Ribosomal Protein S6 Kinase Is a Critical Downstream Effector of the Target of Rapamycin Complex 1 for Long-term Facilitation in Aplysia

Abstract

Long-term facilitation (LTF) in Aplysia is a leading cellular model for elucidating the biochemical mechanisms of synaptic plasticity underlying learning. In Aplysia, LTF requires translational control downstream of the target of rapamycin (TOR) complex 1 (TORC1). The major known downstream targets of TORC1 are 4E binding protein (4E-BP) and S6 kinase (S6K). By removing the site within these regulators required for their interaction with TORC1, we have generated dominant negative proteins that disrupt specific pathways downstream of TORC1. Expression of dominant negative S6K, but not dominant negative 4E-BP, in Aplysia sensory neurons (SNs) blocked 24-h LTF. TORC1 is directly activated by the small GTP-binding protein, Ras homologue enriched in brain (Rheb). To determine the effects of TORC1 activation on translation in Aplysia neurons, we have examined the effects of expressing a constitutively active form of the Aplysia orthologue of Rheb, ApRheb (ApRheb(Q63L)). Expression of ApRheb(Q63L) increased 4E-BP phosphorylation and the level of general, cap-dependent translation within the SN cell soma in a rapamycin-sensitive manner. This increase in cap-dependent translation was blocked neither by dominant negative 4E-BP nor dominant negative S6K. Thus, we demonstrate that S6K is an important downstream target of TORC1 in Aplysia and that it is necessary for 24-h LTF, but not for TORC1-mediated increases in somatic cap-dependent translation.

Introduction

Memories, changes in synaptic strength within a neural network, are composed of mechanistically distinct phases. Whereas short-term memories rely on post-translational modification of pre-existing synaptic proteins, more persistent memories require the production of new proteins and likely involve synaptic growth (1). This de novo protein synthesis within neurons, underlying long-term synaptic plasticity, involves, not only transcriptional regulation, but translational regulation as well (2, 3). Moreover, the protein kinase complex, target of rapamycin (TOR)³ complex 1 (TORC1), a major regulator of translation and growth in eukaryotic cells (4), has been shown to play an essential role in this process (5). In many cell types, TORC1 is activated when conditions are permissive for cell growth through the integration of signaling pathways that sense the presence of these permissive cues (growth factors, amino acids, and energy (ATP)) (4). In neurons, TORC1 is activated during both in vitro models of synaptic plasticity and in vivo models of memory formation and, similar to other cells, acts as a gatekeeper regulating neuronal growth and plasticity (6).

The facilitation of neurotransmitter release at the sensory-to-motor neuron (SN-MN) synapse, in the mollusk, Aplysia californica, is a leading model system for the characterization of the biochemical basis of memory formation, because an increase in the strength of this synapse has been shown to contribute to behavioral sensitization of the reflex (7). In particular, whereas short-term facilitation (STF) is independent of protein synthesis, protein synthesis inhibitors block both long-term facilitation (LTF) of the SN-MN synapse (3) and long-term memory for sensitization (8), making this a relevant system for examining the role of translational control in memory formation. Moreover, in Aplysia SNs, exogenous treatment with the neurotransmitter responsible for inducing facilitation, 5-hydroxytryptamine (5-HT), leads to the activation of TORC1 (9–11), and bath application of the TORC1 inhibitor, rapamycin, blocks LTF measured at 24 h (24-h LTF) (12–13). When specifically applied to the synapse, however, rapamycin spares 24-h LTF but blocks the stabilization of newly grown varicosities as well as a stabilized phase of LTF, measured at 72 h (72-h LTF), at that particular synapse (14). Taken together, these studies suggest that TORC1-mediated translation is required both at the synapse, for the synapse-specific stabilization of new growth and, thereby, the stabilization of LTF, and within the cell soma, for processes that lead to the earlier expression of a phase of LTF, 24-h LTF, that maintains memory until stabilization can occur. While it is known that activation of TORC1 is required for various phases of LTF, the downstream targets of TORC1 required for these forms of plasticity are not known, and, thus, this is an excellent model for elucidating the molecular mechanisms for how TORC1 activation regulates synaptic plasticity.

TORC1 regulates translation through several divergent pathways of which the two best characterized are: 4E-binding protein (4E-BP) and S6 kinase (S6K) (Fig. 1, A and B) (4). 4E-BP specifically inhibits cap-dependent translation by sequestering the cap-binding protein, eukaryotic initiation factor 4E (eIF4E) (15). TORC1-dependent phosphorylation of 4E-BP releases this inhibition, freeing eIF4E and allowing it to bring capped mRNAs to the ribosome through binding to the adaptor eIF4G (4). Although 4E-BP regulates translation of all capped mRNAs, some mRNAs are particularly sensitive to levels of free eIF4E (4, 16) and, therefore, should also be particularly sensitive to 4E-BP regulation. 4E-BP is phosphorylated by stimuli that induce late long-term potentiation (L-LTP), metabotropic glutamate receptor-dependent long-term depression (mGluR-LTD), and, in Aplysia, LTF (11, 17–19). Mice lacking the major isoform of 4E-BP in the brain have a decreased threshold for the induction of L-LTP (18) and display a form of mGluR-LTD that is no longer sensitive to rapamycin (17), suggesting that, in the absence of 4E-BP, proteins important for plasticity are translated in the absence of stimulation. However, these studies do not address whether TORC1-mediated phosphorylation of 4E-BP regulates synaptic plasticity. Conversely, S6K regulates translation indirectly by phosphorylating multiple targets that, themselves, regulate different aspects of translation (20). Similar to 4E-BP, S6K is also activated in a TORC1-dependent manner by stimuli that induce L-LTP, mGluR-LTD, and LTF (9, 21–22). However, S6K knockouts have mild memory impairments and normal L-LTP in rodents (23), although, in this case, there may be compensation from the other mammalian S6K isoform. Importantly, S6K does not directly regulate eIF4E-4E-BP binding, and, thus, these two proteins are independent effectors of TORC1 signaling (4).

FIGURE 1.

Idealized model for how expression of ΔTOS mutants may disrupt specific downstream targets of TORC1. A, in the absence of growth factors/energy (ATP), the Rheb GTPase-activating protein complex, tuberous sclerosis complex 1/2 (TSC1/2), maintains Rheb in its inactive GDP-bound state, and, therefore, TORC1 (composed of the protein kinase TOR, the adaptor protein, raptor, and lethal with sec-thirteen 8 (LST8)) remains inactive. B, when growth factors (or 5-HT) are released, and if ATP is present in the cell, TSC1/2 is inhibited, allowing Rheb to become GTP-charged and active. If amino acids (leucine, in particular) are also present in the cell, Rheb is able to bind TORC1, and this leads to the activation of the complex. Through the raptor-TOS-motif interaction, TORC1 then binds to and regulates 4E-BP and S6K through phosphorylation. In this idealized model, the ability of 4E-BP to bind to eIF4E and prevent initiation of cap-dependent translation is inhibited by TORC1-mediated phosphorylation, while S6K is activated by TORC1-mediated phosphorylation. However, a previous study from our laboratory suggests that 4E-BP binding to eIF4E may not be rate-limiting for the initiation of translation in the Aplysia SN soma unless wild-type 4E-BP is overexpressed (11). In Aplysia, S6K has two identified downstream targets: eEF2K, through which it regulates translational elongation via disinhibition of eEF2, and the ribosomal protein S6. C, expression of 4E-BP(ΔTOS) suppresses the ability of TORC1 to relieve the 4E-BP-mediated restraint on initiation of cap-dependent translation, in this idealized model, because, while TORC1 is still able to bind to and inhibit endogenous 4E-BP, there is an excess of 4E-BP(ΔTOS) that does not interact with TORC1 and thus is not removed from cap-bound eIF4E after TORC1 activation. The ability of TORC1 to interact with, and activate, S6K, however, is unaffected by 4E-BP(ΔTOS) expression. D, expression of S6K(ΔTOS) results in an excess of a form of S6K that is inactive yet still interacts with S6K substrates, preventing them from interacting with, and being regulated by, active forms of the endogenous S6K.

