To test whether LTH of the GWR requires transcription, actinomycin-D (40 μM in ASW with 0.2% DMSO) was infused into the abdominal ganglion. Actinomycin-D's effects are irreversible; therefore, the drug was applied for 1 h, and then washed out with ASW immediately before the start of the pretests. The actinomycin-D treatment (n = 6) blocked LTH (Figure 1A). The difference between the pretest and posttest withdrawal in response to stimulation of the Trained side was not significant (mean normalized posttest response = 99.3 ± 12.5%, p > 0.6). Furthermore, the posttest GWR evoked by stimulation of the Untrained siphon side was not significantly different from the pretest GWR (mean normalized posttest response = 111.2 ± 9.5%, p > 0.1). Finally, there was no statistical difference between the Trained and Untrained posttest GWRs (p > 0.1).

In the control experiments (n = 6) ASW containing 0.2% DMSO was infused into the abdominal ganglion 1 h before the pretests, and washed out immediately before the start of the experiment. The training produced LTH of the GWR to Trained side stimulation (mean normalized posttest response = 47.3 ± 10.2%, p < 0.04 for the comparison with the Trained pretest response) (Figure 1B). By contrast, the posttest GWR to stimulation of the Untrained side of the siphon did not differ significantly from the pretest GWR (mean normalized response = 108 ± 8.8%, p > 0.4). Also, the posttest response evoked by Trained side stimulation was significantly less than that evoked by Untrained side stimulation (p < 0.003). In addition, there was a significant difference between the Trained posttest GWRs of the actinomycin-D-treated and control groups (p < 0.003). The Untrained posttest responses of the two groups, however, did not differ significantly (p > 0.8). Importantly, despite the fact that actinomycin-D treatment blocked LTH, Trained-site stimulation nonetheless produced normal short-term habituation (STH) of the GWR in the actinomycin-treated preparations. (Compare the Trained responses to the Block 1 habituating stimuli in Figures 1A,B).

To control for non-specific effects of actinomycin-D, we tested the effect of a reversible transcriptional inhibitor, 5,6-dichloro-1-β-d-ribofuranosylbenzimidazole (DRB) (Sehgal and Tamm, 1978; Yamaguchi et al., 1998; Apergis-Schoute et al., 2005), on LTH. DRB (150 μM in ASW with 0.2% DMSO) was present in the abdominal ganglion throughout the training period. The presence of DRB during habituation training blocked LTH (n = 9, Figure 2A). There was no significant change in the Trained GWR (mean normalized postest response = 115.1 ± 10.9%, p > 0.2 for the comparison with the pretest response). The GWR to stimulation of the Untrained side also did not change (mean normalized Untrained posttest response = 122.6 ± 19.1%, p > 0.7 for the comparison with the pretest response). In addition, the Trained and Untrained posttests did not differ (p > 0.8). Control experiments, in which ASW with 0.2% DMSO was infused into the abdominal artery, were performed using the same protocol as the DRB experiments. The habituation training produced significant LTH of the Trained GWR in the control experiments (n = 9, Figure 2B). There was a significant decrease in posttest response to Trained side stimulation (mean normalized posttest response = 75.6 ± 8.8% p < 0.04 for the comparison with the pretest response). However, there was a significant increase in the GWR to stimulation of the Untrained side (mean normalized posttest response = 118.4 ± 7.5%, p < 0.03 for the comparison with the Untrained pretest response). Furthermore, the posttest Untrained GWR was significantly greater than the posttest Trained GWR (p < 0.004). Finally, the posttest GWRs to Trained side stimulation differed significantly in the DRB-treated and control preparations (p < 0.02).

The effect of the habituation training on the GWR to Untrained side stimulation in the control experiments suggests that the training produced some generalized sensitization, as well as pathway-specific LTH, in these experiments. In our earlier study (Ezzeddine and Glanzman, 2003) we also observed that the GWR to Untrained side stimulation sometimes exhibited sensitization after habituation training, and that there was often sensitization of the GWR to Trained side stimulation during training in the presence of drugs that blocked LTH. We believe, as discussed in that paper, that these phenomena reflect the fact that the habituating stimuli can activate both habituating and sensitizing neuronal processes, as originally proposed by Groves and Thompson (1970). According to Groves and Thompson, the ultimate behavioral outcome, whether habituation, sensitization, or no change, is determined by the relative strengths of these two countervailing processes. It should be noted, however, that both the sensitization of the GWR on the Untrained side, and the sensitization of the reflex on the Trained side during habituation training in the presence of inhibitors of LTH (e.g., Blocks 2 and 3, Figure 1A), were highly variable and not always observed.

