Two pairs of tentacles and a pair of procerebra: optimized functions and redundant structures in the sensory and central organs involved in olfactory learning of terrestrial pulmonates

The Pulmonata is a subclass of the class Gastropoda in the phylum Mollusca. It contains several orders, such as Stylommatophora (Limax, Achatina, Helix, etc.), Basommatophora (Lymnaea, Helisoma, etc.) and others (Klussmann-Kolb et al., 2008). We focus here on the Stylommatophora, a group of terrestrial pulmonates, i.e. land slugs or snails, possessing slender eye-bearing stalks. Terrestrial pulmonates have two pairs of tentacles: superior and inferior pairs. Their eyes are at the tips of the superior pair, but both pairs are equipped with sensory epithelia. In some studies, the superior and inferior tentacles are also called the posterior and anterior tentacles, respectively. Although the tentacles subserve the detection and transduction of visual, olfactory and mechanosensory stimuli to the central nervous system, terrestrial pulmonates seem to employ olfaction as their primary means of perceiving the external world (Chase, 2002).

Terrestrial pulmonates acquire mammalian-quality learning with odors as conditioned stimuli (CSs) and consolidate what they have learned into long-term memory (Gelperin, 1975; Matsuo et al., 2002). For example, in Limax, olfactory-aversion learning occurs when an odorant CS is paired with an aversive unconditioned stimulus (US) (Gelperin, 1975; Sahley et al., 1981). A natural food odor, such as that of carrot or cucumber, is used as the CS whereas quinidine sulfate, carbon dioxide or electrical shock is used as the US. A long-term memory is formed after a single paired presentation of a CS and a US, and it is retained for more than several weeks (Gelperin, 1975; Matsuo et al., 2002; Matsuo et al., 2010a).

To elucidate the cellular and molecular mechanisms underlying learning and memory in terrestrial pulmonates, researchers have so far focused on the functions of the tentacles and procerebra (e.g. Gelperin, 1999; Watanabe et al., 2008; Matsuo and Ito, 2011). However, we still face the question of why terrestrial pulmonates inevitably possess two pairs of tentacles and a pair of procerebra. In a certain sense, such structures appear futile for the life of pulmonates. Here we exemplify and discuss both the functional optimization and structural redundancy of two pairs of tentacles and a pair of procerebra involved in olfactory learning by terrestrial pulmonates, in particular Limax (Fig. 1). Before starting our commentary, we acknowledge that much of the work of Ronald Chase serves as a solid foundation for the present commentary.

Pulmonates use their chemical senses to find food and to avoid danger, regardless of whether they live on land or in water (Gelperin, 1974). Several locomotive patterns are elicited by chemical senses, e.g. food finding, homing and escape. Chemotaxis is the phenomenon in which pulmonates direct their movements in concentration gradients of certain chemicals, such as odors or pheromones, in their environment. Chemotaxis is called positive if the movement is in the direction of a higher concentration of the chemical in question, such as a chemical in food, and negative if the movement is in the direction of a weakening concentration, such as a source of danger.

Pulmonates use their tentacles primarily to determine the orientation to distant food sources (Croll, 1983) (Figs 1 and 2). Experiments in which chemical stimulants are applied locally to the tentacles and in which receptors in the tentacles are histologically counted showed that tentacles have a special chemosensitivity (Croll, 1983; Emery, 1992; Chase and Tolloczko, 1993; Cummis et al., 2009). Lesion experiments on tentacles have also demonstrated this chemosensitivity: early research by Chase and Croll (Chase and Croll, 1981) showed that the removal of either or both of the superior tentacles prevents the terrestrial snail Achatina from locating food.

The superior tentacles are extremely sensitive to stimulation by winds. In the terrestrial snail Achatina, gentle air puffs to the superior tentacle evoke transient discharges in the tentacular ganglion at both stimulus onset and stimulus offset (Chase, 1981). In Limax, a gentle wind (without odor) induces electro-olfactogram oscillations in the olfactory epithelium of the superior tentacle, and this oscillation affects olfactory information processing downstream (Ito et al., 2006). Therefore, the tentacles function not only as olfactory organs but also as mechanosensors, and there are interactions between these two sensory modalities, such as chemo-sensing and mechano-sensing. In fact, the presence of wind enhances the accuracy of the chemotactic behavior of insects (Willis et al., 2008).

