocytochemistry. Cells plated on coverslips were treated with 100 nM PMA or bryostatin-1 in Krebs–Ringer saline solution for 15 min, whereas control cells were incubated with DMSO (solvent). For detailed procedures see Supporting Materials and Methods. Briefly, a polyclonal antibody diluted 1:50 was used to recognize PKCα or α-tubulin; the mouse biotin-conjugated 16A11 diluted 1:50 was used as anti-nELAV (Molecular Probes). FITC-conjugated rabbit anti-IgG antibody (Calbiochem) was diluted at 1:4,500 to detect PKCα and at 1:200 for α-tubulin; Texas red-conjugated streptavidin (Calbiochem) was diluted at 1:100. Cells were counterstained for DNA with a 0.1 μg/ml Hoechst 33342 solution.

Real-Time Quantitative RT-PCR. Total RNA was extracted from SH-SY5Y and KD SH-SY5Y cells and from their cytosolic, membrane, and cytoskeletal subfractions with TRIzol reagent (Invitrogen), treated with DNase, and subjected to reverse transcription following standard procedures. PCR amplifications were carried out by using a Lightcycler instrument (Roche Diagnostics) as described in ref. 37 with primers designed on the 3′ UTRs of the human GAP-43 and ribosomal protein L6 (RPL6) mRNA sequences by using primer3 software (www-genome.wi.mit.edu/cgibin/primer/primer3_www.cgi). Primer sequences were as follows: GAP-43, 5′-GAGCCTGTCTCTCCCTACCC-3′ (upstream) and 5′-TTGGGATCTTTCCTGCTTTTT-3′ (downstream); RPL6, 5′-CACAAATTTTACCAAAAATCAAA-3′ (upstream) and 5′-TTTAGAACACCAATTTGTGAGGA-3′ (downstream). The RPL6 mRNA was chosen as the reference mRNA on which GAP-43 was normalized because it remained substantially stable in the 8-h time frame of the experiments with DRB (data not shown).

Immunohistochemistry. Rat brains were immediately removed and frozen to –80°C. Twenty-micrometer-thick cryostat sections were processed as described in ref. 37. Nonspecific sites were blocked, and the slices were exposed to the mouse biotin-conjugated 16A11 anti-Hu (Molecular Probes) primary antibody diluted at 1:20 at 4°C overnight, followed by incubation with Streptavidin Alexa Fluor 488 (Molecular Probes) diluted at 1:400.

Data Analysis. Analysis of the data were performed with ANOVA followed by Student's t test as indicated by using the origin 6.1 statistical package (OriginLab, Northampton, MA). Differences were considered statistically significant when P ≤ 0.05.

Results

Diacylglycerol-Mimicking Compounds Promote Nuclear Export, Cytoskeletal Up-Regulation, and Cytoskeletal Binding of nELAV Proteins. Fig. 1A shows that 15 min of treatment with PMA and bryostatin-1, two compounds that mimic diacylglycerol in binding and activating the diacylglycerol-responsive PKC isozymes, induced in SH-SY5Y human neuroblastoma cells a dramatic decrease of nuclear immunostaining by the mAb 16A11, targeted to the nELAV (HuB, HuC, and HuD) proteins. In addition, Western blotting data indicate that PMA and bryostatin-1 at the concentration used (100 nM for both) promote substantial translocation of the PKCα isozyme from the cytosol to either the membrane (+50%, P < 0.0001 for PMA and +42%, P < 0.005 for bryostatin-1) or the cytoskeletal (+69%, P < 0.0001 for PMA and +46%, P < 0.05 for bryostatin-1) compartments, demonstrating their efficacy as activators of conventional PKC isozymes in this cell model. PKC stimulation by both compounds is also associated with an increase in the total cell levels of nELAV proteins detected by Western blotting (+56%, P < 0.01 for PMA and +60%, P < 0.05 for bryostatin-1; Fig. 1B Top). The same Western blotting repeated on SH-SY5Y cell fractions shows that the increased nELAV proteins are confined to the cytoskeletal compartment (+59%, P < 0.001 for PMA and +163%, P < 0.01 for bryostatin-1 in the cytoskeleton; Fig. 1B Bottom). The increase of nELAV proteins in the cytoskeletal fraction induced by PKC stimulation is accompanied by the establishment of their physical association with the cytoskeletal protein α-tubulin, as demonstrated by both immunoprecipitation and immunocytochemistry assays. PMA treatment induced the appearance of the nELAV signal in the cytoskeleton immunoprecipitated with an α-tubulin antibody (+47%, P < 0.05) but not with an isotype-matched irrelevant antibody (Fig. 1C), and the same result (not shown) was obtained by immunoprecipitating with the 16A11 mAb and detecting with the α-tubulin antibody (+40%, P < 0.05). This PKC-dependent interaction was confirmed by confocal imaging of SH-5YSY cells that were coimmunostained for α-tubulin and nELAV proteins, with the appearance of a superimposed signal only in the PMA-treated samples (Fig. 1D).

