Psychopharmacology

, Volume 205, Issue 3, pp 475–487

Delta-9-tetrahydrocannabinol enhances food reinforcement in a mouse operant conflict test

Authors

  • Maria Flavia Barbano
    • Departament de Ciencies Experimentals i de la SalutUniversitat Pompeu Fabra, PRBB
    • Instituto de Investigaciones BiotecnológicasUniversidad de General San Martín, INTI
  • Anna Castañé
    • Departament de Ciencies Experimentals i de la SalutUniversitat Pompeu Fabra, PRBB
    • Department of Neurochemistry and NeuropharmacologyInstitut d’Investigacions Biomèdiques de Barcelona (CSIC), IDIBAPS
  • Elena Martín-García
    • Departament de Ciencies Experimentals i de la SalutUniversitat Pompeu Fabra, PRBB
    • Departament de Ciencies Experimentals i de la SalutUniversitat Pompeu Fabra, PRBB
Original Investigation

DOI: 10.1007/s00213-009-1557-9

Cite this article as:
Barbano, M.F., Castañé, A., Martín-García, E. et al. Psychopharmacology (2009) 205: 475. doi:10.1007/s00213-009-1557-9

Abstract

Rationale

Cannabinoid compounds are known to regulate feeding behavior by modulating the hedonic and/or the incentive properties of food.

Objectives

The aim of this work was to determine the involvement of the cannabinoid system in food reinforcement associated with a conflict situation generated by stress.

Methods

Mice were trained on a fixed ratio 1 schedule of reinforcement to obtain standard, chocolate-flavored or fat-enriched pellets. Once the acquisition criteria were achieved, the reinforced lever press was paired with foot-shock exposure, and the effects of Δ9-tetrahydrocannabinol (THC; 1 mg/kg) were evaluated in this conflict paradigm.

Results

THC did not modify the operant response in mice trained with standard pellets. In contrast, THC improved the instrumental performance of mice trained with chocolate-flavored and fat-enriched pellets. However, the cannabinoid agonist did not fully restore the baseline responses obtained previous to foot-shock delivery. THC ameliorated the performance to obtain high palatable food in this conflict test in both food-restricted and sated mice. The effects of THC on food reinforcement seem to be long-lasting since mice previously treated with this compound showed a better recovery of the instrumental behavior after foot-shock exposure.

Conclusions

These findings reveal that the cannabinoid system is involved in the regulation of goal-directed responses towards high palatable and high caloric food under stressful situations.

Keywords

Food reinforcementTHCOperant behaviorFoot-shockPalatabilityStressMice

Introduction

The psychoactive and orexigenic properties of Cannabis sativa derivatives have been known for centuries (Abel 1975). However, the characterization (Devane et al. 1992; Herkenham et al. 1991) and molecular cloning (Matsuda et al. 1990; Munro et al. 1993; Cravatt et al. 1996) of the different components of the endogenous cannabinoid system have only been recently accomplished. The endocannabinoid system includes the cannabionoid receptors (CB1, CB2, and GPR55), the endogenous lipid ligands (endocannabinoids), and the enzymatic machinery for their synthesis and inactivation. This system is involved in the modulation of several physiological processes including neuronal excitability, nociception, memory, motivation, glucose and lipid metabolism, and food intake, among others (Di Marzo and Matias 2005; Pagotto et al. 2006).

The administration of either exogenous or endogenous cannabinoid agonists increases food intake in humans and in animal models when the doses employed are low or moderate, while these compounds are anorexigenic at higher doses probably due to their sedative properties (Giuliani et al. 2000; Kirkham and Williams 2001; Williams and Kirkham 2002). The effects of cannabinoids on food intake and metabolism are mediated through the activation of the CB1 cannabinoid receptor (Pagotto et al. 2006). Cannabinoid agonists are effective in the clinic to increase food intake in some pathological conditions related to weight loss, namely cancer and AIDS (Cota et al. 2003a; Kirkham and Williams 2001). On the contrary, the acute administration of CB1 cannabinoid antagonists suppresses food intake and food-motivated behavior (Foltin and Haney 2007; Salamone et al. 2007; Sink et al. 2008). The genetic or chronic pharmacological impairment of the endogenous cannabinoid system mainly results in a short-term hypophagia and long-lasting reduction in body weight. Thus, mice chronically treated with the selective CB1 antagonist, rimonabant or lacking the CB1 cannabinoid receptors are leaner, have lower motivation for food and lower plasma leptin levels, as well as a transitory lower caloric intake than their corresponding controls (Cota et al. 2003b; Ravinet-Trillou et al. 2004; Ward and Dykstra 2005). Considering these findings, the endocannabinoid system has recently captured the interest of the scientific community as a promising candidate to develop pharmacological tools to fight against obesity and its related metabolic disorders (Di Marzo and Matias 2005; Di Marzo and Szallasi 2008; Schäfer et al. 2008; Van Gaal et al. 2008), which have dramatically increased in the last years. In this context, rimonabant has emerged as an effective treatment of obesity and metabolic disorders and its beneficial effects have been reported in several clinical trials (Despres et al. 2005; Pi-Sunyer et al. 2006; Scheen et al. 2006; Van Gaal et al. 2008). Accordingly, the European regulatory authorities approved the clinical use of rimonabant in obese patients (BMI > or =30 kg/m or >27 kg/m with complications) in 2006. However, the different clinical trials performed with rimonabant and the data from pharmacovigilance have reported several gastro-intestinal and psychiatric side effects including nausea, anxiety, and depression. Due to these psychiatric side effects, the European Medicines Agency has recommended the suspension of the marketing authorization of rimonabant on October 23, 2008.

