Rapid increase in clusters of synaptophysin at onset of homosynaptic potentiation in Aplysia

Abstract

Imaging studies have shown that even the earliest phases of long-term plasticity are accompanied by the rapid recruitment of synaptic components, which generally requires actin polymerization and may be one of the first steps in a program that can lead to the formation of new stable synapses during late-phase plasticity. However, most of those results come from studies of long-term potentiation in rodent hippocampus and might not generalize to other forms of synaptic plasticity or plasticity in other brain areas and species. For example, recruitment of presynaptic proteins during long-term facilitation by 5HT in Aplysia is delayed for several hours, suggesting that whereas activity-dependent forms of plasticity, such as long-term potentiation, involve rapid recruitment of presynaptic proteins, neuromodulatory forms of plasticity, such as facilitation by 5HT, involve more delayed recruitment. To begin to explore this hypothesis, we examined an activity-dependent form of plasticity, homosynaptic potentiation produced by tetanic stimulation of the presynaptic neuron in Aplysia. We found that homosynaptic potentiation involves presynaptic but not postsynaptic actin and a rapid (under 10 min) increase in the number of clusters of the presynaptic vesicle-associated protein synaptophysin. These results indicate that rapid recruitment of synaptic components is not limited to hippocampal potentiation and support the hypothesis that activity-dependent types of plasticity involve rapid recruitment of presynaptic proteins, whereas neuromodulatory types of plasticity involve more delayed recruitment.

Whereas short-term plasticity involves covalent modifications of existing synapses, long-term plasticity often involves the formation of new synapses (1–8). A good deal is now known about synapse assembly during development (9, 10), but much less is known about the rules, sequence, and mechanisms of synapse assembly during long-term plasticity. Imaging studies have shown that even the earliest phases of long-term plasticity are accompanied by the rapid (within minutes) recruitment of synaptic components (11–16). The initial recruitment of synaptic components generally requires actin polymerization but not protein synthesis, whereas the long-term maintenance of these components requires protein synthesis (7, 8, 14, 16–18). These findings suggest that the early recruitment of synaptic components may be some of the first steps in a program that can lead to the protein synthesis-dependent formation of new stable synapses during late-phase plasticity.

Most of these results come from studies of long-term potentiation (LTP) in rodent hippocampus, however, and whether other forms of synaptic plasticity and plasticity in other brain areas and species are accompanied by a similar rapid recruitment of synaptic proteins is not known. For example, recruitment of presynaptic proteins during long-term facilitation by 5HT in Aplysia is delayed for several hours (19), suggesting that whereas activity-dependent forms of plasticity, such as LTP, involve rapid recruitment of presynaptic proteins, neuromodulatory forms of plasticity, such as facilitation by 5HT, involve more delayed recruitment. To begin to explore that hypothesis, we examined an activity-dependent form of plasticity, homosynaptic potentiation produced by brief tetanic stimulation (20 Hz for 2 s) of the presynaptic neuron at Aplysia sensory motor neuron synapses in isolated cell culture (Fig. 1 A and B). Homosynaptic potentiation was originally thought to be a form of post-tetanic potentiation (20), but subsequent studies have found that it lasts more than 30 min and shares components of the mechanisms of post-tetanic potentiation, NMDA- and mGluR-dependent LTP, and even long-term depression, but is not identical to any of them (21). Like LTP in hippocampal neurons (22–24), homosynaptic potentiation involves both presynaptic and postsynaptic molecular mechanisms, including presynaptic and postsynaptic Ca²⁺ and CamKII (21).

Fig. 1.

