The interstimulus interval and classical conditioning in the marine snail Hermissenda crassicornis

Comparative study of visuo-vestibular conditioning in Lymnaea stagnalis

In this review, we compare the current understanding of visuo-vestibular conditioning in Hermissenda crassicornis and Lymnaea stagnalis on the basis of behavioral, electrophysiologic, and morphologic studies. Paired presentation of a photic conditioned stimulus (CS) and an orbital rotation unconditioned stimulus (US) results in conditioned escape behavior in both species. In Hermissenda, changes in excitability of type B photoreceptors and morphologic modifications at the axon terminals follow conditioning. Caudal hair cells, which detect mechanical turbulence, have reciprocal inhibition with type B photoreceptors. In Lymnaea, the interaction between photoreceptors and hair cells is dependent on statocyst location. Furthermore, the organization of the Lymnaea eye is complex, with more than 100 photoreceptors distributed in a uniquely folded retina. Although the optimal conditions to produce long-term memory (memory persistent for >1 week) are almost identical in Hermissenda and Lymnaea, physiologic and morphologic differences suggest that the neuronal mechanisms underlying learning and memory are distinct.

Ionic Currents Underlying Difference in Light Response Between Type A and Type B Photoreceptors

In Hermissenda crassicornis, the memory of light associated with turbulence is stored as changes in intrinsic and synaptic currents in both type A and type B photoreceptors. These photoreceptor types exhibit qualitatively different responses to light and current injection, and these differences shape the spatiotemporal firing patterns that control behavior. Thus the objective of the study was to identify the mechanisms underlying these differences. The approach was to develop a type B model that reproduced characteristics of type B photoreceptors recorded in vitro, and then to create a type A model by modifying a select number of ionic currents. Comparison of type A models with characteristics of type A photoreceptors recorded in vitro revealed that type A and type B photoreceptors have five main differences, three that have been characterized experimentally and two that constitute hypotheses to be tested with experiments in the future. The three differences between type A and type B photoreceptors previously characterized include the inward rectifier current, the fast sodium current, and conductance of calcium-dependent and transient potassium channels. Two additional changes were required to produce a type A photoreceptor model. The very fast firing frequency observed during the first second after light onset required a faster time constant of activation of the delayed rectifier. The fast spike adaptation required a fast, noninactivating calcium-dependent potassium current. Because these differences between type A and type B photoreceptors have not been confirmed in comparative experiments, they constitute hypotheses to be tested with future experiments.

Paired turbulence and light do not produce a supralinear calcium increase in Hermissenda

The sea slug Hermissenda learns to associate light and hair cell stimulation, but not when the stimuli are temporally uncorrelated. Memory storage, which requires an elevation in calcium, occurs in the photoreceptors, which receive monosynaptic input from hair cells that sense acceleration stimuli such as turbulence. Both light and hair cell activity increase calcium concentration in the photoreceptor, but it is unknown whether paired calcium signals combine supralinearly to initiate memory storage. A correlate of memory storage is an enhancement of the long lasting depolarization (LLD) after light offset, which is attributed to a reduction in voltage dependent potassium currents; however, it is unclear what causes the LLD in the untrained animal. These issues were addressed using a multi-compartmental computer model of phototransduction, calcium dynamics, and ionic currents of the Hermissenda photoreceptor. Simulations of the interaction between light and hair cell activity show that paired stimuli do not produce a greater calcium increase than unpaired stimuli. This suggests that hair cell activity is acting via some other pathway to initiate memory storage. In addition, simulations show that a potassium leak channel, which closes with an increase in calcium, is required to produce both the untrained LLD and the enhanced LLD due to the decrease in voltage dependent potassium currents. Thus, the expression of this correlate of classical conditioning may depend on a leak potassium current.