We previously found that LTH depends on activation of NMDA-type receptors in abdominal ganglion. Activation of postsynaptic NMDA receptors would be expected to produce a rise in intracellular Ca²⁺ in neurons of the GWR circuit. The Ca²⁺/calmodulin-dependent phosphatase calcineurin has been shown to be involved in both long-term synaptic depression (Mulkey et al., 1994) and memory extinction (Baumgartel et al., 2008) in mammals. Given that long-term depression (LTD) of the sensorimotor synapse (Lin and Glanzman, 1996) is a candidate mechanism for LTH in Aplysia (see Ezzeddine and Glanzman, 2003; Glanzman, 2009), we wished to know whether LTH was regulated by calcineurin activity. Accordingly, we tested the effect of ascomycin (10 μM in ASW with 0.2% DMSO), a selective calcineurin inhibitor (Sierra-Paredes and Sierra-Marcuño, 2008), on LTH. Ascomycin application during habituation training blocked LTH (n = 7, Figure 3A). The difference between the pretest and posttest responses for Trained side stimulation was not significant (mean normalized posttest response = 104.2 ± 14%, p > 0.9). However, as was the case for the control experiments in Figure 2B, the posttest response for the Untrained side exhibited significant enhancement (mean normalized posttest response = 143.2 ± 15.7%, p < 0.05 for the comparison with the pretest response). The difference between the posttest responses for the Trained and the Untrained sides, however, was not significant (p > 0.1). Control experiments (n = 5) were performed with rapamycin (10 μM in ASW with 0.2% DMSO) present in the cannula during training (Figure 3B). Rapamycin has the same binding site on calcineurin as ascomycin, but rapamycin does not inhibit dephosphorylation by calcineurin (Schreiber and Crabtree, 1992). In contrast to the effect of ascomycin, rapamycin did not affect LTH. Thus, habituation training resulted in a significantly reduced GWR for the Trained side (mean normalized posttest response = 51.4 ± 13%, p < 0.03 [t-test used here for the comparison with the pretest response]). The difference between the pretest and posttests responses for Untrained side, however, was not significant (mean normalized posttest GWR = 93.4 ± 6.7%, p > 0.3). As would be expected from these results, the posttest responses for the Trained and Untrained sides differed significantly in the experiments involving rapamycin treatment (p < 0.03). Finally, the posttest GWR to Trained side stimulation was significantly less in the rapamycin-treated preparations than in the ascomycin-treated preparations (p < 0.05). Notice that, although blockade of calcineurin activity with ascomycin disrupted LTH, it did not affect STH (Figure 3A).

We previously found that activation of AMPA receptors, as well as of NMDA receptors, contributes to LTH of the GWR (Ezzeddine and Glanzman, 2003). It is therefore possible that depolarization-induced activation of l-type voltage-gated Ca²⁺ channels (LVGCCs) is upstream of calcineurin activity during LTH. To test this possibility, we examined the effect of blockade of LVGCCs on LTH. We infused nitrendipine (100 μM in ASW with 0.14% DMSO), an antagonist of LVGCCs (Bolshakov and Siegelbaum, 1994), into the abdominal ganglion during habituation training (n = 6). Nitrendipine treatment blocked LTH to stimulation of the Trained side of the siphon (mean normalized posttest response = 116.8 ± 13%, p > 0.3 for the comparison with the pretest response) (Figure 4A). In these experiments habituation training did not alter the GWR to stimulation of the Untrained side; the mean normalized Untrained posttest response (157.8 ± 39.4%, was not significantly different from the mean normalized Untrained pretest response (p > 0.4), although the variability in the Untrained responses was admittedly large. Furthermore, the GWR evoked by posttest stimulation of the Trained side was not significantly different from that evoked by posttest stimulation of the Untrained side (p > 0.9). In control experiments (n = 8), performed at the same time as the nitrendipine experiments, the abdominal ganglion was infused ASW plus 0.14% DMSO. Here, habituation training produced significant LTH of the GWR to stimulation of the Trained side (mean normalized posttest response = 55.1 ± 12.4%, p < 0.02 for the comparison with the pretest response) (Figure 4B). There was no significant change in the GWR to stimulation of the Untrained side (mean Untrained normalized posttest response = 124.4 ± 15.3%, p > 0.1 for the comparison with the Untrained pretest response). Furthermore, the Trained side posttest GWR and the Untrained side posttest GWR differed significantly (p < 0.004). Finally, the difference between the posttest response to the Trained side stimulation in the nitrendipine group and that in the control group was highly significant (p < 0.005). These results support the idea that LVGCC activity is essential for LTH, although such activity appears not to be necessary for STH (Figure 4A).