Homosynaptic potentiation involves presynaptic actin. (A) Experimental preparation. (B) General protocol; see Materials and Methods for details. (C) Examples of EPSPs in representative experiments with tetanic stimulation or test-alone (depression) control. (D) Average change in the EPSP in experiments with tetanic stimulation after pretreatment with either the actin polymerization inhibitor cytochalasin D (n = 12) or normal saline control (n = 14), and test-alone (depression) control after pretreatment with cytochalasin D (n = 7) or saline control (n = 7). Significant overall effects of tetanus [F(1,36) = 12,58, P < 0.01] and cytochalasin D (F = 6.14, P < 0.05) were seen on three-way ANOVA with one repeated measure (time). The data have been normalized to the value on the pretest in each experiment. The average pretest value was 10.6 mV, and the average depolarization during the tetanus was 10,818 mVms, with no significant difference between cytochalasin D and control. (E) Average change in the EPSP in experiments with tetanic stimulation after injection of phalloidin, which binds actin, into the sensory neuron (n = 9) or motor neuron (n = 7), and injection of vehicle into the sensory neuron (n = 6) or motor neuron (n = 6). Results with injection of vehicle into the sensory or motor neuron were very similar and have been pooled. A significant overall effect of group [F(2,25) = 2.59, P < 0.05, one-tailed test] was seen on two-way ANOVA with one repeated measure (time). The average pretest value was 7.1 mV, and the average depolarization during the tetanus was 9,496 mVms, with no significant difference between groups. The error bars indicate the SEMs.

LTP also involves both presynaptic and postsynaptic actin, which plays an important role in the recruitment of presynaptic and postsynaptic components during the plasticity (14, 16, 17, 25–31). Likewise, synaptic growth during long-term facilitation in Aplysia requires presynaptic actin (32, 33). However, whether homosynaptic potentiation also involves actin or is accompanied by the rapid recruitment of presynaptic components is not known. We addressed those questions and found that homosynaptic potentiation involves presynaptic but not postsynaptic actin and a rapid (under 10 min) increase in clusters of the presynaptic vesicle-associated protein synaptophysin.

Results

Homosynaptic Potentiation Involves Presynaptic Actin.

Tetanic stimulation of the sensory neuron (20 Hz for 2 s) produced homosynaptic potentiation that lasted more than 30 min [F(1,36) = 4.20, P < 0.05 compared with test-alone (depression) control at 40 min] (Fig. 1 C and D). An inhibitor of actin polymerization, cytochalasin D (5 μM), reduced the potentiation [F(1,36) = 12.59, P < 0.01 compared with normal saline control overall]. There was a significant reduction as early as 1 min after the tetanus and no significant interaction between the effect of cytochalasin D and time, suggesting that actin is important for both the early and late phases of potentiation. Cytochalasin D also increased homosynaptic depression on the test before the tetanus, at 5 min after the first excitatory postsynaptic potential (EPSP) [F(1,38) = 7.72, P < 0.01 for the average of the tetanus and depression groups compared with normal saline controls], perhaps by blocking a transient potentiating effect of the first test stimulation similar to homosynaptic potentiation (21). However, cytochalasin D had no significant effect on the depression from 10 to 40 min, indicating that its overall effect on potentiation is not due to enhanced depression. Cytochalasin D also had no significant effect on the initial amplitude of the EPSP or the postsynaptic depolarization during the tetanus. These results suggest that homosynaptic potentiation involves actin.

To examine whether actin is involved in the presynaptic or postsynaptic neurons, we injected phalloidin, which stabilizes F-actin, into either the sensory neuron or the motor neuron. Injecting phalloidin (1 mM in the electrode) into the sensory neuron reduced homosynaptic potentiation [F(1,25) = 5.08, P < 0.05 compared with vehicle control] (Fig. 1E) without affecting the initial amplitude of the EPSP or postsynaptic depolarization during the tetanus. Again, there was a significant reduction in potentiation as early as 1 min after the tetanus, and no significant interaction between the effect of presynaptic phalloidin and time. In contrast, injecting phalloidin into the motor neuron produced no significant effects, although a slight reduction in potentiation was seen at the early time points. These results suggest that homosynaptic potentiation involves presynaptic, but not postsynaptic, actin.

Homosynaptic Potentiation Is Accompanied by a Rapid Increase in Clusters of the Presynaptic Protein Synaptophysin.

