NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Riddle DL, Blumenthal T, Meyer BJ, et al., editors. C. elegans II. 2nd edition. Cold Spring Harbor (NY): Cold Spring Harbor Laboratory Press; 1997.

Cover of C. elegans II

C. elegans II. 2nd edition.

Show details

Section IIIBehavioral Plasticity in the Adult

Instead of studying the morphology of the nervous system, one can observe changes in the behavior of an animal. In fact, there are a number of ways that an organism can express behavioral plasticity without a morphological change. Neuromodulators can toggle between stereotyped behavioral states. For example, the egg-laying hormone of Aplysia elicits a complex behavioral program by exciting and inhibiting specific neurons throughout the nervous system (Scheller et al. 1982). Alternatively, the strengths of existing synapses can be altered as a result of experience. We present evidence of such complex behavioral changes in worms ranging from examples of neuromodulators to paradigms for learning and memory.

Neuromodulators can instantaneously alter the range of synapses that are active in any particular group of neurons. The modulator can thereby bias a pattern generator toward one of several output patterns. Neuromodulators can act globally as a circulating hormone or they can act on a specific cell via synaptic connections (Dickinson 1989). In invertebrates, aminergic neurotransmitters have been demonstrated to act humorally to control behavioral states. In the lobster, a high ratio of serotonin to octopamine initiates dominant and aggressive behavior, whereas a low ratio initiates submissive behavior (Kravitz 1990). Alternatively, the neurotransmitters can act locally via synaptic contacts to strengthen or weaken a second connection in a circuit. For example, serotonergic or FMRF-amidergic synapses from sensory neurons can, respectively, strengthen or weaken response of the sea slug Aplysia to mechanosensory stimuli (Byrne 1987; Montarolo et al. 1988). In the mollusk Tritonia, a central pattern generator produces rythmic output to the motor neurons that control swimming during escape; serotonergic input dynamically modulates synaptic strengths in this circuit during ongoing behavior (Katz et al. 1994).

A. Aminergic Neuromodulators

The best evidence for neuromodulators in C. elegans comes from studies of aminergic neurotransmission. The most common aminergic neurotransmitters in invertebrates are dopamine, serotonin, histamine, and octopamine. Although a number of behaviors are affected by raising or lowering levels of aminergic neurotransmitters in worms, it is not known whether these neuromodulators are acting humorally as neurohormones or as classical neurotransmitters at discrete synapses. However, in some cases, the actions appear so global that a humoral role seems likely.

Dopamine release might signal the presence of food to a well-fed nematode. Worms alter a number of behaviors upon exiting a lawn of bacteria. They become hyperactive, stop pharyngeal pumping, and do not activate the motor movements of the defecation cycle. Upon entering a bacterial lawn, wild-type worms slow, resume pumping, and resume motor movements of the defecation cycle. Some of these changes in behavior are probably mediated by dopamine transmission. Exogenous application of dopamine causes animals to become inactive, mimicking food abundance (Schafer and Kenyon 1995). Dopamine antagonists such as haloperidol cause animals to become hyperactive, mimicking food depletion. cat-2(e1112) mutants have low levels of dopamine (Sulston et al. 1975), and these animals are hyperactive in the presence of food in comparison to wild-type animals (B. Sawin, pers. comm.). However, they do not lack defecation cycles nor do they exhibit slow pharyngeal pumping, which are the other two behavioral changes that take place when a worm exits food. Laser killing of the dopamine neurons phenocopies the cat-2 defect and thus confirms the role of the dopaminergic nervous system in the suppression of the hyperactive behavior (B. Sawin, pers. comm.). It is not clear whether this behavior is mediated indirectly by a humoral effect or by the direct release of dopamine from sensory neurons onto postsynaptic elements of interneurons. However, the dopamine neurons lack neurosecretory ultrastructure, so the effects of dopamine may be mediated at synapses (White et al. 1986).