Comparison of Hermissenda type a and type B photoreceptors: response to light as a function of intensity and duration

Hermissenda crassicornis is an invertebrate model used to study classical conditioning using light as the conditioned stimulus. The memory of the association is stored in type B photoreceptors, the output of which depends on interactions with type A photoreceptors. To understand the effect of classical conditioning on the output of type B photoreceptors in response to light, we measured the effect of light duration and intensity on membrane potential in both photoreceptor types of Hermissenda. The results show that, independent of light stimulus, the afterhyperpolarization is significantly greater in type A than in type B photoreceptors. In response to light, the generator potential (GP) rises linearly with an increase in either intensity or duration for both type A and type B photoreceptors. However, the difference between type A and type B photoreceptors depends on the time after light onset; the increase in peak GP with intensity is steeper in type A than type B, but by 14 sec after light onset, membrane potential is greater in type B than type A photoreceptors. Similarly, firing frequency increases with intensity and duration in both photoreceptor types but with a difference that is time dependent. During the first second after light onset, type A photoreceptors have a significantly higher firing frequency than type B photoreceptors; after this time, firing frequency is higher in type B than type A photoreceptors. Although membrane potential is correlated with firing frequency, this correlation is much lower in type A than type B photoreceptors, suggesting that some other conductance influences firing frequency in type A photoreceptors.

Associative learning acquisition and retention depends on developmental stage in Lymnaea stagnalis

Associative learning dependent on visual and vestibular sensory neurons and the underlying cellular mechanisms have been well characterized in Hermissenda but not yet in Lymnaea. Three days of conditioning with paired presentations of a light flash (conditional stimulus: CS) and orbital rotation (unconditional stimulus: UCS) in intact Lymnaea stagnalis results in a whole-body withdrawal response (WBWR) to the CS. In the current study, we examined the optimal stimulus conditions for associative learning, including developmental stage, number of stimuli, interstimulus interval, and intertrial interval. Animals with a shell length longer than 18 mm (sexually mature) acquired and retained the associative memory, while younger ones having a shell length shorter than 15 mm acquired but did not retain the memory to the following day. For mature animals, 10 paired presentations of the CS and UCS presented every 2 min were sufficient for the induction of a WBWR to the CS. Furthermore, animals conditioned with the UCS presented simultaneously with the last 2 s of the CS also exhibited a significant WBWR in response to the CS. Blind animals did not acquire the associative memory, suggesting that ocular photoreceptors, and not dermal photoreceptors, detected the CS. These results show that maturity was key to retention of associative learning.

The effect of intensity and duration on the light-induced sodium and potassium currents in the Hermissenda type B photoreceptor

Light duration and intensity influence classical conditioning in Hermissenda through their effects on the light-induced currents. Furthermore, the contribution of voltage-dependent potassium currents to the long-lasting depolarization in type B photoreceptors depends on light-induced currents active at resting potentials. Thus, the present study measures the effect of holding potential, duration, and intensity on the light-induced currents in discontinuous single-electrode voltage clamp mode. Three distinct current components are distinguished by their temporal and voltage characteristics and sensitivity to pharmacological agents. One current component is a transient sodium current, I(Nalgt); another is a plateau sodium current, I(plateau), which persists for the duration of the light stimulus. Substitution of trimethylammonium chloride for sodium reduces both currents equally, suggesting that I(plateau) represents partial inactivation of I(Nalgt). The third current component is a prolonged reduction in potassium currents, I(Klgt); it is accompanied by an increase in input resistance, and it appears at potentials close to rest. An increase in light duration or intensity causes an increase in the peak conductance of both I(Nalgt) and I(Klgt). Latency of I(Nalgt) is decreased by intensity, whereas rise time is increased by duration. An increase in light duration or intensity causes an increase in the time-to-peak and duration of I(Klgt). Characteristics of these currents suggest that I(Klgt) is responsible for the long-lasting depolarization seen after light termination, and thus plays a role in classical conditioning.