Eyes are present only in the superior tentacles, but their functions seem to be used only for phototaxis. Limax use their eyes for negative phototactic behavior (Crozier and Federighi, 1924; Matsuo et al., 2002; Yamagishi et al., 2008). Optic nerves send projections to the cerebral ganglion, although the nerve bundles are narrower than the olfactory pathways in Helix (Ovchinnikov, 1986). Taken together, these results indicate that the two pairs of tentacles of terrestrial pulmonates serve as versatile sensory structures to detect multimodal environmental stimuli.

Because the tentacles are present as bilateral pairs, pulmonates can show tropotaxis. This is the movement of a pulmonate towards or away from a stimulus as the pulmonate compares sensory inputs from the paired receptors on both sides of the body. However, the tentacles can also function independently on each side. This enables klinotaxis, a locomotive mode based on a sampling of external stimuli successively in time during pulmonate movement.

Chase and Croll designed experiments using unilateral lesions of the superior tentacles to determine which of these strategies Achatina uses (Chase and Croll, 1981). The results depended on environmental conditions. When the snails were required to orient themselves in the absence of any wind, they needed the pair of superior tentacles, implying the use of tropotaxis. Because the comparison of olfactory information from the bilateral tentacles is a prerequisite for tropotaxis, these results indicate that the two sides of the olfactory system exchange olfactory information, probably via the cerebral commissures. However, when orienting to an upwind odor source, a single superior tentacle is sufficient, indicating that the snails use klinotaxis in the presence of mechano-stimulation.

A new insight into klinotaxis was recently obtained from Drosophila. In Drosophila larvae, a unilateral olfactory organ is sufficient for chemotactic locomotion along an odor concentration gradient, probably via klinotaxis. But bilateral olfactory inputs enable the larvae to navigate in more complex, challenging environments, such as in a shallow, linear spatial concentration gradient of odorant molecules (Louis et al., 2008). Louis et al. proposed that bilateral olfactory inputs are integrated into the neural coding spikes with a higher signal-to-noise ratio (by a factor of √2) than the case of a unilateral input (Louis et al., 2008). Although it is currently unclear where the bilateral olfactory inputs are integrated in the brain of the slug (or even in Drosophila larvae), a similar enhancement mechanism might work for odor detection in terrestrial pulmonates.

As described earlier, terrestrial pulmonates only use their eyes for phototaxis. Terrestrial pulmonates also use a tropotactic strategy to avoid light places. They seem to compare the intensities of visual inputs from the right and left eyes, and move towards the darker side. If Limax has a unilateral eye removed, it rotates in the direction of the removed eye (Crozier and Federighi, 1924). Similarly, in Helix, if its superior tentacle is anesthetized unilaterally, it often rotates towards the direction of the amputated side (Friedrich and Teyke, 1998). This behavior may be another example of tropotaxis in pulmonates (i.e. negative phototaxis). The possession of bilateral tentacles is obviously a prerequisite for a tropotactic strategy of locomotion, whether it is driven by olfactory or visual information.

The sensory structure of a tentacle includes a sensory epithelium, a tentacular ganglion and a tentacular nerve connecting the tentacular ganglion to the cerebral ganglion (Fig. 3). The sensory epithelium is at the tip of the tentacle. The tentacular ganglion is beneath the sensory epithelium, and it ramifies into digit-like extensions over the sensory epithelium. Studies using silver-stained sections of the tentacles or rhinophores of gastropods (Lane, 1962; Emery and Audesirk, 1978) revealed that bipolar sensory neurons are distributed beneath the sensory epithelium. The majority of these bipolar neurons form synaptic connections (i.e. synaptic glomeruli) with neurons whose cell bodies are located in the digits of the tentacular ganglion, whereas only ∼10% of the axons of these bipolar neurons travel directly to the cerebral ganglia (Chase, 2002).

Retrograde staining of tentacular nerves showed that many sensory neurons and interneurons in the digits of the superior tentacles have axons that terminate directly in the cerebral ganglion (Chase and Tolloczko, 1993). Most of the olfactory inputs from sensory neurons are synaptically relayed at the tentacular ganglion and enter into the procerebrum whereas mechanosensory afferents are thought to directly project into the metacerebrum, a distinct part of the cerebral ganglion (Rogers, 1971; Ierusalimsky and Balaban, 2010). The procerebrum, therefore, serves as a third-order level of olfactory information processing (see A pair of procerebra: Structure and function of procerebra).

