Essential Role of Somatic and Synaptic Protein Synthesis and Axonal Transport in Long-Term Synapse-Specific Facilitation at Distal Sensorimo
http://www.100md.com
《生物学报告》
Department of Bioengineering, University of Utah, Salt Lake City, Utah 84112
Abstract
To investigate further the cellular mechanisms underlying long-term facilitation (LTF) and long-term synapse-specific facilitation (LTSSF), we studied the role of axonal transport and somatic and synaptic protein synthesis at proximal and distal synapses of Aplysia siphon sensory neurons (SNs). The long soma-synapse distances (2.5 to 3 cm) of the SN distal synapses impose important temporal and mechanistic constraints on long-term facilitation and on intracellular signaling. Excitatory postsynaptic potentials (EPSPs) evoked by SNs in central and peripheral siphon motor neurons were used to assay LTF 24–30 h after various pharmacological treatments. Inhibition of protein synthesis via anisomysin application at either the SN soma or distal synapses blocked the induction of LTF and LTSSF normally produced by synaptic application of the facilitating transmitter serotonin (5-hydroxytryptamine). Further, disruption of axonal transport by application of nocodazole to the isolated siphon nerve completely blocked LTF at distal synapses. These results indicate an essential role for somatic and synaptic protein synthesis and active axonal transport in LTSSF at distal synapses, and raise intriguing questions for current synaptic marking/capture models of synapse specificity and LTF.
Abbreviations: ANOVA, analysis of variance ? ASW, artificial seawater ? C-Aniso, central anisomycin ? CMN, central motor neuron ? EPSP, excitatory postsynaptic potential ? 5-HT, 5-hydroxytryptamine, serotonin ? LTF, long-term facilitation ? LTP, long-term potentiation ? LTSSF, long-term synapse-specific facilitation ? MN, motor neuron ? P-Aniso, peripheral anisomycin ? PMN, peripheral motor neuron ? SN, sensory neuron
Introduction
Despite its dependence on genomic regulation and protein synthesis (Montarolo et al., 1986; Kandel, 2001), long-term synaptic facilitation (LTF) at Aplysia sensorimotor connections can be synapse-specific and can occur preferentially at selected synapses of a given siphon sensory neuron (SN), relative to other connections of the same cell (Clark and Kandel, 1993; Gooch and Clark, 1997; Martin et al., 1997a). Further, both short-term and long-term synapse-specific facilitation (LTSSF) can be triggered by applications of the facilitating transmitter serotonin (5-hydroxytryptamine; 5-HT) that are restricted to sensorimotor connections and that do not directly contact the soma, suggesting that communication between the soma and synapse is involved in LTSSF (Clark and Kandel, 1993; Emptage and Carew, 1993; Gooch and Clark, 1997; Martin et al., 1997a; Casadio et al., 1999). LTF at siphon SN connections contributes importantly to long-term behavioral sensitization of the gill- and siphon-withdrawal reflex, which, like LTF, is dependent on protein synthesis (Castellucci et al., 1989).
Because individual neurons can make connections with hundreds or thousands of postsynaptic target cells, synapse-specific facilitation provides a high degree of precision in the modification of neural pathways. It also raises intriguing questions about the mechanisms for LTF, which depends on changes in gene expression and on new protein synthesis.
The first direct demonstration of synapse-specific facilitation in Aplysia examined the connections of individual siphon SNs onto two physically separate sets of postsynaptic targets: at proximal SN synapses onto nearby central motor neurons (CMNs) in the intact abdominal ganglion, and at SN synapses onto peripheral motor neurons (PMNs) at the distal end of the attached siphon nerve, several centimeters away (Clark and Kandel, 1993; Gooch and Clark, 1997). As with short-term facilitation in this system (Clark and Kandel, 1984), application of 5-HT that was restricted to distal synaptic regions (and did not directly involve the soma) was sufficient to induce LTF. Moreover, LTF occurred preferentially at synaptic connections previously treated with 5-HT, compared with untreated connections of the same SNs (Clark and Kandel, 1993; Gooch and Clark, 1997), indicating that LTF, like short-term facilitation, was synapse-specific. LTF can also be induced by restricted 5-HT applications at pleural SN connections onto motor neurons (MNs) in the pedal ganglion (Emptage and Carew, 1993).
In a closely related, elegant series of studies, short-term and long-term synapse-specific facilitation has also been demonstrated and analyzed in vitro (Martin et al., 1997a; Casadio et al., 1999), using a three-cell culture system in which an Aplysia siphon SN bifurcated to synapse onto two nearby MNs. LTF is also synapse-specific in this system, and it requires both local and somatic protein synthesis in the presynaptic neuron. Further results suggest that synapse specificity may arise from synaptic "marking," or "tagging," wherein new gene products are shipped from the soma to all synapses, but are captured or preferentially utilized at synapses marked, or tagged, by previous 5-HT treatment. According to some synaptic marking/capture models (Martin et al., 1997a; Casadio et al., 1999), synaptic application of 5-HT triggers a retrograde intracellular message from synapse to soma that activates transcription and translation. New gene products are then synthesized in the soma and shipped to all synapses, but these products are captured or utilized only at synapses previously marked by molecular events arising from being treated with 5-HT. Hence, LTF is expressed only (or preferentially) at these marked synapses. Although the capture of 24-h LTF is independent of protein synthesis, the initiation of LTF by synaptic application of 5-HT is dependent on protein synthesis, suggesting that the newly synthesized local proteins may serve as, or help generate, the presumed retrograde signal to the nucleus.
An important aspect of synaptic marking/capture models is that both the synaptic mark and the new gene products shipped from soma to synapse must persist sufficiently long for the new gene products to arrive and be captured. However, the reported time windows for both synaptic marking and capture in LTSSF in dissociated cell culture are relatively short, no more than a few hours (Casadio et al., 1999). Similarly, at pleural sensorimotor connections, the time window for coincident, or asymmetric, LTF (arising from combined somatic and synaptic applications of low concentrations of 5-HT, each of which by itself is insufficient to produce LTF) appears to be very short, only about 15 min (Sherff and Carew, 1999). Local protein synthesis and synaptic marking also contribute to hippocampal synapse-specific long-term potentiation (LTP) (Frey et al., 1988; Frey and Morris, 1997, 1998; Steward and Schuman, 2003), which also exhibits a brief time window—less than 3 h—for synaptic marking/capture (Frey and Morris, 1997).
Compared with the situation in dissociated cell culture or pleural sensorimotor connections, the long soma-synapse distance (2.5 to 3 cm) at siphon SN synaptic connections onto PMNs imposes important temporal and mechanistic constraints on the cellular and molecular processes underlying soma-synapse signaling and LTF, thereby raising several key questions. In particular, although a short time window may suffice for LTF at synapses close to the SN soma, how would synaptic marking and capture work at truly distal synapses? The soma-synapse active axonal transport times for macromolecules—perhaps 13–17 h (see Discussion; Goldberg et al., 1978; Schwartz, 1979; Goldberg and Schwartz, 1980; Gunstream et al., 1995)—would presumably be outside the short time windows for synaptic marking and capture observed for proximal synapses. Can a synaptic mark at distal synapses persist long enough to allow retrograde signaling from the synapse to the soma, followed by anterograde signaling back to the synapse? If so, such a long-lasting synaptic mark would require maintenance mechanisms that remain to be elucidated. Further, how are the signals communicated between the soma and distal synapses? If intracellular communication is appreciably faster than conventional axonal transport, then the mechanisms by which this communication could occur are not yet known. Alternatively, are rather different mechanisms involved in mediating LTF at distal synapses in the adult animal, compared with newly developed synapses in dissociated cell culture?
To begin to address these questions, we asked whether LTF and LTSSF produced by restricted 5-HT applications at distal SN connections onto PMNs depend on somatic or local protein synthesis, and on active axonal transport between soma and synapse. We found that blockade of protein synthesis at either the SN soma or the distal synaptic region blocked LTF and LTSSF, indicating that both local and somatic protein synthesis are essential even at distal synapses. Blockade of axonal transport also blocked LTF at distal synapses, implicating active axonal transport and soma-synapse communication. Preliminary reports of some of these results have previously appeared (Guan and Clark, 2001, 2003).
Materials and Methods
Preparation
Wild-caught adult Aplysia californica weighing 100–150 g were supplied by Marinus, Long Beach, California. The slugs were housed in an aquarium tank containing artificial seawater (Tropic Marine) at a temperature of 14 °C. They were fed with romaine lettuce daily to maintain a constant weight.
Prior to dissection, the animal was anesthetized by cooling at 0 °C for 20 min. It was then further anesthetized by injection of cooled (4 °C) isotonic MgCl2 (100 ml MgCl2/100 g animal weight) into the hemocoel. After the skin of the animal was completely relaxed, the animal was pinned to a dissection pan with the ventral side facing up, and was cut open from the caudal to rostral end. The abdominal ganglion together with the attached siphon nerve and the pleuroabdominal connectives were dissected out in a cooled "zero-calcium" solution (NaCl, 338 mM; KCI, 10 mM; MgCl2, 150 mM; 4 °C). The zero-calcium solution (containing no added calcium) was used to minimize the hyperexcitability and synaptic facilitation potentially induced by axon injury during dissection (Ambron et al., 1992). The siphon nerve was kept as long as possible to obtain more PMNs. The connectives and the sheath of the abdominal ganglion were fixed by 25% glutaraldehyde for 30 s.
The excised abdominal ganglia and attached siphon nerve were then pinned onto the Sylgard floor of a polystyrene recording dish in cooled zero-calcium solution. A cold plate was placed underneath the dish to minimize dissection-induced injury. For studies of protein synthesis inhibition, the recording dish was divided into two small chambers, in which the SNs in the abdominal ganglion with their CMN targets were in one chamber and were separated from the SN distal synaptic terminals and PMN postsynaptic targets in the other chamber (see Fig. 1). The siphon nerve was threaded through the slit in the dividing wall, which was then filled with petroleum jelly to ensure that the two chambers were isolated from one another. To test the isolation, a drop of fast green dye was applied to one chamber. If the dye did not pass to the other chamber, the isolation was considered to be complete. This experimental setup separated the SN soma and central synapse from the distal peripheral synapse, thus allowing the perfusion of drugs independently to one site or the other. Similarly, studies of axonal transport inhibition used a three-chamber recording dish, which allowed application of nocodazole to the isolated siphon nerve.
The connective tissue covering the abdominal ganglion was partially removed to allow the penetration of SN and CMN. The distal end of the siphon nerve was treated with 0.1% bovine trypsin (Sigma, type IX) for 5 min at room temperature (22 °C) to facilitate the subsequent penetration of intracellular electrodes into PMNs. The trypsin action was terminated by incubating the treated nerve with 0.25% trypsin inhibitor solution (Sigma, type II-S: soybean) for 15 min.
To allow the dissected neurons to recover from potential injury incurred during dissection, the cooled zero-calcium dissecting solution was replaced by enriched artificial seawater (ASW), and the preparation was allowed to rest in enriched ASW for 5 h before the beginning of electrophysiological testing. Enriched ASW had the components of NaCl, 460 mM; KCl, 10 mM; MgCl2, 55 mM; CaCl2, 11 mM; Hepes-NaOH, 10 mM, along with nutrient, amino acid, vitamin, and sugar supplements (Gibco), and penicillin/streptomycin (Sigma, 100,000 units penicillin and 100 mg streptomycin per liter ASW) to prevent infection (Clark and Kandel, 1993).
Intracellular recording and stimulation
Using standard stimulating and recording techniques, we looked for monosynaptic connections between a given LE siphon SN and an LFs siphon CMN in the abdominal ganglion, a siphon PMN at the distal end of the siphon nerve, or both. A "high Mg2+ and Ca2+" solution was used as the recording medium to hinder the firing of interneurons that might evoke polysynaptic EPSPs in MNs (NaCl, 338 mM; KCI, 10 mM; MgCl2, 100 mM; CaCl2, 50 mM; Hepes-NaOH, 10 mM; also supplemented with vitamins and sugars, amino acids, and penicillin/streptomycin solutions). Typically we recorded from one CMN and one siphon PMN, and then tested about 20 LE siphon SNs per preparation for monosynaptic connections. Neurons were identified on the basis of their morphological and electrophysiological characteristics. Although PMNs are distributed along the distal end of the siphon nerve and are found up to 5 cm from the abdominal ganglion, we typically recorded from PMNs that were relatively close to the SN somata (2.5–3 cm away), just past the first major branching of the siphon nerve, to maximize the probability of finding connections with SNs.
Neurons were sampled by penetration with a single-barreled glass microelectrode (15–25 M) filled with electrolyte (2.5 M KCl). During penetration, hyperpolarizing current was passed through the electrode to help determine the point at which the electrode tip came into physical contact with the cell membrane, and to prevent potential firing of SNs during penetration (and so ensure subsequent accurate sampling of the first synaptic potential evoked in the MN). During testing, MNs were hyperpolarized by 40 mV below resting membrane potential to prevent EPSPs from evoking action potentials. Each SN-synapse was tested a single time during a recording session by passing intracellular current into the SN to evoke an action potential and possible resultant EPSP in the PMN and CMN targets.
After testing was completed on day 1, one of a series of drug treatments (detailed below) was conducted. In order to help find the same neurons on day 2 as on day 1, the cells surrounding the recorded neurons on day 1 were pulse-injected with fast green dye to label their positions, and a diagram was drawn (Clark and Kandel, 1993). Recording sessions and drug treatments were conducted at room temperature (22 °C), which was kept the same on day 1 and day 2. After drug treatments and washout were completed, the nervous system was incubated for 24 h (anisomycin studies) or 30 h (nocodazole studies) in enriched ASW at 18 °C.
The same synaptic connections were retested in "high Mg2+ and Ca2+" solution the following day, day 2. To avoid potential "floor" and "ceiling" effects, only synapses with EPSPs between 2 and 25 mV on day 1 were tested again on day 2. To maximize experimental yield, we retested SNs that had connections onto either a CMN alone, a PMN alone, or both a CMN and PMN.
Pharmacological treatments
We conducted separate studies to examine the effects of protein synthesis inhibition and of axonal transport inhibition.
In the experiments on the inhibition of protein synthesis, after physiological recording was completed on day 1, preparations were exposed to one of six possible drug treatments (see Fig. 1): (1) ASW; (2) C-Aniso alone (anisomycin applied in the central (C) chamber, containing the SN somata); (3) P-Aniso alone (anisomycin applied in the peripheral (P) chamber, containing the SN peripheral synapses); (4) P-5-HT + C-Aniso (peripheral application of 5-HT, combined with central application of anisomycin); (5) P-5-HT + P-Aniso (peripheral application of both 5-HT and anisomycin); or (6) P-5-HT (peripheral application of 5-HT). Aniso (10 μM) was applied 1 h before 5-HT application (5 x 5-min, 50 μM, separated by a 15-min washout period of 10 bath volumes of ASW). For peripheral applications of anisomycin and in the C-Aniso alone condition, anisomycin was washed out after the end of the experiment on day 1. For the P-5-HT + C-Aniso condition, anisomycin remained in the bath until the experiment finished the next day in one subgroup, to allow for potential delay in the critical period for protein synthesis arising from retrograde transport. A separate subgroup received only a 3-h application of central anisomycin, to better match the durations used in the other experimental drug treatments. Results of the two subgroups were nearly identical and hence were combined for the overall analysis of variance (see Results).
In experiments on the inhibition of axonal transport, preparations were exposed to one of three possible drug treatments (see Fig. 3): (1) M-Nocodazole alone (application of the axonal transport inhibitor nocodazole to the middle (M) chamber of a three-chambered dish containing the siphon nerve and SN axons); (2) M-Nocodazole + P-5-HT (application of nocodazole onto the siphon nerve, combined with application of 5-HT at SN peripheral synapses); or (3) P-5-HT alone. The synaptic application of 5-HT was intended to induce LTF; the 5-HT (5 x 5-min, 50 μM, separated by a 15-min washout) was applied into the peripheral chamber only. To examine the role of axonal transport, nocodazole, an axonal transport inhibitor that causes depolymerization of axonal microtubules (Turner and Tartakoff, 1989), was applied only in the middle chamber containing the siphon nerve, including SN axons; hence, nocodazole was not directly applied onto either the SN synapses or SN somata. To avoid disturbing the general metabolism of the neuron, nocodazole concentration was 50 μM. Nocodazole (Sigma) was dissolved in dimethyl sulfoxide (5mg/ml) and culture medium (Gunstream et al., 1995). Because the effect of nocodazole is reversible (Dinter and Berger, 1998), it was applied at least 1 h before 5-HT application and was present until the experiment finished. After completion of the drug treatment and washout following the recording of day 1, the nervous system was incubated for 30 h in enriched ASW at 18 °C. The extension of the incubation time from 24 h (anisomycin experiments) to 30 h (present experiments) was intended to allow potential retrograde and anterograde messengers sufficient time to travel between synapse and soma (Schwartz, 1979), given that 5-HT was applied at a slightly more distal site than in our anisomycin experiments.