Serotonin may signal the presence of food to a hungry worm. Worms that have been removed from food for 30 minutes are very sensitive to the presence of food; upon entering a bacterial lawn, they stop swimming, initiate rapid pharyngeal pumping, begin laying eggs, and suppress the contraction of the enteric muscles when they reinitiate the defecation cycle (B. Sawin, pers. comm.; E. Jorgensen, unpubl.). Application of exogenous serotonin can also induce these behaviors even in a well-fed worm. Bath application of serotonin stimulates pharyngeal pumping (Croll 1975b; Avery and Horvitz 1990), induces egg laying (Trent et al. 1983), causes sluggish locomotion, and inhibits enteric muscle contractions (Ségalat et al. 1995). In addition, hungry worms have altered chemotactic behaviors, and these alterations are abolished by the application of serotonin (C. Bargmann, pers. comm.). cat-4 mutants lack serotonin and dopamine (Sulston et al. 1975; Desai et al. 1988; Weinshenker et al. 1995) and are defective for several of the behaviors that can be induced by serotonin application. These mutants pump slowly (Avery and Horvitz 1990) and are hyperactive (J. Kaplan; B. Sawin; both pers. comm.). Although these mutants are not egg-laying-defective, mutations in cat-4 can enhance egg-laying defects in other mutants (Avery et al. 1993). Finally, cat-2 mutants that express serotonin but not dopamine still have enhanced sensitivity to food after starvation, suggesting that dopamine is not required for this behavior (B. Sawin, pers. comm.).

Serotonin probably acts humorally as well as synaptically to mediate behavior. It is most intensely expressed in two pharyngeal motor neurons, the NSMs, which appear secretory by morphology (Albertson and Thomson 1976; Horvitz et al. 1982). Killing the NSMs along with other serotonergic cells phenocopies at least some of the cat-4 behavioral defects (B. Sawin, pers. comm.). Since NSM synapses are directed toward the nerve ring and do not directly synapse onto the cells that are likely to mediate these behaviors, the NSMs are probably acting at a distance. In other organisms, it is known that serotonin binds seven-pass transmembrane receptors that activate trimeric G-proteins. In C. elegans, serotonin acts via the Go GTPase. The α-subunit of Go is encoded by the gene goa-1 (Lochrie et al. 1991), and mutations in goa-1 disrupt serotonin signaling in several behaviors, including locomotion and defecation (Mendel et al. 1995; Ségalat et al. 1995). Together, these results suggest a model in which a starved worm has very low levels of circulating serotonin; as a consequence, the animal is hyperactive. When it enters the bacterial lawn, serotonin is released humorally, and this induces a number of appropriate behaviors, including active pumping, restricted movement, and egg laying.

Octopamine antagonizes the effects of serotonin in lobsters (Kravitz 1990), but its functions in C. elegans are relatively unexplored. Octopamine is found in C. elegans, but the cells expressing octopamine have not been identified (Horvitz et al. 1982). In contrast to serotonin application, octopamine causes loopy or kinked locomotion and depresses egg laying and pharyngeal pumping. Phentolamine, an octopamine antagonist in invertebrates, stimulates egg laying. Thus, serotonin and octopamine appear to act antagonistically in C. elegans as they appear to do in other invertebrates, but the basic cell biology of octopamine neurotransmission has not yet been investigated.

B. Peptidergic Neuromodulators

Neuropeptides can act as hormones, neuromodulators, or neurotransmitters (Krieger 1983). Although there is circumstantial evidence that neuropeptides can act independently and at a distance to modify the activity of many neurons, it is believed that most neuropeptides act in concert with a classical neurotransmitter to simply modify the output of the primary neurotransmitter on the postsynaptic cell (Cooper et al. 1991). This may make the study of neuropeptides by genetic methods rather difficult because mutants lacking a specific neuropeptide may have only subtle changes in behavior unrecognizable in the rather crude behavioral screens that are presently practical.