Two hypotheses have been proposed to explain the orexigenic/anorexigenic properties of cannabinoids. First, the modulation of CB1 cannabinoid receptors could modify the hedonic value of food since the administration of agonists facilitate palatable food intake while antagonists show opposite effects (Arnone et al. 1997; De Vry et al. 2004; Koch and Matthews 2001; Maccioni et al. 2008; Miller et al. 2004; Rowland et al. 2001). In agreement, CB1 cannabinoid receptors are down-regulated in the hippocampus and striatum of rats consuming palatable food (Harrold et al. 2002). However, cannabinoid compounds are also able to modulate the consumption of standard chow (Thornton-Jones et al. 2005; Williams et al. 1998), suggesting that CB1 cannabinoid receptors can regulate the incentive motivation for food. Accordingly, cannabinoid agonists increase the incentive value of food and food-associated stimuli (Cota et al. 2003a; Kirkham and Williams 2001; Solinas and Goldberg 2005; Thornton-Jones et al. 2005; Williams et al. 1998).

The aim of the present study was to clarify the involvement of the cannabinoid system in food reinforcement. Thus, we evaluated the effects of THC on the stress-induced decrease in the operant responding for different types of food reward. Indeed, stress modifies food reward processing (Ayensu et al. 1995; Howell et al. 1999; Willner et al. 1987) and the endocannabinoid system has been shown to counteract the anhedonic effects induced by stress (Rademacher and Hillard 2007). To evaluate the effects of palatability in these responses, we used two kinds of isocaloric food pellets, one of them rendered highly palatable by the addition of chocolate flavor. The endocannabinoid system has also been involved in the regulation of the energy balance through local effects in several peripheral tissues including the liver and the adipose tissue (Cota et al. 2003b; Di Marzo and Matias 2005; Di Marzo et al. 2001). Therefore, a third type of high caloric food pellets, enriched in fat content, was also used in the present operant paradigm.

Methods

Animals

Five-week-old CD1 male mice (Charles River, France) weighting 20–25 g upon arrival at the laboratory were used in this study. The animals were individually housed in clear plastic cages in an animal vivarium maintained on a 12-h light–dark cycle (lights on at 08:00 h), at constant temperature (21 ± 1°C) and humidity (55 ± 10%). All the experiments were conducted during the light phase of the circadian cycle. Following 1 week of ad libitum access to standard chow, mice were placed on a restricted food regimen designed to reduce their body weight to 90% of their free-feeding values. Animals were maintained on this reduced weight for the whole duration of the experiments, except in experiment 6 where food-sated mice were employed. Water was provided ad libitum, except during experimental sessions. Mice were weighed and handled daily in order to habituate them and minimize handling stress during the experiments. Animal care and experimental procedures were in strict accordance with institutional and international standards (the Guide for the Care and Use of Laboratory Animals adopted and promulgated by the National Institutes of Health, the European Communities Council Directive 86/609/EEC, 24 November 1986) and were approved by the local Ethics Committee (CEEA-PRBB).