Data analyses
EPSPs triggered by SN action potentials were recorded into a computer by a CED MICRO 1401 AD interface card with SPIKE 2 software, ver. 3 (Cambridge Electronic Design). An "LTF score" reflecting the synaptic change from day 1 to day 2 was calculated as follows. First, to determine LTF for a particular synapse, the EPSP amplitude on day 2 at one synapse was divided by the EPSP amplitude on day 1 at the same synapse; a log10 transform was then applied so that potential increases and decreases in synaptic strength would contribute equally. To determine the mean LTF for a particular preparation, the mean of the log10 transform of facilitation across different central (or peripheral) synapses within one preparation was calculated, yielding the LTF score for central (or peripheral) synapses for that individual preparation. Finally, to determine the mean LTF for a particular experimental group receiving a particular drug treatment, the mean LTF across preparations was calculated independently for central and peripheral synapses.
LTF scores were screened for outliers, and any LTF score that was 1.5 interquartile ranges below the first quartile or above the third quartile was removed. This procedure was implemented in part to minimize potential spurious results that might have occurred from misidentifying a neuron between recording sessions on days 1 and 2. Statistical analyses were performed using SPSS statistical software, ver. 11.5.
For studies examining the effects of protein synthesis inhibition, a two-factor analysis of variance (ANOVA) was conducted, in which one factor had two levels to represent changes in synaptic strength at either central or peripheral synapses ("synaptic site" or simply "site"), and the other factor had six levels to represent the six drug treatments ("drug treatment" or simply "drug"). The overall ANOVA was followed by post hoc Tukey HSD tests for individual comparisons among the 12 possible conditions (2 synaptic sites x 6 drug treatments). Individual student’s t tests were used to examine whether LTF scores demonstrated significant changes in EPSP amplitude on day 2 relative to day 1 in each of the 12 treatment conditions. As a more rigorous test of LTSSF in the experiments on protein synthesis inhibition, we separately analyzed LTF at the subset of SNs that had connections onto both CMNs and PMNs.
Comparable analyses were conducted to examine the effects of axonal transport inhibition, using a two-factor ANOVA to compare LTF at central and peripheral synapses for the three experimental treatments, followed by post hoc Tukey tests.
Results
Results indicate an essential role for both somatic and local protein synthesis in LTSSF at distal synapses, and further indicate that active axonal transport between soma and the distal synapse is required.
Inhibition of either somatic or synaptic protein synthesis blocks 5-HT-induced LTF at distal synapses
In dissociated cell culture, LTF and LTSSF at Aplysia sensorimotor connections require protein synthesis in both somatic and synaptic compartments (Montarolo et al., 1986; Martin et al., 1997a; Casadio et al., 1999). However, the close proximity of the siphon sensory cell soma and synapse in those experiments leaves unanswered the question of whether similar requirements exist for synapses that are remote from the soma (e.g., 2.5 to 3 cm, for siphon SN connections onto peripheral siphon MNs), or whether instead such long distances pose a communication challenge that changes the roles of protein synthesis.
To address this question, we applied the protein synthesis inhibitor anisomycin either centrally at the SN cell body and central synapses in the abdominal ganglion, or distally at SN peripheral synapses. We found that blockade of protein synthesis at either the sensory neuron soma or distal synaptic sites completely blocked the LTF and LTSSF that are normally produced by synaptic application of 5-HT. An analysis of variance was conducted on changes in synaptic strength at central and peripheral synaptic connections across each of the six drug treatment conditions in all preparations. Consistent with a role for protein synthesis, LTF varied significantly as a function of drug treatment condition (F(5, 109) = 12.06, P < 0.001), as did differences in LTF at peripheral compared with central synapses (significant drug-by-site interaction, F(5, 109) = 5.83, P < 0.001). The main effect of synaptic site (LTF at peripheral synapses compared with LTF at central synapses, across all six drug treatment groups) approached but did not quite reach significance (F(5, 109) = 3.89, P = 0.051), most likely because, in keeping with experimental hypotheses, only one of the six drug treatments (Peripheral 5-HT alone) showed peripheral-central LTF differences. Subsequent analyses support four main conclusions (Figs. 1 and2).
Synaptic application of 5-HT produces LTF.
First, confirming previous results (Clark and Kandel, 1993; Emptage and Carew, 1993; Gooch, 1997; Gooch and Clark, 1997; Martin et al., 1997a; Casadio et al., 1999; Liu et al., 2003; Grabham et al., 2005), repeated application of 5-HT alone at peripheral synapses, without anisomycin, produced LTF specifically at peripheral synapses, approximately doubling the mean EPSP amplitude from day 1 to day 2 (mean ± SEM LTF score = 0.266 ± 0.027, n = 9 preparations, 11 synapses, t(8) = 9.85, P < 0.001). Consistent with LTF being synapse-specific, there were no significant changes at untreated central synapses (mean LTF score = 0.035 ± 0.031, n = 10 preparations, 18 synapses, t(9) = 1.13, P > 0.25), and facilitation at treated peripheral synapses was greater than at untreated central synapses (P < 0.001, Tukey). These results confirm that synaptic application of 5-HT is sufficient to produce LTF at peripheral synapses, and suggest that this LTF can be synapse-specific.
LTF requires protein synthesis at the soma.
Second, LTF at distal peripheral synapses was completely blocked by the application of anisomycin to the sensory cell soma, as evidenced by the complete lack of facilitation on day 2, relative to day 1 (Peripheral 5-HT + Central Anisomycin, mean LTF score = 0.01 ± 0.01, n = 17 preparations, 32 synapses, t(16) = 1.00, P > 0.95). Further indicating that blockade had occurred, LTF scores at peripheral synapses in these anisomycin-treated preparations (1) were significantly lower than LTF scores at peripheral synapses in the Peripheral 5-HT alone group, without anisomycin (P < 0.001, Tukey); and (2) were nearly identical to LTF scores at central synapses (which had not been treated with 5-HT), as well as to LTF scores at peripheral synapses in either the ASW control group or Central Anisomycin alone control group (all P values > 0.99, Tukey).
The duration of exposure to central anisomycin (3 h or 24 h) made no apparent difference. The blockade of peripheral LTF that occurred in those preparations that received peripheral 5-HT in conjunction with exposure to central anisomycin for only 3 h (mean LTF score = –0.02 ± 0.03, n = 9 preparations, 16 synapses) was nearly identical to blockade in preparations that received central anisomycin for the full 24-h period (mean LTF score = –0.00 ± 0.01, n = 8 preparations, 16 synapses). A separate analysis of variance conducted on changes in synaptic strength at central and peripheral synaptic connections across the two different anisomycin exposure times showed no significant main effect of exposure time (F (1, 29) = 1.66, P > 0.20), synaptic site (F(1, 29) = 0.01, P > 0.85), or time-by-site interaction (F(1, 29) = 0.11, P > 0.70).
Taken together, this second set of results indicates that somatic protein synthesis is essential for LTF at distal synapses. Consequently, these results further suggest that LTF may involve the regulation of genomic events in the soma, as well as anterograde or retrograde messengers between the soma, where protein synthesis was inhibited in these experiments, and the synapse, where LTF was triggered and ultimately expressed.
LTF requires local protein synthesis.
Third, similar to results for anisomycin application to sensory cell somata in these experiments and in previous work (Martin et al., 1997a), LTF at distal peripheral synapses was also completely blocked by the anisomycin application to the sensory cell synapses. Several findings demonstrated this blockade: (1) there was no facilitation on day 2, relative to day 1, for the Peripheral 5-HT + Peripheral Anisomycin group (mean LTF score = 0.01 ± 0.04, n = 8 preparations, 11 synapses, t(7) = 0.40, P > 0.70); (2) facilitation at peripheral synapses in these preparations was significantly lower than in the Peripheral 5-HT alone group, without anisomycin (P < 0.001, Tukey); and (3) LTF scores at peripheral synapses were nearly identical to LTF scores at central synapses (which had not been treated with 5-HT), as well as to LTF scores at peripheral synapses in either the ASW control group or the Peripheral Anisomycin alone control group (all P values > 0.99, Tukey). Taken together, this third set of results indicates that local protein synthesis plays an essential role in LTF at distal synapses.
Blockade of LTF is not due to nonspecific effects.
Finally, synaptic strength was stable from day 1 to day 2 in all three control groups that did not receive 5-HT. Synaptic strength did not change significantly in the ASW control group, thus ruling out potential nonspecific effects of time and of potential cellular responses to the dissection, such as injury-induced synaptic enhancement (Clark and Kandel, 1993) (mean LTF score at central synapses = –0.02 ± 0.03, n = 12 preparations, 23 synapses, t(11) = 0.79, P > 0.40; mean LTF score at peripheral synapses = –0.04 ± 0.04, n = 8 preparations, 9 synapses, t(7) = 0.89, P > 0.35). Similarly, synaptic strength was stable in the presence of central anisomycin alone (mean LTF score at central synapses = –0.03 ± 0.02, n = 11 preparations, 20 synapses, t(10) = 1.12, P > 0.25; mean LTF score at peripheral synapses = –0.02 ± 0.02, n = 8 preparations, 11 synapses, t(7) = 0.69, P > 0.50). Synaptic strength was also stable in the presence of peripheral anisomycin alone (mean LTF score at central synapses = –0.01 ± 0.05, n = 7, t(6) = 0.09, P > 0.90; mean LTF score at peripheral synapses = –0.04 ± 0.02, n = 7, t(6) = 1.93, P > 0.10). This fourth set of results indicates that protein synthesis inhibition per se does not produce nonspecific degradation of synaptic strength.
Overall, these results implicate a critical role for protein synthesis, both at the soma and locally at the synapse, in the induction of LTF at distal synapses.
Inhibition of either somatic or synaptic protein synthesis also blocks synapse-specific LTF
An individual siphon SN may make synapses onto CMNs, PMNs, or both. In the analysis described above, not every SN exhibited a synaptic connection onto both a CMN and a PMN. As a more rigorous examination of whether LTF was synapse-specific, we separately analyzed the subset of individual SNs that had both central and peripheral connections tested on both day 1 and day 2, thus allowing direct comparison of LTF between central and peripheral connections of a given sensory neuron.
Results for this dually connected SN subset were highly comparable with results from the complete data set (Fig. 2B). As with LTF in the complete data set, we found that blockade of protein synthesis at either the sensory neuron soma or distal synaptic sites completely blocked the LTF that is normally produced by synaptic application of 5-HT. A repeated-measures overall analysis of variance was conducted on changes in synaptic strength at central and peripheral synaptic SN-MN connections (within-subjects) across each of the six drug treatment conditions (between-subjects). Consistent with the role of protein synthesis in LTSSF, LTF varied significantly as a function of drug treatment condition (F(5, 25) = 4.27, P < 0.001), as did differences between LTF at peripheral and central synapses of the same individual sensory neurons (significant drug-by-site interaction (F(5, 25) = 2.63, P < 0.05). The main effect of synaptic site (peripheral synapses versus central synapses) collapsed across all six drug treatment conditions was not significant (F(5, 25) = 0.42, P = 0.52), most likely because only one of the six drug treatments (Peripheral 5-HT alone) produced peripheral-central LTF differences, in keeping with experimental hypotheses.
In the Peripheral 5-HT alone condition, strong LTF occurred at peripheral synapses that had been previously treated with 5-HT, as evidenced by an approximate doubling of the mean EPSP amplitude from day 1 to day 2 (mean ± SEM LTF score = 0.242 ± 0.042, n = 5 preparations, 5 synapses, t(4) = 5.70, P < 0.005). In addition, explicitly indicating that LTF was indeed synapse-specific, LTF was greater at treated peripheral synapses than at untreated central synapses of the same sensory neurons (P < 0.05, Tukey), which showed no facilitation (mean LTF score = –0.01 +0.08, t(4) = 0.11, P > 0.90). In contrast with results at 5-HT-treated peripheral synapses, for all five other drug treatment conditions, including the two experimental groups that received 5-HT plus either central or peripheral anisomycin, there were no significant increases or decreases in synaptic strength at either central or peripheral synapses (LTF score ranged from –0.12 ± 0.07 to 0.01 ± 03, all P values > 0.05, t tests). Similarly, there were no significant differences in LTF between peripheral and central synapses (all P values > 0.95, Tukey). These results confirm that LTF produced by synaptic application of 5-HT is synapse-specific, and they explicitly document that LTSSF also requires both somatic and local protein synthesis.
LTF at distal synapses requires active axonal transport
Both somatic and synaptic protein synthesis are essential for LTF triggered at distal sensorimotor synapses, as shown above, as well as at proximal synapses in dissociated cell culture (Martin et al., 1997a; Casadio et al., 1999), suggesting that communication between soma and synapse is essential for LTF. To examine directly the role of active axonal transport in LTF and intracellular signaling between synapse and soma, here we investigated whether incubating SN axons in the axonal transport inhibitor nocodazole, which disrupts microtubules (Dinter and Berger, 1998), would block LTF at distal peripheral SN-MN synapses normally triggered by synaptic application of 5-HT. We found that nocodazole completely blocked LTF at distal synapses, indicating an essential role for active anterograde and/or retrograde transport in LTF at distal synapses.
An analysis of variance was conducted on changes in synaptic strength at central and peripheral synaptic connections across each of three drug treatment conditions in all preparations. Consistent with a role for axonal transport, LTF varied significantly as a function of both drug treatment condition (F(2, 26) = 21.60, P < 0.001) and synaptic site (peripheral synapses vs. central synapses) (F(2, 26) = 28.16, P < 0.001), and there was a significant drug-by-site interaction (F(2, 26) = 35.94, P < 0.001), indicating that peripheral-central differences in facilitation were dependent on the drug treatment condition. Subsequent analyses support the following three main conclusions (Fig. 3).
Synaptic application of 5-HT produces LTF.
First, confirming previously published results (Clark and Kandel, 1993; Emptage and Carew, 1993; Gooch, 1997; Gooch and Clark, 1997; Martin et al., 1997a; Casadio et al., 1999; Liu et al., 2003; Grabham et al., 2005) as well as studies above, repeated application of 5-HT alone at peripheral synapses, without nocodazole, produced LTF specifically at peripheral synapses, approximately doubling the mean EPSP amplitude from day 1 to day 2 (mean ± SEM LTF score = 0.24 ± 0.03, n = 6 preparations, 13 synapses, t5 = 8.02, P < 0.001). Consistent with synapse-specific LTF, facilitation at 5-HT-treated peripheral synapses was also greater than at untreated central synapses (P < 0.001, Tukey), which showed no significant LTF (mean LTF score = 0.022 ± 0.014, n = 6 preparations, 13 synapses, t(5) = 1.57, P > 0.15). Thus, the present results confirm that synaptic application of 5-HT is sufficient to produce LTF and LTSSF at peripheral synapses, and they provide a positive control to examine the effects of axonal transport blockade.
Nocodazole completely blocked LTF at distal synapses.
Second and most importantly, LTF at distal peripheral synapses produced by 5-HT application in the peripheral (P) chamber was blocked completely by the prior application of the active axonal transport inhibitor nocodazole to the middle (M) chamber containing the siphon nerve and SN axons (M-Nocodazole + P-5-HT group; mean LTF score = –0.008 ± 0.017, n = 5 preparations, 11 synapses, t(4) = 0.45, P > 0.65). Further demonstrating blockade of LTF, LTF scores at peripheral synapses in these nocodazole-treated preparations were (1) significantly lower than LTF at peripheral synapses in the 5-HT alone group, which did not receive nocodazole (P < 0.001, Tukey); and (2) nearly identical to LTF scores at central synapses in the same experimental group (P > 0.95, Tukey), and to LTF scores at central and peripheral synapses in the M-Nocodazole alone control group (P > 0.95, Tukey), all of which showed no LTF (LTF score range 0.004 ± 0.009 to 0.012 ± 0.016, all P values > 0.45, t tests). This set of results provides the first direct evidence for an essential role of axonal transport between synapse and soma in LTF and LTSSF.
Blockade of LTF is not due to nonspecific effects.
Finally, synaptic strength was stable from day 1 to day 2 in the M-Nocodazole control group, which did not receive 5-HT, thus ruling out nonspecific effects of time and dissection-induced changes (Clark and Kandel, 1993), as well as nonspecific degradation of synaptic strength produced by blockade of axonal transport (mean LTF score at central synapses = 0.005 ± 0.009, n = 5 preparations, 13 synapses, t(4) = 0.50, P > 0.60; mean LTF score at peripheral synapses = 0.009 ± 0.015, n = 5 preparations, 10 synapses, t(4) = 0.60, P > 0.55). Further, there was no significant difference between LTF at central and peripheral synapses (P > 0.99, Tukey). Because nocodazole was applied to the middle chamber containing only the siphon nerve, and was not applied to the SN soma (in the central chamber) or SN synapses (in the peripheral chamber), it is unlikely that it interfered directly with somatic or synaptic processes involved in LTF.
Taken together, the above results directly demonstrate that active axonal transport is required for LTF.
Discussion
Results support three key conclusions regarding the mechanism of LTF and LTSSF at distal sensorimotor connections. First, proteins synthesized in the SN soma are essential, even for LTF triggered by events at the distal synapse. These proteins could in principle be either induced or constitutively expressed. Second, local protein synthesis at the synapse is also essential for LTF, implying that an immediate translational response might occur at synapses treated with 5-HT. The newly synthesized local proteins might serve as a retrograde messenger, as a synaptic mark that can capture or utilize regulatory factors in LTSSF, or as other direct or indirect factors in the LTF signaling pathway. Third, active anterograde or retrograde axonal transport (or both) are essential for LTF, and they appear to mediate critical intracellular signaling between synapse and soma. However, the mechanisms that might allow for synaptic marking and capture, given the long transport time (perhaps 13–17 h) between the SN distal synapse and soma, remain unknown.
