Programs in Biological Science and Neuroscience, Gallo Center and Department of Neurology, University of California, San Francisco, UCSF School of Medicine, Emeryville, CA, USA
*S. L. McIntire, Programs in Biological Science and Neuroscience, Gallo Center and Department of Neurology, University of California, San Francisco, UCSF School of Medicine, 5858 Horton St. Suite 200, Emeryville, CA 94608, USA. E-mail: firstname.lastname@example.org
*S. L. McIntire, Programs in Biological Science and Neuroscience, Gallo Center and Department of Neurology, University of California, San Francisco, UCSF School of Medicine, 5858 Horton St. Suite 200, Emeryville, CA 94608, USA. E-mail: email@example.com
Memory and the expression of learned behaviors by an organism are often triggered by contextual cues that resemble those that were present when the initial learning occurred. In state-dependent learning, the cue eliciting a learned behavior is a neuroactive drug; behaviors initially learned during exposure to centrally acting compounds such as ethanol are subsequently recalled better if the drug stimulus is again present during testing. Although state-dependent learning is well documented in many vertebrate systems, the molecular mechanisms underlying state-dependent learning and other forms of contextual learning are not understood. Here we demonstrate and present a genetic analysis of state- dependent adaptation in Caenorhabditis elegans. C. elegans normally exhibits adaptation, or reduced behavioral response, to an olfactory stimulus after prior exposure to the stimulus. If the adaptation to the olfactory stimulus is acquired during ethanol administration, the adaptation is subsequently displayed only if the ethanol stimulus is again present. cat-1 and cat-2 mutant animals are defective in dopaminergic neuron signaling and are impaired in state dependency, indicating that dopamine functions in state-dependent adaptation in C. elegans.
Caenorhabditis elegans is able to modulate its olfactory responses based on experience. Worms can detect numerous volatile and soluble odorants that act as attractants or repellents in chemotaxis assays (Bargmann & Horvitz 1991; Dusenbery 1974; Ward 1973). Chemotaxis of a population of animals is determined in these assays by creating a gradient of an odorant from a point source, observing accumulation of animals at the point source over time and quantifying the strength of the behavioral response by calculating a chemotaxis index (CI, Fig. 1) (Bargmann & Horvitz 1991; Colbert & Bargmann 1995). Previous exposure to an odorant can result in a diminished behavioral response to the odorant, a process termed olfactory adaptation (Fig. 1) (Colbert & Bargmann 1995). These adaptive responses occur in an odorant-specific manner. Olfactory discrimination and adaptation may be important for C. elegans to identify food sources in a complex natural environment (Troemel 1999). If an odorant is omnipresent in a given environment, it may not be useful as a cue to a food source. Odorant-specific adaptation would allow C. elegans to ignore an uninformative odorant or modulate its olfactory responses based on experience.
C. elegans is a good model organism for studying simple learning behaviors and has been demonstrated to pair stimuli in several different paradigms (Colbert & Bargmann 1995, 1997; Ishihara et al. 2002; Rankin 2000; Saeki et al. 2001; Wen et al. 1997). We sought to determine whether or not state-dependent modification of olfactory adaptation could be observed after exposure to a drug. State-dependent effects of ethanol are widely appreciated in vertebrate systems (Goodwin et al. 1969; Lowe 1988; Overton 1966, 1972; Shulz et al. 2000), although the molecular mechanisms underlying state-dependency are not understood. We previously demonstrated a depressive effect of ethanol on multiple behaviors of C. elegans (Davies et al. 2003). Similar dose–response relationships were observed for the effects of ethanol on locomotion and egg laying behaviors. The intoxicating effects of ethanol occur at the same internal tissue concentrations that cause intoxication in humans. The similar depressive effects of ethanol on the behavior of C. elegans and other organisms led us to ask if the similarities in ethanol's effects could generalize to its ability to generate state-dependency.