Food and drugs

During the experimental sessions, animals were presented with 20 mg dustless precision standard pellets, chocolate-flavored pellets (TestDiet, Richmond, IN, USA), and fat-enriched pellets (Bio-serv, Frenchtown, NJ, USA). The standard pellet formula was similar to the standard maintenance diet provided to mice (25% protein, 10.8% fat, 64.2% carbohydrate, with a caloric value of 3.26 kcal/g). In a second formula, this diet was modified by the addition of chocolate flavor (2% pure unsweetened cocoa) to render it more palatable to mice (24.3% protein, 10.1% fat, 63.5% carbohydrate, with a caloric value of 3.23 kcal/g). Finally, we also used a fat-enriched formula (14% protein, 60% fat, 26% carbohydrate, with a caloric value of 5.32 kcal/g). Mice were habituated to the different formulas 3 days before the experimental test procedure to avoid food neophobia. The food pellets were presented only during the experimental sessions. Otherwise, animals were maintained on standard chow for their daily food intake. The amount of food consumed during the experimental sessions was discounted from the daily ration (5 g). Mice consumed all the pellets during the sessions.

THC (Sigma, St. Louis, MO, USA) was dissolved in a mixture of 5% ethanol, 5% cremophor-EL, and 90% distilled water. This mixture also served as vehicle control solution. The volume of injection was 1 ml/100 g and the dose used was 1 mg/kg of body weight. This specific dose of THC was chosen based on previous studies showing that it caused hyperphagia but had no effects on locomotor activity and did not produce anxiogenic/anxiolytic-like responses (Onaivi et al. 1990; Valjent et al. 2002; Wiley et al. 2005; Williams et al. 1998). THC and vehicle were administered intraperitoneally (i.p.) 30 min before testing.

Apparatus

Training on instrumental responding for food was conducted in operant chambers (Model ENV-307A-CT, Med Associates, Georgia, VT, USA). Chambers were made of aluminum and clear acrylic, had grid floors connected to an electrical shocker (ENV-414, Med Associates, St Albans, VT, USA), and were housed in sound- and light-attenuated boxes equipped with fans to provide ventilation and ambient noise. They were equipped with a food pellet dispenser, two retractable levers placed on each side of a food tray, a chamber light, and a stimulus light. One of the levers was randomly selected as the reinforced lever and the other as the non-reinforced lever. Active pressing on the reinforced lever resulted in the delivery of a pellet on the food tray, while pressing on the non-reinforced lever had no programmed consequences. The stimulus light located above the reinforced lever was paired contingently with the delivery of the pellet.

Instrumental responding for food

Mice were trained 6 days per week in 1-h sessions to acquire an instrumental responding for food. The chamber light was on at the beginning of the session for 3 s and off during the remaining period. Each daily session started with the presentation of the reinforced and non-reinforced levers, a free pellet delivered in the food tray, and a 2-s presentation of the stimulus light located above the reinforced lever. Responses on the reinforced lever led to a pellet delivery and the presentation of the stimulus light for 2 s. Mice were trained to lever press for food under a fixed ratio 1 (FR1) schedule of reinforcement. A 10-s time-out period was established after the obtainment of each reinforcer. During this 10-s period, the stimulus light was off and no reward was provided after pressing the reinforced lever. Non-reinforced lever presses and all the responses performed during the 10-s time-out period were also recorded. The session was terminated after 100 reinforcers were delivered or after 1 h, whichever occurred first. The criteria for acquisition of the instrumental behavior was achieved when mice maintained a stable responding with less than 20% deviation from the mean of the total number of food pellets earned during three consecutive sessions (80% of stability), with at least 75% responding on the reinforced lever, and a minimum of 10 reinforcers per session. Once the acquisition criteria were achieved, foot-shock exposure began (Fig. 1). During this phase, animals received a 2-s foot-shock (0.21 or 0.23 mA, see below) contingent with food pellet delivery. Foot-shock exposure was continued for 3 days. Following this phase, THC (1 mg/kg) or its vehicle was administered and instrumental responding for food associated with foot-shock was evaluated during four consecutive days. Finally, animals returned to the initial training conditions (no foot-shock, no drug) for two additional days (Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs00213-009-1557-9/MediaObjects/213_2009_1557_Fig1_HTML.gif
Fig. 1

Experimental procedure employed to evaluate THC effects on operant behavior to obtain different kinds of food when associated with stress. Mice were trained on a FR1 schedule of reinforcement to obtain food pellets. Once the acquisition criteria were achieved, reinforced lever presses were paired with an electric foot-shock during 3 days. In the following 4 days, THC (1 mg/kg, i.p.) or vehicle was administered to mice 30 min before testing, which occurred as in the previous 3 days (pellet delivery associated with foot-shock). Finally, animals were allowed to recover for 2 days without shock delivery or drug/vehicle administration

Experiment 1: Evaluation of the effects of different foot-shock intensities

The purpose of this experiment was to determine the appropriate foot-shock intensity to be used in subsequent experiments. The chosen intensity should be able to produce a relevant decrease of the instrumental responding to obtain food in order to reveal any possible effect of THC without abolishing this operant behavior. Fifteen mice were trained to obtain standard pellets as previously described until they reached the acquisition criteria. Subsequently, foot-shock intensities of 0 mA (n = 3), 0.18 mA (n = 4), 0.21 mA (n = 4), or 0.25 mA (n = 4) were assessed during 3 days.