Although there are no essential differences between the repertoires of the types of neurons in the superior and inferior tentacular ganglia (Chase and Kamil, 1983a; Chase and Kamil, 1983b; Ito et al., 2000), Ito et al. found that there are four structural differences between the two pairs of tentacles by means of backfilling of the tentacular nerves: (1) fewer fibers form the thick fiber tracts in neuropils in the superior tentacles than in the inferior tentacles; (2) there are more stained neurons in the superior tentacles than in the inferior tentacles; (3) the ratio of stained neurons to total neurons is lower in the superior tentacles than in the inferior tentacles; and (4) as with the digits, the tentacular ganglion is partially covered by a single cell layer in the inferior tentacles whereas the superior tentacles lack such a cell layer.

The anatomical projection patterns to the cerebral ganglion are roughly identical between the superior and inferior tentacles (Chase and Tolloczko, 1993). However, Ierusalimsky and Balaban (Ierusalimsky and Balaban, 2010) recently demonstrated that there are also some differences in the way superior and inferior tentacles project into the cerebral ganglion of Helix. The entire neuropil area (the terminal mass layer) of the procerebrum receives projections from the tentacular nerves (superior tentacle) whereas only the distal one-third of it receives projections from the medial lip nerves (inferior tentacle). These anatomical differences might explain the functional and physiological differences between the superior and inferior tentacles described below.

Thus far, it has been shown that the superior pair is useful for sensing airborne chemical signals and for directing the pulmonate's locomotion whereas the inferior tentacles are particularly important for following mucus trails and other chemical cues that lie on the substrate. If Achatina has either one or both of its superior tentacles amputated and the air is still, it cannot locate food placed distantly (Chase and Croll, 1981). Thus, the superior tentacles are necessary for tropotaxis. However, the mucus tracing behavior is abolished by inferior tentacle amputation but not by superior tentacle amputation (Chase and Croll, 1981). This suggests that the two pairs of tentacles function differently in the behavior of terrestrial pulmonates, although both serve as chemosensory organs.

Friedrich and Teyke demonstrated in Helix that superior tentacles are necessary and sufficient for locating food to which the snails have been appetitively conditioned whereas the inferior tentacles, but not the superior tentacles, are necessary and sufficient for the acquisition of food-attraction learning (Friedrich and Teyke, 1998). This result is another example of the functional difference between the two pairs of tentacles. Although both pairs can potentially detect the odor molecules derived from food during conditioning and memory retrieval, and can potentially access the procerebrum, where memory formation is thought to take place, the results of Friedrich and Teyke (Friedrich and Teyke, 1998) indicate that each pair is specialized to the distinct pulmonate behavior related to learning and memory.

For the most part, the procerebrum receives olfactory inputs indirectly via the tentacular ganglion, and the oscillatory frequency of the local field potential (LFP) in the procerebrum is modulated by olfactory inputs to the tentacle (Gervais et al., 1996; Kimura et al., 1998b). To determine the primary mechanism for the frequency changes of the LFP in the procerebrum, the tip, middle and basal regions of the digits of the superior and inferior tentacles were electrically stimulated and the LFP was recorded from the procerebrum of Limax (Ito et al., 1999). Stimulation of the middle and basal regions of the digits of the inferior tentacle significantly decreased the frequency of LFP whereas stimulation of the tip region of the digits of the inferior tentacle and of all regions of the digits of the superior tentacle increased the frequency of LFP. These findings suggest that the change in frequency of LFP in the procerebrum depends on the excited region in the digits of the superior and inferior tentacles, providing the physiological differences in olfactory function between the superior and inferior tentacles. These results may explain the previous behavioral data concerning the different roles of the superior and inferior tentacles (Suzuki, 1967; Cook, 1985; Chase, 1986; Friedrich and Teyke, 1998; Kimura et al., 1998a). In particular, Kimura et al. showed that the inferior tentacles play an important role in determining odor preference, because the odor presentation to the inferior tentacle can increase or decrease the frequency of LFP oscillations in the procerebrum, depending on whether the odor is appetitive or aversive to the slugs (Kimura et al., 1998a). The mechanism underlying this frequency change may be used by slugs that are conditioned to food odor.

