Critical role of protein kinase G in the long-term balance between defensive and appetitive behaviors induced by aversive stimuli in Aplysia

Abstract

This study investigated the signaling cascades involved in the long-term storage of the balance between defensive and appetitive behaviors observed when the mollusk Aplysia is exposed to aversive experience. In Aplysia, repeated trials of aversive stimuli induce concurrent sensitization of defensive withdrawal reflexes and suppression of feeding for at least 24 h. This long-term storage of the balance between withdrawal reflexes and feeding is sustained, at least in part, by increased excitability of the tail sensory neurons (SNs) controlling the withdrawal reflexes, and by decreased excitability of feeding decision-making neuron B51. Nitric oxide (NO) is required for the induction of both long-term sensitization and feeding suppression. At the cellular level, NO is also required for long-term decreased B51 excitability but not for long-term increased SN excitability. Here, we characterized the signaling cascade downstream of NO contributing to the long-term storage of the balance between withdrawal reflexes and feeding. We found protein kinase G (PKG) necessary for both long-term sensitization and feeding suppression, indicating that a NO-PKG cascade governs the long-term storage of the balance between defensive and appetitive responses in Aplysia. The role of PKG on feeding suppression was paralleled at the cellular level where a cGMP-PKG pathway was required for long-term decreased B51 excitability. In the defensive circuit, the cGMP-PKG pathway was not necessary for long-term increased SN excitability, suggesting that other cellular correlates of long-term sensitization might depend on the GMP-PKG cascade to sustain the behavioral change.

In order to survive in a dynamic environment, animals need to balance the expression of their behaviors to varying experiences [1,2]. Following exposure to aversive stimuli, this balance manifests as a concurrent increase of defensive responses and decrease of appetitive behaviors [1,3]. Despite growing knowledge of the architecture of survival neural circuits [4–6], and evidence about their short-term interactions [3,7], little is known about the mechanisms underlying changes in neural circuits responsible for the long-term storage of the balance between defensive and appetitive behaviors.

In this study, we utilized a previously-developed paradigm in Aplysia in which repeated applications of noxious stimuli lead to a concurrent enhancement of the tail-induced siphon withdrawal reflex (a form of nonassociative learning known as sensitization) and suppression of the appetitive behavior of feeding for at least 24 h [8,9]. Recent findings identified nitric oxide (NO) as a modulator for the induction of both long-term sensitization (LTS) and long-term feeding suppression (LTFS)[10]. Here, we investigated the downstream targets of NO responsible for these modifications. We focused on the cGMP-PKG pathway because it is the main target of NO and contributes to learning-induced behavioral and cellular plasticity in vertebrates and invertebrates, including Aplysia [11–14].

In the first part of this study, we examined whether PKG activity is required for the training-induced long-term balance between the tail-induced siphon withdrawal reflex (TSWR) and feeding. We utilized the PKG inhibitor KT5823, which prevents forms of classical and operant learning in Aplysia [14,15]. Both TSWR and feeding were measured in the same animals prior to (pre-tests) and 24 h after training (post-tests; Fig. 1A), using previously-established testing and training methodologies [8,10,16]. Briefly, the TSWR was elicited by 20-ms AC electrical stimulation of low current intensity delivered via two electrodes implanted into one side of the tail [16]. Siphon withdrawal duration was used as a measure of TSWR strength. During pre-tests and post-tests (Fig. 1A), the duration of five siphon withdrawal responses was averaged and the change in TSWR duration was calculated as (post-test)/(pre-test)[8]. Feeding was assessed by counting the number of bites evoked during a 5-min exposure to a seaweed extract solution that reliably elicits biting (Fig. 1A)[8]. For each animal, the difference in bites was calculated as bites during post-test minus bites during pre-test [8].

Fig 1.