The actions of modulatory peptides are only beginning to be explored in nematodes. Studies so far have concentrated on a single family of peptides, the FMRFamide-like peptides (FLPs). When C. elegans is stained with an antibody that recognizes all members of this family, about 10% of the neurons express FMRFamide immunoreactivity, including motor neurons and interneurons (Schinkmann and Li 1992). Some of these peptides are likely to originate from the gene, flp-1 , that encodes multiple peptides ending with the amino acid sequence FLRF (Rosoff et al. 1992; see Rand and Nonet, this volume). One neuron class that expresses a FMRFamide-like peptide is the VC class of motor neurons. The VCs synapse to the ventral body muscles and the vulval muscles. Because egg laying requires the contraction of the vulval muscles, the effect of FLRFamide on egg laying was tested. Application of FLRFamide alone caused no change in egg laying, but it was capable of potentiating the induction of egg laying by serotonin. A genetic analysis of these peptides has been complicated by the discovery of at least four genes that could encode FMRFamide-like peptides (C. Li, pers. comm.).

An analysis of the role of peptides in nematodes has been more extensively carried out in the parasitic nematode Ascaris. Because of its large size, Ascaris is more amenable to electrophysiological analyses, and despite the disparity in size, the Ascaris and C. elegans nervous systems are remarkably similar (Stretton et al. 1985). In Ascaris, a large number of neuropeptides have been characterized by immunoreactivity (Stretton et al. 1991), and 12 peptides related to FMRFamide have been purified (AF1–12). Bioactivity of some of these peptides has been tested. AF1 abolishes the spontaneous oscillations of the ventral cord inhibitory motor neurons by reducing the input resistance of the membrane (Cowden et al. 1989). AF2 causes rhythmic contractions of the body muscle (Cowden and Stretton 1993), and AF4 induces continuous contraction of the body muscle (Cowden and Stretton 1995). Despite the extreme differences in lifestyle, the parasitic Ascaris and the free-living C. elegans nervous systems are remarkably similar in cell number, morphology, and neurotransmitter type. However, in C. elegans, less than 10% of the neurons are immunoreactive for FMRFamide-like peptides; in Ascaris, more than 60% of the neurons express FMRFamide-like immunoreactivity (Cowden et al. 1993). The differences in peptide distribution between these two species might be mechanisms to generate very different behaviors in nematode species that share an evolutionarily rigid nervous system (Stretton et al. 1991).

C. Sensory Adaptation

A simple form of behavioral plasticity as a result of experience is sensory adaptation. Sensory adaptation is the decrease or fatigue of sensory neuron response following prolonged exposure to sensory input. In most cases, the sensory response recovers after the stimulus has been removed. In C. elegans, responses to soluble compounds (taste) and responses to volatile compounds (olfaction) show a decrease following prolonged exposure (Ward 1973; Dusenbery 1980b; Colbert and Bargmann 1995). Similarly, worms raised at a specific temperature will avoid other temperatures, but after 2 hours, they no longer avoid these novel temperatures (Hedgecock and Russell 1975). Olfactory adaptation is selective, i.e., an animal will not move toward an adapted odorant, but it will still move toward a novel odorant, even when the two odorants are sensed by the same olfactory neuron (see Bargmann and Mori, this volume). Mutational analysis has shown that the molecular mechanisms for adaptation differ for different odorants. Colbert and Bargmann (1995) characterized two mutants that show normal chemotactic responses to volatile compounds but fail to adapt to different subsets of odors mediated by the AWC neurons: adp-1 mutants fail to adapt to benzaldehyde and butanone but adapt to isoamyl alcohol, and osm-9 mutants do not fully adapt to isoamyl alcohol or butanone but adapt normally to benzaldehyde.