In the present report, we have used a dominant negative approach to specifically remove the TORC1-dependent regulation of 4E-BP and S6K to determine the role of these downstream targets during rapamycin-sensitive, 24-h LTF in SN-MN synapses reconstituted in culture. We show a role for S6K, but not 4E-BP, in 24-h LTF. Moreover, by overexpressing the Aplysia orthologue of Ras homologue enriched in brain (Rheb and ApRheb) to specifically activate TORC1, we show that ApRheb-mediated increases in general, cap-dependent translation require neither the 4E-BP nor the S6K pathway.

EXPERIMENTAL PROCEDURES

Generation of Constructs and Cloning of Aplysia Rheb

The dominant negative 4E-BP construct (4E-BP(ΔTOS)) has previously been described (11). To generate the dominant negative S6K construct (S6K(ΔTOS)-mRFP), S6K was amplified by PCR from BB4-S6K (9) using primers containing XbaI and BamHI sites. The amplified DNA was cut with XbaI and BamHI and ligated to enhanced green fluorescent protein (eGFP) in a pNEX-3 plasmid expression vector cut with these same enzymes. eGFP was then replaced with monomeric red fluorescent protein (mRFP), by cutting out eGFP with AgeI and KpnI and inserting mRFP, using the same sites. The TOS site, located at the amino terminus of S6K, was then removed, leaving the initial methionine intact such that the protein sequence begins at position 13 (MEDRG).

A fragment of ApRheb was cloned using degenerate 5′ primers (5′-TAYGAYCCRACNATHGA-3′) and degenerate 3′ primers (5′-TAYTCRTCYTGNCCNGC-3′), derived from highly conserved sequences of Rheb (YDPTIE and AGQDEY, respectively). Sequence homology was based on protein sequence alignments of Rheb from Schizosaccharomyces pombe, Caenorhabditis elegans, Drosophila melanogaster, and Homo sapiens, obtained using ClustalW software. 5′ RNA ligase-mediated rapid amplification of cDNA ends (Ambion, Austin, TX) as well as 3′ rapid amplification of cDNA ends (Clontech, Palo Alto, CA) were then performed to obtain the full-length ApRheb sequence. A 5′ primer containing a BamHI site and a 3′ primer containing a KpnI site were used, along with Aplysia cDNA, to amplify a 608-bp fragment containing the entire ApRheb open reading frame. This fragment was then digested with BamHI and KpnI and inserted into the pNEX-3 vector. Mutations were made using a two-step overlap PCR technique that has previously been described (24).

To make the enhanced cyan fluorescent protein (eCFP)-tagged protein kinase C (PKC) Apl III construct, eCFP was amplified by PCR using primers containing SphI and XhoI sites. The product of this amplification was then cut with SphI and XhoI and used to replace the mRFP from a pNEX-3-mRFP-PKC Apl III construct, which has previously been described (25), cut with these same enzymes.

The polyhistidine-tagged S6K constructs were generated by amplifying S6K using PCR from pNEX-3-S6K(WT) or pNEX-3-S6K(ΔTOS)-mRFP with primers containing XbaI and SphI sites. The amplified DNA was cut with XbaI and SphI and ligated to the polyhistidine tag in the pFastBac-HT-B plasmid transfer vector (Invitrogen) cut with these same enzymes. All constructs were verified by sequencing.

Aplysia Cell Culture Preparation

SN and SN-MN cultures were prepared as previously described (3, 26). Briefly, adult A. californica were obtained from Marinus Scientific (Long Beach, CA) or University of Miami National Institute of Health Aplysia resource facility (Miami, FL) and maintained in an aquarium containing reconstituted seawater (Crystal Sea, MEI Inc., Baltimore, MD) at 15 °C. Animals were anesthetized with isotonic MgCl₂, and the pleural-pedal and abdominal ganglia were removed and digested with 1% protease type IX (Sigma) in isotonic Leibovitz-15 medium (Sigma). The abdominal and pleural ganglia were desheathed, and siphon (LFS) MNs and/or tail SNs were removed. The dissociated neurons were plated in Leibovitz-15 with 50% hemolymph and 2 mm l-glutamine on glass-bottom well dishes (MatTek Corp., Ashland, MA), pre-coated with poly-l-lysine (molecular weight, >300,000, Sigma). The growth of SN-MN synapses was promoted by plating the neurons with the primary sensory process in contact with the initial segment of the MN. The cultured neurons were left on the stereoscope stage overnight at room temperature and then transferred to a humidity controlled incubator at 18 °C.

Immunocytochemistry

In some experiments, phosphorylation of eukaryotic elongation factor 2 (eEF2) at threonine 57 (homologous to threonine 56 in mammalian eEF2), phosphorylation of 4E-BP at threonines 34/43 (homologous to threonines 37/46 in mammalian 4E-BP1), phosphorylation of eCFP-PKC Apl III at threonine 409, or the total level of S6K was also examined. In experiments examining phosphorylation of eEF2, 2 days after microinjection, a single distal neurite was isolated from each SN cell soma, through transection with a micropipette, and allowed to recover for 2 h. After this recovery period, the SNs, together with their intact and isolated neurites, were treated with either a single, 10-min pulse of 5-HT (10 μm) or a similar mock treatment with artificial sea water (ASW). In experiments examining phosphorylation of eCFP-PKC Apl III, SNs were treated with 5-HT or ASW in a similar manner. SNs were then fixed, either immediately following treatment or in other cases 2 days after microinjection and without treatment, in 4% paraformaldehyde with 30% sucrose for at least 30 min and then immunostained. In all cases, the fixed cells were permeabilized with 0.1% Triton X-100 in PBS with 30% sucrose for 10 min and washed briefly 3× with PBS. Free aldehydes were then quenched with 50 mm ammonium chloride for 15 min. Nonspecific antibody binding was blocked by incubating the cells with 10% normal goat serum (Sigma) plus 0.5% Triton X-100 in PBS for 1 h before incubating with a phospho-specific, anti-eEF2 (T56) antibody (1:800, Cell Signaling Technology, Beverly, MA), a phospho-specific, anti-4E-BP (T37/46) antibody (1:200, Cell Signaling Technology, Beverly, MA), a phospho-specific, anti-PKC Apl III (T409) antibody (1:400) (25), or an antibody to S6K (1:200) (9) within blocking solution overnight at 4 °C. After washing away the primary antibody 4× with PBS for 10 min, the cells were incubated with a secondary Cy3-conjugated, goat anti-rabbit antibody (1:1000, Invitrogen, Carlsbad, CA), for phospho-4E-BP, a fluorescein isothiocyanate-conjugated, goat anti-rabbit antibody (1:150, Invitrogen), for phospho-eEF2 and S6K, or an Alexa-Fluor-647-conjugated, goat anti-rabbit antibody (1:200, Invitrogen), for phospho-eCFP-PKC Apl III, for 1 h and 30 min. The cells were washed again 4× in PBS for 10 min and imaged using an Eclipse TE200 inverted fluorescence microscope (Nikon, Tokyo, Japan) equipped with GFP-BP v2 (cyan), GFP(Y)-BP (yellow), and Texas-Red Y-2E/C (red) filters, a 20×, numerical aperture (NA) 0.40 or 40×, NA 0.75 objective, and a charge-coupled device camera controlled by MetaVue Imaging software (Universal Imaging Co., Downingtown, PA).