Because actin plays an important role in structural alterations and the redistribution of synaptic proteins during other forms of plasticity (14, 16, 17, 28–33), we next examined the possible redistribution of synaptic proteins during homosynaptic potentiation. Because that form of plasticity involves presynaptic actin but not postsynaptic actin, we focused on the presynaptic protein synaptophysin. We cloned a DNA construct for a synaptophysin-GFP fusion protein into the Aplysia expression vector pNEX3 (19, 34) and microinjected the purified plasmid DNA into the nucleus of a sensory neuron cocultured with a motor neuron. The next day, the synaptophysin-GFP fluorescence appeared in discrete puncta along the processes of the motor neuron (Fig. 2A, Right). Most of those puncta colocalized with puncta of synaptophysin immunofluorescence (Fig. 2A, Left); however, a few synaptophysin-GFP puncta did not colocalize, indicating that the synaptophysin antibody did not recognize synaptophysin-GFP. These results suggest that the subcellular distribution of overexpressed synaptophysin is very similar to that of endogenous synaptophysin.

Fig. 2.

Homosynaptic potentiation is accompanied by a rapid increase in clusters of the presynaptic protein synaptophysin. (A) (Left) Example of synaptophysin-GFP fluorescent puncta (green), synaptophysin-immunoreactive puncta (red), and colocalization (yellow). (Right) Example of the distribution of synaptophysin-GFP puncta (green) on processes of the initial segment of the motor neuron (phase contrast). (Scale bars: 10 and 15 μm.) (B) Examples of synaptophysin-GFP fluorescent puncta before (pre, green) superimposed on those 10 min after (post, red) test-alone control (Left) or tetanic stimulation of the sensory neuron (20 Hz for 2 s; Right). Puncta that were present in the posttest only are in red, those present in the pretest only are in green, and those present at both times are in yellow. The pre-tetanus fluorescent image is the same as the dashed area in A. (Scale bar: 10 μm.) (C) Percentage changes in the EPSP and in the total fluorescence of the synaptophysin-GFP puncta after tetanic stimulation (n = 22) or test-alone control (n = 13), and the correlation between them in experiments like those shown in B. The solid line indicates the linear regression, and the dashed lines indicate the 95% confidence limits. (D) Average time course of changes in the total number of synaptophyin-GFP fluorescent puncta. (E) Average number of synaptophysin-GFP puncta present in the 10-min posttest only (Left), the pretest only (Middle), and both tests (Right) after tetanic stimulation or test-alone control. In D and E, the number of puncta in the field (235 μm × 355 μm) has been normalized to the total number on the pretest in each experiment. The average pretest value was 22, not significantly different between tetanus and control. (F) Example of changes in subthreshold fluorescent signals. The image is the same as that in B, Right (which was generated with an intensity threshold to facilitate recognition and counting of the puncta), but without the threshold. The arrowheads indicate subthreshold fluorescence on the pretest near superthreshold puncta on the posttest. (G) Average changes in fluorescence of the puncta and the surrounding area within 5 μm for puncta that were counted in the posttest only (Left), in the pretest only (Middle), and in both tests (Right). **P < 0.01; *P < 0.05; ⁺P < 0.05 one-tailed test, in planned comparisons with control (in D and E) or pretest (in G).

We next examined the distribution of the synaptophysin-GFP fluorescence before and after the induction of homosynaptic potentiation. We acquired an image, tested the sensory-motor neuron PSP, delivered tetanic stimulation (20 Hz, 2 s) to the sensory neuron, tested the PSP again, and acquired images at ∼10, 25, and 40 min after the pretest. In control experiments, we omitted the tetanic stimulation to check for any changes due to the test stimulation, photo damage, bleaching, etc. The images were analyzed with custom software that automatically identifies puncta and measures their size, intensity, and number as well as changes in these parameters over time (14).