LTF and LTSSF at distal synapses require both somatic and synaptic protein synthesis
Protein synthesis has been shown to be required both in long-term behavioral sensitization of the gill-withdrawal reflex evoked by a tactile siphon stimulus in Aplysia, and in LTF induced by cell-wide applications of 5-HT at Aplysia sensorimotor connections in dissociated cell culture (Montarolo et al., 1986; Castellucci et al., 1989). Studies in dissociated cell cultures in which a single sensory neuron contacts two nearby motor neurons have further indicated that both local and somatic protein synthesis are required for the establishment of LTSSF (Martin et al., 1997a).
The present work extends these previous findings to indicate that synaptic and local protein synthesis are also required for LTF and LTSSF at truly distal SN synapses, located several centimeters from the SN soma. The long soma-synapse distance used in the present studies imposes important temporal and mechanistic constraints on the processes underlying LTF and on the intracellular signals by which soma and synapse communicate, and thus raises questions regarding proposed models based on LTF experiments at proximal synapses.
As previously suggested (Martin et al., 1997a; Casadio et al., 1999), proteins newly synthesized at synapses in response to 5-HT may, among other possibilities, serve as (1) retrograde messengers or signals to generate retrograde messengers (Martin et al., 1997a; Casadio et al., 1999); (2) synaptic tags, or marks, that enable capture and/or stabilization of the LTF at 5-HT treated synapses (Martin et al., 1997a; Casadio et al., 1999); or (3) factors that directly or indirectly mediate LTF (Liu et al., 2003).
Our studies by themselves do not directly indicate whether the expression of the critical proteins is constitutive or is induced or modulated by 5-HT; they also do not discriminate between synthesis in presynaptic SN processes and postsynaptic motor neurons. However, LTF produced by repeated synaptic 5-HT applications in dissociated cell culture and at intact pleural-pedal sensorimotor connections requires presynaptic but not postsynaptic protein synthesis, although the requirements are different for coincident (asymmetric) LTF involving combined somatic/synaptic 5-HT applications (Martin et al., 1997a; Sherff and Carew, 2002, 2004).
According to the first scenario proposed above, retrograde messengers that are newly synthesized in SN processes in response to 5-HT may be transported to the SN soma to modify gene transcription and translation. In turn, newly synthesized somatic proteins may be transported anterogradely back to synapses, where they are captured and utilized preferentially at synapses previously treated with 5-HT, thus producing LTSSF. Several previous findings are consistent with this interpretation. First, in cultures containing a single SN that bifurcates to two different postsynaptic targets, 24-h LTF is induced at one set of SN synapses treated with five pulses of 5-HT, but not at other, untreated synapses of the same SN; further, this LTSSF is blocked by protein synthesis inhibition at synapses being treated with 5-HT. Second, if one SN branch receives five pulses of 5-HT, and the other branch receives only a single pulse of 5-HT, then both branches show 24-h LTF, even though one pulse of 5-HT is usually not sufficient to produce LTF. Critically, protein synthesis inhibition at the branch receiving one pulse of 5-HT does not block LTF at that branch. Taken together, these results suggest that induction of 24-h LTF requires local protein synthesis (e.g., to generate a retrograde message), but 24-h synaptic capture does not, at least under these conditions.
An additional possibility is that locally synthesized proteins could serve as the synaptic tag, or mark, at least when the synaptic stimulus substantively precedes the nuclear events, as was the case in the present studies. Such a mark would allow the synapse to capture or utilize newly synthesized macromolecules subsequently shipped from the soma. In the present studies, such a mark would need to be reasonably long-lasting (to allow arrival of products from the soma), and hence might involve a protein-synthesis-dependent step, even though capture per se is independent of protein synthesis. It is also possible that there are constitutive inhibitory constraints to LTF (as described below), which are relieved cell-wide by 5-HT application and the resultant local protein synthesis. With these inhibitory constraints relieved, a single pulse of 5-HT might be sufficient to mark a synapse for LTF, and enable it to capture proteins either induced or constitutively expressed. Further, whereas local protein synthesis is not required for synaptic capture at 24 h (Martin et al., 1997a), it is required for capture at 72 h, and for stabilization and persistence of synaptic growth for 72-h LTF (Casadio et al., 1999).
After removal of the SN cell body, isolated SN synapses in dissociated cell culture express protein-synthesis-dependent LTF lasting 24 h (Liu et al., 2003), indicating that local protein synthesis can contribute to 24-h LTF relatively directly, independently of a potential role in retrograde signaling or synaptic marking. However, LTF decays more rapidly than usual and is diminished at 48 h (Liu et al., 2003), suggesting that gene expression in the SN soma is required to maintain LTF indefinitely. Curiously, at 24 h, LTF expressed at isolated synapses is greater that that expressed by SN synapses connected to their cell bodies (Liu et al., 2003). These results raise the possibility (among others) that removing the SN soma eliminates or reduces constraints that normally repress LTF. Hence, one additional possible role for changes in transcription and subsequent translation in the SN soma that are triggered by a retrograde message is to remove constraints on the induction or expression of LTF. In the present experiments, potential retrograde messages conveyed from soma to synapse may in principle contribute to these processes as well.
It is unlikely that anisomycin itself was transported retrogradely from synapse to soma, where it inhibited somatic protein synthesis and thereby blocked LTF. First and most convincingly, application of protein synthesis inhibitors to one branch of a given Aplysia SN does not block LTF at another branch treated with 5-HT (Martin et al., 1997a). If anisomycin had been transported from the synapse to the soma where it blocked protein synthesis and consequently LTF, then it would have blocked LTF at the other synapse as well. Second, anisomycin is lipophilic, so it diffuses across cell membrane easily. Therefore, it might be expected to diffuse out from axons once the extracellular anisomycin had been removed, unless somehow bound intracellularly. Finally, in dissociated cell culture, anisomycin inhibits protein synthesis by Aplysia neurons only transiently (Montarolo et al., 1986). This near-immediate recovery of protein synthesis also suggests that anisomycin applied at synapses does not block somatic protein synthesis and LTF by being retrogradely transported to the soma, which is a time-dependent process.
LTF at distal synapses requires active axonal transport
The requirement for protein synthesis both in the soma and locally at the synapse for LTF (Martin et al., 1997a; Guan and Clark, 2001), as well as for coincident induction of LTF (Sherff and Carew, 1999; Sherff and Carew, 2002, 2004), implies an essential role for intracellular communication between the SN synapse and soma. However, the mechanisms by which this communication occurs were not well documented empirically.
Active axonal retrograde transport and anterograde transport between synapse and soma have been proposed as attractive candidate mechanisms for intracellular communication in LTF and have been indirectly implicated in a variety of experiments (Bacskai et al., 1993; Clark and Kandel, 1993; Martin et al., 1997b; Giustetto et al., 2003; Ormond et al., 2004; Thompson et al., 2004). Axonal transport also plays a critical role in other, potentially related phenomena, such as some aspects of injury-induced long-term hyperexcitability in Aplysia SN (Walters et al., 1991; Ambron et al., 1995; Gunstream et al., 1995; Walters and Ambron, 1995), but possibly not all aspects thereof (Weragoda et al., 2004). Axonal transport is also involved in synaptic responses to nerve growth factor (Thoenen et al., 1979). However, although its involvement seems plausible, the requirement for active axonal transport in LTF and LTSSF produced by 5-HT applications to Aplysia SNs had never been directly demonstrated. Further, other signaling pathways, such as calcium signaling or chains of phosphorylation, have also been proposed to be involved in at least some forms of LTF (Sherff and Carew, 1999).
In the present experiments, we found that the axonal transport inhibitor nocodazole, applied selectively to nerve (and not to the SN soma or synapse), completely blocked LTF at distal synapses, indicating that active anterograde axonal transport, retrograde axonal transport, or both, are in fact essential for LTF. Combined with previous results on the role of somatic and synaptic protein synthesis, these findings suggest that molecules generated directly or indirectly via protein synthesis in the synapse in response to 5-HT might be transported retrogradely back to soma to trigger LTF-related cellular events, including modulation of gene expression. In turn, macromolecules synthesized in the soma, which are either constitutively expressed or induced by retrograde signals from the 5-HT-treated synapse, might be transported anterogradely back to the synapse to produce and sustain LTF.
Although axonal transport is relatively slow, it is not necessarily limiting in the present experiments, which examined LTF at 24 or 30 h. As explained below, total round-trip time from synapse to soma and back again might require on the order of 13 h in our anisomycin experiments, or 17 h in our nocodazole experiments, which is well within our 24-h and 30-h time windows.
Estimated transport time can be inferred by extrapolation from retrograde transport rates of 36 mm/day at 15 °C in Aplysia mechanosensory neurons, which yields a retrograde axonal transport rate of 50 mm/day at 18 °C (our overnight incubation temperature), or 78 mm/day at 22 °C (during 2 h of drug treatment), given a Q10 of about 3 in Aplysia neurons (Goldberg et al., 1978) and other systems (Heslop and Howes, 1972). In our experiments, although our postsynaptic sites were 2.5 to 3 cm from the soma, 5-HT application occurred throughout the peripheral chamber and hence was slightly closer to the soma, 1 cm (anisomycin experiments) to 2 cm (nocodazole experiments), thus potentially reducing retrograde transport distance and transport time if functioning 5-HT receptors exist in more proximal regions of SN axons. Retrograde transport over a distance of 1–2 cm might require about 4 to 8.5 h, or less if transport rates were increased by cAMP induced by 5-HT application (Azhderian et al., 1994), by substrate concentration (Goldberg et al., 1978), or other factors (Azhderian et al., 1994).
Similarly, given an estimated basal anterograde transport rate of about 75 mm/d at 18 °C (our overnight incubation temperature), anterograde transport from soma to synapse (2.5 cm) might require about 8 h.
If the retrograde signal induced expression (or repression) of new macromolecules in the soma that were subsequently shipped to the synapse, additional time might be required for these intermediate steps. However, given that the round-trip time for axonal transport may be considerably less than our 24-h or 30-h time windows, in principle there might be sufficient time for these additional transcriptional and translational modifications to occur as well.
Consistent with the present results, retrograde/anterograde transport times also did not appear to be limiting for LTF and LTSSF at proximal sensorimotor connections in previous experiments in dissociated cell culture, which are only about 500 μm from the SN soma (Casadio et al., 1999).
It has been suggested that the temporal window for coincident (asymmetric) LTF, involving both synaptic and somatic 5-HT application, precludes retrograde axonal transport as a viable mechanism for some (but not all) models of coincident LTF (Sherff and Carew, 1999). However, this conclusion was based on several assumptions that could substantively influence transport time (Guan, 2004), and mechanisms for coincident LTF appear to be somewhat different from LTF evoked by repeated 5-HT applications (Sherff and Carew, 2002, 2004). Hence our present results are not necessarily in conflict with this previous work.
Axonal transport time and blockade of somatic protein synthesis
The duration of exposure to central anisomycin made no apparent difference in the ability to block LTF induced by synaptic application of 5-HT. Preparations that received peripheral 5-HT in conjunction with exposure to central anisomycin for only 3 h at the time of 5-HT exposure exhibited LTF blockade that was nearly identical to LTF blockade in preparations that received central anisomycin for the full 24-h period.
The 3-h anisomycin treatment provided a control condition that was well matched to our other anisomycin treatments, thus ruling out nonspecific degradation from protein synthesis inhibition that might have resulted from longer, 24-h anisomycin exposure (though empirically this did not appear to occur). However, it is perhaps somewhat surprising that the 3-h anisomycin treatment of the soma was sufficient to block LTF. In dissociated cell culture, LTF requires protein synthesis only briefly, during and immediately after 5-HT application (Montarolo et al., 1986). Further, in culture, inhibition of protein synthesis by anisomycin is transient: after anisomycin is removed, protein synthesis returns to normal or above-normal levels within 2 h (Montarolo et al., 1986). However, if in our experiments LTF triggered by synaptic application of 5-HT to distal terminals requires retrograde transport to the soma and subsequent somatic protein synthesis, then one might have expected that protein synthesis in the soma could have returned to normal by the time the retrograde signal arrived, and hence LTF blockade might not have occurred.
There are at least two possible explanations for why even relatively brief 3-h central anisomycin blocked LTF. First, perhaps recovery of protein synthesis in the abdominal ganglion after anisomycin washout is not as rapid as in dissociated cell culture. Potentially, the greater neuropil and connective tissue that are present in the intact abdominal ganglion could have served as a trap for anisomycin and slowed its removal, thus prolonging inhibition of protein synthesis.
The second reason involves the time for the retrograde signal to reach the soma. Given that the actual retrograde message remains unknown, relatively fast signaling pathways, such as calcium waves or a chain of phosphorylation (Sherff and Carew, 1999), might provide a means for 5-HT-treated synapses to communicate with the SN soma, other than the active retrograde transport of molecules that was previously suggested. Alternatively, given the experimental distances involved and potential axonal transport rates, even molecules carried by conventional retrograde axonal transport may have been able to reach the soma within a sufficiently brief time, as described above. Molecules arriving 4 h after start of 5-HT application could perhaps reach the soma only 2 h after start of anisomycin washout, at which time protein synthesis may not have fully recovered.
Potential roles for retrograde and anterograde axonal transport
Nocodazole disrupts microtubules and blocks both anterograde and retrograde transport (Dinter and Berger, 1998). Hence, although our experiments indicate that active axonal transport is essential for LTF, they do not determine whether anterograde transport alone, retrograde transport alone, or both, are critically involved. Our experiments also do not by themselves discriminate between transport of newly synthesized macromolecules and those constitutively expressed.
Synaptic application of 5-HT induces local protein synthesis that is essential for LTF (Martin et al., 1997a). One potential role for retrograde transport is to carry these newly synthesized proteins or other molecules from synapse to soma, where they would in turn modulate transcription and/or translation and induce or upregulate somatic protein synthesis. Alternatively, retrograde messengers could interact with somatic constitutive protein synthesis by placing an inhibitory constraint on it, or by relieving its inhibitory constraint on other machinery so that LTF is expressed. An essential role for retrograde transport in LTF would thus be consistent with synaptic protein synthesis. Certain other modes of synapse-soma signaling, such as calcium waves or phosphorylation cascades (Sherff and Carew, 1999), could not themselves directly carry newly synthesized synaptic proteins from the synapse to the soma. However, it is possible that synaptic protein synthesis activates other reactions that could be transmitted to the soma via alternative signaling pathways. It is also possible that the newly synthesized local proteins mediate functions other than serving as the retrograde signal.
Anterograde transport is probably the most straightforward mechanism by which the soma might transmit newly synthesized molecules to the synapse, where they could lead to the expression of LTF. Again, in principle these molecules could be either induced by 5-HT or constitutively expressed. Anterograde transport is consistent with an essential role of somatic protein synthesis shown here and elsewhere (Martin et al., 1997a), as well as with the presence of elevated levels of several gene products in LTF expression (Goelet et al., 1986; Castellucci et al., 1988, 1989; Barzilai et al., 1989; Eskin et al., 1989; Bergold et al., 1990; Dash et al., 1990; Bailey and Kandel, 1993; Byrne et al., 1993; Kaang et al., 1993; Alberini et al., 1994; Ghirardi et al., 1995; Alberini, 1999). Evidence for a constitutively expressed factor suppressing facilitation and potentially carried by anterograde transport comes from findings that excising the synaptic processes from the soma before 5-HT treatment leads to an increase in 24-h LTF (Liu et al., 2003). However, this LTF is not as persistent as normal, suggesting that anterograde transport of somatic molecules is necessary for stabilization of long-term changes.
Limitations of current models of synaptic marking and capture
Our findings involving the long soma-synapse distances of the distal SN synapses onto PMNs pose intriguing challenges not yet fully explained by the current models of synaptic marking and capture.
One challenge arises from the proposed time windows for synaptic marking and capture, each of which is on the order of a few hours (Casadio et al., 1999). To examine how long the synaptic marking signal persists in SNs in dissociated cell culture, a single pulse of 5-HT was applied to one branch of the SN at various times prior to application of five pulses of 5-HT to the other branch. The single pulse was able to capture LTF only when it occurred 1–2 h before the five pulses, indicating that the duration of the effectiveness of the mark was relatively short. In complementary experiments, a single pulse to one branch could capture the LTF when it was given 1 h, but not 4 h, after the application of the five pulses to the other branch, indicating that the duration of the capture period was also relatively short. Short time windows in the range of minutes to 3 h or less have also been reported for coincident LTF at Aplysia pleural-pedal SN-MN connections (Sherff and Carew, 1999), and for hippocampal LTP (Frey and Morris, 1997).
Short time windows are problematic for the distal synapses used in the present experiments. Retrograde messengers newly synthesized in the synapse in response to 5-HT would not arrive at the soma for several hours if they are carried by retrograde axonal transport. Similarly, proteins newly synthesized in the soma and carried by anterograde transport would not reach the distal synaptic terminals until after a comparable time period, producing a potential round trip approaching a day. By present synaptic marking/capture models, LTSSF would thus require that a synaptic mark be able to persist and remain effective for almost 24 h (by mechanisms that remain to be determined), and the ability of molecules to be captured would also need to persist for several hours. However, neither of these time frames seems consistent with the short time frames presently proposed for synaptic marking and capture.