Fig. 1.

Up-regulation of nELAV proteins and nuclear export after PKC stimulation. (A) Confocal fluorescence microscopy representative images of SH-SY5Y neuroblastoma cells treated with biotin-conjugated anti-nELAV (red) and Hoechst 33342 dye (blue), which stains the nuclei. Control (CTR), PMA-treated, and bryostatin-1-treated (BRY) cells are shown; both treatments were at a concentration of 100 nM for 15 min. (Scale bars: 8 μm.) (B) Representative Western blots of nELAV and α-tubulin (αTUB) proteins in whole cells and after biochemical fractionation of control, PMA-, and bryostatin-1-treated cells (n = 6 for all experiments). (C) Representative Western blots of nELAV proteins after immunoprecipitation with either an anti-α-tubulin antibody or an irrelevant IgG2 isotype antibody (as negative control) in the cytoskeletal fractions of control and PMA-treated cells (n = 4). (D) Confocal fluorescence microscopy representative images show α-tubulin (green) and nELAV (red) proteins in control and PMA-treated (100 nM for 15 min) SH-SY5Y neuroblastoma cells. Merged images (yellow) indicate the degree of colocalization of α-tubulin and nELAV proteins. (Scale bars: 8 μm.)

Activated PKCα Associates with nELAV Proteins and Increases Their Threonine Phosphorylation. We next sought to determine whether the biological effects on nELAV proteins induced by PKC stimulation could be due to a direct action of specific members of this serine/threonine protein kinase family on the nELAV targets. Of the PKC isozymes regulated by diacylglycerol and its mimicking compounds (for a review, see ref. 38), we focused our attention on PKCα, because down-regulation of this isozyme has been shown to be incompatible with neuronal differentiation of SH-SY5Y cells (39), a process that is known to depend on the nELAV proteins in other neuronal differentiation models (29, 33). Immunocytochemical analysis (Fig. 2A) shows a clear PMA-induced colocalization of the PKCα and nELAV signals. This result is confirmed by the experiments reported in Fig. 2B, where the cytoskeletal fraction of SH-SY5Y cells is immunoprecipitated with the anti-ELAV 16A11 mAb, and PKCα is then detected in the pellet in the presence and absence of PKC stimulation by PMA. The association with PKCα was dramatically increased when PKC was activated (+222%, P < 0.05 in the cytoskeleton), as also confirmed by the reverse experiment (not shown). Taken together, these results and the previous ones suggest that treatment of neuroblastoma cells with diacylglycerol-mimicking compounds promotes an increase and a cytoplasmic recruitment of nELAV proteins specifically at the level of the cytoskeletal network, where they associate with the translocated PKCα isoform.

Fig. 2.

Redistribution of PKCα and association with nELAV proteins after stimulation. (A) Confocal fluorescence microscopy images show PKCα (green) and nELAV (red) proteins in control and PMA-treated (100 nM for 15 min) SH-SY5Y neuroblastoma cells. Merged images indicate a colocalization of PKCα and nELAV proteins after PMA exposure (increase of the yellow signal). (Scale bars: 8 μm.) (B) Representative Western blots of the PKCα protein after immunoprecipitation with the nELAV antibody in the cytoskeletal fractions of control and PMA-treated cells (n = 3). (C) Representative Western blots of nELAV proteins after immunoprecipitation with a phosphothreonine-specific antibody in the cytoskeletal fractions of control and PMA-treated cells (n = 3). For immunoprecipitations in B and C, an irrelevant isotype-matched IgG was used as negative control (not shown).