State-dependent effects of ethanol on olfactory adaptation were observed for multiple odorants. We analyzed state-dependency in mutants that are defective in dopaminergic signaling, and found that dopamine is required for the generation of state-dependency in C. elegans. Neuroadaptive processes in dopamine neurons have been linked to alcohol dependence in vertebrate systems (reviewed in Weiss & Porrino 2002).
Materials and methods
Nematodes were maintained as described (Brenner 1974). Strains used were: Wildtype Bristol N2, cat-1(e1111), cat-2(e1112), glr-1(n2461).
Chemotaxis assay plates were prepared as follows: 10 ml of assay agar (2% agar, 5 mM KPO4[pH 6], 1 mM CaCl2, 1 mM MgSO4) were aliquoted into 10 cm Petri plates, and allowed to dry overnight. The day of the experiment, all plates were dried at 37 °C without lids for 1 h. If ethanol was to be used in the experiment, the contents of one representative plate were melted to determine volume, and 100% ethanol was added to the desired final concentration (generally 200 mM). Adding 95% ethanol to the plates to the desired final concentration generated similar results (data not shown). The plates were sealed with parafilm, and ethanol was allowed to equilibrate into the agar for 1.5–2 h. Animals were washed twice in S Basal (0.1 M NaCl, 0.05 M KPO4[pH 6], 5 mg/l cholesterol [in 100% ethanol]) and once in assay buffer (5 mM KPO4[pH 6], 1 mM CaCl2, 1 mM MgSO4). A spot of diluted odorant (for benzaldehyde, 1 µl of 1:200 benzaldehyde:EtOH, for butanone 1 µl of 1:1000 butanone:EtOH) was placed on one side of the plate. Exactly opposite was placed a spot of diluent (1 µl EtOH). To each spot was added the anaesthetic sodium azide (1 µl of 1 M Na Azide) to immobilize the worms when they reached a spot. Worms were considered to be at a spot during counting if they were within 1 cm of the center of the spot of odorant or diluent. Between 100 and 300 washed worms were then placed onto a spot on the plate that was exactly between the odorant and diluent spots and slightly off center (see Fig. 1), and excess liquid was wicked off. After 1 h, worms were counted, and a CI was calculated as follows: CI = (number of worms at the odorant spot – number of worms at the diluent spot)/number of worms in the assay. A high CI (close to 1) indicates that the odorant acted as a strong attractant, a lower CI indicates that the odorant was a less effective attractant or did not act as an attractant.
Adaptation plates were prepared as follows: 10 ml of adaptation agar (3% agar, 5 mM KPO4[pH 6], 1 mM CaCl2, 1 mM MgSO4) or assay agar (2% agar, 5 mM KPO4[pH 6], 1 mM CaCl2, 1 mM MgSO4) was aliquoted into 10 cm Petri plates, and allowed to dry overnight. The day of the experiment, all plates were dried at 37 °C without lids for 1 h, and the contents of a representative plate of each type was melted to determine volume. Ethanol was added to appropriate plates to the desired concentration, the plates were sealed with parafilm and allowed to equilibrate for 1.5 h. Odorant was aliquoted onto 5 agar plugs on the lid of adaptation plates (for benzaldehyde adaptation, 2 µl of 100% benzaldehyde was used; for butanone, 5 µl of 100% butanone was used). Typically, in our hands, butanone adaptation generated lower CIs than we generally generated to benzaldehyde. This occurred when animals were treated in the presence or absence of ethanol, indicating that this feature was intrinsic to butanone adaptation, and did not reflect a combination of effects of butanone and ethanol. Animals were washed twice in S Basal (0.1 M NaCl, 0.05 M KPO4[pH 6], 5 mg/l cholesterol [in 100% ethanol]) and once in assay buffer (5 mM KPO4[pH 6], 1 mM CaCl2, 1 mM MgSO4), then placed on an adaptation plate, and the plate was sealed with parafilm. Animals were incubated in all adaptation conditions for 90 min, then washed twice with S Basal and once with assay buffer and placed on chemotaxis plates. Chemotaxis was allowed to proceed for 1 h, worms were counted, and a CI was calculated. Adaptation was observed as a decrease in chemotaxis to an odorant after pretreatment with the odorant. Animals were considered to have adapted if the CI (adapted with no drug treatment, tested in the absence of drug) was less than 0.6. In all experiments in which animals were pretreated in the same way but tested in different conditions, populations were treated on a single pre-exposure plate, then washed and the population was divided only when it was placed onto different chemotaxis assay plates (Fig. 2).