In some experiments, cells were imaged using an Axiovert 100 inverted microscope (Zeiss) equipped with a 25×, NA 0.80 or 63×, NA 1.4 objective and mounted on an LSM 510 (Zeiss) laser confocal scanning microscope. In the case of phospho-eEF2 staining, intact and healthy-looking isolated neurites were imaged as previously described (10).

Electrophysiology

After 2 days in culture, the SNs within the SN-MN pairs were microinjected with a construct encoding either 4E-BP(ΔTOS) (67 ng/μl), S6K(ΔTOS)-mRFP (67 ng/μl), or mRFP (45 ng/μl) or with empty pNEX-3 vector (59 ng/μl) along with 0.2% fast green. A bicistronic reporter construct (27) (67 ng/μl) was co-injected in all groups to monitor expression of the constructs. After microinjection, the cells were returned to the incubator (18 °C) to allow synapses to continue to grow. On day 4, the baseline synaptic strengths of SN-MN pairs were measured through intracellular recording using an Axoclamp 2B amplifier (Molecular Devices, Sunnyvale, CA) in bridge mode and microelectrodes (15–30 MΩ) filled with 2 m potassium acetate. A hyperpolarizing current was applied while penetrating both the SNs and MNs to prevent them from firing action potentials. The MNs were maintained at −80 mV, and the SNs were maintained at −50 mV. Input resistances were measured, and a single excitatory post-synaptic potential (EPSP) was evoked in the MN through intracellular injection of a depolarizing current into the SN. The co-cultures were, then, treated either with a single 5-min pulse of 5-HT (10 μm) or five pulses of 5-HT with a 20-min interpulse interval. In between pulses, 5-HT was washed away using a 1:1 mixture (v/v) of Leibovitz-15 and ASW. Input resistances and EPSPs were re-recorded 10 min after one pulse of 5-HT or 24 h after five pulses of 5-HT. As a control, some cultures were mock treated with ASW instead of 5-HT. The slopes of the rising phase of the EPSPs were measured and adjusted for changes in MN input resistance, and the percent change in EPSP slope was calculated. Data from the SN-MN recordings were discarded if any of the following parameters were over two standard deviations from the means of the accumulated data set: the input resistance of the MN (>129 mΩ), the input resistance of the SN (>217 mΩ), the increase in the input resistance of the MN (>60%), the increase in the input resistance of the SN (>160%), the resting membrane potential of the MN (<−48 mV), the amplitude of the action potential evoked in the MN (<50 mV), the depolarizing current required to evoke an action potential in the SN (>5 nA), and the slope of the initial EPSP (<1.5 mV/ms).

Translation Assay

After 1 day in culture, SNs were microinjected with 0.2% fast green and various combinations of constructs encoding ApRheb(WT) (100 ng/μl), ApRheb(Q63L) (100 ng/μl), ApRheb(I38K) (100 ng/μl), 4E-BP(ΔTOS) (67 ng/μl), S6K(ΔTOS)-mRFP (67 ng/μl), and mRFP (45 ng/μl) as well as empty pNEX-3 vector (amount varied such that the total molar amount of construct injected was approximately equal across all groups within an experiment; 0, 59, 84, or 143 ng/μl depending on the group). The SNs were also co-injected with a bicistronic reporter construct (67 ng/μl) encoding eCFP, preceded by a generic 5′-untranslated region (5′UTR), as well as enhanced yellow fluorescent protein (eYFP), preceded by an internal ribosomal entry site (IRES) derived from Aplysia egg-laying hormone (ELH) (27). The cells were returned to the incubator, and, 2 days later, levels of eCFP (cap-dependent translation) and eYFP (IRES-dependent translation) and/or levels of mRFP were determined by imaging the fluorescence intensity within the sensory cell soma. In experiments involving the drug rapamycin, SNs were left for 1 h after microinjection, then treated with either 20 nm rapamycin in 0.1% DMSO or 0.1% DMSO alone for 24 h, and imaged immediately following treatment. Images were captured as described above.

GST Pulldown Assay

High titer baculoviruses (5 × 10⁸ plaque forming units/ml) encoding His-S6K(WT) and His-S6K(ΔTOS)-mRFP were generated using Bac-to-Bac (Invitrogen). Sf9 cells were infected with these stocks, and extracts were harvested after 3 days of infection.

Glutathione S-transferase (GST) and GST-S6 (9) constructs, present in pGEX-2T and pGEX-5X-1, respectively, were transformed and expressed in BL21 cells. The resultant protein was purified using glutathione-Sepharose 4B beads (Amersham Biosciences) and stored attached to the beads on ice. Meanwhile, the Sf9 cells that were previously infected were lysed using a lysis buffer containing 50 mm Tris, pH 7.5, 10 mm MgCl₂, 1 mm EGTA, 2.5 mm β-mercaptoethanol, 10% glycerol, 1% Tergitol Nonidet P-40, and EDTA-free protease inhibitor mixture (Roche Applied Science). Non-solubilized proteins were spun down at 11,000 × g for 10 min at 4 °C, and the supernatant was incubated with beads containing either purified GST-S6 (10 μg) or the same molar amount of purified GST alone (4.4 μg) for 1 h at 4 °C. The beads were then pelleted and washed 3× for 10 min with lysis buffer at 4 °C. Laemmli buffer was added to the samples, which were subsequently boiled at 94 °C for 10 min, subjected to SDS-PAGE, transferred onto nitrocellulose membranes, and immunoblotted with mouse anti-His antibody (1:500, Invitrogen) and rabbit anti-S6 antibody (1:1000) (9).