Within 10 min after tetanic stimulation, there were increases in the PSP [F(1,33) = 12.46, P < 0.01 compared with the no tetanus controls] and total fluorescence in the puncta (F = 13.55, P < 0.01) (Fig. 2 B and C). Furthermore, the increase in fluorescence correlated with the increase in the PSP (r = 0.34, P < 0.05) and was due primarily to an increase in the number of puncta (Fig. 2D), with no significant changes in their size or intensity (although modest changes in those parameters might have been missed due to the thresholding procedure used to identify the puncta). In the control experiments, most synaptophysin-GFP puncta remained stable (“pre and post,” yellow), but there were also both losses of preexisting puncta (“pre only,” green) and gains of new puncta (“post only,” red) (Fig. 2 B and E). After tetanic stimulation, more puncta were gained [F(1,33) = 3.40, P < 0.05, one-tailed test] and fewer puncta were lost (F = 4.86, P < 0.05) than in the control group, resulting in a net increase in the number of puncta. This increase was then maintained at approximately the same level for at least another 30 min (F = 8.28, P < 0.01 overall) (Fig. 2D), suggesting that the increase could contribute to the maintenance as well as induction of the potentiation.

We identified the puncta based in part on a threshold set to maximize the discrimination of puncta from background, which we also used to generate the images in Fig. 2B. We next examined the effect of tetanic stimulation on subthreshold changes in the fluorescence, which might provide information about the origin of the new puncta (Fig. 2F is the same image without the threshold). In some cases, puncta appeared to move along a process with little change in size or intensity. This movement did not contribute to the increase in number and was not affected by tetanic stimulation. In other cases, however, subthreshold puncta appeared to move and increase in intensity, or subthreshold areas of diffuse fluorescence appeared to aggregate into nearby superthreshold puncta (arrowheads). To quantify those effects, we measured changes in the total fluorescence of the puncta and the surrounding area within 5 μm of the puncta (Fig. 2G). For puncta that were gained (“post only”), the increase in fluorescence of the puncta was accompanied by a decrease in fluorescence of the surrounding area [F(1,25) = 9.17, P < 0.01]. Conversely, for puncta that were lost (“pre only”), the decrease in fluorescence of the puncta was accompanied by an increase in fluorescence of the surrounding area [F(1,31) = 6.38, P < 0.05]. There was no significant difference in these effects between the tetanus and control groups, and thus the results were pooled. These results suggest that under control conditions, the puncta undergo continual increases and decreases in intensity or aggregation and disaggregation, and that tetanic stimulation temporarily shifts the balance so that a greater number of puncta intensify or aggregate.

Discussion

Role of Actin in Synaptic Plasticity and Recruitment of Synaptic Proteins.

We have found that, like hippocampal LTP (14, 17, 28, 29), homosynaptic potentiation involves presynaptic actin. Long-term facilitation in Aplysia also involves presynaptic actin, whereas short-term facilitation does not (32, 33). These results are parallel to those for recruitment of presynaptic components (see below) and support a role for presynaptic actin in that process. Hippocampal LTP also involves recruitment of postsynaptic components and postsynaptic actin (14, 16, 17, 25–28, 30, 31). However, although homosynaptic potentiation may be accompanied by recruitment of postsynaptic proteins (21), it does not involve postsynaptic actin, suggesting a possible difference in postsynaptic mechanisms of recruitment in Aplysia. Whether intermediate- and long-term facilitation in Aplysia, which are also accompanied by recruitment of postsynaptic proteins (35, 36), involve postsynaptic actin is not known.

Rules for Rapid Recruitment of Synaptic Proteins.