If instead, intracellular communication is sufficiently fast to convey proteins between synapse and soma without raising these difficulties, the mechanisms by which this communication would occur are not yet known.
An additional challenge for synaptic marking models of 24-h LTF and LTSSF is that the mechanism for maintaining such a synaptic mark remains unknown. A novel series of studies recently indicated that a neuron-specific isoform of cytoplasmic polyadenylation element binding protein (CPEB) has many of the characteristics appropriate for a self-perpetuating, long-lasting synaptic mark for maintaining late-phase, 72-h LTF ( Si et al., 2003a,b, 2004; Bailey et al., 2004). However, local inhibition of CPEB in neural processes does not affect 24-h LTF. Thus, the 24-h form of LTF and LTSSF investigated in the present experiments would require a different mark. Such a mark could be an anatomical mark rather than, or in addition to, being a molecular mark. SN processes show protein-synthesis-dependent morphological growth that persists for 12 h but not 24 h when transcription is blocked, and can persist for 24 h but not 48 h when the SN soma is removed (Grabham et al., 2005). Local synaptic growth that is stabilized by interactions with gene products shipped from the soma could in principle contribute to synaptic marking.
ApTrkI, a trk-like receptor that contributes to LTF and responds to both 5-HT and sensorin, a neuropeptide released by 5-HT (Hu et al., 2004), has properties that imply it may function in the induction of a synaptic tag; further, translocation of activated trk receptors from synapses to the soma could be a critical step that depends on axonal transport (Ormond et al., 2004). Again, however, it is not known whether it could mediate synaptic marking and capture for the time frames involved in the present study.
Among additional alternative possibilities are the following. (1) Different time frames for synaptic marking and capture could exist for synapses at different distances from the soma, or at synapses onto different MNs (the present studies used siphon motor neurons, whereas many culture experiments use the gill motor neuron L7); however, this seems inelegant and would be difficult to coordinate biologically. (2) Different mechanisms could mediate LTF at mature, intact synapses, compared with mechanisms at newly formed synapses in dissociated cell culture. (3) Synaptic marking/capture in circumstances involving two different synaptic sites (Martin et al., 1997a) may be fundamentally different from synaptic capture/marking in circumstances involving round-trip signaling between the synapse and soma. (4) Time-window requirements could vary parametrically according to stimulus conditions, so that marking or capture with five pulses of 5-HT (rather than one pulse) persists considerably longer.
In any event, the results reported here suggest that our understanding of the relationships between synaptic marking, capture, and LTF remains incomplete.
Conclusions
We found that inhibition of protein synthesis at either the soma or synaptic regions blocks LTF and LTSSF at distal sensorimotor terminals, suggesting important communication between soma and synapse in establishment of LTF. Consistent with this idea, blockade of axonal transport also blocked LTF. In principle, the essential somatic or synaptic proteins could be either induced or constitutively expressed. One possible explanation is that here, as elsewhere (Martin et al., 1997a; Casadio et al., 1999), an immediate translational response occurs at synapses treated with 5-HT. The newly synthesized synaptic proteins might then serve as a retrograde messenger, as a synaptic mark to capture or utilize somatically generated molecules in LTF, or as other direct or indirect factors in the LTF signaling pathways (Fig. 4). Similar phenomena have been observed, and their underlying mechanisms analyzed, at more proximal synapses in dissociated cell culture (Martin et al., 1997a; Casadio et al., 1999). However, the long distance between soma and synapse in the present preparation imposes temporal and mechanistic constraints on the intracellular signals by which synapse and soma communicate. A particular challenge is maintaining a synaptic mark long enough for communication to occur between a distal synapse and the soma and back again, given the short time window proposed to be involved for synaptic marking and capture. Possibly, longer-lasting synaptic marks, longer time windows, or unknown, faster transport mechanisms could help solve this conundrum. Alternatively, perhaps the mechanisms involved in LTF at mature distal synapses are different from those at newly developed synapses in dissociated cell culture. In any case, a complete understanding of LTF and LTSSF at distal synapses appears to require further information about mechanisms of synaptic marking and capture and their relationship to LTF.
Acknowledgments
We thank Robert D. Hawkins, Mary T. Lucero, Edwin M. Maynard, Richard A. Normann, and Doju Yoshikami for helpful comments on an earlier version of this manuscript. Supported by NSF IBN-9796235 and the University of Utah.
Footnotes
Received 8 November 2005; accepted 4 April 2006.
Literature Cited
Alberini, C. M. 1999. Genes to remember. J. Exp. Biol.202:2887–2891.
Alberini, C. M., M. Ghirardi, R. Metz, and E. R. Kandel. 1994. C/EBP is an immediate-early gene required for the consolidation of long-term facilitation in Aplysia. Cell 76:1099–1114.
Ambron, R. T., R. Schmied, C. C. Huang, and M. Smedman. 1992. A signal sequence mediates the retrograde transport of proteins from the axon periphery to the cell body and then into the nucleus. J. Neurosci.12:2813–2818.
Ambron, R. T., M. F. Dulin, X. P. Zhang, R. Schmied, and E. T. Walters. 1995. Axoplasm enriched in a protein mobilized by nerve injury induces memory-like alterations in Aplysia neurons. J. Neurosci.15:3440–3446.
Azhderian, E. M., D. Hefner, C. H. Lin, L. K. Kaczmarek, and P. Forscher. 1994. Cyclic AMP modulates fast axonal transport in Aplysia bag cell neurons by increasing the probability of single organelle movement. Neuron12:1223–1233.
Bacskai, B. J., B. Hochner, M. Mahaut-Smith, S. R. Adams, B. K. Kaang, E. R. Kandel, and R. Y. Tsien. 1993. Spatially resolved dynamics of cAMP and protein kinase A subunits in Aplysia sensory neurons. Science260:222–226.
Bailey, C. H., and E. R. Kandel. 1993. Structural changes accompanying memory storage. Annu. Rev. Physiol.55:397–426.
Bailey, C. H., E. R. Kandel, and K. Si. 2004. The persistence of long-term memory: a molecular approach to self-sustaining changes in learning-induced synaptic growth. Neuron44:49–57.
Barzilai, A., T. E. Kennedy, J. D. Sweatt, and E. R. Kandel. 1989. 5-HT modulates protein synthesis and the expression of specific proteins during long-term facilitation in Aplysia sensory neurons. Neuron2:1577–1586.
Bergold, P. J., J. D. Sweatt, I. Winicov, K. R. Weiss, E. R. Kandel, and J. H. Schwartz. 1990. Protein synthesis during acquisition of long-term facilitation is needed for the persistent loss of regulatory subunits of the Aplysia cAMP-dependent protein kinase. Proc. Natl. Acad. Sci. USA87:3788–3791.
Byrne, J. H., R. Zwartjes, R. Homayouni, S. D. Critz, and A. Eskin. 1993. Roles of second messenger pathways in neuronal plasticity and in learning and memory: insights gained from Aplysia. Adv. Second Messenger Phosphoprot. Res. 27:47–108.
Casadio, A., K. C. Martin, M. Giustetto, H. Zhu, M. Chen, D. Bartsch, C. H. Bailey, and E. R. Kandel. 1999. A transient, neuron-wide form of CREB-mediated long-term facilitation can be stabilized at specific synapses by local protein synthesis. Cell99:221–237.
Castellucci, V. F., T. E. Kennedy, E. R. Kandel, and P. Goelet. 1988. A quantitative analysis of 2-D gels identifies proteins in which labeling is increased following long-term sensitization in Aplysia. Neuron 1:321–328.
Castellucci, V. F., H. Blumenfeld, P. Goelet, and E. R. Kandel. 1989. Inhibitor of protein synthesis blocks long-term behavioral sensitization in the isolated gill-withdrawal reflex of Aplysia. J. Neurobiol. 20:1–9.
Clark, G. A., and E. R. Kandel. 1984. Branch-specific heterosynaptic facilitation in Aplysia siphon sensory cells. Proc. Natl. Acad. Sci. USA205:417–419.
Clark, G. A., and E. R. Kandel. 1993. Induction of long-term facilitation in Aplysia sensory neurons by local application of serotonin to remote synapses. Proc. Natl. Acad. Sci. USA90:11411–11415.
Dash, P. K., B. Hochner, and E. R. Kandel. 1990. Injection of the cAMP-responsive element into the nucleus of Aplysia sensory neurons blocks long-term facilitation. Nature345:718–721.
Dinter, A., and E. G. Berger. 1998. Golgi-disturbing agents. Histochem. Cell Biol.109:571–590.
Emptage, N. J., and T. J. Carew. 1993. Long-term synaptic facilitation in the absence of short-term facilitation in Aplysia neurons. Science262:253–256.
Eskin, A., K. S. Garcia, and J. H. Byrne. 1989. Information storage in the nervous system of Aplysia: specific proteins affected by serotonin and cAMP. Proc. Natl. Acad. Sci. USA86:2458–2462.
Frey, U., and R. G. Morris. 1997. Synaptic tagging and long-term potentiation. Nature385:533–536.
Frey, U., and R. G. Morris. 1998. Synaptic tagging: implications for late maintenance of hippocampal long-term potentiation. Trends Neurosci.21:181–188.
Frey, U., M. Krug, K. G. Reymann, and H. Matthies. 1988. Anisomycin, an inhibitor of protein synthesis, blocks late phases of LTP phenomena in the hippocampal CA1 region in vitro. Brain. Res. 452:57–65.
Ghirardi, M., P. G. Montarolo, and E. R. Kandel. 1995. A novel intermediate stage in the transition between short- and long-term facilitation in the sensory to motor neuron synapse of Aplysia. Neuron 14:413–420.
Giustetto, M., A. N. Hegde, K. Si, A. Casadio, K. Inokuchi, W. Pei, E. R. Kandel, and J. H. Schwartz, 2003. Axonal transport of eukaryotic translation elongation factor 1 alpha mRNA couples transcription in the nucleus to long-term facilitation at the synapse. Proc. Natl. Acad. Sci. USA100:13680–13685.
Goelet, P., V. F. Castellucci, S. Schacher, and E. R. Kandel. 1986. The long and the short of long-term memory—a molecular framework. Nature322:419–422.
Goldberg, D. J., and J. H. Schwartz. 1980. Fast axonal transport of foreign transmitters in an identified serotonergic neurone of Aplysia californica. J. Physiol. 307:259–272.
Goldberg, D. J., J. H. Schwartz, and A. A. Sherbany. 1978. Kinetic properties of normal and perturbed axonal transport of serotonin in a single identified axon. J. Physiol.281:559–579.
Gooch, C. M. 1997. Synapse-specific heterosynaptic long-term facilitation at Aplysia sensorimotor connections. Ph.D. dissertation, Neuroscience Program, Princeton University.
Gooch, C. M., and G. A. Clark. 1997. Synapse-specific long-term facilitation at Aplysia sensorimotor connections. Soc. Neurosci. Abstr.23:1415.
Grabham, P. W., F. Wu, S. Schacher, and D. J. Goldberg. 2005. Initiating morphological changes associated with long-term facilitation in Aplysia is independent of transcription or translation in the cell body. J. Neurobiol.64:202–212.
Guan, X. 2004. Essential role of somatic and synaptic protein synthesis and axonal transport in long-term facilitation at distal sensorimotor connections in Aplysia californica. Ph.D. dissertation, Department of Bioengineering, University of Utah.
Guan, X., and G. A. Clark. 2001. Role of protein synthesis in long-term facilitation at distal sensorimotor connections in Aplysia. Abstract 644.8 [Online: Abstract Viewer/Itinerary Planner]. Society for Neuroscience, Washington, DC.
Guan, X., and G. A. Clark. 2003. Essential role of axonal transport at distal sensorimotor connections in Aplysia. Abstract 291.220 [Online: Abstract Viewer/Itinerary Planner]. Society for Neuroscience, Washington, DC.
Gunstream, J. D., G. A. Castro, and E. T. Walters. 1995. Retrograde transport of plasticity signals in Aplysia sensory neurons following axonal injury. J. Neurosci.15:439–448.
Heslop, J. P., and E. A. Howes. 1972. Temperature and inhibitor effects on fast axonal transport in a molluscan nerve. J. Neurochem.19:1709–1716.
Hu, J. Y., L. Glickman, F. Wu, and S. Schacher. 2004. Serotonin regulates the secretion and autocrine action of a neuropeptide to activate MAPK required for long-term facilitation in Aplysia. Neuron 43:373–385.
Kaang, B. K., E. R. Kandel, and S. G. Grant. 1993. Activation of cAMP-responsive genes by stimuli that produce long-term facilitation in Aplysia sensory neurons. Neuron10:427–435.
Kandel, E. R. 2001. The molecular biology of memory storage: a dialogue between genes and synapses. Science294:1030–1038.
Liu, K., J. Y. Hu, D. Wang, and S. Schacher. 2003. Protein synthesis at synapse versus cell body: enhanced but transient expression of long-term facilitation at isolated synapses. J. Neurobiol.56:275–286.
Martin, K. C., A. Casadio, H. Zhu, E. Yaping, J. C. Rose, M. Chen, C. H. Bailey, and E. R. Kandel. 1997a. Synapse-specific, long-term facilitation of Aplysia sensory to motor synapses: a function for local protein synthesis in memory storage. Cell91:927–938.
Martin, K. C., D. Michael, J. C. Rose, M. Barad, A. Casadio, H. Zhu, and E. R. Kandel. 1997b. MAP kinase translocates into the nucleus of the presynaptic cell and is required for long-term facilitation in Aplysia. Neuron 18:899–912.
Montarolo, P. G., P. Goelet, V. F. Castellucci, J. Morgan, E. R. Kandel, and S. Schacher. 1986. A critical period for macromolecular synthesis in long-term heterosynaptic facilitation in Aplysia. Science 234:1249–1254.
Ormond, J., J. Hislop, Y. Zhao, N. Webb, F. Vaillaincourt, J. R. Dyer, G. Ferraro, P. Barker, K. C. Martin, and W. S. Sossin. 2004. ApTrkl, a Trk-like receptor, mediates serotonin-dependent ERK activation and long-term facilitation in Aplysia sensory neurons. Neuron44:715–728.
Schwartz, J. H. 1979. Axonal transport: components, mechanisms, and specificity. Annu. Rev. Neurosci.2:467–504.
Sherff, C. M., and T. J. Carew. 1999. Coincident induction of long-term facilitation in Aplysia: cooperativity between cell bodies and remote synapses. Science285:1911–1914.
Sherff, C. M., and T. J. Carew. 2002. Coincident induction of long-term facilitation at sensory-motor synapses in Aplysia: presynaptic and postsynaptic factors. Neurobiol. Learn. Mem.78:498–507.
Sherff, C. M., and T. J. Carew. 2004. Parallel somatic and synaptic processing in the induction of intermediate-term and long-term synaptic facilitation in Aplysia. Proc. Natl. Acad. Sci. USA 101:7463–7468.
Si, K., M. Giustetto, A. Etkin, R. Hsu, A. M. Janisiewicz, M. C. Miniaci, J. H. Kim, H. Zhu, and E. R. Kandel. 2003a. A neuronal isoform of CPEB regulates local protein synthesis and stabilizes synapse-specific long-term facilitation in Aplysia. Cell 115:893–904.
Si, K., S. Lindquist, and E. R. Kandel. 2003b. A neuronal isoform of the Aplysia CPEB has prion-like properties. Cell115:879–891.
Si, K., S. Lindquist, and E. Kandel. 2004. A possible epigenetic mechanism for the persistence of memory. Cold Spring Harbor Symp. Quant. Biol.69:497–498.
Steward, O., and E. M. Schuman. 2003. Compartmentalized synthesis and degradation of proteins in neurons. Neuron40:347–359.
Thoenen, R. T., U. Otten, and M. Schwab. 1979. Orthograde and retrograde signals for the regulation of neuronal gene expression: the peripheral sympathetic nervous system as a model. Pp. 911–928 in The Neurosciences: Fourth Study Program, F. O Schmitt and F. G. Worden, eds. MIT Press, Cambridge, MA.
Thompson, K. R., K. O. Otis, D. Y. Chen, Y. Zhao, T. J. O’Dell, and K. C. Martin. 2004. Synapse to nucleus signaling during long-term synaptic plasticity: a role for the classical active nuclear import pathway. Neuron44:997–1009.
Trudeau, L. E., and V. F. Castellucci. 1995. Postsynaptic modifications in long-term facilitation in Aplysia: upregulation of excitatory amino acid receptors. J. Neurosci.15:1275–1284.
Turner, J. R., and A. M. Tartakoff. 1989. The response of the Golgi complex to microtubule alterations: the roles of metabolic energy and membrane traffic in Golgi complex organization. J. Cell Biol.109:2081–2088.
Walters, E. T., and R. T. Ambron. 1995. Long-term alterations induced by injury and by 5-HT in Aplysia sensory neurons: convergent pathways and common signals? Trends Neurosci.18:137–142.
Walters, E. T., H. Alizadeh, and G. A. Castro. 1991. Similar neuronal alterations induced by axonal injury and learning in Aplysia. Science 253:797–799.