Quantitation of Fluorescence

Fluorescence intensity within the SN cell soma or neurites was quantified using ImageJ software (National Institutes of Health). The mean pixel intensity within the entire cell soma or, in the case of phospho-4E-BP staining and S6K(ΔTOS)-mRFP fluorescence, just within the cytoplasm was measured, and the mean pixel intensity of non-expressing cells within the same experiment and the mean pixel intensity of the surrounding background were subtracted to account for endogenous fluorescence within the cell and fluorescence from the surrounding media. The mean of this net fluorescence in control (vector alone/mRFP construct)-injected cells was calculated, and the values of all individual cells from all groups, including the control group, were normalized to this mean. In the case of phospho-PKC-Apl III staining, the mean pixel intensity of phospho-PKC-Apl III staining across all cells that were not expressing eCFP-PKC Apl III was subtracted, along with the mean pixel intensity of the surrounding background, from the mean pixel intensity of phospho-PKC-Apl III staining only from cells expressing low levels of eCFP-PKC Apl III (25). This net phospho-PKC-Apl III staining was used to calculate the ratio of phospho-PKC-Apl III staining to eCFP-PKC Apl III expression for each cell, and the phospho-ratio of each individual cell, from each group (including the cells from the mock treated groups), was normalized to the mean ratio of the mock treated cells expressing a similar construct. In the case of phospho-eEF2 staining, the weighted mean pixel intensity of punctate fluorescent staining within isolated and intact neurites was measured using the threshold function with the lower threshold set to two standard deviations above the mean total neuritic staining of intact neurites from mock treated, control-injected SNs. Puncta were defined as having an area greater or equal to 0.04 μm² using the analyze particle function. The weighted mean pixel intensity was calculated by weighting the mean pixel intensity of each puncta according to the area of each puncta and dividing the sum of this by the total area of all puncta. The mean pixel intensity of the surrounding background was then subtracted from this weighted mean pixel intensity. The product of this corrected value and the total area of all puncta were divided by the total area of the neurites to obtain a measure of the net eEF2 phosphorylation per unit of neurite area. The net eEF2 phosphorylation within isolated neurites was then divided by the net fluorescence of intact neurites from the same SN to obtain a measure of the ratio of eEF2 phosphorylation between isolated and intact neurites. The ratio of eEF2 phosphorylation of each individual cell, from each group (including the cells from the mock treated groups), was normalized to the mean ratio of the mock treated cells expressing a similar construct.

Statistical Analysis

The statistical significances of the differences between means was tested by using Student's t tests. Bonferroni's correction was applied when multiple, pre-planned comparisons between pre-determined groups were made within a single experiment. Welch's correction was applied to all Student's t tests, because the ratios of the various population variances of the samples being compared were unknown, and performing a Student's t test with Welch's correction alone maintains a size close to the prescribed alpha value across a wide range of variance ratios, whereas performing a preliminary variance test followed by the appropriate Student's t test or variant does not (28–29). One-tailed t tests were used only when specifically testing a hypothesis that the means would differ in a given direction, otherwise a two-tailed t test was employed.

RESULTS

4E-BP is a negative regulator of translation. Thus, knocking down levels of 4E-BP does not block its activity but leads to constitutive activation of this pathway. Moreover, knocking out or knocking down one target of TORC1 leads to increased activation of the other targets, presumably due to their competition for regulatory-associated protein of TOR (raptor) binding sites (30–31). To block TORC1 signaling through 4E-BP, a strategy that blocks the ability of TORC1 to remove the repression mediated by 4E-BP is required. Both S6K and 4E-BP have a TOR signaling (TOS) motif that binds to the TORC1 adaptor protein raptor (32, 33). This site is required for TORC1-mediated regulation of both these proteins (Fig. 1B) (33–34). We have previously characterized a form of Aplysia 4E-BP that lacks the carboxyl-terminal TOS motif (4E-BP(ΔTOS)), and expression of this construct is capable of inhibiting the initiation of cap-dependent translation, thus, providing a means of specifically interfering with TORC1 signaling through 4E-BP (Fig. 1C) (11). Here we use a similar strategy to generate a dominant negative S6K construct that should specifically disrupt the S6K pathway but not other TORC1 downstream effectors (Fig. 1D). Previous dominant negative strategies for disrupting S6K activity blocked 4E-BP phosphorylation and 5′-terminal-oligopyrimidine tract (5′TOP)-dependent translation by sequestering raptor and/or TOR (35), but this does not occur if the TOS site is removed (33). Removing the TOS site in S6K also greatly decreases kinase activity, suggesting that removing this site should generate a dominant negative kinase (33). Therefore, we generated an Aplysia S6K mutant construct that lacks the amino-terminal TOS motif and is tagged with an mRFP at the non-conserved carboxyl-terminal end to monitor expression (S6K(ΔTOS)-mRFP) (see “Experimental Procedures”).

Characterization of Dominant Negative Constructs

To test whether the S6K(ΔTOS)-mRFP construct, but not the 4E-BP(ΔTOS) construct, acted as a dominant negative to S6K substrates, we examined eEF2 phosphorylation at threonine 57. We have previously shown that, in a rapamycin-sensitive manner, 5-HT can reverse the increase in eEF2 phosphorylation at threonine 57 that occurs after cutting neurites in Aplysia SNs, presumably through phosphorylation and inactivation of the calcium/calmodulin-dependent eEF2 kinase (eEF2K) by S6K (10). Although this assay is somewhat indirect, phosphopeptide antibodies that work in immunocytochemistry to direct S6K substrates in Aplysia are not available. We, therefore, expressed either S6K(ΔTOS)-mRFP, 4E-BP(ΔTOS), and mRFP, or the mRFP control alone in Aplysia SNs, transected one of each of the SNs neurites, and examined the effect of a single 10-min pulse of 5-HT on the increased eEF2 phosphorylation observed within isolated neurites as compared with intact neurites from the same SN. Similar to our previous findings, application of 5-HT was able to decrease eEF2 phosphorylation at threonine 57 in isolated neurites of SNs expressing mRFP alone (Fig. 2, A and B). Although this was also the case in the isolated neurites of SNs expressing 4E-BP(ΔTOS) and mRFP, the 5-HT-induced decrease in eEF2 phosphorylation was completely blocked in the isolated neurites of SNs expressing S6K(ΔTOS)-mRFP (Fig. 2, A and B). Therefore, the S6K(ΔTOS)-mRFP mutant is a dominant negative form of S6K that can be used to disrupt TORC1 signaling through S6K. Furthermore, a ΔTOS mutant of a different downstream target of TORC1, 4E-BP(ΔTOS), did not interfere with TORC1 signaling through S6K, which indicates that our strategy of removing the TOS site to create dominant negative proteins prevents nonspecific disruption of other TORC1 pathways.

FIGURE 2.