D. Learning and Memory

Another way that behavior can change as a result of experience is through learning. Traditionally, theorists have divided learning into two categories: nonassociative and associative. Nonassociative learning occurs when an individual is exposed to a single type of stimulus and behavior is changed as a result of that exposure. Examples of nonassociative learning include habituation and sensitization. Like sensory adaptation, habituation is a simple decrement in response to a repeated stimulus, but it can be distinguished from sensory adaptation by a number of features (see below). Sensitization is an increase in response to a wide variety of stimuli following a noxious stimulus. Associative learning occurs when animals learn to link a stimulus or behavior with a second temporally associated stimulus. Associative learning includes classical conditioning and operant conditioning. The most prominent example of classical conditioning is Pavlov's experiments with dogs in which the animal learns to associate the ringing of a bell with food. In operant conditioning, an animal learns to associate one of its own behaviors with a stimulus. For example, in B.F. Skinner's classic operant conditioning experiments, a rat learns to press a lever for a reward of food.

Whereas learning is a change in behavior as a result of experience, memory is the ability to store and recall those changes to behavior. Research on both vertebrates and invertebrates has suggested that there may be a number of phases of memory (ranging from two in Aplysia to three in rats and birds to four in flies; for review, see DeZazzo and Tully 1995). Memory can last in these various phases from as short as seconds as is found in short-term memory or as long as hours to a lifetime as is found in long-term memory. The cellular and molecular mechanisms behind these phases of memory seem to be distinct. For example, long-term but not short-term memory can be disrupted by treatments such as electroconvulsive shock or inhibitors of protein synthesis (Davis and Squire 1984).

1. Habituation

Perhaps the simplest and most ubiquitous form of learning is habituation, which is a decrease in a response to a given stimulus after repeated trials. Observations of worms that bumped into glass beads (Croll 1975a) or that had been touched with a fine hair (Chalfie et al. 1985) demonstrated that the backing response declined with repeated mechanosensory stimulation. However, to distinguish this decrement in response from sensory adaptation or fatigue, a number of features of habituation must be observed (Groves and Thompson 1970). For example, habituation occurs more slowly with more intense stimuli or with longer interstimulus intervals. Habituation can be built up with repeated training sessions. Habituation, sensory adaptation, and fatigue all diminish gradually with time, but the rate of recovery from habituation depends on the interstimulus intervals of training. Finally, only habituation can be rapidly abolished with the application of a novel or noxious stimulus in a phenomenon known as dishabituation.

One simple stimulus that can be used to study habituation in C. elegans is a controlled tap to the side of the petri dish (Rankin et al. 1990). Such a tap causes an animal that is motionless or moving forward to move backward. As taps are repeated, the average distance a worm moves backward decreases. Following habituation training, an electrical stimulus delivered to the agar on either side of the worm causes dishabituation, i.e., the shock restores the normal response to tap.

In other organisms such as Aplysia (Rankin and Carew 1987) and rat (Davis 1970), the speed and degree of response decrement are dependent on the interstimulus interval. Similarly, worms rapidly habituate to stimuli delivered at 10-second intervals but slowly and less completely to stimuli delivered at 60-second intervals (Fig. 4) (Rankin and Broster 1992). Spontaneous recovery from habituation is also dependent on the interstimulus interval. Worms recover more rapidly from habituation induced by short interstimulus intervals than they do from long interstimulus intervals (Rankin and Broster 1992). These data indicate that the different intervals are recorded in the nervous system at least 1 hour after the delivery of the last stimulus and can continue to influence behavior differentially. They also suggest that habituation to short and long stimulus intervals may recruit different cellular mechanisms.

Figure 4. Habituation in C.

Figure 4

Habituation in C. elegans. (A) Habituation of backing to trains of taps for 30 stimuli delivered at a 10-sec interstimulus interval (ISI). (B) Habituation (more...)