The strength of olfactory adaptation in our experiments varied considerably from day to day. When we used all data from all experiments we found statistically significant state-dependency (Tukey post hoc test: P < 0.05). However, we found that the degree of olfactory adaptation strongly impacted the expression of state dependency, such that our olfactory adaptation assays naturally fell into two distinct classes. We observed a natural demarkation between the behavior of animals. On days when adaptation yielded very low CIs (less than 0.35), we noted that state-dependency was never total (see Table 1). On days when olfactory adaptation was more modest (CIs between 0.4 and 0.6), expression of state-dependency was more complete, such that there was no evidence of adaptation when animals were tested in the absence of ethanol (Fig. 2c). Thus, we chose to analyse the data in two groups: All experiments that yielded CI (adapted with no drug treatment, tested in the absence of drug) < 0.35 were considered to be ‘overadapted’, and were grouped together for analysis. The results of these experiments are shown in Table 1. All experiments that yielded CI (adapted with no drug treatment, tested in the absence of drug) 0.35 < CI < 0.6 were grouped together for analysis. The results of these experiments are shown in Fig. 2. For all subsequent analysis, we chose to use conditions that generated complete state-dependency in the wild-type control groups.
Table 1. : State-dependency in wild-type and mutant animals, and effects of overtraining on state-dependency in wild-type animals
Chemotaxis Indices (CIs) (± SEM)
Benz., benzaldehyde; But., butanone; n = number of experiments.
All values for chemotaxis tested on and off ethanol for animals pre-exposed to an odorant are significantly different from each other (P < 0.05) except for N2 pre-exposed to benzaldehyde and 50 mM ethanol, and cat-1 and cat-2 animals pre-exposed to benzaldehyde and 200 mM ethanol.
All numbers except overadapted were tested by t-test. Overadapted numbers were tested by anova with Tukey post hoc tests (see Materials and methods).
Values for overadapted animals are significantly different for animals adapted on ethanol and off ethanol tested off ethanol, and for animals adapted on ethanol and off ethanol tested on ethanol as tested by anova with Tukey post hoc tests (P < 0.05).
Concentration of ethanol was the same for pre-exposure and for chemotaxis assay when ethanol was present during the chemotaxis assay.
When only a pairwise comparison was made, we used two-tailed t-tests. When more than two treatments were compared, we used a one-way anova with Tukey post hoc tests.
We first characterized the acute effects of ethanol intoxication on chemotaxis. Although moderate slowing of locomotion was observed at an exogenous dose of ethanol of 200 mM, chemotaxis to the volatile odorant benzaldehyde was not inhibited by this treatment (Fig. 2a). As a result, we could analyze the effect of ethanol on olfactory adaptation. Adaptation to benzaldehyde is normally observed as a decrease in the chemotaxis response of the animals to the odorant after a 90-minute period of pre-exposure to a high dose of this attractant (Fig. 2b) (Colbert & Bargmann 1995).
In order to determine whether or not ethanol is capable of producing state-dependent effects, we treated animals with ethanol during pre-exposure to benzaldehyde and subsequently tested them in the presence or absence of ethanol (Fig. 2c). No benzaldehyde adaptation was observed when these animals were tested in the absence of ethanol. However, if the animals were tested in the presence of ethanol, normal benzaldehyde adaptation was evident, indicating that ethanol exposure during pretreatment did not prevent the animals from adapting to benzaldehyde. Rather, these data suggest that the olfactory adaptation is state-dependent: if ethanol is present during the adaptation phase then it must be present during the chemotaxis phase for the learning to be recalled. We noted that the state-dependency we observed is asymmetric; ethanol treatment only during the chemotaxis assay after adaptation had no significant effect on the degree of adaptation demonstrated (Fig. 2b). Asymmetry in state-dependency has also been observed in mammalian models of state-dependent learning (Overton 1987).