Memory in Aplysia has been divided into three stages: short-term, intermediate-term and long-term (Goelet et al., 1986; Sutton and Carew, 2002). These three stages are distinguished by both their temporal and mechanistic properties. Temporally – although their time courses overlap somewhat – short-term memory (STM) lasts from seconds to ∼30 min; intermediate-term memory (ITM) lasts from ∼30 min to 3 h; and long-term memory lasts for >10 h. Mechanistically, STM depends exclusively on posttranslational changes; ITM requires protein synthesis, but not gene transcription; and LTM requires both translation and transcription. However, most of the previous work characterizing the mechanisms underlying the different phases of memory in Aplysia has concerned only one type of memory – sensitization of the withdrawal reflex. Although an early report demonstrated that a neural correlate of STH of the GWR did not require protein synthesis (Schwartz et al., 1971), there has been little attempt until recently to determine whether there are different phases of memory for habituation that correspond mechanistically to those for sensitization. We previously showed that the habituation of the GWR that persisted for 1–6 h following a 5 h period of spaced training required protein synthesis (Ezzeddine and Glanzman, 2003). But in that study we did not test whether this form of habituation memory also required gene transcription. Here, we have shown that the habituation resulting from a similar training protocol requires transcription, because it was blocked by actinomycin-D and DRB. Therefore, this form of habituation fits the definition of LTM as it has commonly been defined in Aplysia (Goelet et al., 1986). It will be interesting in the future to determine whether there is a training protocol that produces an intermediate form of habituation memory, one that requires translation, but not transcription.

It has long been believed that the mechanism of habituation of the gill- and siphon-withdrawal reflex is due exclusively to presynaptic changes. This mechanism, homosynaptic depression (HSD), is thought to involve decreased transmitter release from presynaptic terminals due to repeated activation of sensory neurons (Castellucci and Kandel, 1974; Armitage and Siegelbaum, 1998). Currently, the mechanism underlying HSD in Aplysia is controversial. It was originally proposed that HSD was caused by depletion of the readily releasable pool of presynaptic vesicles (Gingrich and Byrne, 1985; Hochner et al., 1986; Bailey and Chen, 1988b). More recent evidence, however, indicates that HSD results from all-or-none silencing of presynaptic release sites (Gover and Abrams, 2009).

Although HSD may indeed be a major mechanism of STH, the data from this and our previous study (Ezzeddine and Glanzman, 2003) argue strongly against the notion that LTH can be explained by exclusively presynaptic changes. First, HSD does not depend on glutamate receptor activity (Armitage and Siegelbaum, 1998), whereas LTH does (Ezzeddine and Glanzman, 2003). Second, LTH requires activation of LVGCCs (present study), which is most likely caused by AMPA receptor-mediated postsynaptic depolarization (see Ezzeddine and Glanzman, 2003). Studies by Bailey and Chen (1983, 1988a) indicate that LTH is accompanied by presynaptic morphological changes, including fewer vesicles in sensory neuron-associated active zones, and a decrease in the number of varicosities per sensory neuron. These morphological data, together with our data, support the conclusion that LTH involves both pre- and postsynaptic long-term cellular changes. If correct, this conclusion raises the intriguing possibility that the pre- and postsynaptic changes are coordinated by transsynaptic signals (see Glanzman, 2009, for further discussion).