S6K(ΔTOS)-mRFP, but not 4E-BP(ΔTOS), disrupts regulation of eEF2 phosphorylation downstream of S6K. A, SNs were microinjected with a construct (or constructs) encoding either mRFP (Control), 4E-BP(ΔTOS) and mRFP, or S6K(ΔTOS)-mRFP and control amounts of empty vector such that the amount of DNA injected was held constant. Two days after injection, a distal neurite was isolated from each SN cell soma, and, 2 h later, the isolated and intact neurites were either treated with 5-HT or mock treated with ASW. The cells were fixed immediately following treatment and stained with an antibody that recognizes eEF2 phosphorylated at threonine 57 (P-eEF2 (T57)) (10). Representative neurites are shown for each group. B, the quantitation of the mean ± S.E. ratio of net eEF2 phosphorylation between isolated and intact neurites from neurites in A normalized to the mean of the mock treated groups (-fold difference in the ratio of eEF2 phosphorylation; see “Experimental Procedures”) are shown for each group. The n for each group is shown under the corresponding bar. Data were analyzed by one-tailed Student's t tests with Welch's correction (**, p ≤ 0.01; *, p ≤ 0.05; NS, p > 0.05). C, SNs were microinjected with a construct encoding mRFP (Control) or S6K(ΔTOS)-mRFP. Two days after injection, levels of mRFP expression were measured, and then cells were fixed and stained with an antibody that recognizes both endogenous S6K and the exogenously expressed mutant form of S6K (S6K (Total)) (9). Representative cells and the quantitation of the mean ± S.E. of the net fluorescence values normalized to the mean of the control group (relative fluorescence; see “Experimental Procedures”) are shown for each group. The n for each group is shown under the corresponding bar. Data were analyzed by one-tailed (S6K(Total)) and two-tailed (mRFP) Student's t tests with Welch's correction (***, p ≤ 0.001; NS, p > 0.05).

Although the dominant negative construct blocks a downstream event of S6K (eEF2 dephosphorylation), it is possible that it does this by blocking signaling pathways upstream of S6K. This could represent a complication as some of these pathways may be important for LTF independently of their effect on translation. To test for this, we examined an important upstream activator of S6K, phosphoinositide-dependent protein kinase 1 (9). This kinase is also important for phosphorylation of exogenously expressed PKC Apl III by 5-HT (25), and, thus, if the dominant negative acts by sequestering upstream activators, we would expect that expression of S6K(ΔTOS)-mRFP would block 5-HT-mediated phosphorylation of PKC Apl III. However, expression of S6K(ΔTOS)-mRFP did not block phosphorylation of eCFP-tagged PKC Apl III and, thus, does not act by sequestering upstream activators (supplemental Fig. 1). Moreover, this construct can compete with binding to downstream targets of S6K as shown by both the decrease in dephosphorylation of eEF2 (Fig. 2) and competitive binding to S6 (supplemental Fig. 2). Therefore, the dominant negative S6K(ΔTOS)-mRFP acts by competing with endogenous S6K for phosphorylating substrates, not by sequestering upstream activators.

To determine the extent to which the S6K(ΔTOS)-mRFP mutant is expressed relative to endogenous S6K, we immunostained a separate group of injected cells with an antibody to S6K that recognizes both endogenous S6K and the exogenously expressed mutant form of S6K. Expression of S6K(ΔTOS)-mRFP led to an ∼4-fold increase in staining compared with SNs injected with mRFP alone (Fig. 2C). This is about the same level of expression seen when 4E-BP is overexpressed (11). We also looked at levels of mRFP expression and found that the S6K(ΔTOS)-mRFP construct expressed at similar levels as mRFP alone (Fig. 2C).

24-h LTF Requires S6K-, but Not 4E-BP-, Signaling

To investigate whether the 4E-BP and/or S6K signaling pathways downstream of TORC1 are involved in LTF, we expressed either the dominant negative 4E-BP(ΔTOS) mutant that we previously characterized (11) or the dominant negative S6K(ΔTOS)-mRFP mutant, characterized above, in the SNs of SN-MN co-cultures. As a control, we either expressed mRFP or injected the SNs with the empty control vector. LTF was observed in both types of control-injected SNs 24 h after induction via bath application of 5 pulses of 5-HT, and, therefore, these controls were pooled. LTF was also seen when 4E-BP(ΔTOS) was expressed (Fig. 3A). Expression of S6K(ΔTOS)-mRFP, on the other hand, blocked 24-h LTF (Fig. 3A). Therefore, disregulation of S6K signaling, but not 4E-BP signaling, disrupts 24-h LTF.

FIGURE 3.

24-h LTF requires S6K but not 4E-BP. A, SN-MN pairs were cultured, and, on the second day of culture, SNs were microinjected with either empty vector or a construct encoding mRFP (Control), with a construct encoding 4E-BP(ΔTOS), or with a construct encoding S6K(ΔTOS)-mRFP. On day 4 of culture, an initial EPSP slope was measured, followed by bath application of five pulses of 5-HT or mock treatment with ASW, and 24 h later the EPSP slope was measured again to determine the amount of LTF (% change in EPSP slope). Representative recordings of baseline (gray) and 24 h (black) EPSPs along with the quantitation of the mean % change in EPSP slope ± S.E. (see “Experimental Procedures”) are shown for each group. The n for each group is shown under the corresponding bar. Data were analyzed by one-tailed Student's t test with Welch's correction (**, p ≤ 0.01; *, p ≤ 0.05; NS, p > 0.05). B, same as A, but, after the initial EPSP slope was measured, there was a bath application of one pulse of 5-HT or mock treatment with ASW, and 10 min later the EPSP slope was measured again to determine the amount of STF (% change in EPSP slope). Representative recordings of baseline (gray) and 24 h (black) EPSPs along with the quantitation of the mean % change in EPSP slope ± S.E. (see “Experimental Procedures”) are shown for each group. The n for each group is shown under the corresponding bar. Data were analyzed by a one-tailed Student's t test with Welch's correction (**, p ≤ 0.01; *, p ≤ 0.05).

The expression of S6K and 4E-BP dominant negatives did not disrupt basal properties of the SNs as measured by initial synaptic strength, input resistance, and holding current (Table 1). Moreover, it did not disrupt the ability of 5-HT to signal effectively as STF was unaffected by the dominant negatives (Fig. 3B). Thus, signaling through the S6K pathway is specifically required for the expression of 24-h LTF but not STF.

TABLE 1.

Comparison of intrinsic and synaptic properties of sensory neurons expressing control or dominant negative constructs

SN property	Control	4E-BP(ΔTOS)	S6K(ΔTOS)-mRFP	Student's t testsa
Ihold (nA)b	−0.13 ± 0.02c	−0.07 ± 0.02	−0.12 ± 0.01	NSd
Rinput (MΩ)e	109.7 ± 6.0	103.1 ± 8.4	103.9 ± 7.0	NS
EPSPi (mV/ms)f	9.9 ± 1.1	7.6 ± 0.9	8.8 ± 1.1	NS

^a Student's t tests with both Welch's and Bonferroni's corrections.

^b Current needed to hold SN at −50 mV during baseline recording.

^c All values in the table are the mean ± S.E.

^d NS, p > 0.0167.

^e Input resistance during baseline recording.

^f Initial EPSP slope during baseline recording.

Overexpression of ApRheb Increases General, Cap-dependent Translation through TORC1

Next, we set out to determine the effects that expression of our dominant negatives have on TORC1 signaling. This required a means of specifically activating TORC1 signaling within SNs. In many systems, the immediate activator of TORC1 is the GTP-bound form of Rheb (Fig. 1, A and B) (36–39). We, therefore, cloned the Aplysia orthologue of Rheb. This protein, which we refer to as ApRheb, has a high degree of homology with other Rheb proteins within the core effector domain and flanking regions (Fig. 4). ApRheb also contains a conserved arginine residue at position 14, which distinguishes Rheb from other Ras-like GTPases (40), as well as a characteristic Ras-like CAAX farnesylation motif at its carboxyl terminus, which anchors Rheb to the cellular membrane (41).