PKG activity is required for the induction of both LTS and LTFS following four-trial in vivo training. (A) The TSWR and feeding were measured prior to and 24 h after four trials of aversive training (lightning bolts). Injection of the PKG inhibitor KT5823 (KT) or vehicle is indicated by the syringe. KT5823 prevented both LTS (B) and LTFS (C). In this and in the following figures, values are expressed as mean ± SEM. Statistical significance was set at p < 0.05. In this and in the following figures, shared capitalized letters denote statistical similarity among groups.

After the pre-tests, animals were randomly assigned to be injected systemically with 1 mL per 100-g body mass of either KT5823 (6.5 μM in 0.65% DMSO), or 0.65% DMSO (vehicle control). The estimated final KT5823 hemolymph concentration was 100 nM, which prevented the induction of Aplysia long-term operant learning in a previous study [14]. One h after injection, animals were randomly assigned to receive an aversive training protocol consisting of four trials of electrical stimuli or serve as untrained controls [8,10]. The experimenter performing behavioral tests was kept blind to the training history of the animals. Four groups of 12 animals were utilized and compared: trained injected with vehicle (T-V), untrained injected with vehicle (UT-V), trained injected with KT5823 (T-KT) and untrained injected with KT5823 (UT-KT). Changes in TSWR duration and differences in bites were compared among the four groups using the Kruskal-Wallis test (H), followed by Student-Newman-Keuls pairwise post-hoc comparisons (q) to isolate the sources of significance [10].

Analysis of the TSWR revealed an overall statistical significance among the four groups (H₃ = 16.26; p < 0.05; Fig. 1B). Post-hoc analysis revealed that the T-V group was significantly greater than the UT-V (q = 5.14; p < 0.05), T-KT (q = 3.45; p < 0.05) and UT-KT (q = 5.69; p < 0.05) groups. The UT-KT group was significantly smaller than the UT-V (q = 4.11; p < 0.05) and the T-KT (q = 6.12; p < 0.05) groups. Because pre-test TSWR durations were not significantly different among the four groups employed (H₃ = 3.49; p = 0.32), we attribute the small reduction of TSWR duration in the UT-KT group to an unspecific effect of KT5823 on the TSWR duration. Within-group comparison of the pre-test and post-test TSWR duration data with the Wilcoxon signed-rank test also revealed significant increased TSWR duration in the T-V group (pre-test: 3.23 s ± 0.36; post-test: 5.59 s ± 1.05; W = 68.00; p < 0.05), but not in the T-KT group (pre-test: 2.81 s ± 0.24; post-test: 2.82 s ± 0.14; W = 8.00; p = 0.79). These results indicate that KT5823 prevented LTS induction.

Analysis of feeding indicates an overall statistical significance among the four groups (H₃ = 17.80; p < 0.05; Fig. 1C). Post-hoc analysis revealed that significant differences were observed only between the T-V group and the UT-V (q = 5.43; p < 0.05), the T-KT (q = 8.17; p < 0.05) and the UT-KT (q = 6.21; p < 0.05) groups. These findings indicate that KT5823 prevented LTFS induction.

Combined with previous results [10], the above findings describe the requirement of a NO-PKG dependent signaling cascade for both LTFS and LTS, thus identifying this pathway as a common modulatory system in the induction of the long-term balance between defensive and appetitive behaviors in Aplysia. These findings are consistent with previous studies reporting a role for the NO-PKG pathway in learning paradigms and learning-dependent plasticity in mollusks, including Aplysia [13–15], insects [12,17] and rodents [11]. In particular, in Aplysia, a NO-PKG pathway appears necessary for the induction of long-term downregulation of feeding by both associative [14] and nonassociative (this study) learning paradigms.