The data from the present study suggest that increased intracellular Ca²⁺plays a key role in LTH. One likely source of elevated intracellular Ca²⁺ is influx through postsynaptic NMDA receptors (Ezzeddine and Glanzman, 2003). Activated LVGCCs, due to AMPA receptor-mediated postsynaptic depolarization, represent another potential source of the elevated intracellular Ca²⁺ necessary for LTH. Our finding that ascomycin blocks the induction of LTH indicates that calcineurin is one of the proteins whose activity is stimulated by increased intracellular Ca²⁺. Interestingly, calcineurin activity has been implicated in both LTD (Mulkey et al., 1994) and depotentiation (Jouvenceau et al., 2003) in the mammalian brain, as well as in some forms of extinction in mammals (Lin et al., 2003; Baumgartel et al., 2008). We do not yet know what downstream pathways are stimulated by calcineurin during LTH. One action linked with calcineurin activity in the mammalian brain is endocytosis of AMPA receptors, which is believed to be a key event in LTD (Beattie et al., 2000). Endocytosis of AMPA-type receptors could well be one effect of calcineurin activity during habituation of the GWR (see Glanzman, 2009), because exocytosis of AMPA-type receptors has previously been shown to play a role in synaptic facilitation and enhancement of the withdrawal reflex in Aplysia (Chitwood et al., 2001; Li et al., 2005, 2009).

The synaptic mechanisms that underlie LTH of the withdrawal reflex remain to be identified. One potential mechanism that is consistent with the data in this and our previous study (Ezzeddine and Glanzman, 2003) is LTD induced by elevated postsynaptic Ca²⁺ (Malenka and Bear, 2004). A form of LTD of the sensorimotor synapse that depends partly on elevated postsynaptic Ca²⁺ has been demonstrated in dissociated cell culture (Lin and Glanzman, 1996); but this form of synaptic depression has not yet been experimentally linked to LTH. A problem for the idea that activity-dependent LTD is a mechanism of LTH is that induction of activity-dependent LTD requires 1 Hz electrical stimulation, which is quite different from the rate of behavioral stimulation that results in LTH. Further complicating the issue of the synaptic mechanism of LTH is the demonstration by Montarolo et al. (1988) that homosynaptic activation of the sensorimotor synapse with a rate of stimulation designed to mimic the synaptic stimulation that occurs during LTH training fails to induce long-term (24 h) depression of the synapse. The apparent failure of naturalistic homosynaptic activity to induce LTD of the sensorimotor synapse suggests that homosynaptic activation of the sensorimotor pathway, by itself, may be insufficient for LTH in Aplysia. The possibility that LTH may depend, at least in part, on one or more heterosynaptic pathways is supported by the finding that spaced applications of the endogenous inhibitory neuropeptide, FMRFamide, can produce depression of the sensorimotor synapse that persists for ≥24 h (Montarolo et al., 1988). Both heterosynaptic, inhibitory (GABAergic), and homosynaptic pathways have been shown to play roles in habituation of the crayfish tailflip escape response (Krasne and Teshiba, 1995).

Habituation, like extinction, is the waning of a response due to repeated presentation of an unchanging stimulus. The major phenomenological distinction between habituation and extinction is that habituation is the waning of an innate response, whereas extinction is the waning of a conditioned response. Despite this distinction, the behavioral similarity between habituation and extinction has led to the speculation that these two forms of memory may share common mechanisms (McSweeney and Swindell, 2002). Although controversial, there is experimental support for this idea (Kamprath et al., 2006). We have found that LTH in Aplysia requires the activity of calcineurin and LVGCCs. Both of these processes have also been implicated in extinction (Lin et al., 2003; Barad et al., 2004; Suzuki et al., 2004, 2008; Cain et al., 2005; Baumgartel et al., 2008) (but see Schafe, 2008). Possibly, therefore, an understanding of the cellular mechanisms that mediate LTH of the GWR in Aplysia will contribute important mechanistic insights into the considerably more complex phenomenon of extinction.