Experiments 2–5: Effects of THC on food-reinforced instrumental responding associated with foot-shock in food-restricted mice

In these experiments, the effects of THC on the instrumental responding to obtain different kinds of food pellets were investigated when food delivery was associated with foot-shock. Mice were trained to obtain food pellets until reaching the acquisition criteria, as previously described. In experiment 2, THC (n = 13) or its vehicle (n = 10) was administered to mice trained to obtain standard pellets, and the foot-shock intensity was 0.21 mA. In experiment 3, THC (n = 9) or its vehicle (n = 9) was administered to mice trained to obtain chocolate-flavored pellets, and the foot-shock intensity was also 0.21 mA. Experiment 4 was conducted in the same way as experiment 3, but a foot-shock intensity of 0.23 mA was used (THC: n = 10; vehicle: n = 9). In experiment 5, THC (n = 11) or its vehicle (n = 12) was administered to mice trained to obtain fat-enriched pellets and the foot-shock intensity was 0.23 mA.

Experiment 6: Effects of THC on food-reinforced instrumental responding associated with foot-shock in ad libitum mice

This experiment was conducted in the same way of the previous experiments, but food sated mice (THC: n = 7, vehicle: n = 7) were used instead of food-restricted animals. Foot-shocks of 0.21 mA intensity and chocolate-flavored pellets were employed.

Statistical analysis

The results were analyzed using a multivariate analysis of variance (MANOVA), with time (day) and lever (reinforced versus non-reinforced) as within-subjects factors, when analyzing acquisition of instrumental behavior to obtain food. In the case of foot-shock experiments, phase of the experiment (mean of 3 days of acquisition, mean of 3 days of foot-shock alone, mean of 4 days of THC/vehicle plus shock) as well as lever (reinforced versus non-reinforced) were the within-subjects factors. In this case, the between-subjects factors were drug (vehicle or THC) and/or food (standard, chocolate-flavored and fat-enriched pellets). In all the cases, the dependent variable was the number of lever presses accomplished. When significant overall interactions were found, further analyses of partial interactions were carried out. Post hoc analyses were performed with Newman–Keuls test when the initial p value was significant. In the case of experiment 6, the amount of pellets earned by mice treated with vehicle or THC was compared using a Student’s t test. Given that percentages are arranged in ordinal scales, a Mann–Whitney U test was employed to compare the percentage of discrimination between the reinforced and the non-reinforced levers. The percentage of discrimination was calculated as the number of reinforced lever presses divided by the sum of reinforced and non-reinforced lever-presses, and the resulting value multiplied by 100. All data were analyzed with Statistica software (StatSoft Inc., France) and are expressed as mean ± SEM. A result was considered significant if p < 0.05.

Results

Effects of different shock intensities on food-reinforced instrumental behavior

Three different intensities of electrical foot-shock were first evaluated on operant responding to obtain standard pellets in order to select the appropriate conditions for the following experiments. A decrease in reinforced lever pressing of 51%, 53%, and 82% with regard to control conditions (0 mA) was observed with foot-shocks of 0.18, 0.21, and 0.25 mA, respectively (Fig. 2). The discrimination between levers was 88.28 ± 2.61% in the control conditions, 76.92 ± 4.56% in the case of 0.18 mA, 54.92 ± 4.84% in the case of 0.21 mA and 80.80 ± 4.59% for 0.25 mA. We considered that a reduction of half the level of operant responses in the reinforced lever observed when no shock was applied and a discrimination close to the one observed only by chance (50%) would be the most appropriate conditions for the following experiments. Accordingly, the subsequent experiments in mice trained to obtain standard pellets were conducted using a foot-shock intensity of 0.21 mA.
https://static-content.springer.com/image/art%3A10.1007%2Fs00213-009-1557-9/MediaObjects/213_2009_1557_Fig2_HTML.gif
Fig. 2