To understand the neural circuits in the tentacles, LFP oscillations in the digits and the tentacular ganglia were analyzed in Limax. For example, the changes in LFP oscillations were recorded from the tentacular nerves after odor stimulation to the sensory epithelium. Recordings from the inferior tentacular nerves (medial lip nerves) connected to the inferior tentacular ganglia and sensory epithelia show spontaneous oscillatory activity, primarily at frequencies of 0.6–6 Hz (Ito et al., 2001). Similarly, ca. 1.5 Hz oscillatory activity was recorded from the neuropil region of the tentacular ganglion of the superior tentacle (Inokuma et al., 2002). Further, the application of various odors (garlic, carrot and rat chow) changed the oscillatory properties of tentacular ganglion recorded from the superior tentacular nerve in Limax (Ito et al., 2003a). For example, the appearance of high-amplitude spontaneous oscillations was accompanied by a decrease in odor-evoked spike activity in the superior tentacular nerve and the appearance of odor-evoked spike activity was accompanied by a decrease in the amplitude of spontaneous oscillations, demonstrating a significant negative correlation between spontaneous oscillations and odor-evoked spike activity (Ito et al., 2003a). These results suggest that the intrinsic oscillatory activity contributes to olfactory processing in pulmonates in a manner different from that reported in mammalian olfactory bulbs (Adrian, 1950).

To analyze the dynamics of odor-processing circuits in the digits and tentacular ganglia, Ito et al. (Ito et al., 2004) studied the effects of γ-aminobutyric acid (GABA), glutamate and acetylcholine – which are neurotransmitters present in the tentacular ganglion (Ito et al., 2003b) – on the circuit dynamics of the oscillatory network(s) in the superior and inferior tentacular ganglia. Application of GABA to the cell masses decreased the tentacular nerve oscillation whereas the application of glutamate and acetylcholine to the digits increased the oscillations recorded at the digits. These results suggest that there are two intrinsic oscillatory circuits that respond differentially to neurotransmitters in the tentacular ganglion in Limax. However, no differences have been found in the responsiveness to these neurotransmitters between the superior and inferior tentacles.

Recently, the metacerebrum of the cerebral ganglion has been identified as one of the central loci of tentacular inputs in another terrestrial slug, Incilaria (Makinae et al., 2008). The stimulation of the superior tentacular nerves activates the medial and lateral halves of the medial neuropil region of the metacerebrum almost evenly whereas the stimulation of the inferior tentacular nerves activates the lateral half of this region more strongly than the medial half. Makinae et al. also demonstrated that the activation of the metacerebrum precedes that of the procerebrum by ca. 50 ms when a superior or an inferior tentacle is electrically stimulated, implicating the existence of at least two types of fibers with different conduction velocities within both the tentacular and medial lip nerves (Makinae et al., 2008). Together with the fact that the metacerebrum receives direct sensory inputs from the mechanosensory neurons located in the sensory pad (Ierusalimsky and Balaban, 2010), the differential timing of the activation of the metacerebrum and the procerebrum may reflect the different modalities of the inputs, i.e. the mechanosensory and olfactory inputs.

We have gained an understanding of the physiological and morphological differences between the superior and inferior tentacles. We thus turn our attention to the role of each pair of tentacles in olfactory-aversion learning. In previous studies, slugs have been easily conditioned using carrot juice and 1% quinidine sulfate solution as the CS and the US, respectively (Sahley et al., 1981; Nakaya et al., 2001; Matsuo et al., 2002). In these conditioning experiments, the slugs crawl towards carrot juice put in their path. When the slugs touch the carrot juice, quinidine sulfate solution is applied to the mouth. The memory retention test is performed later with the carrot juice. As described below, we examined the pole of each pair of tentacles by its amputation in the olfactory-aversion experiments.

A previous study showed that the amputation of inferior tentacles after conditioning degrades memory retrieval whereas the amputation of superior tentacles has no such effect, suggesting that the olfactory inputs from the inferior tentacles are important for the retrieval of aversive memory (Kimura et al., 1999). However, several aspects of that study needed to be evaluated again. We thus re-examined the olfactory ability of Limax following tentacle amputation (Yamagishi et al., 2008). The results showed that memory formation was not altered by the amputation of either of the pairs before or after olfactory-aversion learning; likewise, the odor sensibility of Limax was maintained. These data suggested that either pair of tentacles is sufficient for the acquisition and retrieval of aversive olfactory memory.