Weragoda, R. M., E. Ferrer, and E. T. Walters. 2004. Memory-like alterations in Aplysia axons after nerve injury or localized depolarization. J. Neurosci.24:10393–10401.(Xin Guan and Gregory A. C)
Abstract
To investigate further the cellular mechanisms underlying long-term facilitation (LTF) and long-term synapse-specific facilitation (LTSSF), we studied the role of axonal transport and somatic and synaptic protein synthesis at proximal and distal synapses of Aplysia siphon sensory neurons (SNs). The long soma-synapse distances (2.5 to 3 cm) of the SN distal synapses impose important temporal and mechanistic constraints on long-term facilitation and on intracellular signaling. Excitatory postsynaptic potentials (EPSPs) evoked by SNs in central and peripheral siphon motor neurons were used to assay LTF 24–30 h after various pharmacological treatments. Inhibition of protein synthesis via anisomysin application at either the SN soma or distal synapses blocked the induction of LTF and LTSSF normally produced by synaptic application of the facilitating transmitter serotonin (5-hydroxytryptamine). Further, disruption of axonal transport by application of nocodazole to the isolated siphon nerve completely blocked LTF at distal synapses. These results indicate an essential role for somatic and synaptic protein synthesis and active axonal transport in LTSSF at distal synapses, and raise intriguing questions for current synaptic marking/capture models of synapse specificity and LTF.
Abbreviations: ANOVA, analysis of variance ? ASW, artificial seawater ? C-Aniso, central anisomycin ? CMN, central motor neuron ? EPSP, excitatory postsynaptic potential ? 5-HT, 5-hydroxytryptamine, serotonin ? LTF, long-term facilitation ? LTP, long-term potentiation ? LTSSF, long-term synapse-specific facilitation ? MN, motor neuron ? P-Aniso, peripheral anisomycin ? PMN, peripheral motor neuron ? SN, sensory neuron
Introduction
Despite its dependence on genomic regulation and protein synthesis (Montarolo et al., 1986; Kandel, 2001), long-term synaptic facilitation (LTF) at Aplysia sensorimotor connections can be synapse-specific and can occur preferentially at selected synapses of a given siphon sensory neuron (SN), relative to other connections of the same cell (Clark and Kandel, 1993; Gooch and Clark, 1997; Martin et al., 1997a). Further, both short-term and long-term synapse-specific facilitation (LTSSF) can be triggered by applications of the facilitating transmitter serotonin (5-hydroxytryptamine; 5-HT) that are restricted to sensorimotor connections and that do not directly contact the soma, suggesting that communication between the soma and synapse is involved in LTSSF (Clark and Kandel, 1993; Emptage and Carew, 1993; Gooch and Clark, 1997; Martin et al., 1997a; Casadio et al., 1999). LTF at siphon SN connections contributes importantly to long-term behavioral sensitization of the gill- and siphon-withdrawal reflex, which, like LTF, is dependent on protein synthesis (Castellucci et al., 1989).
Because individual neurons can make connections with hundreds or thousands of postsynaptic target cells, synapse-specific facilitation provides a high degree of precision in the modification of neural pathways. It also raises intriguing questions about the mechanisms for LTF, which depends on changes in gene expression and on new protein synthesis.
The first direct demonstration of synapse-specific facilitation in Aplysia examined the connections of individual siphon SNs onto two physically separate sets of postsynaptic targets: at proximal SN synapses onto nearby central motor neurons (CMNs) in the intact abdominal ganglion, and at SN synapses onto peripheral motor neurons (PMNs) at the distal end of the attached siphon nerve, several centimeters away (Clark and Kandel, 1993; Gooch and Clark, 1997). As with short-term facilitation in this system (Clark and Kandel, 1984), application of 5-HT that was restricted to distal synaptic regions (and did not directly involve the soma) was sufficient to induce LTF. Moreover, LTF occurred preferentially at synaptic connections previously treated with 5-HT, compared with untreated connections of the same SNs (Clark and Kandel, 1993; Gooch and Clark, 1997), indicating that LTF, like short-term facilitation, was synapse-specific. LTF can also be induced by restricted 5-HT applications at pleural SN connections onto motor neurons (MNs) in the pedal ganglion (Emptage and Carew, 1993).
In a closely related, elegant series of studies, short-term and long-term synapse-specific facilitation has also been demonstrated and analyzed in vitro (Martin et al., 1997a; Casadio et al., 1999), using a three-cell culture system in which an Aplysia siphon SN bifurcated to synapse onto two nearby MNs. LTF is also synapse-specific in this system, and it requires both local and somatic protein synthesis in the presynaptic neuron. Further results suggest that synapse specificity may arise from synaptic "marking," or "tagging," wherein new gene products are shipped from the soma to all synapses, but are captured or preferentially utilized at synapses marked, or tagged, by previous 5-HT treatment. According to some synaptic marking/capture models (Martin et al., 1997a; Casadio et al., 1999), synaptic application of 5-HT triggers a retrograde intracellular message from synapse to soma that activates transcription and translation. New gene products are then synthesized in the soma and shipped to all synapses, but these products are captured or utilized only at synapses previously marked by molecular events arising from being treated with 5-HT. Hence, LTF is expressed only (or preferentially) at these marked synapses. Although the capture of 24-h LTF is independent of protein synthesis, the initiation of LTF by synaptic application of 5-HT is dependent on protein synthesis, suggesting that the newly synthesized local proteins may serve as, or help generate, the presumed retrograde signal to the nucleus.
An important aspect of synaptic marking/capture models is that both the synaptic mark and the new gene products shipped from soma to synapse must persist sufficiently long for the new gene products to arrive and be captured. However, the reported time windows for both synaptic marking and capture in LTSSF in dissociated cell culture are relatively short, no more than a few hours (Casadio et al., 1999). Similarly, at pleural sensorimotor connections, the time window for coincident, or asymmetric, LTF (arising from combined somatic and synaptic applications of low concentrations of 5-HT, each of which by itself is insufficient to produce LTF) appears to be very short, only about 15 min (Sherff and Carew, 1999). Local protein synthesis and synaptic marking also contribute to hippocampal synapse-specific long-term potentiation (LTP) (Frey et al., 1988; Frey and Morris, 1997, 1998; Steward and Schuman, 2003), which also exhibits a brief time window—less than 3 h—for synaptic marking/capture (Frey and Morris, 1997).
Compared with the situation in dissociated cell culture or pleural sensorimotor connections, the long soma-synapse distance (2.5 to 3 cm) at siphon SN synaptic connections onto PMNs imposes important temporal and mechanistic constraints on the cellular and molecular processes underlying soma-synapse signaling and LTF, thereby raising several key questions. In particular, although a short time window may suffice for LTF at synapses close to the SN soma, how would synaptic marking and capture work at truly distal synapses? The soma-synapse active axonal transport times for macromolecules—perhaps 13–17 h (see Discussion; Goldberg et al., 1978; Schwartz, 1979; Goldberg and Schwartz, 1980; Gunstream et al., 1995)—would presumably be outside the short time windows for synaptic marking and capture observed for proximal synapses. Can a synaptic mark at distal synapses persist long enough to allow retrograde signaling from the synapse to the soma, followed by anterograde signaling back to the synapse? If so, such a long-lasting synaptic mark would require maintenance mechanisms that remain to be elucidated. Further, how are the signals communicated between the soma and distal synapses? If intracellular communication is appreciably faster than conventional axonal transport, then the mechanisms by which this communication could occur are not yet known. Alternatively, are rather different mechanisms involved in mediating LTF at distal synapses in the adult animal, compared with newly developed synapses in dissociated cell culture?
To begin to address these questions, we asked whether LTF and LTSSF produced by restricted 5-HT applications at distal SN connections onto PMNs depend on somatic or local protein synthesis, and on active axonal transport between soma and synapse. We found that blockade of protein synthesis at either the SN soma or the distal synaptic region blocked LTF and LTSSF, indicating that both local and somatic protein synthesis are essential even at distal synapses. Blockade of axonal transport also blocked LTF at distal synapses, implicating active axonal transport and soma-synapse communication. Preliminary reports of some of these results have previously appeared (Guan and Clark, 2001, 2003).
Materials and Methods
Preparation
Wild-caught adult Aplysia californica weighing 100–150 g were supplied by Marinus, Long Beach, California. The slugs were housed in an aquarium tank containing artificial seawater (Tropic Marine) at a temperature of 14 °C. They were fed with romaine lettuce daily to maintain a constant weight.
Prior to dissection, the animal was anesthetized by cooling at 0 °C for 20 min. It was then further anesthetized by injection of cooled (4 °C) isotonic MgCl2 (100 ml MgCl2/100 g animal weight) into the hemocoel. After the skin of the animal was completely relaxed, the animal was pinned to a dissection pan with the ventral side facing up, and was cut open from the caudal to rostral end. The abdominal ganglion together with the attached siphon nerve and the pleuroabdominal connectives were dissected out in a cooled "zero-calcium" solution (NaCl, 338 mM; KCI, 10 mM; MgCl2, 150 mM; 4 °C). The zero-calcium solution (containing no added calcium) was used to minimize the hyperexcitability and synaptic facilitation potentially induced by axon injury during dissection (Ambron et al., 1992). The siphon nerve was kept as long as possible to obtain more PMNs. The connectives and the sheath of the abdominal ganglion were fixed by 25% glutaraldehyde for 30 s.
The excised abdominal ganglia and attached siphon nerve were then pinned onto the Sylgard floor of a polystyrene recording dish in cooled zero-calcium solution. A cold plate was placed underneath the dish to minimize dissection-induced injury. For studies of protein synthesis inhibition, the recording dish was divided into two small chambers, in which the SNs in the abdominal ganglion with their CMN targets were in one chamber and were separated from the SN distal synaptic terminals and PMN postsynaptic targets in the other chamber (see Fig. 1). The siphon nerve was threaded through the slit in the dividing wall, which was then filled with petroleum jelly to ensure that the two chambers were isolated from one another. To test the isolation, a drop of fast green dye was applied to one chamber. If the dye did not pass to the other chamber, the isolation was considered to be complete. This experimental setup separated the SN soma and central synapse from the distal peripheral synapse, thus allowing the perfusion of drugs independently to one site or the other. Similarly, studies of axonal transport inhibition used a three-chamber recording dish, which allowed application of nocodazole to the isolated siphon nerve.
The connective tissue covering the abdominal ganglion was partially removed to allow the penetration of SN and CMN. The distal end of the siphon nerve was treated with 0.1% bovine trypsin (Sigma, type IX) for 5 min at room temperature (22 °C) to facilitate the subsequent penetration of intracellular electrodes into PMNs. The trypsin action was terminated by incubating the treated nerve with 0.25% trypsin inhibitor solution (Sigma, type II-S: soybean) for 15 min.
To allow the dissected neurons to recover from potential injury incurred during dissection, the cooled zero-calcium dissecting solution was replaced by enriched artificial seawater (ASW), and the preparation was allowed to rest in enriched ASW for 5 h before the beginning of electrophysiological testing. Enriched ASW had the components of NaCl, 460 mM; KCl, 10 mM; MgCl2, 55 mM; CaCl2, 11 mM; Hepes-NaOH, 10 mM, along with nutrient, amino acid, vitamin, and sugar supplements (Gibco), and penicillin/streptomycin (Sigma, 100,000 units penicillin and 100 mg streptomycin per liter ASW) to prevent infection (Clark and Kandel, 1993).
Intracellular recording and stimulation
Using standard stimulating and recording techniques, we looked for monosynaptic connections between a given LE siphon SN and an LFs siphon CMN in the abdominal ganglion, a siphon PMN at the distal end of the siphon nerve, or both. A "high Mg2+ and Ca2+" solution was used as the recording medium to hinder the firing of interneurons that might evoke polysynaptic EPSPs in MNs (NaCl, 338 mM; KCI, 10 mM; MgCl2, 100 mM; CaCl2, 50 mM; Hepes-NaOH, 10 mM; also supplemented with vitamins and sugars, amino acids, and penicillin/streptomycin solutions). Typically we recorded from one CMN and one siphon PMN, and then tested about 20 LE siphon SNs per preparation for monosynaptic connections. Neurons were identified on the basis of their morphological and electrophysiological characteristics. Although PMNs are distributed along the distal end of the siphon nerve and are found up to 5 cm from the abdominal ganglion, we typically recorded from PMNs that were relatively close to the SN somata (2.5–3 cm away), just past the first major branching of the siphon nerve, to maximize the probability of finding connections with SNs.
Neurons were sampled by penetration with a single-barreled glass microelectrode (15–25 M) filled with electrolyte (2.5 M KCl). During penetration, hyperpolarizing current was passed through the electrode to help determine the point at which the electrode tip came into physical contact with the cell membrane, and to prevent potential firing of SNs during penetration (and so ensure subsequent accurate sampling of the first synaptic potential evoked in the MN). During testing, MNs were hyperpolarized by 40 mV below resting membrane potential to prevent EPSPs from evoking action potentials. Each SN-synapse was tested a single time during a recording session by passing intracellular current into the SN to evoke an action potential and possible resultant EPSP in the PMN and CMN targets.
After testing was completed on day 1, one of a series of drug treatments (detailed below) was conducted. In order to help find the same neurons on day 2 as on day 1, the cells surrounding the recorded neurons on day 1 were pulse-injected with fast green dye to label their positions, and a diagram was drawn (Clark and Kandel, 1993). Recording sessions and drug treatments were conducted at room temperature (22 °C), which was kept the same on day 1 and day 2. After drug treatments and washout were completed, the nervous system was incubated for 24 h (anisomycin studies) or 30 h (nocodazole studies) in enriched ASW at 18 °C.
The same synaptic connections were retested in "high Mg2+ and Ca2+" solution the following day, day 2. To avoid potential "floor" and "ceiling" effects, only synapses with EPSPs between 2 and 25 mV on day 1 were tested again on day 2. To maximize experimental yield, we retested SNs that had connections onto either a CMN alone, a PMN alone, or both a CMN and PMN.
Pharmacological treatments
We conducted separate studies to examine the effects of protein synthesis inhibition and of axonal transport inhibition.
In the experiments on the inhibition of protein synthesis, after physiological recording was completed on day 1, preparations were exposed to one of six possible drug treatments (see Fig. 1): (1) ASW; (2) C-Aniso alone (anisomycin applied in the central (C) chamber, containing the SN somata); (3) P-Aniso alone (anisomycin applied in the peripheral (P) chamber, containing the SN peripheral synapses); (4) P-5-HT + C-Aniso (peripheral application of 5-HT, combined with central application of anisomycin); (5) P-5-HT + P-Aniso (peripheral application of both 5-HT and anisomycin); or (6) P-5-HT (peripheral application of 5-HT). Aniso (10 μM) was applied 1 h before 5-HT application (5 x 5-min, 50 μM, separated by a 15-min washout period of 10 bath volumes of ASW). For peripheral applications of anisomycin and in the C-Aniso alone condition, anisomycin was washed out after the end of the experiment on day 1. For the P-5-HT + C-Aniso condition, anisomycin remained in the bath until the experiment finished the next day in one subgroup, to allow for potential delay in the critical period for protein synthesis arising from retrograde transport. A separate subgroup received only a 3-h application of central anisomycin, to better match the durations used in the other experimental drug treatments. Results of the two subgroups were nearly identical and hence were combined for the overall analysis of variance (see Results).
In experiments on the inhibition of axonal transport, preparations were exposed to one of three possible drug treatments (see Fig. 3): (1) M-Nocodazole alone (application of the axonal transport inhibitor nocodazole to the middle (M) chamber of a three-chambered dish containing the siphon nerve and SN axons); (2) M-Nocodazole + P-5-HT (application of nocodazole onto the siphon nerve, combined with application of 5-HT at SN peripheral synapses); or (3) P-5-HT alone. The synaptic application of 5-HT was intended to induce LTF; the 5-HT (5 x 5-min, 50 μM, separated by a 15-min washout) was applied into the peripheral chamber only. To examine the role of axonal transport, nocodazole, an axonal transport inhibitor that causes depolymerization of axonal microtubules (Turner and Tartakoff, 1989), was applied only in the middle chamber containing the siphon nerve, including SN axons; hence, nocodazole was not directly applied onto either the SN synapses or SN somata. To avoid disturbing the general metabolism of the neuron, nocodazole concentration was 50 μM. Nocodazole (Sigma) was dissolved in dimethyl sulfoxide (5mg/ml) and culture medium (Gunstream et al., 1995). Because the effect of nocodazole is reversible (Dinter and Berger, 1998), it was applied at least 1 h before 5-HT application and was present until the experiment finished. After completion of the drug treatment and washout following the recording of day 1, the nervous system was incubated for 30 h in enriched ASW at 18 °C. The extension of the incubation time from 24 h (anisomycin experiments) to 30 h (present experiments) was intended to allow potential retrograde and anterograde messengers sufficient time to travel between synapse and soma (Schwartz, 1979), given that 5-HT was applied at a slightly more distal site than in our anisomycin experiments.