FIGURE 4.

Amino acid sequence of Aplysia Rheb. The ApRheb amino acid sequence from A. californica was aligned with orthologous sequences from D. melanogaster, H. sapiens, S. pombe, and C. elegans using ClustalW software. The conserved effector domain and CAAX motif are labeled along with the characteristic arginine residue at position 14 (*). The glutamine residue at position 63, which was mutated to a leucine to produce the ApRheb(Q63L) mutant construct (†), and the isoleucine residue at position 38, which was mutated to a lysine to produce the ApRheb(I38K) mutant construct (#), are also labeled.

In the mammalian system, a glutamine-64-to-leucine (Q64L) point mutation in the switch 2 segment of Rheb is analogous to the Q61L mutation in Ras, which abolishes its ability to hydrolize GTP (42), and renders Rheb constitutively active (38). Conversely, an isoleucine 39 to lysine (I39K) mutation within the core effector domain prevents Rheb association with downstream effectors (43). We used site-directed mutagenesis to introduce homologous mutations into ApRheb (ApRheb(Q63L) and ApRheb(I38K), respectively). We then expressed these mutants or overexpressed wild-type ApRheb (ApRheb(WT)) in Aplysia SNs and examined their effects on both translation, using a bicistronic translational reporter construct (Fig. 5A) (27), and TORC1-mediated phosphorylation, using antibodies to phosphorylated 4E-BP (11). Our bicistronic construct reports rates of both cap- and IRES-dependent translation (through levels of eCFP and eYFP, respectively). Translation of most mRNAs is initiated at the 5′-m⁷-GTP cap of the transcript. However, under certain physiological conditions (i.e. mitosis, apoptosis, hypoxia, and some viral infections), it can be advantageous to initiate translation independently of the cap, and, for this reason, translation of a subset of transcripts can be initiated at a site within their 5′UTR known as an IRES (44). The presence of a second cap-independent translational reporter on the same transcript as our generic reporter of cap-dependent translation provides an internal reference with which we can control for cell-to-cell variation in the amount of plasmid injected. It also rules out any effect of manipulations on transcription of the reporter, because both eCFP and eYFP are encoded on the same mRNA. Thus, the ratio of the level of cap-dependent translation of eCFP to the level of IRES-dependent translation of eYFP (hereafter referred to as the cap:IRES ratio) is a sensitive measure of changes in general, cap-dependent translation (11, 27, 45).

FIGURE 5.

Overexpression of Rheb causes an increase in markers of TORC1 activation in the Aplysia SN cell soma. A, the bicistronic translational reporter construct used to measure rates of cap-dependent and IRES-dependent translation (27) consists of an open reading frame encoding eCFP, translation of which is initiated at a 5′-m⁷-GTP cap (Cap) located at the 5′ terminus of a generic 5′UTR. The eCFP open reading frame is followed by at least one stop codon in each reading frame and, then, a second 5′UTR that contains an IRES derived from Aplysia ELH. This ELH IRES drives IRES-dependent translation of a second open reading frame encoding eYFP. The nucleotide sequence depicted in this schematic diagram does not reflect the actual sequence of the reporter construct. B, SNs were microinjected with the bicistronic (eCFP-eYFP) construct (27) together with either empty vector (Control) or constructs encoding ApRheb(WT), ApRheb(Q63L), or ApRheb(I38K). Two days after injection, levels of eCFP (Cap) and eYFP (IRES) expression were measured. The cells were fixed and stained with a phospho-specific antibody that recognizes 4E-BP phosphorylated at threonines 34/43 (P-4E-BP (T34/43)) (11). Representative cells and the quantitation of the mean ± S.E. of the net fluorescence values normalized to the mean of the control group (relative fluorescence; see “Experimental Procedures”) are shown for each group. The n for each group is shown under the corresponding bar. Data were analyzed by two-tailed (Cap:IRES) and one-tailed (P-4E-BP (T34/43)) Student's t tests with both Welch's and Bonferroni's corrections (α = 0.0167; ***, p ≤ 0.001; **, p ≤ 0.01; NS, p > 0.0167). C, SNs were microinjected with the bicistronic (eCFP-eYFP) construct (27) together with either empty vector or a construct encoding ApRheb(Q63L). 1 h after injection, the cells were treated with either 0.1% DMSO or 20 nm rapamycin with 0.1% DMSO (Rapamycin). One day after injection, levels of eCFP (Cap) and eYFP (IRES) expression were measured. Representative cells and the quantitation of the mean ± S.E. of the net fluorescence values normalized to the mean of the control group (relative fluorescence; see “Experimental Procedures”) are shown for each group. The n for each group is shown under the corresponding bar. Data were analyzed by one-tailed (Cap) and two-tailed (IRES and Cap:IRES) Student's t tests with both Welch's and Bonferroni's corrections (α = 0.0167; **, p ≤ 0.01; *, p ≤ 0.0167; NS, p > 0.0167).

Both expression of ApRheb(Q63L) and overexpression of ApRheb(WT) were found to activate TORC1 signaling. Expression of either of these constructs increased the cap:IRES ratio to a level significantly different from that in control-injected SNs and also caused a significant increase in 4E-BP phosphorylation at threonine 34/43 relative to control-injected SNs (Fig. 5B). Inactive ApRheb(I38K), on the other hand, had no effect on either of these measures (Fig. 5B). The fact that overexpression of wild-type ApRheb increased TORC1 signaling is consistent with previous findings within the mammalian system that overexpression of wild-type Rheb is sufficient to increase the percent of GTP-bound Rheb (46). This is likely due to the fact that Rheb has little intrinsic GTPase activity (47).

To verify that the effects of expressing constitutively active ApRheb on translation are indeed mediated by TORC1, we repeated this experiment in the presence and absence of the TORC1 inhibitor, rapamycin. We found that rapamycin treatment blocked the ability of ApRheb(Q63L) to increase cap-dependent translation as well as the effect of ApRheb(Q63L) on the cap:IRES ratio (Fig. 5C). There was also an unexpected effect of ApRheb(Q63L) and rapamycin on IRES-dependent translation that may depend on specific regulation of this IRES sequence (see “Discussion”). In summary, expression of constitutively active ApRheb and overexpression of wild-type ApRheb increased TORC1 signaling, which in turn stimulated general, cap-dependent, but not IRES-dependent, translation within the SN cell soma.