Although the data by Yamagishi et al. demonstrated that the two pairs of tentacles are functionally redundant with respect to olfactory-aversion learning (Yamagishi et al., 2008), do slugs have two pairs only in preparation for injury? Can each pair substitute for the other's function, whatever that may be? In Achatina, each pair of tentacles serves a function in some tasks other than olfactory-aversion learning: trail following is exclusively dependent upon the inferior tentacles whereas orientation towards distant odor source depends on the superior tentacles (Chase and Croll, 1981). Further, in food-attraction learning in Helix, the acquisition of olfactory memory requires sensory inputs conveyed by the inferior tentacles whereas the recall of memory requires intact superior tentacles, indicating the functional specialization of each pair of tentacles in appetitive learning (Friedrich and Teyke, 1998).

Such inconsistencies with Yamagishi et al. (Yamagishi et al., 2008) might be explained by differences in the learning paradigm used in the studies, that is, aversive versus appetitive conditioning. Another explanation is that the manner of tentacle inactivation was different: Friedrich and Teyke acutely anesthetized either pair of tentacular epithelia with lidocaine (Friedrich and Teyke, 1998) and Kimura et al. also ‘acutely’ amputated either pair of tentacles (Kimura et al., 1999) whereas Yamagishi et al. amputated either pair of tentacles 7 days before the behavioral experiments (Yamagishi et al., 2008), allowing for functional compensation by the remaining pair of tentacles, if any. Tentacle amputation also affects the cautiousness of the slug. Slugs that have had their superior tentacles amputated tend to be very cautious in approaching any odorant sources (Kimura, 2000; Yamagishi et al., 2008). Such changes in behavior make it difficult to interpret the behavioral data in learning experiments. However, the data of Yamagishi et al. at least demonstrated that slugs could manage to acquire and retrieve olfactory-aversion memory in the absence of either pair of tentacles (Yamagishi et al., 2008).

The procerebrum is one of the loci that receive olfactory inputs from the tentacles. The procerebrum is a division of the cerebral ganglion unique to terrestrial pulmonates and is specialized to the processing of olfactory information (Ratté and Chase, 1997; Ratté and Chase, 2000; Chase, 2000) (Fig. 4). The number of neurons present in the combined left and right procerebra is estimated to be ca. 10⁵ in Limax, comprising approximately half of all the neurons in the brain (Gelperin and Tank, 1990; Chase, 2000). The neurons are small – 5 to 12 μm in diameter – and densely packed, with their neuropils extending into the core region within the procerebrum.

The olfactory function of the procerebrum is indicated by the fact that it receives axons of the tentacular nerve and the medial lip nerve from the superior and inferior tentacular ganglions, respectively (Chase and Tolloczko, 1993; Kawahara et al., 1997; Ierusalimsky and Balaban, 2010). Procerebrum neurons also receive direct inputs from olfactory sensory neurons, bypassing the tentacular ganglion (Chase and Tolloczko, 1993), indicating that the procerebrum serves not only as a secondary olfactory center but also as a primary olfactory center. Recently, however, Ierusalimsky and Balaban reported that such a bypassing pathway is not observed in the olfactory ascending pathway in Helix and that few such direct pathways are mechanosensory tracts (Ierusalimsky and Balaban, 2010).

Injection of dyes into the cell bodies of procerebrum neurons showed that some cells possess long neurites that extend into the contralateral side of the cerebral ganglion as well as into the mesocerebrum in the ipsilateral cerebral ganglion (Ratté and Chase, 1997; Ratté and Chase, 2000). Because these neurites have mostly output synapses on their distal processes, whereas they have mostly input synapses near the cell body, the centrally projecting cells probably carry processed outputs from the procerebrum to motor control neurons, including those responsible for food finding or locomotion. In fact, some of the procerebrum neurons send outputs to the pedal ganglion as well as to the other regions of the cerebral ganglion, and the membrane potential in these neurons synchronizes with the LFP in the procerebrum (Chase and Tolloczko, 1989; Ratte and Chase, 1997; Gelperin and Flores, 1997; Shimozono et al., 2001).