Effects of different stressful conditions on food-motivated instrumental behavior. Several intensities of foot-shock were assessed in mice trained to obtain standard pellets in order to choose the most appropriate conditions for the following experiments. While the performance on the reinforced lever was equivalent in mice exposed to 0.18 and 0.21 mA foot-shock intensities, the discrimination between levers was significant only in the case of 0.18 mA. The exposure to a foot-shock intensity of 0.25 mA produced a general impairment of the instrumental performance revealed by a decrease in the responses in both the reinforced and non-reinforced levers, which was considered too strong. **p < 0.01; ***p < 0.001, differences between reinforced and non-reinforced lever presses (Newman–Keuls test)

Acquisition of instrumental responding for different kinds of food

The instrumental performance of mice varied in accordance with the kind of pellets used, as revealed by a significant main effect of food (F2, 111 = 13.28, p < 0.001). The mean time to acquire a reliable instrumental behavior following the previously defined criteria was 6.27 ± 0.48 days in the case of standard pellets, 5.58 ± 0.35 days in the case of chocolate-flavored pellets and 5.48 ± 0.32 days for fat-enriched pellets. One-way ANOVA conducted on these data did not reveal significant differences on the acquisition time observed with the different kinds of pellets (F2, 55 = 1.26, n.s.). At the end of the training phase, 23 out of 25 mice (92%) achieved the acquisition criteria when trained with standard pellets, 37 out of 40 mice (92.5%) in the case of chocolate flavored pellets, and 23 out of 28 mice (82.1%) in the case of fat-enriched pellets. Only those animals acquiring a reliable food instrumental behavior continued the experiment. The operant performance of the different experimental groups during the acquisition training is shown in Fig. 3. Mice trained to obtain chocolate-flavored pellets showed the best operant performance. Indeed, the number of instrumental responses for chocolate pellets during the last 3 days of training was significantly higher than the responses for standard pellets. The number of operant responses for fat-enriched pellets was similar to the responses for standard pellets (Table 1). When training was over, mice were randomly divided in two groups that would receive either vehicle or THC later on. Table 2 shows that no differences in lever pressing were revealed between groups before exposure to the different experimental conditions, as shown by the lack of main effect of drug and the absence of interaction between drug and lever factors.
https://static-content.springer.com/image/art%3A10.1007%2Fs00213-009-1557-9/MediaObjects/213_2009_1557_Fig3_HTML.gif
Fig. 3

Acquisition of operant behavior to self-administer different kinds of food. Mice were trained daily in 1-h sessions using standard (a), chocolate-flavored (b), and fat-enriched (c) pellets. Mice reliably discriminated between reinforced and non-reinforced levers from the second day of training and thereafter. ***p < 0.001, differences between reinforced and non-reinforced lever presses (Newman–Keuls test)

Table 1

Mean number of lever presses for different kinds of pellets during the last 3 days of instrumental training

Type of pellets

Reinforced lever

Non-reinforced lever

Standard pellets

76.64 ± 3.14

14.06 ± 2.62+++

Chocolate pellets

94.79 ± 1.76*

5.16 ± 1.18+++

Fat-enriched pellets

76.44 ± 4.28

11.10 ± 1.52+++

*p < 0.001, significantly different from the reinforced lever responses for standard pellets; +++p < 0.001, significantly different from the reinforced lever responses for the same kind of food

Table 2

Presses on the reinforced (R) and non-reinforced (NR) levers before the exposure to foot-shock and/or drug injections

 

Standard pellets

Chocolate pellets

Fat-enriched pellets

R lever

NR lever

R lever

NR lever

R lever

NR lever

VEH

77.21 ± 5.11

11.29 ± 5.48

95.55 ± 2.78

4.07 ± 1.24

71.78 ± 6.11

13.33 ± 2.18

THC

76.28 ± 4.13

15.77 ± 2.67

94.10 ± 2.33

6.13 ± 1.96

81.52 ± 5.86

8.67 ± 1.96

D

F1, 19 = 0.17, p = n.s.

F1, 17 = 0.02, p = n.s.

F1, 21 = 0.28, p = n.s.

L

F1, 19 = 221.22, p < 0.001

F1, 17 = 1,432, p < 0.001

F1, 21 = 245.31, p < 0.001

D×L

F1, 19 = 0.40, p = n.s.

F1, 17 = 0.55, p = n.s.

F1, 21 = 2.95, p = n.s.