We sought to exclude other possible explanations for these observations. A prolonged exposure to ethanol in the absence of benzaldehyde did not itself result in an apparent adaptation to benzaldehyde or a decrease in chemotaxis (Fig. 2d). Sequential rather than simultaneous pre-exposures to ethanol and benzaldehyde also failed to produce state-dependency (for 20 minute ethanol exposure followed by 90 minute benzaldehyde exposure CI [benzaldehyde + ethanol] = 0.35 ± 0.18, CI [benzaldehyde – ethanol] = 0.18 ± 0.29 P > 0.05 [n = 3]; for 90 minute benzaldehyde exposure followed by 20 minute ethanol exposure CI [benzald-heyde + ethanol] = 0.32 ± 0.07, CI [benzaldehyde – ethanol]= 0.47 ± 0.09 P > 0.05 [n = 14]), suggesting that the olfactory stimulus must be presented simultaneously with the drug in order for the adaptive response to the odorant to become state-dependent.
Benzaldehyde olfactory adaptation has been reported to be disrupted following a strong centrifugation that appears to act as a dishabituating stimulus (Nuttley et al. 2001). We asked if the dependency of olfactory adaptation on continued ethanol treatment could reflect a disruption of olfactory adaptation where withdrawal from ethanol constitutes a non-specific dishabituating physiological stress. In the course of the state-dependent learning experiments, worms were washed in the absence of ethanol between the adaptation and chemotaxis steps for no longer than 15 min, so any withdrawal shock must occur within this interval. We tested for a withdrawal effect by allowing for a thorough 35 minute withdrawal from ethanol between the pre-exposure period and the subsequent chemotaxis assay (Fig. 2e). As before, adaptation was not expressed unless the animals were treated with ethanol during the subsequent chemotaxis assay period. This result suggests that the dependence on ethanol reflects an association of olfactory adaptation with ethanol rather a shock response to withdrawal from the drug.
Does the ethanol dependency of olfactory adaptation reflect the intoxicating actions of ethanol or other stimulus properties of the compound? Ethanol itself can act as a weak attractant in chemotaxis assays (Bargmann & Horvitz 1991). We tested an exogenous dose of ethanol (50 mM) that is insufficient to induce intoxication in C. elegans but acts as a volatile odorant (Davies et al. 2003). No effect on olfactory adaptation was observed (Table 1), suggesting that the relevant ethanol stimulus is its ability to change intoxication state, not its odorant properties. Doses of ethanol higher than 200 mM were also tested and shown to generate state-dependency (400 mM, data not shown), although these doses also produced greater impairment of locomotion which made the experimental results more difficult to interpret. The doses of ethanol required for generating state-dependent effects correlated with doses required for producing the neurodepressive effects, but not with lower doses that act only as olfactory stimuli.
State-dependency should result in a change in response to olfactory cues that are specifically paired with ethanol during the pre-exposure period rather than a generalized change in olfactory response. The volatile odorant butanone is sensed by the two AWC neurons, the same chemosensory neurons that sense benzaldehyde (Bargmann & Horvitz 1991). We demonstrated that state-dependency is not unique to benzaldehyde adaptation by showing state-dependent effects of ethanol on butanone adaptation (Table 1). Although benzaldehyde and butanone are sensed by the same neurons, adaptation to these odorants is odorant-specific; that is, adaptation to benzaldehyde does not result in a change in response to butanone. State-dependency also appears to be odorant-specific. Animals pretreated with benzaldehyde and ethanol do not show significant adaptation to butanone, with or without ethanol treatment during the chemotaxis assay (after adaptation to butanone + ethanol, CI [butanone + ethanol] = 0.29 ± 0.05 [n = 11]; after adaptation to benzaldehyde + ethanol, CI [butanone + ethanol] = 0.59 ± 0.03; P < 0.001 [n = 7]). While these results do not rule out the possibility that there is some small general effect of combined odorant and ethanol treatment on AWC function, they do preclude the possibility that state-dependency is due to overall AWC dysfunction.