Data analyses
EPSPs triggered by SN action potentials were recorded into a computer by a CED MICRO 1401 AD interface card with SPIKE 2 software, ver. 3 (Cambridge Electronic Design). An "LTF score" reflecting the synaptic change from day 1 to day 2 was calculated as follows. First, to determine LTF for a particular synapse, the EPSP amplitude on day 2 at one synapse was divided by the EPSP amplitude on day 1 at the same synapse; a log10 transform was then applied so that potential increases and decreases in synaptic strength would contribute equally. To determine the mean LTF for a particular preparation, the mean of the log10 transform of facilitation across different central (or peripheral) synapses within one preparation was calculated, yielding the LTF score for central (or peripheral) synapses for that individual preparation. Finally, to determine the mean LTF for a particular experimental group receiving a particular drug treatment, the mean LTF across preparations was calculated independently for central and peripheral synapses.
LTF scores were screened for outliers, and any LTF score that was 1.5 interquartile ranges below the first quartile or above the third quartile was removed. This procedure was implemented in part to minimize potential spurious results that might have occurred from misidentifying a neuron between recording sessions on days 1 and 2. Statistical analyses were performed using SPSS statistical software, ver. 11.5.
For studies examining the effects of protein synthesis inhibition, a two-factor analysis of variance (ANOVA) was conducted, in which one factor had two levels to represent changes in synaptic strength at either central or peripheral synapses ("synaptic site" or simply "site"), and the other factor had six levels to represent the six drug treatments ("drug treatment" or simply "drug"). The overall ANOVA was followed by post hoc Tukey HSD tests for individual comparisons among the 12 possible conditions (2 synaptic sites x 6 drug treatments). Individual student’s t tests were used to examine whether LTF scores demonstrated significant changes in EPSP amplitude on day 2 relative to day 1 in each of the 12 treatment conditions. As a more rigorous test of LTSSF in the experiments on protein synthesis inhibition, we separately analyzed LTF at the subset of SNs that had connections onto both CMNs and PMNs.
Comparable analyses were conducted to examine the effects of axonal transport inhibition, using a two-factor ANOVA to compare LTF at central and peripheral synapses for the three experimental treatments, followed by post hoc Tukey tests.
Results
Results indicate an essential role for both somatic and local protein synthesis in LTSSF at distal synapses, and further indicate that active axonal transport between soma and the distal synapse is required.
Inhibition of either somatic or synaptic protein synthesis blocks 5-HT-induced LTF at distal synapses
In dissociated cell culture, LTF and LTSSF at Aplysia sensorimotor connections require protein synthesis in both somatic and synaptic compartments (Montarolo et al., 1986; Martin et al., 1997a; Casadio et al., 1999). However, the close proximity of the siphon sensory cell soma and synapse in those experiments leaves unanswered the question of whether similar requirements exist for synapses that are remote from the soma (e.g., 2.5 to 3 cm, for siphon SN connections onto peripheral siphon MNs), or whether instead such long distances pose a communication challenge that changes the roles of protein synthesis.
To address this question, we applied the protein synthesis inhibitor anisomycin either centrally at the SN cell body and central synapses in the abdominal ganglion, or distally at SN peripheral synapses. We found that blockade of protein synthesis at either the sensory neuron soma or distal synaptic sites completely blocked the LTF and LTSSF that are normally produced by synaptic application of 5-HT. An analysis of variance was conducted on changes in synaptic strength at central and peripheral synaptic connections across each of the six drug treatment conditions in all preparations. Consistent with a role for protein synthesis, LTF varied significantly as a function of drug treatment condition (F(5, 109) = 12.06, P < 0.001), as did differences in LTF at peripheral compared with central synapses (significant drug-by-site interaction, F(5, 109) = 5.83, P < 0.001). The main effect of synaptic site (LTF at peripheral synapses compared with LTF at central synapses, across all six drug treatment groups) approached but did not quite reach significance (F(5, 109) = 3.89, P = 0.051), most likely because, in keeping with experimental hypotheses, only one of the six drug treatments (Peripheral 5-HT alone) showed peripheral-central LTF differences. Subsequent analyses support four main conclusions (Figs. 1 and2).
Synaptic application of 5-HT produces LTF.
First, confirming previous results (Clark and Kandel, 1993; Emptage and Carew, 1993; Gooch, 1997; Gooch and Clark, 1997; Martin et al., 1997a; Casadio et al., 1999; Liu et al., 2003; Grabham et al., 2005), repeated application of 5-HT alone at peripheral synapses, without anisomycin, produced LTF specifically at peripheral synapses, approximately doubling the mean EPSP amplitude from day 1 to day 2 (mean ± SEM LTF score = 0.266 ± 0.027, n = 9 preparations, 11 synapses, t(8) = 9.85, P < 0.001). Consistent with LTF being synapse-specific, there were no significant changes at untreated central synapses (mean LTF score = 0.035 ± 0.031, n = 10 preparations, 18 synapses, t(9) = 1.13, P > 0.25), and facilitation at treated peripheral synapses was greater than at untreated central synapses (P < 0.001, Tukey). These results confirm that synaptic application of 5-HT is sufficient to produce LTF at peripheral synapses, and suggest that this LTF can be synapse-specific.
LTF requires protein synthesis at the soma.
Second, LTF at distal peripheral synapses was completely blocked by the application of anisomycin to the sensory cell soma, as evidenced by the complete lack of facilitation on day 2, relative to day 1 (Peripheral 5-HT + Central Anisomycin, mean LTF score = 0.01 ± 0.01, n = 17 preparations, 32 synapses, t(16) = 1.00, P > 0.95). Further indicating that blockade had occurred, LTF scores at peripheral synapses in these anisomycin-treated preparations (1) were significantly lower than LTF scores at peripheral synapses in the Peripheral 5-HT alone group, without anisomycin (P < 0.001, Tukey); and (2) were nearly identical to LTF scores at central synapses (which had not been treated with 5-HT), as well as to LTF scores at peripheral synapses in either the ASW control group or Central Anisomycin alone control group (all P values > 0.99, Tukey).
The duration of exposure to central anisomycin (3 h or 24 h) made no apparent difference. The blockade of peripheral LTF that occurred in those preparations that received peripheral 5-HT in conjunction with exposure to central anisomycin for only 3 h (mean LTF score = –0.02 ± 0.03, n = 9 preparations, 16 synapses) was nearly identical to blockade in preparations that received central anisomycin for the full 24-h period (mean LTF score = –0.00 ± 0.01, n = 8 preparations, 16 synapses). A separate analysis of variance conducted on changes in synaptic strength at central and peripheral synaptic connections across the two different anisomycin exposure times showed no significant main effect of exposure time (F (1, 29) = 1.66, P > 0.20), synaptic site (F(1, 29) = 0.01, P > 0.85), or time-by-site interaction (F(1, 29) = 0.11, P > 0.70).
Taken together, this second set of results indicates that somatic protein synthesis is essential for LTF at distal synapses. Consequently, these results further suggest that LTF may involve the regulation of genomic events in the soma, as well as anterograde or retrograde messengers between the soma, where protein synthesis was inhibited in these experiments, and the synapse, where LTF was triggered and ultimately expressed.
LTF requires local protein synthesis.
Third, similar to results for anisomycin application to sensory cell somata in these experiments and in previous work (Martin et al., 1997a), LTF at distal peripheral synapses was also completely blocked by the anisomycin application to the sensory cell synapses. Several findings demonstrated this blockade: (1) there was no facilitation on day 2, relative to day 1, for the Peripheral 5-HT + Peripheral Anisomycin group (mean LTF score = 0.01 ± 0.04, n = 8 preparations, 11 synapses, t(7) = 0.40, P > 0.70); (2) facilitation at peripheral synapses in these preparations was significantly lower than in the Peripheral 5-HT alone group, without anisomycin (P < 0.001, Tukey); and (3) LTF scores at peripheral synapses were nearly identical to LTF scores at central synapses (which had not been treated with 5-HT), as well as to LTF scores at peripheral synapses in either the ASW control group or the Peripheral Anisomycin alone control group (all P values > 0.99, Tukey). Taken together, this third set of results indicates that local protein synthesis plays an essential role in LTF at distal synapses.
Blockade of LTF is not due to nonspecific effects.
Finally, synaptic strength was stable from day 1 to day 2 in all three control groups that did not receive 5-HT. Synaptic strength did not change significantly in the ASW control group, thus ruling out potential nonspecific effects of time and of potential cellular responses to the dissection, such as injury-induced synaptic enhancement (Clark and Kandel, 1993) (mean LTF score at central synapses = –0.02 ± 0.03, n = 12 preparations, 23 synapses, t(11) = 0.79, P > 0.40; mean LTF score at peripheral synapses = –0.04 ± 0.04, n = 8 preparations, 9 synapses, t(7) = 0.89, P > 0.35). Similarly, synaptic strength was stable in the presence of central anisomycin alone (mean LTF score at central synapses = –0.03 ± 0.02, n = 11 preparations, 20 synapses, t(10) = 1.12, P > 0.25; mean LTF score at peripheral synapses = –0.02 ± 0.02, n = 8 preparations, 11 synapses, t(7) = 0.69, P > 0.50). Synaptic strength was also stable in the presence of peripheral anisomycin alone (mean LTF score at central synapses = –0.01 ± 0.05, n = 7, t(6) = 0.09, P > 0.90; mean LTF score at peripheral synapses = –0.04 ± 0.02, n = 7, t(6) = 1.93, P > 0.10). This fourth set of results indicates that protein synthesis inhibition per se does not produce nonspecific degradation of synaptic strength.
Overall, these results implicate a critical role for protein synthesis, both at the soma and locally at the synapse, in the induction of LTF at distal synapses.
Inhibition of either somatic or synaptic protein synthesis also blocks synapse-specific LTF
An individual siphon SN may make synapses onto CMNs, PMNs, or both. In the analysis described above, not every SN exhibited a synaptic connection onto both a CMN and a PMN. As a more rigorous examination of whether LTF was synapse-specific, we separately analyzed the subset of individual SNs that had both central and peripheral connections tested on both day 1 and day 2, thus allowing direct comparison of LTF between central and peripheral connections of a given sensory neuron.
Results for this dually connected SN subset were highly comparable with results from the complete data set (Fig. 2B). As with LTF in the complete data set, we found that blockade of protein synthesis at either the sensory neuron soma or distal synaptic sites completely blocked the LTF that is normally produced by synaptic application of 5-HT. A repeated-measures overall analysis of variance was conducted on changes in synaptic strength at central and peripheral synaptic SN-MN connections (within-subjects) across each of the six drug treatment conditions (between-subjects). Consistent with the role of protein synthesis in LTSSF, LTF varied significantly as a function of drug treatment condition (F(5, 25) = 4.27, P < 0.001), as did differences between LTF at peripheral and central synapses of the same individual sensory neurons (significant drug-by-site interaction (F(5, 25) = 2.63, P < 0.05). The main effect of synaptic site (peripheral synapses versus central synapses) collapsed across all six drug treatment conditions was not significant (F(5, 25) = 0.42, P = 0.52), most likely because only one of the six drug treatments (Peripheral 5-HT alone) produced peripheral-central LTF differences, in keeping with experimental hypotheses.
In the Peripheral 5-HT alone condition, strong LTF occurred at peripheral synapses that had been previously treated with 5-HT, as evidenced by an approximate doubling of the mean EPSP amplitude from day 1 to day 2 (mean ± SEM LTF score = 0.242 ± 0.042, n = 5 preparations, 5 synapses, t(4) = 5.70, P < 0.005). In addition, explicitly indicating that LTF was indeed synapse-specific, LTF was greater at treated peripheral synapses than at untreated central synapses of the same sensory neurons (P < 0.05, Tukey), which showed no facilitation (mean LTF score = –0.01 +0.08, t(4) = 0.11, P > 0.90). In contrast with results at 5-HT-treated peripheral synapses, for all five other drug treatment conditions, including the two experimental groups that received 5-HT plus either central or peripheral anisomycin, there were no significant increases or decreases in synaptic strength at either central or peripheral synapses (LTF score ranged from –0.12 ± 0.07 to 0.01 ± 03, all P values > 0.05, t tests). Similarly, there were no significant differences in LTF between peripheral and central synapses (all P values > 0.95, Tukey). These results confirm that LTF produced by synaptic application of 5-HT is synapse-specific, and they explicitly document that LTSSF also requires both somatic and local protein synthesis.
LTF at distal synapses requires active axonal transport
Both somatic and synaptic protein synthesis are essential for LTF triggered at distal sensorimotor synapses, as shown above, as well as at proximal synapses in dissociated cell culture (Martin et al., 1997a; Casadio et al., 1999), suggesting that communication between soma and synapse is essential for LTF. To examine directly the role of active axonal transport in LTF and intracellular signaling between synapse and soma, here we investigated whether incubating SN axons in the axonal transport inhibitor nocodazole, which disrupts microtubules (Dinter and Berger, 1998), would block LTF at distal peripheral SN-MN synapses normally triggered by synaptic application of 5-HT. We found that nocodazole completely blocked LTF at distal synapses, indicating an essential role for active anterograde and/or retrograde transport in LTF at distal synapses.
An analysis of variance was conducted on changes in synaptic strength at central and peripheral synaptic connections across each of three drug treatment conditions in all preparations. Consistent with a role for axonal transport, LTF varied significantly as a function of both drug treatment condition (F(2, 26) = 21.60, P < 0.001) and synaptic site (peripheral synapses vs. central synapses) (F(2, 26) = 28.16, P < 0.001), and there was a significant drug-by-site interaction (F(2, 26) = 35.94, P < 0.001), indicating that peripheral-central differences in facilitation were dependent on the drug treatment condition. Subsequent analyses support the following three main conclusions (Fig. 3).
Synaptic application of 5-HT produces LTF.
First, confirming previously published results (Clark and Kandel, 1993; Emptage and Carew, 1993; Gooch, 1997; Gooch and Clark, 1997; Martin et al., 1997a; Casadio et al., 1999; Liu et al., 2003; Grabham et al., 2005) as well as studies above, repeated application of 5-HT alone at peripheral synapses, without nocodazole, produced LTF specifically at peripheral synapses, approximately doubling the mean EPSP amplitude from day 1 to day 2 (mean ± SEM LTF score = 0.24 ± 0.03, n = 6 preparations, 13 synapses, t5 = 8.02, P < 0.001). Consistent with synapse-specific LTF, facilitation at 5-HT-treated peripheral synapses was also greater than at untreated central synapses (P < 0.001, Tukey), which showed no significant LTF (mean LTF score = 0.022 ± 0.014, n = 6 preparations, 13 synapses, t(5) = 1.57, P > 0.15). Thus, the present results confirm that synaptic application of 5-HT is sufficient to produce LTF and LTSSF at peripheral synapses, and they provide a positive control to examine the effects of axonal transport blockade.
Nocodazole completely blocked LTF at distal synapses.
Second and most importantly, LTF at distal peripheral synapses produced by 5-HT application in the peripheral (P) chamber was blocked completely by the prior application of the active axonal transport inhibitor nocodazole to the middle (M) chamber containing the siphon nerve and SN axons (M-Nocodazole + P-5-HT group; mean LTF score = –0.008 ± 0.017, n = 5 preparations, 11 synapses, t(4) = 0.45, P > 0.65). Further demonstrating blockade of LTF, LTF scores at peripheral synapses in these nocodazole-treated preparations were (1) significantly lower than LTF at peripheral synapses in the 5-HT alone group, which did not receive nocodazole (P < 0.001, Tukey); and (2) nearly identical to LTF scores at central synapses in the same experimental group (P > 0.95, Tukey), and to LTF scores at central and peripheral synapses in the M-Nocodazole alone control group (P > 0.95, Tukey), all of which showed no LTF (LTF score range 0.004 ± 0.009 to 0.012 ± 0.016, all P values > 0.45, t tests). This set of results provides the first direct evidence for an essential role of axonal transport between synapse and soma in LTF and LTSSF.
Blockade of LTF is not due to nonspecific effects.
Finally, synaptic strength was stable from day 1 to day 2 in the M-Nocodazole control group, which did not receive 5-HT, thus ruling out nonspecific effects of time and dissection-induced changes (Clark and Kandel, 1993), as well as nonspecific degradation of synaptic strength produced by blockade of axonal transport (mean LTF score at central synapses = 0.005 ± 0.009, n = 5 preparations, 13 synapses, t(4) = 0.50, P > 0.60; mean LTF score at peripheral synapses = 0.009 ± 0.015, n = 5 preparations, 10 synapses, t(4) = 0.60, P > 0.55). Further, there was no significant difference between LTF at central and peripheral synapses (P > 0.99, Tukey). Because nocodazole was applied to the middle chamber containing only the siphon nerve, and was not applied to the SN soma (in the central chamber) or SN synapses (in the peripheral chamber), it is unlikely that it interfered directly with somatic or synaptic processes involved in LTF.
Taken together, the above results directly demonstrate that active axonal transport is required for LTF.
Discussion
Results support three key conclusions regarding the mechanism of LTF and LTSSF at distal sensorimotor connections. First, proteins synthesized in the SN soma are essential, even for LTF triggered by events at the distal synapse. These proteins could in principle be either induced or constitutively expressed. Second, local protein synthesis at the synapse is also essential for LTF, implying that an immediate translational response might occur at synapses treated with 5-HT. The newly synthesized local proteins might serve as a retrograde messenger, as a synaptic mark that can capture or utilize regulatory factors in LTSSF, or as other direct or indirect factors in the LTF signaling pathway. Third, active anterograde or retrograde axonal transport (or both) are essential for LTF, and they appear to mediate critical intracellular signaling between synapse and soma. However, the mechanisms that might allow for synaptic marking and capture, given the long transport time (perhaps 13–17 h) between the SN distal synapse and soma, remain unknown.