The main effect drug (D) was not significant in either of the experimental conditions. However, the main effect lever (L) resulted highly significant in all the cases because mice pressed more the reinforced lever, independently of the kind of food tested. The lack of a significant interaction between drug and lever factors (D×L) for each type of pellets employed indicates that the operant performance of mice was equivalent between vehicle (VEH) and THC-treated groups (THC) before the exposure to the different experimental conditions

Effects of THC on the instrumental performance to obtain standard pellets in a conflict situation

The effects of THC on the operant performance to obtain food associated with the delivery of foot-shock (0.21 mA) were evaluated in mice trained with standard pellets (Fig. 4). MANOVA showed a significant effect of the experimental phase (F2, 42 = 5.15, p < 0.01) since the level of instrumental behavior was higher in the baseline conditions compared to the foot-shock and drug injection phases. A main effect of lever (F1, 21 = 10.80, p < 0.01) was also revealed since mice pressed more the reinforced lever during the baseline conditions. Thus, the interaction phase × lever also resulted significant (F2, 42 = 27.36, p < 0.001). However, no effect of drug was observed (F1, 21 = 0.22, n.s.) nor interaction between drug and phase (F2, 42 = 2.73, n.s.), drug × lever (F1, 21 = 0.0005, n.s.), nor drug × phase × lever (F2, 42 = 0.26, n.s.). The discrimination between levers did not significantly differ between vehicle- (63.06 ± 13.87%) and THC-injected (57.42 ± 1.35%) mice as evidenced by the Mann–Whitney U test (U13 = 39.00, n.s.).
https://static-content.springer.com/image/art%3A10.1007%2Fs00213-009-1557-9/MediaObjects/213_2009_1557_Fig4_HTML.gif
Fig. 4

Effects of vehicle (a) or THC (b) administration on the instrumental performance to obtain standard pellets in a conflict test. The foot-shock intensity employed was 0.21 mA. Abscissa axis: baseline refers to the mean of 3 days in which mice acquired the criteria for food-reinforced instrumental behavior, foot-shock refers to the mean of 3 days of foot-shock exposure alone, vehicle (n = 10) or THC (n = 13) refers to the mean of 4 days of foot-shock administration associated with the respective injections. THC did not modify the operant performance of mice trained to obtain standard pellets. ***p < 0.001, differences between reinforced and non-reinforced lever presses. ++p < 0.01, differences between experimental phases when considering the same lever (Newman–Keuls test)

Effects of THC on the instrumental performance to obtain chocolate-flavored pellets in a conflict situation

The effects of THC were first evaluated on the operant performance to obtain chocolate-flavored pellets using a foot-shock intensity of 0.21 mA (data not shown). Foot-shock administration only diminished reinforced lever-pressing about 35% when compared with the values obtained during the acquisition period (from 86.44 ± 2.76 to 56.19 ± 8.09), and mice were still able to discriminate between the reinforced and non-reinforced levers (72.29 ± 8.91% of discrimination). Considering these results, the intensity of the electrical foot-shock was enhanced in order to obtain an operant performance similar to that observed in the previous experiment with standard pellets. When a foot-shock intensity of 0.25 mA was used, the number of reinforced lever pressing was dramatically diminished (decrease at about 82% with regard to the values obtained during the acquisition period, see experiment 1). Therefore, an intermediate intensity of 0.23 mA was used for this experiment.

When the effects of THC were evaluated on the operant performance to obtain chocolate pellets associated with the delivery of a 0.23-mA foot-shock, a significant interaction among the three main factors (phase × lever × drug, F2, 34 = 4.01, p < 0.05) was revealed by MANOVA, which permitted the independent analysis of vehicle- and THC-treated animals. Figure 5a depicts the effects of vehicle administration on the instrumental behavior of mice under these experimental conditions. MANOVA showed a main effect of lever (F1, 8 = 6.99, p < 0.05), not an effect of the experimental phase (F2, 16 = 0.66, n.s.) and a significant interaction between lever and experimental phase (F2, 16 = 6.91, p < 0.01). This interaction was due to the significant difference between reinforced and non-reinforced lever presses that was only observed in the baseline phase. Differences between reinforced and non-reinforced lever responses were not observed during foot-shock exposure nor during the vehicle injection phase. MANOVA also revealed a main effect of lever (F1, 9 = 32.16, p < 0.001) and a significant interaction between lever and experimental phase (F2, 18 = 19.38, p < 0.001) in THC-treated mice (Fig. 5b). THC significantly improved the discrimination between the reinforced and non-reinforced levers (82.16 ± 10.03% against 58.23 ± 9.82% in vehicle-injected mice) as evidenced by the Mann–Whitney U test (U9 = 18.00; p < 0.05).
https://static-content.springer.com/image/art%3A10.1007%2Fs00213-009-1557-9/MediaObjects/213_2009_1557_Fig5_HTML.gif
Fig. 5