We also found that, like hippocampal LTP (14), homosynaptic potentiation in Aplysia involves rapid (under 10 min) recruitment of the presynaptic vesicle-associated protein synaptophysin. Both LTP and homosynaptic potentiation also are thought to involve rapid recruitment of postsynaptic glutamate receptors, although the evidence is indirect for homosynaptic potentiation (11, 21). Rapid recruitment does not occur for all types of plasticity or synaptic proteins, however; for example, application of dopamine to hippocampal neurons produces potentiation accompanied by rapid recruitment of postsynaptic glutamate receptors but not of presynaptic vesicle-associated proteins (37, 38). Similarly, intermediate- and long-term facilitation by 5HT in Aplysia are accompanied by rapid recruitment of postsynaptic glutamate receptors but more delayed (hours) recruitment of presynaptic vesicle-associated proteins (19, 35, 36), and short-term facilitation is not accompanied by recruitment of either (35, 39). In fact, short-term facilitation is accompanied by no change in the presynaptic vesicle-associated membrane protein (VAMP) and a dispersion of synapsin I, which is thought to reflect dissociation of synapsin I from synaptic vesicles (39, 40).

Collectively, these results suggest that activity-dependent types of plasticity, such as homosynaptic potentiation and LTP, involve rapid recruitment of presynaptic proteins, whereas neuromodulatory types of plasticity, such as facilitation by 5HT or dopamine, may involve more delayed recruitment. However, both activity-dependent and neuromodulatory plasticity may involve rapid recruitment of postsynaptic proteins. Thus, during neuromodulatory plasticity, the rapid recruitment of postsynaptic proteins could provide a target for later recruitment of the presynaptic proteins. More generally, for both types of plasticity, recruitment of synaptic proteins could be an early step in synapse assembly during long-term plasticity (8, 19, 35).

Recruitment of Synaptic Proteins and Synaptic “Tagging.”

A seeming difficulty with the foregoing idea is that homosynaptic potentiation, which is accompanied by rapid recruitment of synaptophysin, does not have a long-term (24 h) effect. But if homosynaptic potentiation is combined with several brief applications of 5HT that produce modest long-term facilitation, the potentiation enhances the facilitation at synapses from the active presynaptic neuron, but not at synapses from an inactive presynaptic neuron onto the same postsynaptic neuron (41). This hybrid form of plasticity thus combines the synapse specificity of activity-dependent plasticity with the long-term persistence of neuromodulatory plasticity (42).

Long-term plasticity can be synapse-specific even though it usually requires protein and RNA synthesis in the nucleus, which would be expected to produce cell-wide effects. A proposed explanation is that the active synapses are somehow “tagged” so that they can capture gene products that arrive from the nucleus later (43, 44). The identity of this tag is unknown, however (45, 46). One possible function of the rapid recruitment of synaptic components is to serve as such a tag for synapse-specific long-term growth (8). Consistent with that idea, actin polymerization, which is important for rapid changes in synaptic puncta and structures (14, 16, 17, 28–31), is also known to be important for synaptic tagging (47).

Recruitment of either presynaptic or postsynaptic proteins might serve as such a tag, depending on the experimental protocol (43, 44). For example, our results suggest that pairing homosynaptic potentiation with bath application of 5HT should produce rapid recruitment of postsynaptic proteins at all synapses, but recruitment of presynaptic proteins only at synapses from the active presynaptic neuron. Thus, recruitment of presynaptic proteins seems more likely to serve as a synapse-specific tag in that case. However, local application of 5HT to a set of synapses (44) should produce rapid recruitment of postsynaptic, but not presynaptic, proteins at those synapses. Thus, recruitment of postsynaptic proteins seems more likely to serve as a tag in that case. Of course, other presynaptic or postsynaptic molecules might contribute to tagging as well (45, 46), and these presynaptic and postsynaptic mechanisms also may interact through transynaptic signaling (36, 41, 48–50).

Materials and Methods

Electrophysiology.

Aplysia cocultures (an L7 gill motor neuron and one or two pleural sensory neurons) were prepared as described previously (20, 51) (Fig. 1A). Electrophysiological methods were applied as described previously (21) (Fig. 1B). In brief, 4–6 d after cell plating, the motor neuron was impaled with a sharp microelectrode (10–20 MΩ) filled with 2.5 M KCl, and extracellular stimulation was used to stimulate one of the sensory neurons and produce an EPSP. After 30 min of rest or drug incubation, the EPSP was tested once every 5 min for a total of 40 min. A tetanus (20 Hz for 2 s) was applied to the sensory neuron at 1 min before the third EPSP. In some experiments (test alone or homosynaptic depression control), the tetanus was omitted.