During the course of these experiments, we noted that the degree of state-dependency we saw was strongly affected by the degree of olfactory adaptation we induced. In experiments in which animals responded extremely strongly to the pre-exposure stimulus such that they lost all attractive response to the odorant (CI < 0.35), state-dependency was significant but less robust (Table 1). In these cases, animals still showed significantly stronger adaptation when tested in the presence of ethanol, but adaptation was also observed in the absence of the drug. These results indicate that olfactory adaptation is not entirely state-dependent under these conditions, and that the animal can completely modify its response to an odorant state-dependently only if the odorant cue is not overwhelmingly large. This characteristic of state-dependency is shared by classic state-dependent learning paradigms; overtraining of the learned response appears to abolish state-dependent learning in mammals (Overton 1987).
In other systems, changes in dopaminergic function have been implicated in drug dependent behavioral changes and cue-conditioned responses (Berke & Hyman 2000). We analyzed mutants with known abnormalities in dopaminergic function to determine if dopamine is required for state-dependency. cat-1 mutants are defective in the monoamine vesicular transporter that packages dopamine and serotonin into synaptic vesicles (Duerr et al. 1999). cat-2 animals are deficient in the synthetic enzyme for dopamine, tyrosine hydroxylase (Lints & Emmons 1999). Norepinephrine and epinephrine, both of which are enzymatically derived from dopamine, have not been identified as neurotransmitters in C. elegans (Riddle et al. 1997). cat-1 and cat-2 mutant animals exhibit wild-type sensitivity to ethanol and show normal chemotaxis and adaptation to benzaldehyde (for cat-2: CI [benzaldehyde] = 0.95 ± 0.03 [n = 4], after adaptation to benzaldehyde CI [benzaldehyde] = 0.48 ± 0.07 [n = 4]; for cat-1: CI [benzaldehyde] = 0.88 ± 0.04 [n = 5], after adaptation to benzaldehyde CI [benzaldehyde] = 0.23 ± 0.07 [n = 4]). Intriguingly, both cat-1 and cat-2 animals are deficient in the state-dependent effect. Treatment of these mutant animals with ethanol during the benzaldehyde pre-exposure period had no significant effect on subsequent adapted responses to benzaldehyde observed in the absence of ethanol (Table 1). Abnormalities in state-dependency were not observed in animals deficient in glr-1, an AMPA type glutamate receptor subunit, that is known to play a role in another nematode learning paradigm, habituation to tap (Hart et al. 1995; Maricq et al. 1995; Rose et al. 2003) (Table 1). The abnormalities in cat-1 and cat-2 animals indicate a role for dopamine in the generation of state-dependency in C. elegans.
The finding of state-dependent effects on olfactory adaptation indicates that C. elegans is able to modify its response to an olfactory stimulus in the presence of a drug stimulus. Pre-exposure to an odorant and ethanol results in an ethanol-dependent change in the olfactory response to the odorant. The recall of the adaptation, or the expression of the changed response to the odorant, requires the same ethanol stimulus that was present when the adaptation to the odorant occurred. If the odorant is presented in the absence of the associated ethanol stimulus, the animal exhibits no decrease in response to the odorant, and acts as if it had not encountered the odorant previously.
The development of state-dependency is distinct from previously established associative learning phenomena in worms. Nuttley et al. (2002) showed that worms can form an association between olfactory adaptation and the absence of food. Rather than a simple association between olfactory adaptation and the presence of ethanol, we have observed an association between olfactory adaptation and an internal state of intoxication. The presence of a subintoxicating dose of ethanol is insufficient to generate this association. The state-altering properties of ethanol act as the stimulus recognized by the worm, and this change in internal state is what is being used as a cue in the development of the association with olfactory adaptation.