Effects of vehicle (a) or THC (b) administration on the instrumental performance to obtain chocolate-flavored pellets in a conflict test. The foot-shock intensity employed was 0.23 mA. Abscissa axis: baseline refers to the mean of 3 days in which mice acquired the criteria for food-reinforced instrumental behavior, foot-shock refers to the mean of 3 days of foot-shock exposure alone, vehicle (n = 10) or THC (n = 9) refers to the mean of 4 days of foot-shock administration associated with the respective injections. THC greatly improved the discrimination between reinforced and non-reinforced levers. ***p < 0.001, differences between reinforced and non-reinforced lever presses. +p < 0.05, differences between experimental phases when considering the same lever (Newman–Keuls test)

Effects of THC on the instrumental performance to obtain fat-enriched pellets in a conflict situation

The effects of THC were evaluated on the operant performance to obtain fat-enriched pellets associated with the delivery of foot-shock of an intensity of 0.23 mA. Considering that the caloric content of the fat-enriched pellets may enhance food motivation (Ward and Dykstra 2005), foot-shock intensity of 0.23 mA was also used in this experiment. A significant interaction between phase, lever, and drug was revealed by MANOVA (F2, 42 = 3.38, p < 0.05), which allows to analyze vehicle-injected animals separately from THC-treated mice (Fig. 6). A significant main effect of phase (F2, 22 = 7.50, p < 0.01), no effect of lever (F1, 11 = 2.56, n.s.), and a significant interaction between phase and lever (F2, 22 = 18.22, p < 0.001) were revealed by MANOVA in vehicle-injected mice (Fig. 6a). This interaction was due to the significant difference between reinforced and non-reinforced lever presses that was only observed during the baseline phase. In THC-treated mice (Fig. 6b), MANOVA revealed a main effect of the experimental phase (F2, 20 = 11.39, p < 0.001), a main effect of lever (F1, 10 = 18.23, p < 0.01) and a significant interaction between phase and lever (F2, 20 = 16.38, p < 0.001). As in the previous experiment, THC significantly improved the discrimination between the reinforced and the non-reinforced levers (74.61 ± 10.67% against 35.99 ± 10.56% in vehicle-injected mice) as evidenced by the Mann–Whitney U test (U11 = 25.00; p < 0.05).
https://static-content.springer.com/image/art%3A10.1007%2Fs00213-009-1557-9/MediaObjects/213_2009_1557_Fig6_HTML.gif
Fig. 6

Effects of vehicle (a) or THC (b) administration on the instrumental performance to obtain fat-enriched pellets in a conflict test. The foot-shock intensity employed was 0.23 mA. Abscissa axis: baseline refers to the mean of 3 days in which mice acquired the criteria for food-reinforced instrumental behavior, foot-shock refers to the mean of 3 days of foot-shock exposure alone, vehicle (n = 12) or THC (n = 11) refers to the mean of 4 days of foot-shock administration associated with the respective injections. THC greatly improved the discrimination between reinforced and non-reinforced levers. *p < 0.05; ***p < 0.001, differences between reinforced and non-reinforced lever presses. ++p < 0.01; +++p < 0.001, differences between experimental phases when considering the same lever (Newman–Keuls test)

Recovery of the instrumental performance after 2 days without foot-shock in the absence of treatment

After the foot-shock phase, mice were trained during 2 days to obtain the corresponding food pellets in the operant boxes without foot-shock exposure or drug treatment (Fig. 7). An independent two-way ANOVA was conducted for each kind of food. In mice trained with standard pellets (Fig. 7a), no main effect of drug (F1, 13 = 0.14, n.s.) or lever (F1, 13 = 4.12, n.s.), nor the interaction between these two factors was revealed significantly by ANOVA (F1, 13 = 0.34, n.s.). In mice trained with chocolate-flavored pellets (Fig. 7b), ANOVA revealed a main effect of lever (F1, 17 = 11.95, p < 0.01), without main effect of drug (F1, 17 = 0.76, n.s.) nor interaction between these two factors (F1, 17 = 1.47, n.s.). In mice trained with fat-enriched pellets (Fig. 7c), ANOVA revealed a significant interaction between drug and lever (F1, 20 = 6.90, p < 0.05) without main effects of these two factors (drug: F1, 20 = 2.18, n.s.; lever: F1, 20 = 0.02, n.s.) since mice formerly treated with THC presented a higher level of reinforced lever presses than vehicle-treated mice. A trend for a better recovery was observed in animals previously treated with THC for the three kind of pellets used since the discrimination between reinforced and non-reinforced lever was better in this group when compared with animals previously treated with vehicle. Thus, the discrimination of vehicle-treated mice trained with standard pellets was 66.60 ± 9.55%, while the discrimination of mice treated with THC was 85.99 ± 4.28%. In mice trained with chocolate-flavored pellets, the previous treatment with vehicle resulted in a discrimination of 77.65 ± 8.92%, while this value was of 92.89 ± 4.24% in the case of THC pretreatment. Finally, the discrimination of vehicle-treated mice trained with fat-enriched pellets was 15.50 ± 9.65%, while this discrimination was 69.43 ± 10.42% in THC-treated mice.
https://static-content.springer.com/image/art%3A10.1007%2Fs00213-009-1557-9/MediaObjects/213_2009_1557_Fig7_HTML.gif
Fig. 7