Cytochalasin D (EMD Chemicals) was prepared as a stock solution in DMSO and diluted in the perfusion medium before use. It was then included in the perfusion solution from the beginning of the 30-min rest period until the end of the experiment. In some experiments, phalloidin (EMD Chemicals) was pressure-injected into the sensory or motor neuron from a low-resistance (3–6 MΩ) electrode. The vehicle solution consisted of 0.5 M potassium acetate, 10 mM Tris-HCl to adjust the pH to 7.5, and 0.2% Fast Green to visualize the injection. In control experiments, the vehicle solution alone was injected into the neuron. The injection was terminated when the cell body became visibly green and began to swell. The electrode was then removed and replaced with a stimulating (sensory neuron) or recording (motor neuron) electrode, and the preparation was rested for 30 min before starting the experiment.

Drug and inhibitor experiments were interleaved with control experiments from the same culture batch, usually on the same day. The data were analyzed by two- or three-way ANOVA with one repeated measure (time), followed by planned comparisons of the individual groups. The overall ANOVA results are presented in the figure legend, and individual comparisons of interest are presented in Results.

Imaging.

A DNA construct for a synaptophysin-GFP fusion protein was cloned into the Aplysia expression vector pNEX3 (19, 34). Purified plasmid DNA was microinjected into the sensory neuron, and 1 d later, expression of the fluorescent protein was examined with a Bio-Rad MRC-1000 laser confocal scanning system coupled to a Zeiss Axiovert inverted microscope. The techniques for imaging and data analysis were the same as described previously for cultured hippocampal neurons (14). In brief, a field around the initial segment of the motor neuron containing synaptophysin-GFP–expressing processes of the sensory neuron was selected for analysis by an observer who was blinded to the experimental treatment. A Z series consisting of 10–15 optical sections (with Kalman averages of three scans for each section) was collected for the entire thickness of the tissue and stacked to create a projected image. The number, size, and intensity of synaptophysin-GFP puncta were then measured automatically using a computer program written in IDL (Research Systems). The puncta were identified based on a fluorescence intensity that exceeded a threshold set to maximize discrimination from the background, along with a diameter of 1–10 μm.

Once the laser intensity, photomultiplier gain, and fluorescence intensity threshold were set, they were not changed for the remainder of the experiment. We acquired an image, tested the sensory motor neuron PSP, delivered tetanic stimulation (20 Hz, 2 s) to the sensory neuron, tested the PSP again, and acquired images at ∼10, 25, and 40 min after the pretest or at ∼5, 20, and 35 min after the tetanus. In control experiments, the tetanic stimulation was omitted. The number, size, and intensity of puncta at each time were normalized to the pretest values. The data were analyzed by two-way ANOVA with one repeated measure (time), followed by planned comparisons of the experimental treatments at each test time.

We also used immunofluorescence techniques to examine the localization of endogenous synaptophysin and compare it with overexpressed synaptophysin-GFP. The immunofluorescence techniques were similar to those used previously for hippocampal neurons (14), modified for Aplysia (39, 52). In brief, the cells were fixed in 4% paraformaldehyde in PBS (pH 7.4) containing 30% sucrose and permeabilized with 0.1% Triton X-100. Nonspecific antibody binding was blocked with 10% normal goat serum. The cells were then incubated in mouse monoclonal anti-synaptophysin (5 μg/mL; Boehringer Mannheim Biochemicals), which recognizes the Aplysia protein (53), followed by goat anti-mouse conjugated with Cy5 (diluted 1:200; Jackson Immunoresearch). The cells were washed with PBS between each step and viewed using the confocal scanning system described above.