The procerebrum consists of three layers: the cell mass, terminal mass and internal mass layers (Fig. 5). The cell mass layer contains a large number of cell bodies of small neurons. The terminal mass layer and internal mass layer are neuropil layers. The procerebrum neurons receive inputs from the tentacular nerves at the terminal mass layer (Kawahara et al., 1997; Matsuo et al., 2010c). They also receive serotonergic and Phe-Met-Arg-Phe-NH₂ergic (FMRFamidergic) projections (Inoue et al., 2004; Kobayashi et al., 2010), as well as GABAergic projections (S.K., R.M. and E.I., unpublished observation), in the terminal and/or internal mass layers in Limax, suggesting that the procerebrum is under complex neuromodulatory controls from the outside (Gelperin et al., 1993).

The procerebrum spontaneously produces a periodic slow oscillation of LFP (ca. 0.7 Hz), even in the absence of olfactory inputs (Gelperin and Tank, 1990). The LFP oscillation is produced by the concerted actions of two types of neurons that are thought to be the major constituents of the procerebrum. A small percentage of procerebrum neurons fire periodic bursts of action potentials (bursting neurons) (Kleinfeld et al., 1994; Watanabe et al., 1998). These bursting neurons are slightly larger than nonbursting neurons, which constitute the majority of neurons in the procerebrum (Watanabe et al., 1998). The nonbursting neurons are synaptically inhibited by the bursting neurons via glutamatergic inputs in Limax (Watanabe et al., 1999; Matsuo et al., 2009). The potential oscillations arise from the summed synaptic currents in the nonbursting neurons triggered by inhibitory glutamatergic inputs from bursting neurons whereas the phase gradient along the longitudinal length of the procerebrum is caused by differences in the excitability of bursting neurons in the apical and basal regions (Watanabe et al., 2003).

The precise function of the procerebrum and the role of its oscillations have been controversial. In some experiments, stimulation of a tentacle with odors produces changes in spontaneous oscillations, including amplitude and frequency effects (Gelperin and Tank, 1990; Gervais et al., 1996; Kimura et al., 1998b; Nikitin and Balaban, 2000; Cooke and Gelperin, 2001; Samarova and Balaban, 2007; Samarova and Balaban, 2009). In addition, odor stimulation causes a collapse of the phase gradient across the procerebrum in the optical recordings; that is, the oscillations momentarily cease to propagate as a wave (Delaney et al., 1994; Gervais et al., 1996). Schütt et al. reported that several odors each cause specific changes in the oscillations, with each odor producing oscillations at a characteristic peak frequency (Schütt et al., 1999).

Despite all these reports of the effects of odor on spontaneous procerebrum oscillations, suppression of the oscillations does not completely eliminate odor-evoked responses in efferent cerebral nerves (Teyke and Gelperin, 1999). This last result, with others, led Gelperin (Gelperin, 1999) to propose that the procerebrum is primarily a learning machine, it is more involved in odor discrimination than in odor recognition and the function of its oscillations is to spatially segregate learned odor representations. Supporting this notion, Teyke and Gelperin (Teyke and Gelperin, 1999) and Sakura et al. (Sakura et al., 2004) demonstrated that suppression of the procerebrum's oscillatory activity by an inhibitor of nitric oxide synthase, which has been vigorously investigated in Limax (Gelperin et al., 1996; Gelperin et al., 2000; Fujie et al., 2002; Fujie et al., 2005; Matsuo et al., 2008; Matsuo and Ito, 2009), abolishes the ability to discriminate similar odors even while the detection of odor itself remains intact in Limax. In many studies, the mere application of an odor did not cause a frequency change (Kimura et al., 1998a; Inoue et al., 2006; Samarova and Balaban, 2009). It is only the odors that have some meaning to pulmonates that can elicit frequency changes in LFP oscillation; that is, only odors that the pulmonates approach or avoid can induce a frequency change in the procerebrum. Samarova and Balaban recently showed that the timing of the frequency change in the procerebrum coincides with decision making (approach or escape) in freely moving snails (Samarova and Balaban, 2009).