The identification of a NO-PKG signaling cascade necessary for the induction of LTS raises the issue of how this pathway fits into the current molecular model that sustains this form of long-term memory in Aplysia, which involves 5-HT-dependent recruitment of multiple kinases, including PKA and ERK [18]. Examples of NO modulation of PKA activity via a cGMP-PKG mechanism have been described in long-term memory formation in honeybees [12] and crickets [17]. In addition, it has been postulated that, in classical conditioning of siphon withdrawal in Aplysia, a NO-dependent pathway might work synergistically with 5-HT to amplify PKA activity [15]. However, future investigations are required to understand the means by which NO-PKG and 5-HT-PKA cascades interact to sustain LTS induction.

In Aplysia, LTS and LTFS are sustained, at least in part, by concomitant long-term changes in the corresponding neural circuits, which include increased excitability of the tail sensory neurons (SNs) controlling the TSWR and decreased excitability of feeding decision-making neuron B51 [9,16]. These two cellular correlates of LTS and LTFS can be induced in vitro in a reduced ganglion preparation by repeated electrical stimulation of afferent nerves that mimics aversive training in vivo [19]. In the second part of this study, we utilized this in vitro analog to investigate the contribution of the cGMP-PKG pathway to the induction of long-term increased SN excitability and long-term decreased B51 excitability. It was previously reported that NO is necessary for the induction of long-term decreased B51 excitability [10], thus suggesting that in vitro aversive training might recruit a downstream cGMP-PKG pathway for long-term B51 plasticity. Although long-term increased SN excitability is NO independent [10], it was still measured in this study to identify/rule out the contribution of the cGMP-PKG pathway to this form of plasticity.

The design of the long-term in vitro analog of the balance between TSWR and feeding has been extensively described previously [10,19]. Briefly, the pleural, pedal, cerebral and buccal ganglia were removed, with interganglionic connections and 2–3 cm of pedal nerves p8 and p9 retained [19]. Two-electrode current clamp was used to measure B51 resting potential, input resistance and excitability [20]. B51 excitability was assessed as the threshold to elicit a plateau potential, defined as the minimum amount of depolarizing current necessary to elicit a burst of activity that outlasted the duration of a 5-s pulse [20]. Burst threshold was measured by injecting 5-s depolarizing current pulses of incremental intensity at 10-s intervals, beginning at 5 nA, until the cell fired a plateau potential [20]. One-electrode current clamp was used to measure SN resting potential and excitability [19]. Excitability was determined by counting the number of action potentials elicited by a 1-s, 2-nA depolarizing pulse [19].

B51 and SN membrane properties were measured prior to (pre-tests) and 24-h after (post-tests) the administration of an in vitro aversive training protocol consisting of four trials of electrical stimuli delivered simultaneously to nerves P8 and P9 (Figs. 2A, 3A)[19,21]. Controls for training consisted of untrained preparations that did not receive electrical stimulation [10,19]. For each membrane property of each neuron recorded, the percent change was calculated as [(post-pre/pre) x 100] to assess modifications due to training and/or treatment [10].

Fig. 2.

sGC activity is necessary for training-induced long-term decreased B51 excitability, but not for long-term increased SN excitability. (A) Protocol of incubation with the sGC inhibitor ODQ or vehicle and subsequent measurements of SN and B51 properties prior to and 24 h after the four-trial in vitro training (lightning bolts). (B) Sample traces of B51 burst threshold. (C) Summary data illustrate that ODQ blocked the long-term decreased B51 excitability. (D) Sample traces of SN firing. (E) Summary data illustrate that ODQ did not block the long-term increased SN excitability.

Fig. 3.

PKG activity is necessary for training-induced long-term decreased B51 excitability, but not for long-term increased SN excitability. (A) SN and B51 membrane properties were measured prior to and 24 h after the four-trial in vitro training (lightning bolts). Treatment with KT5823 or vehicle was administered 10 min prior to the beginning of training (A). (B) Sample traces of B51 burst threshold. (C) Summary data illustrate that KT5823 blocked the long-term decreased B51 excitability. (D) Sample traces of SN firing. (E) Summary data illustrate that KT5823 did not block the long-term increased SN excitability.