The association of olfactory adaptation with ethanol appears to require dopaminergic function. Dopamine is found in only eight cells in C. elegans. No direct synapses exist between the chemosensory neurons (in which olfactory adaptation is presumed to occur) and the dopaminergic neurons of C. elegans (Ward et al. 1975). Thus, the role for dopaminergic input suggests that state-dependency is not due solely to a direct effect of ethanol on the AWC neurons, but requires activation of a neural circuit.
Dopamine may play a role in learning in other systems. Midbrain dopaminergic neurons respond to natural rewards such as food and liquid as well as to conditioned reward-predicting stimuli (Schultz 1998). In vertebrate learning paradigms, dopaminergic neurons are active during the early learning phase of conditioned reinforcement, but cease to respond as the predicted stimuli and reward are repeated (Hollerman & Schultz 1998). These experiments have led to the hypothesis that dopamine could act during learning by signaling errors in the prediction of reward. A selective increase in dopamine release has also been demonstrated in the nucleus accumbens when an association is formed between two stimuli, neither of which independently affects dopamine release (Spanagel & Weiss 1999; Young et al. 1998). Hence, dopamine may more generally modulate learning that does not involve reinforcement. These experiments indicate that dopaminergic neuron activity correlates with learned behavioral changes. Our results now reveal that genetic elimination of dopaminergic function disrupts a simple form of learning in C. elegans.
Worms are able to modify their responses to stimuli by developing contextual associations in other paradigms. Rankin has shown that the habituation to tap response can be modified by the context in which the habituation regimen occurs (Rankin 2000). Worms will respond to a tap by moving away from the stimulus, but this response habituates in response to repeated tapping. If the habituation occurs in an environment with a contextual cue (an odorant), the habituation is strengthened if tested in the presence of the cue. This is somewhat reminicent of the development of state-dependency to olfactory adaptation, and may reflect a commonality in C. elegans between these two behaviors. Worms may have to signal salience of particular previously-encountered stimuli in their complex natural environment, and this could be accomplished by development of associations, represented, in these cases, as the development of associations between stimuli and external (contextual) cues or internal (state) cues. The mechanisms of the development of these associations may not overlap at the molecular level because these two behaviors do differ in their requirements for neurotransmitters. The glutamate receptor glr-1 is required for long-term habituation to the tap response (although it was not required for short-term habituation to the tap response) (Rose et al. 2003) but not for olfactory adaptation. Furthermore, dopamine is required for state-dependency and has not been reported to affect the contextual conditioning paradigm.
The effect of ethanol on olfactory adaptation in C. elegans has provided an opportunity to study the mechanisms underlying state-dependency in this genetically manipulable model. Previous studies of state-dependent learning have not elucidated the molecular mechanisms that are responsible for the generation of state-dependency. State-dependency may be important in addictive behavior. Chronic drug use and alcoholism are frequently linked to neuroadaptive processes within the dopaminergic system (Berke & Hyman 2000; Koob & Le Moal 1997; Phillips et al. 1998; Rubinstein et al. 1997; Spanagel & Weiss 1999). Ethanol alters the levels of dopamine in specific brain regions in mammals, and alterations in dopamine levels are thought to be important for the addictive properties of ethanol (reviewed in Weiss & Porrino 2002).
In both vertebrates and C. elegans, intoxicating doses of ethanol are required for state-dependent effects. State-dependent learning provides a mechanism whereby behavioral changes evolving during drug exposure become at least partially dependent on the drug for continued expression. Such a mechanism could contribute to the addictive properties of ethanol and other drugs of abuse.
We thank members of the McIntire and Bargmann laboratories for helpful discussions and comments on the manuscript, and Catharine Eastman for technical assistance. Strains used were provided by the Caenorhabditis Genetics Center (funded by the NIH National Center for Research Support). Support was provided by the State of California for medical research on alcohol and substance abuse through UCSF, and by the NIAAA, NIH (J.C.B).