Neurophysiological substrates of context conditioning in Hermissenda suggest a temporally invariant form of activity-dependent neuronal facilitation

The neurophysiological basis for context conditioning is conceptually problematic because neurophysiological descriptions of activity-dependent (associative) forms of neuronal plasticity uniformly assume that a specific temporal relationship between signals is necessary for memory induction. In the present experiments, this problem is addressed empirically by presenting, as a temporally diffuse contextual signal, a stimulus that results in known neural modifications following punctate (temporally contiguous) pairings with an aversive unconditioned stimulus. Hermissenda were trained to discriminate between adjoining contexts that were distinguished only in that one was lit and one was dark. Thirty unsignaled rotations were presented during each of three 15-min sessions in one of the two (lit or dark) contexts. Prior to training, animals displayed a slight preference for the lit context. After exposure to unsignaled rotation, animal's preferences shifted strongly to the dark context if unsignaled rotations were presented in the light, and tended (nonsignificantly) to the lit context if unsignaled rotations were presented in the dark. The B photoreceptors of the Hermissenda eye undergo several forms of activity-dependent facilitation (e.g., an increase in neuronal input resistance and evoked spike frequency) following pairings of punctate light (CS) and presynaptic vestibular stimulation (US). Similar facilitation in the B photoreceptor was observed following in vitro training that mimicked context conditioning in which presynaptic vestibular stimulation was presented repetitively during a continuous 7.5-min light. Subsequently, Ca(2+)-imaging experiments were conducted with Fura-2AM. It was determined that intracellular Ca(2+), the CS-induced second messenger critical for the induction of activity-dependent facilitation, was elevated in the B photoreceptor throughout the 7.5-min light presentation. These results indicate that activity-dependent facilitation within similar neural structures can underlie learning about both temporally diffuse contextual stimuli and temporally punctate CS-US pairings. These results suggest that a common mechanism may underlie learning about diffuse contextual stimuli as well as punctate-conditioned stimuli, provided that the stimuli are processed similarly in each type of conditioning arrangement. Consequently, the expression of different responses to contextual and discrete stimuli are likely to reflect a higher property of the neural network, and do not necessarily arise from unique underlying mechanisms.

Associative learning of visual and vestibular stimuli in Lymnaea

A conditioned withdrawal response was characterized in the pond snail Lymnaea stagnalis. Using light as the conditioned stimulus and high-speed orbital rotation as the unconditioned stimulus, experimental animals were trained with 30 paired presentations of light and orbital rotation per day for 3 days. After training, all experimental animals responded to light with a withdrawal response, the conditioned response. Control animals exposed to the same number of explicitly unpaired presentations of light and orbital rotation, light alone, or no stimulation did not respond to light. Thirty paired presentations per day for 2 days produced less than optimal acquisition of the conditioned withdrawal response. Neither 45 paired presentations per day for 2 days nor 90 paired presentations for 1 day resulted in complete acquisition of the conditioned withdrawal response. The conditioned withdrawal response observed following 30 paired presentations per day for 3 to 5 days persisted to Day 10, regardless of the number of training days. As a measure of savings, reacquisition of the conditioned response after extinction was investigated. After the conditioned withdrawal response was extinguished, only 2 to 5 paired presentations of light and orbital rotation were required for reacquisition of the conditioned response for most animals. This study further establishes Lymnaea as an animal model of basic associative learning.

Associative learning in a network model of Hermissenda crassicornis. II. Experiments

A companion paper in a previous issue of this journal presented a resistance-capacitance circuit computer model of the four-neuron visual-vestibular network of the invertebrate marine mollusk Hermissenda crassicornis. In the present paper, we demonstrate that changes in the model's output in response to simulated associative training is quantitatively similar to behavioral and electrophysiological changes in response to associative training of Hermissenda crassicornis. Specifically, the model demonstrates many characteristics of conditioning: sensitivity to stimulus contingency, stimulus specificity, extinction, and savings. The model's learning features also are shown to be devoid of non-associative components. Thus, this computational model is an excellent tool for examining the information flow and dynamics of biological associative learning and for uncovering insights concerning associative learning, memory, and recall that can be applied to the development of artificial neural networks.