TORC1-mediated Translational Regulation Requires Neither 4E-BP nor S6K Signaling

Dominant negative S6K, but not dominant negative 4E-BP, blocked 24-h LTF. To investigate whether these results reflect their differential ability to mediate the effects of somatic TORC1 signaling on overall cap-dependent translation, we determined the effect of these dominant negatives on ApRheb-mediated increases in cap-dependent translation of our reporter within the SN cell soma. Expression of 4E-BP(ΔTOS) alone significantly decreased cap-dependent translation and resulted in a decreased cap:IRES ratio that was significantly different from the cap:IRES ratio in control-injected SNs (Fig. 6A), similar to what was previously described (11). However, even in the presence of 4E-BP(ΔTOS), expression of ApRheb(Q63L) was still able to significantly increase cap-dependent translation and to raise the cap:IRES ratio to a level significantly different from that observed in SNs expressing 4E-BP(ΔTOS) alone (Fig. 6A). The TORC1-mediated increase in general, cap-dependent translation is, therefore, not dependent on 4E-BP signaling. ApRheb(Q63L) also caused increased IRES-dependent translation in the presence of 4E-BP(ΔTOS) that was significantly different from that of control-injected SNs, although the mechanism underlying this is not understood (see “Discussion”) (Fig. 6A).

FIGURE 6.

The ApRheb-mediated increase in cap-dependent translation neither requires S6K nor 4E-BP regulation. A, SNs were microinjected with the bicistronic (eCFP-eYFP) construct (27) together with either empty vector or a construct encoding ApRheb(Q63L) and either empty vector or a construct encoding 4E-BP(ΔTOS). Two days after injection, levels of eCFP (Cap) and eYFP (IRES) expression were measured. Representative cells and the quantitation of the mean ± S.E. of the net fluorescence values normalized to the mean of the control group (relative fluorescence; see “Experimental Procedures”) are shown for each group. The n for each group is shown under the corresponding bar. Data were analyzed by one-tailed (Cap) and two-tailed (IRES and Cap:IRES) Student's t tests with both Welch's and Bonferroni's corrections (α = 0.0167; ***, p ≤ 0.001; **, p ≤ 0.01; *, p ≤ 0.0167; NS, p > 0.0167). B, SNs were microinjected with the bicistronic (eCFP-eYFP) construct (27) together with either empty vector or a construct encoding ApRheb(Q63L) and either a construct encoding mRFP or a construct encoding S6K(ΔTOS)-mRFP. Two days after injection, levels of eCFP (Cap) and eYFP (IRES) expression were measured. Representative cells and the quantitation of the mean ± S.E. of the net fluorescence values normalized to the mean of the control group (relative fluorescence; see “Experimental Procedures”) are shown for each group. The n for each group is shown under the corresponding bar. Data were analyzed by one-tailed (Cap: Control versus ApRheb(Q63L) & mRFP, and S6K(ΔTOS)-mRFP versus ApRheb(Q63L) & S6K(ΔTOS)-mRFP) and two-tailed (Cap: Control versus S6K(ΔTOS)-mRFP, IRES, and Cap:IRES) Student's t tests with both Welch's and Bonferroni's corrections (α = 0.0167; ***, p ≤ 0.001; **, p ≤ 0.01; *, p ≤ 0.0167; NS, p > 0.0167).

Interestingly, when S6K(ΔTOS)-mRFP alone was expressed, both cap-dependent translation and IRES-dependent translation were increased to levels significantly different from those in control-injected SNs (Fig. 6B). Despite these increases in the presence of S6K(ΔTOS)-mRFP expression, ApRheb(Q63L) expression still significantly increased cap-dependent translation beyond the extent to which this measure was already increased by S6K(ΔTOS)-mRFP expression alone (Fig. 6B). Also, despite the expression of S6K(ΔTOS)-mRFP, ApRheb(Q63L) still increased the cap:IRES ratio to a level significantly different from that in SNs expressing S6K(ΔTOS)-mRFP alone (Fig. 6B). The TORC1-mediated increase in general, cap-dependent translation, thus, does not require S6K signaling. There was, again, a puzzling effect on IRES-dependent translation; in the presence of S6K(ΔTOS)-mRFP expression, ApRheb(Q63L) expression decreased IRES-dependent translation to a level significantly different from S6K(ΔTOS)-mRFP expression alone, an effect opposite to that of ApRheb(Q63L) in the presence of 4E-BP(ΔTOS) expression. Despite this, from the above data, it can be concluded that, within the Aplysia SN cell soma, TORC1-mediated increases in general, cap-dependent translation are independent of both S6K and 4E-BP.

DISCUSSION

Here we show that a dominant negative form of S6K blocks 24-h LTF. We have previously demonstrated that 5-HT activates S6K through the TORC1 pathway (9), and, together, these results strongly implicate TORC1-mediated activation of S6K in 24-h LTF. In contrast, despite the ability of dominant negative 4E-BP to decrease basal cap-dependent translation, this construct did not block 24-h LTF. Thus, we have been able to determine the important downstream target of TORC1 for a physiologically important form of synaptic plasticity. On the other hand, ApRheb-mediated increases in general translation were neither blocked by dominant negative 4E-BP nor dominant negative S6K, suggesting that TORC1 increases general translation through a third, as yet unidentified effector.

Dominant Negative S6K and LTF

Expression of a form of S6K that lacks the TOS site abolished the ability of 5-HT to decrease eEF2 phosphorylation within isolated neurites (Fig. 2, A and B), a rapamycin-sensitive process that has been shown to be mediated by S6K in other systems (48). This provides good evidence that this S6K mutant acts as a dominant negative. Because it cannot bind raptor, this form of S6K should not be regulated by TORC1 and, thus, should be inactive, and, because it is expressed at a level higher than endogenous S6K (Fig. 2C), it should effectively compete for binding to substrates (Fig. 1D and supplemental Fig. 2). There was no effect of the dominant negative on the ability of an upstream activator of S6K, phosphoinositide-dependent protein kinase 1, to phosphorylate a different substrate, PKC Apl III (supplemental Fig. 1). Thus, it is likely that the main action of the S6K dominant negative is to block phosphorylation of downstream targets of S6K (Fig. 1D).

Rapamycin blocks 24-h LTF when applied to the bath but does not block 24-h LTF when applied specifically to the synapse (13, 14). From these data, it is likely that rapamycin-sensitive translational events in the soma are important for 24-h LTF. Whereas 24-h LTF was blocked by expression of dominant negative S6K (Fig. 3A), the increase in general, cap-dependent translation due to activation of the TORC1 system, through ApRheb, was not (Fig. 6B). Thus, S6K is not required for TORC1-mediated increases in general, cap-dependent translation within the SN cell soma and, therefore, it does not promote 24-h LTF by stimulating general somatic translation. Rather, S6K likely plays a role in 24-h LTF by stimulating somatic translation of a specific transcript or a specific group of transcripts. It should be noted that the fact that the S6K dominant negative did not block the increase in general, cap-dependent translation does not imply that a general increase in translation is not important for LTF. LTF requires a number of signaling events, and not all of them are dependent on each other. Whether or not the increase in general, cap-dependent translation is important for LTF will require identification of the downstream target of TORC1 required for this step.

The targets downstream of S6K that would be required for specific up-regulation of certain transcripts during 24-h LTF remain to be identified. There are a number of S6K substrates that are known. The best known target is the ribosomal protein S6. Evidence suggests that phosphorylation of S6 is required for TORC1 regulation of cell size (49), but the mechanism of action of S6 phosphorylation is unknown. S6 is a good candidate for S6K regulation in Aplysia, because S6K sites in S6 are conserved in Aplysia, and, indeed, we have shown that S6 is phosphorylated by 5-HT in a rapamycin-sensitive manner during LTF (9).