In 2006, evidence suggested that the procerebrum is essential for olfactory-aversion learning in Limax (Kasai et al., 2006). In a series of olfactory-aversion learning experiments, the paired procerebra were bilaterally lesioned 7 days prior to conditioning. Most of the lesioned slugs did not avoid carrot juice (CS) in the memory retention test performed 1 day after the conditioning. That is, either the acquisition or retrieval of memory was abolished by lesioning of the bilateral procerebra. The slugs whose procerebra were lesioned after conditioning also did not avoid carrot juice in the memory retention test, even if the procerebra were lesioned 7 days after the conditioning. These results clearly demonstrate that the procerebrum is a necessary component in the retention and/or retrieval of olfactory-aversion memory in Limax. However, Kasai et al. also demonstrated that procerebrum lesion does not damage odor-sensing ability because procerebrum-lesioned slugs can avoid innately aversive odorants such as garlic or onion and yet can approach their everyday food (Kasai et al., 2006).

A subsequent investigation, using an experiment to exploit the spontaneous recoverability of the procerebrum from injury, revealed that the procerebrum is the storage site of olfactory-aversion memory and is not a mere conductor of neuronal information about memory stored elsewhere (Matsuo et al., 2010a). Other experiments (Matsuo et al., 2010a) showed that the surgically lesioned procerebrum spontaneously regenerates within a 1-month recovery period through neurogenesis from neuronal progenitor cells remaining within the procerebrum (Zakharov et al., 1998; Watanabe et al., 2008), and that the procerebrum substantially recovers its volume. The regenerated procerebrum also restores the oscillatory activity of LFP. The slugs can achieve olfactory learning after recovery but they cannot retrieve olfactory memory acquired prior to surgery (Matsuo and Ito, 2008; Matsuo et al., 2010a). These results suggest that surgical damage to the procerebrum irreversibly extinguishes the memory stored in the procerebrum, and support the notion that the procerebrum is the storage site of olfactory-aversion memory (Gelperin, 1999).

However, to date it has been hypothesized that only the unilateral procerebrum is used for olfactory-aversion learning. This is because the unilateral labeling of procerebrum neurons was observed after an injection of Lucifer Yellow into the body cavity following conditioning in Limax (Kimura et al., 1998b; Ermentrout et al., 2001; Sekiguchi et al., 2010). Although the precise meaning of this band-shaped labeling is still elusive, Sekiguchi et al. recently proposed a model with two layered oscillators (neuropil and cell mass layers) to explain this phenomenon (Sekiguchi et al., 2010), and considered this band to be the most strongly depolarized region during conditioning by the synchronized activity in the two layers. However, their model did not refer to laterality in the labeling of the procerebrum.

Matsuo et al. recently presented data supporting the notion of unilateral memory storage through unilateral procerebrum lesion experiments (Matsuo et al., 2010b). The number of slugs with intact memory performance was reduced by ∼50% after the procerebrum was surgically lesioned unilaterally before or after the conditioning. There were no differences in the memory performance of slugs whose right versus left procerebrum was lesioned. Those authors also showed that there was no lateral memory transfer from one procerebrum to the other up to 7 days after the conditioning. The results demonstrated that either the left or right procerebrum is randomly used for olfactory learning and memory storage, and that the side of use is determined at the level of the olfactory ascending pathway to the procerebrum. This is explained by the fact that the unilaterally lesioned slugs can learn normally if the tentacles ipsilateral to the lesioned procerebrum are amputated at the same time, indicating that there is some competitive mechanism between the right and left olfactory inputs before they reach the procerebrum.

Unilateral tentacle amputation experiments (Matsuo et al., 2010b) suggested that the procerebrum ipsilateral to the remaining tentacle is always used for memory storage. These results also support the view that the olfactory projection to the procerebrum from the tentacles is primarily ipsilateral (Chase and Tolloczko, 1989; Matsuo et al., 2010c). In the presence of tentacles on only one side, the slugs inevitably use the procerebrum on the same side for memory acquisition and storage. This means that the slug brain somehow monitors the state of olfactory inputs from both sides, and one side of the olfactory system can take over the mnemonic function in case the other side is functionally disabled.

It is uncertain how the brain decides which side of the procerebrum to use for processing and storing memory in the presence of bilateral tentacles. If this decision is made before olfactory inputs reach the procerebrum, the metacerebrum might play a role in this process because the timing of sensory inputs arrival to this region precedes that to the procerebrum (Makinae et al., 2008). Otherwise, lateral inhibition between the right and left procerebra might result in the unilateral use of the procerebrum in a winner-take-all manner through an unidentified mechanism (Teyke et al., 2000).