Work with other organisms (e.g., Aplysia; Carew and Kandel 1973) has shown that with repeated habituation training sessions, two things happen: (1) a build up of habituation over blocks of trials and (2) the possibility of the formation of long-term memory. A paradigm used to study long-term habituation in Aplysia was modified and applied to C. elegans (Beck and Rankin 1995). In this paradigm, worms were given three blocks of 20 stimuli each and stimuli within blocks were delivered every 60 seconds, with an hour rest between blocks. To determine whether the training produced long-term habituation, these same worms were given a block of 20 stimuli at 60-second intervals on the second day. The results showed that there was a build up of habituation over the course of the three training blocks on the first day. In addition, the worms were capable of long-term memory as demonstrated by retaining memory of habituation training for at least 24 hours. This long-term memory was disrupted by heat shock during the rest intervals (Beck and Rankin 1995). The current hypothesis is that heat shock disrupts cellular processes such as protein synthesis that are necessary for memory formation (Davis and Squire 1984).

Although habituation is a form of learning found in many organisms, surprisingly, little is known about the cellular processes underlying this form of learning. Studies in Aplysia (Bailey and Chen 1983) have suggested that there may be a decrease in the amount of neurotransmittor available for release in terminals that have undergone habituation training. Although studies in C. elegans have not yet progressed to the molecular level, C. elegans with its simple nervous system may offer new insights into the cellular mechanisms underlying habituation. Two interacting neural circuits are activated during response to head and tail touch (Chalfie et al. 1985; see Driscoll and Kaplan, this volume): Stimulation of the tail touch receptors activates the interneurons that direct the worm to move forward, and stimulation of the head touch receptors activate the interneurons that direct the animal to move backward. In response to tap, these neural circuits are activated simultaneously and the behavior results from an integration of two competing outputs (Wicks and Rankin 1995).

Given that the observed response to tap in intact animals is actually an integration of two competing responses, what is the response of each of the competing circuits alone to habituation training? The two circuits do not produce the same pattern of behavioral outputs in response to habituation training (Wicks and Rankin 1996). When the posterior touch neurons (PLM) are killed, the circuit that moves the animal backward can be viewed in isolation. In the operated animals, the reversals habituate more slowly and less completely than they do in the wild-type worms. Ablation of the anterior touch neurons ALM and AVM produce animals that respond to tap by accelerating forward. When such ablated animals are given habituation training, a different pattern of response was observed. With short interstimulus intervals (10 seconds), accelerations first increase in magnitude (sensitization) before habituating. With long interstimulus intervals (60 seconds), there is no evidence of sensitization. Again, the data suggest that habituation in unoperated animals is the result of a balance of two competing behaviors: reversals and accelerations. For example, with short interstimulus intervals, animals rapidly habituate within the first few stimuli. This rapid habituation may reflect the increased input of the sensitized accelerations that decrease the magnitude of the reversals. As habituation continues, reversals become infrequent and the animals often accelerate forward in response to tap. Presumably, the rapid habituation of reversals and the slower habituation of accelerations are integrated as a net movement forward. These results suggest that there might not be a single mechanism underlying habituation, nor might all cells involved in a behavior respond in the same way to repeated stimulation. Instead, each cell type may have a unique response to repeated stimulation, and the behavior that is observed is the integrated output of all of the cell types.

In the future, genetic analyses of the mechanisms involved in the long- and short-term memory phases of habituation should lead to additional insights into the similarities and differences between memory processes in this simple nervous system and in more complex organisms such as Drosophila, Aplysia, and mammals.

2. Sensitization

A second form of nonassociative learning is sensitization (Groves and Thompson 1970), which refers to the increase in reflexive responses due to the application of a noxious stimulus. Sensitization is not a form of associative learning because the stimulus is not specifically paired with another stimulus. The stimulus merely raises the arousal level of the animal so that all reflex pathways are facilitated. Sensitization in C. elegans has been demonstrated in several ways, but it has not been investigated to the same extent as habituation. Sensitization was first shown by presenting worms with a single tap, then a stronger stimulus, in the form of trains of taps, and then looking at the response to a single tap again (Rankin et al. 1990). Worms showed larger responses to the single tap following the train of taps than they did to the initial single tap.