Another attractive target of S6K is eEF2K (50). In fact, we have shown that the downstream target of eEF2K, eEF2 phosphorylation, is regulated by the dominant negative S6K (Fig. 2) (10). Although eEF2K regulation should affect overall translation, certain transcripts have recently been identified that are specifically regulated downstream of eEF2K (51, 52).

Dominant Negatives and Translational Regulation

TORC1-mediated increases in general translation within the SN cell soma were not blocked by either dominant negative 4E-BP or dominant negative S6K (Fig. 6, A and B), which suggests that a third, unknown TORC1 effector is responsible for this increase in translation. However, we cannot rule out that this increase may be due to the effects on a substrate of S6K that is less sensitive to the effects of the dominant negative than either eEF2K or the substrate important for LTF. A group of transcripts, encoding ribosomal proteins and translation factors, that contain a highly suppressive regulatory element in their 5′UTR, known as a 5′TOP, are translationally up-regulated by TORC1 (53). Translational up-regulation of TOP mRNAs by TORC1 has been shown to be independent of both 4E-BP and S6K (54–56). Furthermore, in Aplysia, 5-HT treatment translationally up-regulates TOP mRNA encoding eEF2 in a rapamycin-sensitive manner (10) and up-regulates levels of S6 and eEF1A (9, 57), two other proteins encoded by TOP mRNAs. Thus, TOP mRNA regulation is a good candidate for mediating the observed effects of TORC1 activation on general translation. Indeed, an increase in the production of ribosomal proteins and translation factors, in other words, the translational machinery itself, would be an effective way to increase general translation, although it is not clear that it can explain the specific increase in cap-dependent translation seen after expression of ApRheb in this study. Other possible targets downstream of TORC1, such as proline-rich Akt substrate of 40 kDa (31, 58), could also mediate the effects of TORC1 on general, cap-dependent translation.

Expression of dominant negative S6K alone increased both cap-dependent and IRES-dependent translation (Fig. 6B). These results indicate that, in Aplysia SNs, S6K signaling is a negative regulator of both general translation and translation of at least one IRES-dependent transcript. A study by Ruvinsky et al. (49) supports this conclusion. They found that mouse embryonic fibroblasts with a non-phosphorylatable form of the ribosomal protein S6 knocked-in showed higher rates of [³⁵S]methionine and [³⁵S]cysteine incorporation into total cytoplasmic proteins and higher rates of protein accumulation than wild-type mouse embryonic fibroblasts. Therefore, dominant negative S6K expression may relieve a restraint on general translation by allowing ribosomal S6 to become dephosphorylated.

In our experiments looking at IRES-dependent translation, specifically, several results were surprising. For example, the decrease in IRES-dependent translation with rapamycin treatment obtained only in the presence of ApRheb(Q63L) expression (Fig. 5C) was unexpected. The converse increase in IRES-dependent translation with dominant negative 4E-BP expression seen only with co-expression of ApRheb(Q63L) (Fig. 6A) was also surprising. Lastly, we did not expect the reversal of the increased IRES-dependent translation caused by expression of dominant negative S6K observed when ApRheb(Q63L) is co-expressed with dominant negative S6K (Fig. 6B). At this time, we are unable to explain these findings. One possibility is that RNA-binding proteins that specifically regulate the IRES used in these studies, the IRES from the 5′UTR of the mRNA encoding the Aplysia ELH, are translationally regulated by these manipulations. Because ELH is not naturally expressed in tail SNs (59), these findings may not be physiologically relevant and have not been explored further.

Separating TORC1 Regulation from Other Translational Regulation in Aplysia

Overexpression of ApRheb increases cap-dependent translation, but 5-HT initially decreases translation rates in SNs, despite the fact that it activates TORC1 (60, 61). This is probably due to additional effects of 5-HT on other translational control pathways. For example, 5-HT initially decreases translation by activating PKC, and this is not affected by rapamycin (61). However, whereas PKC appears to dominate over TORC1 in regulating general translation at earlier time points, translation rates of specific transcripts at these time points may still be more dependent on TORC1 activity. Moreover, spaced applications of 5-HT, as used for LTF in this study, may favor activation of the TORC1 system, compared with the continuous application of 5-HT used in the studies that measured translation rate (11, 61).

4E-BP and Somatic Translation in Aplysia Sensory Neurons

The general rate of cap-dependent translation was stimulated by TORC1 independently of 4E-BP regulation. This is consistent with our previous results demonstrating that within the cell soma of Aplysia SNs the level of free eIF4E is not rate-limiting for translation as overexpressing eIF4E did not increase cap-dependent translation (11). As discussed above, because 24-h LTF appears to require translational control in the cell soma, it is, therefore, not surprising that 4E-BP regulation is not required for 24-h LTF. Other phases of facilitation, however, may require TORC1 signaling out at the synapse, because 72-h LTF is blocked by local, synaptic application of rapamycin (14) and the translational requirements for the intermediate term facilitation induced by spaced applications of 5-HT have not yet been determined. Also, increased phosphorylation of 4E-BP is observed in synaptosomes treated with 5-HT, and 4E-BP is hypophosphorylated in synaptosomes as compared with cytosol (11). Therefore, it is important to stress that, whereas somatic 4E-BP signaling may lack a role in 24-h LTF, this does not exclude synaptic 4E-BP signaling from playing a role in other phases of facilitation.

Comparison to a Mammalian Model

The roles of S6K have also been examined in TORC1-dependent forms of synaptic plasticity and learning in mice. It was found that mGluR-LTD was intact in S6K2 knock-out mice and enhanced in S6K1/S6K2 double knock-out mice (22). Removal of different isoforms of S6K in rodents affected several memory tests and early long-term potentiation, although it did not affect protein synthesis-dependent L-LTP (23). This may be due to different translational mechanisms being involved in mGluR-LTD and L-LTP, versus 24-h LTF in Aplysia. Also, the effect on L-LTP was not examined in double knock-out mice. Therefore, the lack of an effect may be due to compensation from the other S6K isoform. Moreover, in these studies, the complete removal of S6K over the lifetime of the animal may have had additional affects, such as the loss of feedback regulation onto Akt signaling as removing S6K increases Akt activation, and this may have compensated for the loss of S6K activity (62). It will be interesting, once detailed mechanisms underlying the role of S6K in LTF are elucidated, to determine if these mechanisms are conserved in mammals.

Summary

In summary, we have demonstrated that S6K, but not 4E-BP, plays a critical role during 24-h LTF and have shown that neither S6K nor 4E-BP are involved in TORC1-mediated increases in general, somatic translation. Thus, we have begun to unravel the various routes through which TORC1 orchestrates distinct types and phases of plasticity and, therefore, the targets of its translational regulation, whether general or specific. By continuing this process, we will gain a clearer understanding of the significance of this key switch that helps our brains grow and generate imprints of our experiences.