Morphological and neuroethological analyses revealed that the superior and inferior tentacles differ from each other so that each pair can function in optimized and specialized ways. In the case of injury, however, either pair can substitute for the function of the other in olfactory-aversion learning tasks. This might be enabled by the overall similarity in the structure and cellular components between the superior and inferior tentacles. Such flexibility is, however, restricted to the case when either tentacle is damaged, because selective damages to the unilateral procerebrum result in the reduction of learning performance in that half without compensation by the remaining procerebrum (Matsuo et al., 2010b). In this sense, only the tentacles have functional redundancy for olfactory learning. This may be because the tentacles, appendages serving as multimodal sensory organs, are more prone to suffer from damage occurring in a pulmonate's daily life in comparison to the procerebra, which are protected in the pulmonate body. However, both the tentacle and the procerebrum have an exquisite ability to regenerate spontaneously after a recovery period. We do not yet have a clear explanation why the pulmonates use only a unilateral procerebrum for mnemonic function every time the slug learns. The doubling of memory capacity is one possibility (Gelperin, 1999). Further, the existence of right and left procerebra may provide greater abilities to express tropotaxis by processing information conveyed from bilateral tentacles.

 
• Cerebral commissures

  Nerve bundles connecting the right and left cerebral hemiganglia of pulmonates. Serotonergic, GABAergic, glutamatergic and dopaminergic nerves pass within these bundles (Matsuo et al., 2009; Makino and Yano, 2010).

 
• Chemotaxis

  An animal's directional movement strategy based on an external chemical substance. An animal may move towards (positive chamotaxis) or away from (negative chemotaxis) the chemical source.

 
• Conditioned stimulus

  A chemical or physical stimulus that evokes a response acquired by Pavlovian classical conditioning. Usually, a conditioned stimulus does not trigger any response in an animal before learning. However, it evokes a behavioral response after the simultaneous (or temporally close) presentation of another stimulus (an unconditioned stimulus) that always evokes that behavioral response.

 
• Klinotaxis

  An animal's behavioral strategy during directional movement. In klinotaxis, an animal determines its locomotive orientation by successively sampling the external environmental cues, thereby monitoring its own ongoing movement. In contrast to tropotaxis, in which the animal monitors spatial gradient of environmental cues, the animal monitors temporal change in klinotaxis.

 
• Local field potential

  Extracellular electrical potential reflecting the membrane potential changes of the cells near the tip of a recording electrode. This potential change is produced by the sum of the ionic currents passing through the membranes of the neighboring cells.

 
• Metacerebrum

  A part of the cerebral ganglion of a terrestrial pulmonate. The metacerebrum is situated at the posterior part of the cerebral ganglion near the cerebropleural and cerebropedal connectives, and constitutes the major part of the cerebral ganglion. Motor neurons, sensory neurons and interneurons are all contained in this structure (for details, see Chase, 2000).

 
• Phototaxis

  An animal's directional movement strategy based on light. In positive phototaxis an animal moves towards a light source whereas in negative phototaxis an animal moves away from the light source.

 
• Procerebrum

  A specialized structure in the cerebral ganglion of a terrestrial pulmonate consisting of ca. 10⁵ small interneurons involved in olfactory information processing. This structure is absent from aquatic pulmonates. Procerebral neurons receive direct and indirect olfactory inputs from tentacles and send outputs to various brain structures including the pedal ganglia. The procerebrum is involved in higher olfactory functions such as odor discrimination and olfactory learning.

 
• Tropotaxis

  An animal's behavioral strategy during directional movement. In tropotaxis, an animal determines its locomotive orientation by comparing the sensory inputs received from bilateral sensory organs.

 
• Unconditioned stimulus

  A chemical or physical stimulus that always evokes a behavioral response in an animal. It has some significance for the animal's survival in most cases. Once an unconditioned stimulus is presented (in some cases repeatedly) in combination with a neutral stimulus (a conditioned stimulus) that does not trigger any response, the animal learns that the conditioned stimulus is a predictive sign of the unconditioned stimulus, and the former evokes some response (a conditioned response) after learning.