LTF and LTSSF at distal synapses require both somatic and synaptic protein synthesis
Protein synthesis has been shown to be required both in long-term behavioral sensitization of the gill-withdrawal reflex evoked by a tactile siphon stimulus in Aplysia, and in LTF induced by cell-wide applications of 5-HT at Aplysia sensorimotor connections in dissociated cell culture (Montarolo et al., 1986; Castellucci et al., 1989). Studies in dissociated cell cultures in which a single sensory neuron contacts two nearby motor neurons have further indicated that both local and somatic protein synthesis are required for the establishment of LTSSF (Martin et al., 1997a).
The present work extends these previous findings to indicate that synaptic and local protein synthesis are also required for LTF and LTSSF at truly distal SN synapses, located several centimeters from the SN soma. The long soma-synapse distance used in the present studies imposes important temporal and mechanistic constraints on the processes underlying LTF and on the intracellular signals by which soma and synapse communicate, and thus raises questions regarding proposed models based on LTF experiments at proximal synapses.
As previously suggested (Martin et al., 1997a; Casadio et al., 1999), proteins newly synthesized at synapses in response to 5-HT may, among other possibilities, serve as (1) retrograde messengers or signals to generate retrograde messengers (Martin et al., 1997a; Casadio et al., 1999); (2) synaptic tags, or marks, that enable capture and/or stabilization of the LTF at 5-HT treated synapses (Martin et al., 1997a; Casadio et al., 1999); or (3) factors that directly or indirectly mediate LTF (Liu et al., 2003).
Our studies by themselves do not directly indicate whether the expression of the critical proteins is constitutive or is induced or modulated by 5-HT; they also do not discriminate between synthesis in presynaptic SN processes and postsynaptic motor neurons. However, LTF produced by repeated synaptic 5-HT applications in dissociated cell culture and at intact pleural-pedal sensorimotor connections requires presynaptic but not postsynaptic protein synthesis, although the requirements are different for coincident (asymmetric) LTF involving combined somatic/synaptic 5-HT applications (Martin et al., 1997a; Sherff and Carew, 2002, 2004).
According to the first scenario proposed above, retrograde messengers that are newly synthesized in SN processes in response to 5-HT may be transported to the SN soma to modify gene transcription and translation. In turn, newly synthesized somatic proteins may be transported anterogradely back to synapses, where they are captured and utilized preferentially at synapses previously treated with 5-HT, thus producing LTSSF. Several previous findings are consistent with this interpretation. First, in cultures containing a single SN that bifurcates to two different postsynaptic targets, 24-h LTF is induced at one set of SN synapses treated with five pulses of 5-HT, but not at other, untreated synapses of the same SN; further, this LTSSF is blocked by protein synthesis inhibition at synapses being treated with 5-HT. Second, if one SN branch receives five pulses of 5-HT, and the other branch receives only a single pulse of 5-HT, then both branches show 24-h LTF, even though one pulse of 5-HT is usually not sufficient to produce LTF. Critically, protein synthesis inhibition at the branch receiving one pulse of 5-HT does not block LTF at that branch. Taken together, these results suggest that induction of 24-h LTF requires local protein synthesis (e.g., to generate a retrograde message), but 24-h synaptic capture does not, at least under these conditions.
An additional possibility is that locally synthesized proteins could serve as the synaptic tag, or mark, at least when the synaptic stimulus substantively precedes the nuclear events, as was the case in the present studies. Such a mark would allow the synapse to capture or utilize newly synthesized macromolecules subsequently shipped from the soma. In the present studies, such a mark would need to be reasonably long-lasting (to allow arrival of products from the soma), and hence might involve a protein-synthesis-dependent step, even though capture per se is independent of protein synthesis. It is also possible that there are constitutive inhibitory constraints to LTF (as described below), which are relieved cell-wide by 5-HT application and the resultant local protein synthesis. With these inhibitory constraints relieved, a single pulse of 5-HT might be sufficient to mark a synapse for LTF, and enable it to capture proteins either induced or constitutively expressed. Further, whereas local protein synthesis is not required for synaptic capture at 24 h (Martin et al., 1997a), it is required for capture at 72 h, and for stabilization and persistence of synaptic growth for 72-h LTF (Casadio et al., 1999).
After removal of the SN cell body, isolated SN synapses in dissociated cell culture express protein-synthesis-dependent LTF lasting 24 h (Liu et al., 2003), indicating that local protein synthesis can contribute to 24-h LTF relatively directly, independently of a potential role in retrograde signaling or synaptic marking. However, LTF decays more rapidly than usual and is diminished at 48 h (Liu et al., 2003), suggesting that gene expression in the SN soma is required to maintain LTF indefinitely. Curiously, at 24 h, LTF expressed at isolated synapses is greater that that expressed by SN synapses connected to their cell bodies (Liu et al., 2003). These results raise the possibility (among others) that removing the SN soma eliminates or reduces constraints that normally repress LTF. Hence, one additional possible role for changes in transcription and subsequent translation in the SN soma that are triggered by a retrograde message is to remove constraints on the induction or expression of LTF. In the present experiments, potential retrograde messages conveyed from soma to synapse may in principle contribute to these processes as well.
It is unlikely that anisomycin itself was transported retrogradely from synapse to soma, where it inhibited somatic protein synthesis and thereby blocked LTF. First and most convincingly, application of protein synthesis inhibitors to one branch of a given Aplysia SN does not block LTF at another branch treated with 5-HT (Martin et al., 1997a). If anisomycin had been transported from the synapse to the soma where it blocked protein synthesis and consequently LTF, then it would have blocked LTF at the other synapse as well. Second, anisomycin is lipophilic, so it diffuses across cell membrane easily. Therefore, it might be expected to diffuse out from axons once the extracellular anisomycin had been removed, unless somehow bound intracellularly. Finally, in dissociated cell culture, anisomycin inhibits protein synthesis by Aplysia neurons only transiently (Montarolo et al., 1986). This near-immediate recovery of protein synthesis also suggests that anisomycin applied at synapses does not block somatic protein synthesis and LTF by being retrogradely transported to the soma, which is a time-dependent process.
LTF at distal synapses requires active axonal transport
The requirement for protein synthesis both in the soma and locally at the synapse for LTF (Martin et al., 1997a; Guan and Clark, 2001), as well as for coincident induction of LTF (Sherff and Carew, 1999; Sherff and Carew, 2002, 2004), implies an essential role for intracellular communication between the SN synapse and soma. However, the mechanisms by which this communication occurs were not well documented empirically.
Active axonal retrograde transport and anterograde transport between synapse and soma have been proposed as attractive candidate mechanisms for intracellular communication in LTF and have been indirectly implicated in a variety of experiments (Bacskai et al., 1993; Clark and Kandel, 1993; Martin et al., 1997b; Giustetto et al., 2003; Ormond et al., 2004; Thompson et al., 2004). Axonal transport also plays a critical role in other, potentially related phenomena, such as some aspects of injury-induced long-term hyperexcitability in Aplysia SN (Walters et al., 1991; Ambron et al., 1995; Gunstream et al., 1995; Walters and Ambron, 1995), but possibly not all aspects thereof (Weragoda et al., 2004). Axonal transport is also involved in synaptic responses to nerve growth factor (Thoenen et al., 1979). However, although its involvement seems plausible, the requirement for active axonal transport in LTF and LTSSF produced by 5-HT applications to Aplysia SNs had never been directly demonstrated. Further, other signaling pathways, such as calcium signaling or chains of phosphorylation, have also been proposed to be involved in at least some forms of LTF (Sherff and Carew, 1999).
In the present experiments, we found that the axonal transport inhibitor nocodazole, applied selectively to nerve (and not to the SN soma or synapse), completely blocked LTF at distal synapses, indicating that active anterograde axonal transport, retrograde axonal transport, or both, are in fact essential for LTF. Combined with previous results on the role of somatic and synaptic protein synthesis, these findings suggest that molecules generated directly or indirectly via protein synthesis in the synapse in response to 5-HT might be transported retrogradely back to soma to trigger LTF-related cellular events, including modulation of gene expression. In turn, macromolecules synthesized in the soma, which are either constitutively expressed or induced by retrograde signals from the 5-HT-treated synapse, might be transported anterogradely back to the synapse to produce and sustain LTF.
Although axonal transport is relatively slow, it is not necessarily limiting in the present experiments, which examined LTF at 24 or 30 h. As explained below, total round-trip time from synapse to soma and back again might require on the order of 13 h in our anisomycin experiments, or 17 h in our nocodazole experiments, which is well within our 24-h and 30-h time windows.
Estimated transport time can be inferred by extrapolation from retrograde transport rates of 36 mm/day at 15 °C in Aplysia mechanosensory neurons, which yields a retrograde axonal transport rate of 50 mm/day at 18 °C (our overnight incubation temperature), or 78 mm/day at 22 °C (during 2 h of drug treatment), given a Q10 of about 3 in Aplysia neurons (Goldberg et al., 1978) and other systems (Heslop and Howes, 1972). In our experiments, although our postsynaptic sites were 2.5 to 3 cm from the soma, 5-HT application occurred throughout the peripheral chamber and hence was slightly closer to the soma, 1 cm (anisomycin experiments) to 2 cm (nocodazole experiments), thus potentially reducing retrograde transport distance and transport time if functioning 5-HT receptors exist in more proximal regions of SN axons. Retrograde transport over a distance of 1–2 cm might require about 4 to 8.5 h, or less if transport rates were increased by cAMP induced by 5-HT application (Azhderian et al., 1994), by substrate concentration (Goldberg et al., 1978), or other factors (Azhderian et al., 1994).
Similarly, given an estimated basal anterograde transport rate of about 75 mm/d at 18 °C (our overnight incubation temperature), anterograde transport from soma to synapse (2.5 cm) might require about 8 h.
If the retrograde signal induced expression (or repression) of new macromolecules in the soma that were subsequently shipped to the synapse, additional time might be required for these intermediate steps. However, given that the round-trip time for axonal transport may be considerably less than our 24-h or 30-h time windows, in principle there might be sufficient time for these additional transcriptional and translational modifications to occur as well.
Consistent with the present results, retrograde/anterograde transport times also did not appear to be limiting for LTF and LTSSF at proximal sensorimotor connections in previous experiments in dissociated cell culture, which are only about 500 μm from the SN soma (Casadio et al., 1999).
It has been suggested that the temporal window for coincident (asymmetric) LTF, involving both synaptic and somatic 5-HT application, precludes retrograde axonal transport as a viable mechanism for some (but not all) models of coincident LTF (Sherff and Carew, 1999). However, this conclusion was based on several assumptions that could substantively influence transport time (Guan, 2004), and mechanisms for coincident LTF appear to be somewhat different from LTF evoked by repeated 5-HT applications (Sherff and Carew, 2002, 2004). Hence our present results are not necessarily in conflict with this previous work.
Axonal transport time and blockade of somatic protein synthesis
The duration of exposure to central anisomycin made no apparent difference in the ability to block LTF induced by synaptic application of 5-HT. Preparations that received peripheral 5-HT in conjunction with exposure to central anisomycin for only 3 h at the time of 5-HT exposure exhibited LTF blockade that was nearly identical to LTF blockade in preparations that received central anisomycin for the full 24-h period.
The 3-h anisomycin treatment provided a control condition that was well matched to our other anisomycin treatments, thus ruling out nonspecific degradation from protein synthesis inhibition that might have resulted from longer, 24-h anisomycin exposure (though empirically this did not appear to occur). However, it is perhaps somewhat surprising that the 3-h anisomycin treatment of the soma was sufficient to block LTF. In dissociated cell culture, LTF requires protein synthesis only briefly, during and immediately after 5-HT application (Montarolo et al., 1986). Further, in culture, inhibition of protein synthesis by anisomycin is transient: after anisomycin is removed, protein synthesis returns to normal or above-normal levels within 2 h (Montarolo et al., 1986). However, if in our experiments LTF triggered by synaptic application of 5-HT to distal terminals requires retrograde transport to the soma and subsequent somatic protein synthesis, then one might have expected that protein synthesis in the soma could have returned to normal by the time the retrograde signal arrived, and hence LTF blockade might not have occurred.
There are at least two possible explanations for why even relatively brief 3-h central anisomycin blocked LTF. First, perhaps recovery of protein synthesis in the abdominal ganglion after anisomycin washout is not as rapid as in dissociated cell culture. Potentially, the greater neuropil and connective tissue that are present in the intact abdominal ganglion could have served as a trap for anisomycin and slowed its removal, thus prolonging inhibition of protein synthesis.
The second reason involves the time for the retrograde signal to reach the soma. Given that the actual retrograde message remains unknown, relatively fast signaling pathways, such as calcium waves or a chain of phosphorylation (Sherff and Carew, 1999), might provide a means for 5-HT-treated synapses to communicate with the SN soma, other than the active retrograde transport of molecules that was previously suggested. Alternatively, given the experimental distances involved and potential axonal transport rates, even molecules carried by conventional retrograde axonal transport may have been able to reach the soma within a sufficiently brief time, as described above. Molecules arriving 4 h after start of 5-HT application could perhaps reach the soma only 2 h after start of anisomycin washout, at which time protein synthesis may not have fully recovered.
Potential roles for retrograde and anterograde axonal transport
Nocodazole disrupts microtubules and blocks both anterograde and retrograde transport (Dinter and Berger, 1998). Hence, although our experiments indicate that active axonal transport is essential for LTF, they do not determine whether anterograde transport alone, retrograde transport alone, or both, are critically involved. Our experiments also do not by themselves discriminate between transport of newly synthesized macromolecules and those constitutively expressed.
Synaptic application of 5-HT induces local protein synthesis that is essential for LTF (Martin et al., 1997a). One potential role for retrograde transport is to carry these newly synthesized proteins or other molecules from synapse to soma, where they would in turn modulate transcription and/or translation and induce or upregulate somatic protein synthesis. Alternatively, retrograde messengers could interact with somatic constitutive protein synthesis by placing an inhibitory constraint on it, or by relieving its inhibitory constraint on other machinery so that LTF is expressed. An essential role for retrograde transport in LTF would thus be consistent with synaptic protein synthesis. Certain other modes of synapse-soma signaling, such as calcium waves or phosphorylation cascades (Sherff and Carew, 1999), could not themselves directly carry newly synthesized synaptic proteins from the synapse to the soma. However, it is possible that synaptic protein synthesis activates other reactions that could be transmitted to the soma via alternative signaling pathways. It is also possible that the newly synthesized local proteins mediate functions other than serving as the retrograde signal.
Anterograde transport is probably the most straightforward mechanism by which the soma might transmit newly synthesized molecules to the synapse, where they could lead to the expression of LTF. Again, in principle these molecules could be either induced by 5-HT or constitutively expressed. Anterograde transport is consistent with an essential role of somatic protein synthesis shown here and elsewhere (Martin et al., 1997a), as well as with the presence of elevated levels of several gene products in LTF expression (Goelet et al., 1986; Castellucci et al., 1988, 1989; Barzilai et al., 1989; Eskin et al., 1989; Bergold et al., 1990; Dash et al., 1990; Bailey and Kandel, 1993; Byrne et al., 1993; Kaang et al., 1993; Alberini et al., 1994; Ghirardi et al., 1995; Alberini, 1999). Evidence for a constitutively expressed factor suppressing facilitation and potentially carried by anterograde transport comes from findings that excising the synaptic processes from the soma before 5-HT treatment leads to an increase in 24-h LTF (Liu et al., 2003). However, this LTF is not as persistent as normal, suggesting that anterograde transport of somatic molecules is necessary for stabilization of long-term changes.
Limitations of current models of synaptic marking and capture
Our findings involving the long soma-synapse distances of the distal SN synapses onto PMNs pose intriguing challenges not yet fully explained by the current models of synaptic marking and capture.
One challenge arises from the proposed time windows for synaptic marking and capture, each of which is on the order of a few hours (Casadio et al., 1999). To examine how long the synaptic marking signal persists in SNs in dissociated cell culture, a single pulse of 5-HT was applied to one branch of the SN at various times prior to application of five pulses of 5-HT to the other branch. The single pulse was able to capture LTF only when it occurred 1–2 h before the five pulses, indicating that the duration of the effectiveness of the mark was relatively short. In complementary experiments, a single pulse to one branch could capture the LTF when it was given 1 h, but not 4 h, after the application of the five pulses to the other branch, indicating that the duration of the capture period was also relatively short. Short time windows in the range of minutes to 3 h or less have also been reported for coincident LTF at Aplysia pleural-pedal SN-MN connections (Sherff and Carew, 1999), and for hippocampal LTP (Frey and Morris, 1997).
Short time windows are problematic for the distal synapses used in the present experiments. Retrograde messengers newly synthesized in the synapse in response to 5-HT would not arrive at the soma for several hours if they are carried by retrograde axonal transport. Similarly, proteins newly synthesized in the soma and carried by anterograde transport would not reach the distal synaptic terminals until after a comparable time period, producing a potential round trip approaching a day. By present synaptic marking/capture models, LTSSF would thus require that a synaptic mark be able to persist and remain effective for almost 24 h (by mechanisms that remain to be determined), and the ability of molecules to be captured would also need to persist for several hours. However, neither of these time frames seems consistent with the short time frames presently proposed for synaptic marking and capture.