To explore the functional consequence of the formation of this complex, we looked at the nELAV levels in the cytoskeletal fraction of cell lysates that were immunoprecipitated with anti-phosphoserine and phosphothreonine antibodies or reversely pulled down with the 16A11 mAb and analyzed with phosphoserine and phosphothreonine antibodies. Only the threonine phosphorylation of ELAV proteins increased after PMA treatment (+106%, P < 0.01; Fig. 2C). Therefore, PKC activation by PMA enhances threonine phosphorylation of nELAV proteins in SH-SY5Y neuroblastoma cells.

Recruitment of nELAV Proteins Requires PKCα Activation. To substantiate the correlative evidence of PKCα involvement in nELAV protein activation, we wanted to establish the requirement of this specific isozyme for the effects of PMA treatment on nELAV proteins. We adopted two models of PKCα knock-down, a genetic model represented by a clone of the same SH-SY5Y cells obtained by stable antisense expression of the PKCα ORF (34) and a pharmacological model of inhibition by the selective PKCα inhibitor Gö6976 (40). In the untreated, knocked-down SH-SY5Y neuroblasts (PKCα KD), as expected, PKCα protein levels measured by Western blotting were reduced in all cell fractions, being almost undetectable in the cytoskeleton (–53%, P < 0.0001 in the cytosol; –42%, P < 0.0001 in the membrane; and –65%, P < 0.005 in the cytoskeleton; Fig. 3A), whereas the level of nELAV proteins was unchanged. In the PKCα KD SH-SY5Y cells and in the WT SH-SY5Y cells preincubated for 30 min with 2 μMGö6976, PMA treatment was ineffective in inducing an increase of nELAV protein levels in all of the cellular fractions examined (Fig. 3 B and C), producing no detectable change in the knocked-down model and a statistically insignificant change with the PKCα inhibitor. From these experiments, we conclude that the enhancing effect of PMA on nELAV protein levels is mediated by PKCα.

Fig. 3.

PKCα inactivation affects nELAV protein up-regulation. (A) Representative Western blots relative to the basal levels of PKCα and nELAV proteins in WT and PKCα antisense-expressing knocked-down (KD) SH-SY5Y neuroblastoma cells after biochemical fractionation. (B) Mean gray level ratios (mean ± SEM) of nELAV protein/α-tubulin immunoreactivity measured by Western blots in control and PMA-treated transfected SH-SY5Y cells. cyt., cytosol; memb., membrane; cytosk., cytoskeleton. (C) Mean gray level ratios (mean ± SEM) of nELAV proteins/α-tubulin immunoreactivity measured by Western blots in control and PMA-treated WT SH-SY5Y cells after 30 min of preincubation with 2 μMGö6976.

Stabilization of the GAP-43 mRNA and Increased Translation of the GAP-43 Protein at the Cytoskeleton by PKCα Activation. Among the neuronal genes whose mRNA is a target of the nELAV proteins, GAP-43 is by far the most well characterized. From in vitro and in vivo experiments, we know that the binding of nELAV proteins to GAP-43 mRNA induces its stabilization and an increase in the whole-cell GAP-43 protein levels, with profound biological effects. Therefore, we measured kinetically the GAP-43 mRNA decay rate in our cell model after the addition of DRB, a molecule able to specifically block transcription (41). The 15-min PMA treatment induced a marked, persistent stabilization of the GAP-43 transcript, raising the remaining intact mRNA from 30% to 50% of the initial levels after 8 h (Fig. 4A). Again, the dependence of this effect on PKCα was clearly shown with the demonstration of its absence in the PKCα-deficient KD SH-SY5Y cells, which bear a basically more unstable GAP-43 mRNA whose decay rate remains unchanged after PMA treatment (dotted lines in Fig. 4A). We then examined GAP-43 protein levels by Western blotting after 15 min of PMA-mediated PKC activation. We found that there was an increase in GAP-43 protein content, which, unexpectedly, was limited to the cytoskeletal fraction (+38%, P < 0.005; Fig. 4 B and C). This increase was prevented by Gö6976 pretreatment (Fig. 4D) and did not occur in the KD SH-SY5Y cells (not shown), again demonstrating the need of expressed and enzymatically active PKCα for its occurrence.