Classical conditioning and protein kinase C activation regulate the same single potassium channel in Hermissenda crassicornis photoreceptors

The patch-clamp technique was used to study the effects of classical conditioning and protein kinase C (PKC) activation on K+ channels of identified neurons in the snail Hermissenda crassicornis. Here we present evidence that classical conditioning and PKC activation similarly modify the same K+ channel. K+ channels were recorded in cells from animals with different training experience. The 64-pS K+ channel appeared with significantly lower frequency in the conditioned group compared to the frequencies in control animals (naive and unpaired). In addition, when present, the 64-pS channel exhibited a lower percentage of open time and an increased interval between opening bursts in cells from conditioned animals. The 42-pS K+ channel was observed with about the same frequency in all three groups, and its percentage of open time was invariant, regardless of the animal's experience. Incubation of the photoreceptor with the PKC activator phorbol 12,13-dibutyrate (PDBu) led to a profound decrease in the percentage of open time of the 64-pS K+ channel, from 35.7% in the control group to 2.5% in the PDBu-treated group. The inactive phorbol 4 alpha-phorbol 12-myristate 13-acetate had no effect. The use of the PKC inhibitor H-7 significantly blocked the phorbol effect. Inside-out patches obtained from phorbol preincubated cells likewise showed the same effect of PDBu on K+ channels, but the effect was not observed when phorbol was added after the cell-free patches were obtained from nontreated cells. By contrast, the percentage of open time of the 42-pS K+ channel remained unchanged after phorbol treatment.

GABA-induced potentiation of neuronal excitability occurs during contiguous pairings with intracellular calcium elevation

The temporal convergence of neuronal signals is commonly considered as a likely prerequisite for enhanced neuronal excitability underlying the induction of associative memories. Here we report that transmitter application on presynaptic terminals of the Hermissenda Type B photoreceptors, when paired with depolarization, results in a potentiation of the excitability of the B-cell which derives from an increase in input resistance across the B-cell soma membrane. Pressure microapplication of gamma-aminobutyric acid (GABA) (12.5 microM) on the terminal branches of the Hermissenda Type B photoreceptors results in the fast (less than 1 s) activation of an inward Cl- conductance, characterized by a decrease in neuronal membrane resistance and an accompanying hyperpolarization (3-6 mV) of the B-cell. A slower effect of GABA, characterized by a slight depolarization (2-4 mV) and increase in resistance was observed approximately 2 min after GABA application. Following bath application of the Cl- channel blocker picrotoxin (100 microM), this increase in resistance was observed within 20 s of GABA application, suggesting that it was normally masked by the faster Cl- conductance. The magnitude of the resistance increase in response to GABA was enhanced when the B-cell was held at depolarized membrane potentials (-40 to -20 mV), but was eliminated if Ca2+ was removed from the extracellular bath, or if the non-specific protein kinase inhibitor H7 (100 microM) was added to the extracellular bath. In a final experiment, GABA application was paired with a transient (10 s) depolarization of the B-cell (to -20 mV).(ABSTRACT TRUNCATED AT 250 WORDS)

Sequential changes of potassium currents in Hermissenda type B photoreceptor during early stages of classical conditioning

Classical conditioning of the marine snail Hermissenda can be produced in a single session of 50 pairings of light and rotation stimuli. Voltage clamp measurements of two outward K+ currents, IA and ICa2(+)-K+ were obtained from medial Type B photoreceptors that were isolated from the nervous system 1 day after animals were exposed to paired light and rotation stimuli or control procedures (Unpaired, or no exposure to light and rotation), ICa2(+)-K+ was found to be unchanged 18-30 h after 50 training trials. This result is consistent with a previous study where ICa2(+)-K+ was found to be unchanged after 50 light and rotation trials, although significantly reduced by 100 trials. In the present study 50 pairings of light and rotation produced a significant reduction in IA, suggesting an important role for this current in the earliest stages of classical conditioning.