3. Associative Learning

In the simple nonassociative forms of learning, an animal alters its behavior to a single stimulus; in contrast, in associative learning, an animal learns to use a previously neutral stimulus to predict the presence or absence of a second more significant stimulus. C. elegans is capable of this more advanced form of learning as demonstrated by several different paradigms. Examples of associative learning by C. elegans come from classical conditioning paradigms in response to chemosensory stimuli (J.Y.M. Wen et al., in prep.). In this discriminative classical conditioning assay, one ion is associated with food and a second ion is associated with the absence of food; the conditioned animals will then selectively migrate to the ion paired with food. First, adult hermaphrodites are deprived of food for 5 hours, and then an ion, either sodium or chloride, is presented to the animals with bacteria for the first hour and the other ion is presented without bacteria for a second hour. In the test phase, the animals are then given a choice between diffusive gradients of sodium and chloride for 1.5 hours. The results show that conditioned animals display significant preference for the ion paired with food and that the preference lasts up to 7 hours after training (Fig. 5). Switching the order of the conditioning stimuli did not affect the results. Presentation of the ion paired with no food before presentation of the ion paired with food resulted in identical degrees of learning.

Figure 5. Classical conditioning in the C.

Figure 5

Classical conditioning in the C. elegans wild type and in lrn-1 and lrn-2 mutants. (more...)

Learning can be assayed in individual animals as well. In this paradigm, worms are conditioned in liquid medium in test tubes containing solutions of the ions and Escherichia coli. To test for learning, individuals are placed on test plates with a gradient of each of the ions, and the initial heading of the worm is assayed. Conditioned animals show initial headings similar to their final accumulation; thus, learning can be assayed within 30 seconds, rather than waiting the 1.5 hours required in the chemotaxis assay.

C. elegans can also learn aversive associations. In this type of experiment, an ion, such as sodium or chloride, is paired with a noxious stimulus, and thereafter the worms avoid the conditioned ion. For example, adult hermaphrodites can first be conditioned with an ion associated with an aversive stimulus such as garlic. Subsequently, when tested in a chemotaxis assay, these animals avoided the ion that had been paired with garlic (Fig. 5C). Aversive learning can also be tested in individuals as well as in populations. In this kind of experiment, worms are exposed to an attractive ion and the aversive stimulus, for example, copper ion. To test for learned aversion, conditioned worms are placed on a spot containing the paired or unpaired ion in the absence of copper, and the time required for the animal to leave the spot is measured.

The eventual goal is to employ genetic techniques in C. elegans to identify molecules essential for learning and memory. Having established that C. elegans shows discriminative classical conditioning in a variety of paradigms, van der Kooy and colleagues screened for mutants defective in associative learning (J.Y.M. Wen et al., in prep.). They isolated two lines of ethylmethanesulfonate (EMS)-induced learning-deficient mutants that show normal chemosensory responses but no evidence for classical conditioning in any of the discriminitive classical conditioning paradigms. These mutations define two loci, lrn-1 and lrn-2 (learn).

Another possible case of associative learning is a phenomenon first described by Hedgecock and Russell (1975) who demonstrated not only that C. elegans could detect thermal gradients and selectively migrate along an isothermic contour, but also that such thermotaxis could be modified by experience. The temperature at which the animals were raised and fed determines the temperature to which they migrate. A brief starvation of 2 hours induces strong dispersion from the starvation temperature; a 4-hour period of starvation decreases these responses (Hedgecock and Russell 1975; Mori and Ohshima 1995). Although it is possible that dispersion from the starvation temperature involves a learned association between a specific temperature and a lack of food, new data indicate that starvation may simply suppress thermotaxis (see Bargmann and Mori, this volume).

Copyright © 1997, Cold Spring Harbor Laboratory Press.
Bookshelf ID: NBK20011


  • PubReader
  • Print View
  • Cite this Page

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...