Fig. 4.

Effect of activated PKCα on GAP-43 mRNA stability and protein levels. (A) Kinetic determination by real-time quantitative PCR of the GAP-43 mRNA level in both WT and PKCα-knocked-down (KD) SH-SY5Y cells after the transcriptional block induced by DRB, with or without PMA stimulation (100 nM for 15 min). The levels of GAP-43 mRNA were normalized to those of the ribosomal protein L6 mRNA and expressed as a percentage of the initial steady-state GAP-43 mRNA levels. The data are reported in a semilogarithmic scale. (B) Representative Western blots of the GAP-43 protein after biochemical fractionation in control and PMA-treated (100 nM for 15 min) WT SH-SY5Y cells (n = 6). (C) Mean gray level ratios (mean ± SEM) of GAP-43/α-tubulin immunoreactivity measured by Western blots in control and PMA-treated WT SH-SY5Y cells. *, P < 0.005, Student's t test; n = 7. cyt., cytosol; memb., membrane; cytosk., cytoskeleton. (D) Mean gray level ratios (mean ± SEM) of GAP-43/α-tubulin immunoreactivity measured by Western blots in control and PMA-treated WT SH-SY5Y cells after 30 min of preincubation with 2 μMGö6976.

Pharmacological Modulation of the PKCα-Dependent nELAV Recruitment Pathway in the Mammalian Brain. Unlike phorbol esters, which are tumor promoters and therefore do not represent viable options for drug development, PKC activators belonging to the class of diacylglycerol-mimicking compounds have been proposed as potential drugs for applications in oncology (42) and neurodegeneration (43, 44). Bryostatin-1 is one of these compounds and has already been shown to be devoid of important side effects and actively studied in phase I and II clinical trials as an anticancer agent (45). Given the ability of bryostatin-1 to recruit nELAV proteins in our neuroblastoma cell model (Fig. 1), we checked whether this agonist would be effective in doing the same when administered in vivo into the mammalian brain. Fig. 5 shows that after 24 h of intraventricular injection of bryostatin-1 in rats, nELAV proteins undergo a sustained up-regulation in all of the hippocampal subregions, in a way reminiscent of their activation after spatial learning (46).

Fig. 5.

nELAV protein up-regulation after PKC stimulation in vivo. Confocal fluorescence microscopy images show nELAV proteins in control (CTR) and bryostatin-1-treated (BRY, 2 μM) rat hippocampal subregions. DG+, dentate gyrus.

Discussion

The dramatic biological activity of nELAV proteins has been clear since the first description of these genes, the report on the Drosophila elav locus (47) whose mutants dye as embryos for the absence of a fully developed nervous system. Subsequently, studies of ectopic expression in vertebrates (48) and experiments of gene overexpression/knock-down in “neural” model cell cultures (33, 49, 50) have revealed a key role for nELAV proteins in the differentiation of the neuronal lineage. In parallel with investigations about their biological activity, biochemical studies have established the molecular function of ELAV proteins as stabilizers of the target mRNAs, an activity demonstrated both for the neuronal and the ubiquitous (HuR) members of the family (for reviews, see refs. 23 and 24). Nevertheless, neither the mechanism(s) by which ELAV proteins counteract decay of bound mRNAs nor the cellular pathways by which this activity is triggered are known.