If instead, intracellular communication is sufficiently fast to convey proteins between synapse and soma without raising these difficulties, the mechanisms by which this communication would occur are not yet known.
An additional challenge for synaptic marking models of 24-h LTF and LTSSF is that the mechanism for maintaining such a synaptic mark remains unknown. A novel series of studies recently indicated that a neuron-specific isoform of cytoplasmic polyadenylation element binding protein (CPEB) has many of the characteristics appropriate for a self-perpetuating, long-lasting synaptic mark for maintaining late-phase, 72-h LTF ( Si et al., 2003a,b, 2004; Bailey et al., 2004). However, local inhibition of CPEB in neural processes does not affect 24-h LTF. Thus, the 24-h form of LTF and LTSSF investigated in the present experiments would require a different mark. Such a mark could be an anatomical mark rather than, or in addition to, being a molecular mark. SN processes show protein-synthesis-dependent morphological growth that persists for 12 h but not 24 h when transcription is blocked, and can persist for 24 h but not 48 h when the SN soma is removed (Grabham et al., 2005). Local synaptic growth that is stabilized by interactions with gene products shipped from the soma could in principle contribute to synaptic marking.
ApTrkI, a trk-like receptor that contributes to LTF and responds to both 5-HT and sensorin, a neuropeptide released by 5-HT (Hu et al., 2004), has properties that imply it may function in the induction of a synaptic tag; further, translocation of activated trk receptors from synapses to the soma could be a critical step that depends on axonal transport (Ormond et al., 2004). Again, however, it is not known whether it could mediate synaptic marking and capture for the time frames involved in the present study.
Among additional alternative possibilities are the following. (1) Different time frames for synaptic marking and capture could exist for synapses at different distances from the soma, or at synapses onto different MNs (the present studies used siphon motor neurons, whereas many culture experiments use the gill motor neuron L7); however, this seems inelegant and would be difficult to coordinate biologically. (2) Different mechanisms could mediate LTF at mature, intact synapses, compared with mechanisms at newly formed synapses in dissociated cell culture. (3) Synaptic marking/capture in circumstances involving two different synaptic sites (Martin et al., 1997a) may be fundamentally different from synaptic capture/marking in circumstances involving round-trip signaling between the synapse and soma. (4) Time-window requirements could vary parametrically according to stimulus conditions, so that marking or capture with five pulses of 5-HT (rather than one pulse) persists considerably longer.
In any event, the results reported here suggest that our understanding of the relationships between synaptic marking, capture, and LTF remains incomplete.
Conclusions
We found that inhibition of protein synthesis at either the soma or synaptic regions blocks LTF and LTSSF at distal sensorimotor terminals, suggesting important communication between soma and synapse in establishment of LTF. Consistent with this idea, blockade of axonal transport also blocked LTF. In principle, the essential somatic or synaptic proteins could be either induced or constitutively expressed. One possible explanation is that here, as elsewhere (Martin et al., 1997a; Casadio et al., 1999), an immediate translational response occurs at synapses treated with 5-HT. The newly synthesized synaptic proteins might then serve as a retrograde messenger, as a synaptic mark to capture or utilize somatically generated molecules in LTF, or as other direct or indirect factors in the LTF signaling pathways (Fig. 4). Similar phenomena have been observed, and their underlying mechanisms analyzed, at more proximal synapses in dissociated cell culture (Martin et al., 1997a; Casadio et al., 1999). However, the long distance between soma and synapse in the present preparation imposes temporal and mechanistic constraints on the intracellular signals by which synapse and soma communicate. A particular challenge is maintaining a synaptic mark long enough for communication to occur between a distal synapse and the soma and back again, given the short time window proposed to be involved for synaptic marking and capture. Possibly, longer-lasting synaptic marks, longer time windows, or unknown, faster transport mechanisms could help solve this conundrum. Alternatively, perhaps the mechanisms involved in LTF at mature distal synapses are different from those at newly developed synapses in dissociated cell culture. In any case, a complete understanding of LTF and LTSSF at distal synapses appears to require further information about mechanisms of synaptic marking and capture and their relationship to LTF.
Acknowledgments
We thank Robert D. Hawkins, Mary T. Lucero, Edwin M. Maynard, Richard A. Normann, and Doju Yoshikami for helpful comments on an earlier version of this manuscript. Supported by NSF IBN-9796235 and the University of Utah.
Footnotes
Received 8 November 2005; accepted 4 April 2006.
Literature Cited
Alberini, C. M. 1999. Genes to remember. J. Exp. Biol.202:2887–2891.
Alberini, C. M., M. Ghirardi, R. Metz, and E. R. Kandel. 1994. C/EBP is an immediate-early gene required for the consolidation of long-term facilitation in Aplysia. Cell 76:1099–1114.
Ambron, R. T., R. Schmied, C. C. Huang, and M. Smedman. 1992. A signal sequence mediates the retrograde transport of proteins from the axon periphery to the cell body and then into the nucleus. J. Neurosci.12:2813–2818.
Ambron, R. T., M. F. Dulin, X. P. Zhang, R. Schmied, and E. T. Walters. 1995. Axoplasm enriched in a protein mobilized by nerve injury induces memory-like alterations in Aplysia neurons. J. Neurosci.15:3440–3446.
Azhderian, E. M., D. Hefner, C. H. Lin, L. K. Kaczmarek, and P. Forscher. 1994. Cyclic AMP modulates fast axonal transport in Aplysia bag cell neurons by increasing the probability of single organelle movement. Neuron12:1223–1233.
Bacskai, B. J., B. Hochner, M. Mahaut-Smith, S. R. Adams, B. K. Kaang, E. R. Kandel, and R. Y. Tsien. 1993. Spatially resolved dynamics of cAMP and protein kinase A subunits in Aplysia sensory neurons. Science260:222–226.
Bailey, C. H., and E. R. Kandel. 1993. Structural changes accompanying memory storage. Annu. Rev. Physiol.55:397–426.
Bailey, C. H., E. R. Kandel, and K. Si. 2004. The persistence of long-term memory: a molecular approach to self-sustaining changes in learning-induced synaptic growth. Neuron44:49–57.
Barzilai, A., T. E. Kennedy, J. D. Sweatt, and E. R. Kandel. 1989. 5-HT modulates protein synthesis and the expression of specific proteins during long-term facilitation in Aplysia sensory neurons. Neuron2:1577–1586.
Bergold, P. J., J. D. Sweatt, I. Winicov, K. R. Weiss, E. R. Kandel, and J. H. Schwartz. 1990. Protein synthesis during acquisition of long-term facilitation is needed for the persistent loss of regulatory subunits of the Aplysia cAMP-dependent protein kinase. Proc. Natl. Acad. Sci. USA87:3788–3791.
Byrne, J. H., R. Zwartjes, R. Homayouni, S. D. Critz, and A. Eskin. 1993. Roles of second messenger pathways in neuronal plasticity and in learning and memory: insights gained from Aplysia. Adv. Second Messenger Phosphoprot. Res. 27:47–108.
Casadio, A., K. C. Martin, M. Giustetto, H. Zhu, M. Chen, D. Bartsch, C. H. Bailey, and E. R. Kandel. 1999. A transient, neuron-wide form of CREB-mediated long-term facilitation can be stabilized at specific synapses by local protein synthesis. Cell99:221–237.
Castellucci, V. F., T. E. Kennedy, E. R. Kandel, and P. Goelet. 1988. A quantitative analysis of 2-D gels identifies proteins in which labeling is increased following long-term sensitization in Aplysia. Neuron 1:321–328.
Castellucci, V. F., H. Blumenfeld, P. Goelet, and E. R. Kandel. 1989. Inhibitor of protein synthesis blocks long-term behavioral sensitization in the isolated gill-withdrawal reflex of Aplysia. J. Neurobiol. 20:1–9.
Clark, G. A., and E. R. Kandel. 1984. Branch-specific heterosynaptic facilitation in Aplysia siphon sensory cells. Proc. Natl. Acad. Sci. USA205:417–419.
Clark, G. A., and E. R. Kandel. 1993. Induction of long-term facilitation in Aplysia sensory neurons by local application of serotonin to remote synapses. Proc. Natl. Acad. Sci. USA90:11411–11415.
Dash, P. K., B. Hochner, and E. R. Kandel. 1990. Injection of the cAMP-responsive element into the nucleus of Aplysia sensory neurons blocks long-term facilitation. Nature345:718–721.
Dinter, A., and E. G. Berger. 1998. Golgi-disturbing agents. Histochem. Cell Biol.109:571–590.
Emptage, N. J., and T. J. Carew. 1993. Long-term synaptic facilitation in the absence of short-term facilitation in Aplysia neurons. Science262:253–256.
Eskin, A., K. S. Garcia, and J. H. Byrne. 1989. Information storage in the nervous system of Aplysia: specific proteins affected by serotonin and cAMP. Proc. Natl. Acad. Sci. USA86:2458–2462.
Frey, U., and R. G. Morris. 1997. Synaptic tagging and long-term potentiation. Nature385:533–536.
Frey, U., and R. G. Morris. 1998. Synaptic tagging: implications for late maintenance of hippocampal long-term potentiation. Trends Neurosci.21:181–188.
Frey, U., M. Krug, K. G. Reymann, and H. Matthies. 1988. Anisomycin, an inhibitor of protein synthesis, blocks late phases of LTP phenomena in the hippocampal CA1 region in vitro. Brain. Res. 452:57–65.
Ghirardi, M., P. G. Montarolo, and E. R. Kandel. 1995. A novel intermediate stage in the transition between short- and long-term facilitation in the sensory to motor neuron synapse of Aplysia. Neuron 14:413–420.
Giustetto, M., A. N. Hegde, K. Si, A. Casadio, K. Inokuchi, W. Pei, E. R. Kandel, and J. H. Schwartz, 2003. Axonal transport of eukaryotic translation elongation factor 1 alpha mRNA couples transcription in the nucleus to long-term facilitation at the synapse. Proc. Natl. Acad. Sci. USA100:13680–13685.
Goelet, P., V. F. Castellucci, S. Schacher, and E. R. Kandel. 1986. The long and the short of long-term memory—a molecular framework. Nature322:419–422.
Goldberg, D. J., and J. H. Schwartz. 1980. Fast axonal transport of foreign transmitters in an identified serotonergic neurone of Aplysia californica. J. Physiol. 307:259–272.
Goldberg, D. J., J. H. Schwartz, and A. A. Sherbany. 1978. Kinetic properties of normal and perturbed axonal transport of serotonin in a single identified axon. J. Physiol.281:559–579.
Gooch, C. M. 1997. Synapse-specific heterosynaptic long-term facilitation at Aplysia sensorimotor connections. Ph.D. dissertation, Neuroscience Program, Princeton University.
Gooch, C. M., and G. A. Clark. 1997. Synapse-specific long-term facilitation at Aplysia sensorimotor connections. Soc. Neurosci. Abstr.23:1415.
Grabham, P. W., F. Wu, S. Schacher, and D. J. Goldberg. 2005. Initiating morphological changes associated with long-term facilitation in Aplysia is independent of transcription or translation in the cell body. J. Neurobiol.64:202–212.
Guan, X. 2004. Essential role of somatic and synaptic protein synthesis and axonal transport in long-term facilitation at distal sensorimotor connections in Aplysia californica. Ph.D. dissertation, Department of Bioengineering, University of Utah.
Guan, X., and G. A. Clark. 2001. Role of protein synthesis in long-term facilitation at distal sensorimotor connections in Aplysia. Abstract 644.8 [Online: Abstract Viewer/Itinerary Planner]. Society for Neuroscience, Washington, DC.
Guan, X., and G. A. Clark. 2003. Essential role of axonal transport at distal sensorimotor connections in Aplysia. Abstract 291.220 [Online: Abstract Viewer/Itinerary Planner]. Society for Neuroscience, Washington, DC.
Gunstream, J. D., G. A. Castro, and E. T. Walters. 1995. Retrograde transport of plasticity signals in Aplysia sensory neurons following axonal injury. J. Neurosci.15:439–448.
Heslop, J. P., and E. A. Howes. 1972. Temperature and inhibitor effects on fast axonal transport in a molluscan nerve. J. Neurochem.19:1709–1716.
Hu, J. Y., L. Glickman, F. Wu, and S. Schacher. 2004. Serotonin regulates the secretion and autocrine action of a neuropeptide to activate MAPK required for long-term facilitation in Aplysia. Neuron 43:373–385.
Kaang, B. K., E. R. Kandel, and S. G. Grant. 1993. Activation of cAMP-responsive genes by stimuli that produce long-term facilitation in Aplysia sensory neurons. Neuron10:427–435.
Kandel, E. R. 2001. The molecular biology of memory storage: a dialogue between genes and synapses. Science294:1030–1038.
Liu, K., J. Y. Hu, D. Wang, and S. Schacher. 2003. Protein synthesis at synapse versus cell body: enhanced but transient expression of long-term facilitation at isolated synapses. J. Neurobiol.56:275–286.
Martin, K. C., A. Casadio, H. Zhu, E. Yaping, J. C. Rose, M. Chen, C. H. Bailey, and E. R. Kandel. 1997a. Synapse-specific, long-term facilitation of Aplysia sensory to motor synapses: a function for local protein synthesis in memory storage. Cell91:927–938.
Martin, K. C., D. Michael, J. C. Rose, M. Barad, A. Casadio, H. Zhu, and E. R. Kandel. 1997b. MAP kinase translocates into the nucleus of the presynaptic cell and is required for long-term facilitation in Aplysia. Neuron 18:899–912.
Montarolo, P. G., P. Goelet, V. F. Castellucci, J. Morgan, E. R. Kandel, and S. Schacher. 1986. A critical period for macromolecular synthesis in long-term heterosynaptic facilitation in Aplysia. Science 234:1249–1254.
Ormond, J., J. Hislop, Y. Zhao, N. Webb, F. Vaillaincourt, J. R. Dyer, G. Ferraro, P. Barker, K. C. Martin, and W. S. Sossin. 2004. ApTrkl, a Trk-like receptor, mediates serotonin-dependent ERK activation and long-term facilitation in Aplysia sensory neurons. Neuron44:715–728.
Schwartz, J. H. 1979. Axonal transport: components, mechanisms, and specificity. Annu. Rev. Neurosci.2:467–504.
Sherff, C. M., and T. J. Carew. 1999. Coincident induction of long-term facilitation in Aplysia: cooperativity between cell bodies and remote synapses. Science285:1911–1914.
Sherff, C. M., and T. J. Carew. 2002. Coincident induction of long-term facilitation at sensory-motor synapses in Aplysia: presynaptic and postsynaptic factors. Neurobiol. Learn. Mem.78:498–507.
Sherff, C. M., and T. J. Carew. 2004. Parallel somatic and synaptic processing in the induction of intermediate-term and long-term synaptic facilitation in Aplysia. Proc. Natl. Acad. Sci. USA 101:7463–7468.
Si, K., M. Giustetto, A. Etkin, R. Hsu, A. M. Janisiewicz, M. C. Miniaci, J. H. Kim, H. Zhu, and E. R. Kandel. 2003a. A neuronal isoform of CPEB regulates local protein synthesis and stabilizes synapse-specific long-term facilitation in Aplysia. Cell 115:893–904.
Si, K., S. Lindquist, and E. R. Kandel. 2003b. A neuronal isoform of the Aplysia CPEB has prion-like properties. Cell115:879–891.
Si, K., S. Lindquist, and E. Kandel. 2004. A possible epigenetic mechanism for the persistence of memory. Cold Spring Harbor Symp. Quant. Biol.69:497–498.
Steward, O., and E. M. Schuman. 2003. Compartmentalized synthesis and degradation of proteins in neurons. Neuron40:347–359.
Thoenen, R. T., U. Otten, and M. Schwab. 1979. Orthograde and retrograde signals for the regulation of neuronal gene expression: the peripheral sympathetic nervous system as a model. Pp. 911–928 in The Neurosciences: Fourth Study Program, F. O Schmitt and F. G. Worden, eds. MIT Press, Cambridge, MA.
Thompson, K. R., K. O. Otis, D. Y. Chen, Y. Zhao, T. J. O’Dell, and K. C. Martin. 2004. Synapse to nucleus signaling during long-term synaptic plasticity: a role for the classical active nuclear import pathway. Neuron44:997–1009.
Trudeau, L. E., and V. F. Castellucci. 1995. Postsynaptic modifications in long-term facilitation in Aplysia: upregulation of excitatory amino acid receptors. J. Neurosci.15:1275–1284.
Turner, J. R., and A. M. Tartakoff. 1989. The response of the Golgi complex to microtubule alterations: the roles of metabolic energy and membrane traffic in Golgi complex organization. J. Cell Biol.109:2081–2088.
Walters, E. T., and R. T. Ambron. 1995. Long-term alterations induced by injury and by 5-HT in Aplysia sensory neurons: convergent pathways and common signals? Trends Neurosci.18:137–142.
Walters, E. T., H. Alizadeh, and G. A. Castro. 1991. Similar neuronal alterations induced by axonal injury and learning in Aplysia. Science 253:797–799.
Weragoda, R. M., E. Ferrer, and E. T. Walters. 2004. Memory-like alterations in Aplysia axons after nerve injury or localized depolarization. J. Neurosci.24:10393–10401.(Xin Guan and Gregory A. C)