The first experiment tested the contribution of the soluble guanylyl cyclase (sGC) to training-induced B51 and SN long-term excitability changes by using the selective blocker H-(1,2,4) oxidazole (4,3-a) quinoxalin-1-one (ODQ; Millipore Sigma). It must be noted that an in vivo experiment was initially conducted in which ODQ was systemically injected at a concentration (1 mL of 1 mM in 1% DMSO per 100-g body mass) that was previously used in the pond snail Lymnaea without any reported effects on behaviors including feeding [13]. However, 90% of the ODQ injections (but not DMSO alone injections) culminated with the release of ink and/or opaline, which is an indicator that the injection was perceived as an aversive stimulus that may have itself induced sensitization [8,16]. Even with the ODQ concentration decreased by half, the problem persisted. Therefore, we focused our investigation of ODQ effects on the long-term in vitro balance. We used a bath concentration of 25 μM ODQ, which is known to successfully inhibit sGC activity in SNs and B51 [22,23]. Ganglia were incubated in artificial seawater (ASW) containing either 25 μM ODQ in 0.1% DMSO, or 0.1% DMSO (vehicle control) 30 min prior to the pre-test (Fig 2A)[23]. After pre-test measurements, cells were labeled for post-test identification [10] and the ASW with ODQ/vehicle was exchanged with modified L15 medium [19] containing either ODQ or vehicle (15 min) for overnight ganglia storage. Ten min later, in vitro training was administered (Fig 2A). 24-h post-tests were conducted in ASW with ODQ/vehicle. The second experiment tested the necessity of PKG for training-induced long-term B51 and SN excitability changes. After pre-test measurements in ASW, cells were labeled for post-test identification and the ASW was exchanged with modified L15 medium (15 min; Fig. 3A) for overnight ganglia storage. A bolus of KT5823 (Calbiochem; final concentration: 2 μM in 0.1% DMSO) or vehicle (0.1% DMSO)[15] was then applied to the recording chamber 10 min before in vitro training (Fig. 3A). 24-h post-tests were conducted in ASW.

In both experiments, four groups of preparations were utilized and compared: trained treated with vehicle (T-V), untrained treated with vehicle (UT-V), trained treated with blocker (T-ODQ/KT) and untrained treated with blocker (UT-ODQ/KT). Statistical analyses were conducted using the Kruskal-Wallis test, followed by Student-Newman-Keuls post-hoc comparisons [10].

Results indicate that long-term decreased B51 excitability was prevented by blocking either sGC (Fig. 2) or PKG (Fig. 3). When ODQ was used to block the sGC, analysis of B51 excitability revealed an overall statistical significance among four groups of 11 cells (H₃ = 8.38; p < 0.05; Fig. 2B1–B4, 2C). Post-hoc analysis revealed that significant differences were observed only between the T-V group and the UT-V (q = 4.05; p < 0.05), the T-ODQ (q = 3.76; p < 0.05) and the UT-ODQ (q = 5.67; p < 0.05) group. No overall statistical significance was detected among the four groups for either B51 resting potential (H₃ = 1.47; p = 0.69) or input resistance (H₃ = 0.51; p = 0.92). When KT5823 was used to block PKG activity, analysis of B51 excitability revealed an overall statistical significance among four groups of 11 cells (H₃ = 9.18; p < 0.05; Fig. 3B1–B4, 3C). Post-hoc analysis revealed that significant differences were observed only between the T-V group and the UT-V (q = 3.64; p < 0.05), the T-KT (q = 6.62; p < 0.05) and the UT-KT (q = 4.57; p < 0.05) groups. No overall statistical significance was detected among the four groups for either B51 resting potential (H3 = 0.76; p = 0.86) or input resistance (H₃ = 3.13; p = 0.37).