The present work sheds light on the second problem, demonstrating the existence of a pathway involving PKCα for nELAV protein recruitment and activity. The evidence comes from the demonstration that in cultured neuroblastoma cells, PKC activators induce nELAV proteins to be exported from the nucleus and to increase their total cell content at the level of the cytoskeleton, up-regulating at the same time and in the same cell compartment the GAP-43 protein, a nELAV target. Both of these increases can be prevented by the specific inactivation of the PKCα isozyme, suggesting that this isozyme, among the others of the family, plays the major role in the process. PKCα colocalization with nELAV proteins after stimulation and the increase in their threonine phosphorylation are also in agreement with a direct kinase activity of PKCα on HuB, HuC, and HuD proteins, which could take place at the five conserved putative threonine PKC phosphorylation sites that they share in the primary amino acid sequence. Given the very short time frame of nELAV and GAP-43 protein up-regulation induced by PKC stimulation (15 min), it is tempting to speculate that this PKC-induced control of gene expression is exerted at the translational level, which would be in agreement with the polysomal localization of nELAV proteins in the neuronal cytoplasm (30, 31, 49, 51) and with their proposed activity of translational enhancement (49, 52, 53). The up-regulation of the nELAV proteins themselves could be an autoregulation mechanism: We already know that the Drosophila ELAV protein is autoregulated at the mRNA level (54) and that murine HuB mRNA is bound by the HuB protein itself (55). An autologous binding activity is described for other RBPs acting at the translational level, such as fragile X mental retardation protein (56), poly(A)-binding protein (57), and ribosomal protein S15 (58). The action of PKC on nELAV protein targets at the translational level is also suggested by experiments of disruption of this activity by translational inhibitors (unpublished data).

But what are the possible biological outcomes of PKC-induced activation of nELAV proteins? First of all, nELAV activation by PKC agonists could be a key event in the program of neuronal differentiation during development. Expression of nELAV proteins during the shaping of the mammalian nervous system is a tightly regulated process (59, 60), and PKC-dependent signal transduction events, fundamental for specific aspects of nervous system development (61, 62), can provide, in view of our findings, an additional level of modulation of nELAV activity. In vitro models recapitulating neuronal differentiation provide correlative evidence of this interaction. In PC12 pheochromocytoma cells exposed to nerve growth factor, neurite outgrowth appears to be both PKC- (63, 64) and HuD-dependent (33), whereas neural-like differentiation of teratocarcinoma NT2 cells involves PKC function (65, 66) and is promoted by HuB overexpression (49).

Moreover, the functional link between PKC and ELAV proteins could be involved in another important cell program of nondevelopmental neural plasticity, nerve regeneration. All diacylglycerol-dependent PKC isoforms are intensely up-regulated in the growth cone of regenerating axons in the rat peripheral nervous system, an increase that takes place within hours of nerve injuries and co-stimulates axonal regeneration (67, 68), whereas PKC inhibition has been found to decrease the in vitro regenerative nerve growth potential (69, 70). The nELAV HuD protein was shown to be overexpressed in rat dorsal root ganglia neurons during peripheral nerve regeneration (71).

Finally, a substantial body of evidence supports a functional role for PKC activation and redistribution in invertebrate (72, 73), avian (74), and mammalian (75–81) models of memory. We previously proposed a physiological role for nELAV proteins in controlling gene expression in memory formation (37, 46). The evidence came from region-specific up-regulation of the nELAV proteins in hippocampal neurons of rodents trained in spatial discrimination tasks, which was associated with increased binding and increased steady-state levels of the GAP-43 mRNA. In vivo knock-down of HuC impaired learning performance and specifically prevented GAP-43 mRNA up-regulation, demonstrating the need of these biochemical events for memory formation (37). Interestingly, these effects were particularly evident in the cytoskeletal compartment of hippocampal lysates and were characterized by increased colocalization of the HuD protein and GAP-43 mRNA (46). A report extended hippocampal HuD up-regulation to another nonspatial learning paradigm, fear conditioning, demonstrating the generality of nELAV involvement in memory formation (82).

Therefore, the PKCα-dependent nELAV protein recruitment and activation described here could sustain nELAV protein function in the differentiation, regeneration, and learning nervous system programs. Our in vivo demonstration of hippocampal nELAV protein up-regulation by bryostatin-1 represents experimental evidence that the role of nELAV proteins in these programs could be pharmacologically modulated by acting on PKC by compounds already shown to be effective in models of neurodegeneration (43).

Further experiments are needed to address the relevance of this pathway for each of the mentioned major neural programs involving nELAV proteins and also for its role in nonneural systems involving the HuR protein and characterized by ARE-dependent modulation of gene expression. Moreover, the generality of the PKC signaling on ELAV protein recruitment and activation could allow the proposal of pharmacological approaches aimed both at the enhancement of this function in conditions of decreased neuronal plasticity and at its inhibition in conditions such as paraneoplastic syndromes (83) and colorectal cancer (84, 85) that are already associated with improper expression, respectively, of the neuronal and ubiquitous ELAV proteins.