These results indicate that both sGC and PKG are required for the induction of long-term decreased B51 excitability. Previous work reported the necessity of sGC in B51 decreased excitability induced by exogenous NO release and the sufficiency of increased cytosolic cGMP levels to sustain B51 decreased excitability for at least 24 h [23]. Our findings now demonstrate the recruitment of the NO-cGMP-PKG pathway in the induction of long-term plasticity of B51 excitability. One possible biophysical site of modulation by the NO-cGMP-PKG pathway might be the voltage-gated Na+ channels, which contribute, at least in part, to the long-term decreased excitability in B51 [24].

In contrast to decreased B51 excitability, blocking neither the sGC (Fig 2) nor PKG (Fig. 3) prevented the long-term increase in SN excitability induced by in vitro training. When ODQ was used to block the sCG, analysis of SN excitability revealed an overall statistical significance among the four groups of 13 cells (H₃ = 10.98; p < 0.05; Fig. 2D1–D4, 2E). Post-hoc analysis revealed that SN excitability increased in both T-V and T-ODQ preparations (T-V vs. UT-V; q = 4.12; p < 0.05; T-ODQ vs. UT-ODQ; q = 4.00; p < 0.05). The increased SN excitability was not significantly different between the T-V and T-ODQ groups (q = 2.50; p ≥ 0.05), indicating that ODQ did not block the induction of long-term increased SN excitability. Analysis of SN resting potential revealed no overall statistical significance among the four groups (H₃ = 3.10; p = 0.38). When KT5823 was used to block PKG activity, analysis of SN excitability revealed an overall statistical significance among the four groups of 13 cells (H₃ = 9.57; p < 0.05; Fig. 3D1–D4, 3E). Post-hoc analysis revealed that SN excitability increased in both T-V and T-KT preparations (T-V vs. UT-V; q = 3.75; p < 0.05; T-KT vs. UT-KT; q = 4.39; p < 0.05). The increased SN excitability was not significantly different between the T-V and T-KT groups (q = 1.41; p ≥ 0.05), indicating that KT5823 did not block the induction of long-term increased SN excitability. Analysis of SN resting potential revealed no overall statistical significance among the four groups (H₃ = 3.37; p = 0.34).

Confirming previous findings [10], these data highlight a divergence between behavioral and cellular results: despite the necessity of the NO-PKG pathway for the induction of LTS, long-term increased SN excitability induced by multiple in vitro aversive trials does not depend on either sGC or PKG activity. Conversely, a NO-cGMP-PKG pathway is known to mediate a form of long-term SN hyperexcitability caused in Aplysia by peripheral injury [22]. The nature of the noxious stimuli employed might be the discriminating factor that triggers different signaling cascades inducing otherwise phenomenologically similar forms of long-term increased SN excitability: repeated electrical shocks that do not lesion the skin [16] activate a 5-HT-cAMP-PKA pathway [18], whereas pinches with forceps causing tissue injury activate a NO-cGMP-PKG pathway [22].

The independence of the long-term increased SN excitability from the NO-PKG pathway, which is necessary for LTS, signifies that this signaling cascade might contribute to another form of plasticity within the TSWR circuit in order to sustain the enhancement of the siphon withdrawal reflex observed in LTS. This form of learning-induced plasticity is likely the long-term synaptic facilitation of the sensorimotor synapses mediating the TSWR, which is a well-established neural correlate of LTS [18]. Indeed, it has been shown in Aplysia that exogenous NO release facilitates sensorimotor synapses and that a NO-PKG cascade is required for classical conditioning of siphon withdrawal [15]. Future experiments will determine the involvement of a NO-PKG pathway in long-term synaptic facilitation following repetitive in vitro aversive training.

Highlights.

• Repeated aversive stimuli induce long-term feeding suppression (LTFS) and sensitization (LTS) in Aplysia

• We found that PKG is required for both LTFS and LTS

• cGMP-PKG signaling is required for decreased excitability of decision-making neuron B51 mediating LTFS

• cGMP-PKG signaling does not contribute to increased sensory neuron excitability mediating LTS