Recovery of the instrumental performance after 2 days without foot-shock or vehicle/THC administration. Lever presses in a 1-h session using standard (a), chocolate-flavored (b), and fat-enriched (c) pellets are depicted. In the abscissa axis, VEH (vehicle) and THC refers to the treatment administered to mice in the previous phase. The results here presented were obtained in a drug-free state. Mice that had received THC showed a better recovery, independent of the kind of pellets employed. RL: reinforced lever; NRL: non-reinforced lever. +p < 0.05, differences between experimental phases when considering the same lever (Newman–Keuls test)

Effects of THC on the instrumental performance to obtain chocolate-flavored pellets in ad libitum mice

The effects of THC were evaluated on the operant performance to obtain chocolate-flavored pellets in food-sated mice. Ad libitum mice may present a lower motivation to obtain food because of an absence of metabolic needs. This situation could render the mice too sensitive to strong foot-shock intensity. Consequently, a foot-shock intensity of 0.21 mA was employed in this experiment. THC administration did not modify the amount of pellets earned and consumed by sated animals when compared with vehicle-treated mice, as revealed by Student’s t test (t13 = −0.69, p = 0.50; Fig. 8a). However, the discrimination between levers was significantly increased by THC as evidenced by a significant Mann–Whitney U test (U13 = 30.00; p < 0.01; Fig. 8b).
https://static-content.springer.com/image/art%3A10.1007%2Fs00213-009-1557-9/MediaObjects/213_2009_1557_Fig8_HTML.gif
Fig. 8

Effects of vehicle (VEH) or THC administration on the instrumental performance of ad libitum mice (n = 7 per group) to obtain chocolate-flavored pellets in a conflict test. a Number of pellets obtained under the effects of the respective injection. b Discrimination between the reinforced and the non-reinforced levers when mice were under the effects of the respective injection. The foot-shock intensity employed was 0.21 mA. THC did not modify the amount of pellets obtained but greatly improved the discrimination between reinforced and non-reinforced levers. **p < 0.01, differences in the percentage of discrimination between reinforced and non-reinforced levers (Mann–Whitney test)

Discussion

In this study, we showed a selective effect of THC that was dependent on the kind of food in an operant task performed by mice to obtain food reinforcers under a conflict situation. Thus, THC did not modify the operant performance when mice were trained to obtain standard food pellets associated to foot-shock delivery. However, THC improved the discrimination between the reinforced and the non-reinforced levers when the pellets were rendered more palatable by the addition of chocolate flavor or presented a higher caloric value by increasing the amount of fat.

Standard pellets had the same content and caloric value than chocolate-flavored pellets except for the addition of chocolate flavor. This flavor increases the palatability without affecting any other property of the food pellet. As a consequence of this increased palatability, the number of operant responses to obtain chocolate-flavored pellets was higher than to obtain fat-enriched or standard pellets. Both fat-enriched and normal pellets were equally reinforcing since mice responded similarly to obtain these pellets. However, fat-enriched pellets have a higher caloric content than standard pellets (5.32 kcal/g versus 3.23 kcal/g, respectively). The high caloric content of fat-enriched pellets could have two opposite effects on operant performance: it could enhance the reinforcing value of this food pellet, but it could also decrease the maximum amount of pellets required during a short time period due to the higher ability to induce satiety. This last effect may be masking the increased palatability of fat-enriched pellets.

Operant performance in conflict paradigms similar to the one used in the present study can be improved by the administration of anxiolytic compounds (Geller and Seifter 1960; Gomita et al. 2003; Smith and Barrett 1997). The vehicle of THC, which contains 5% ethanol, is devoid of anxiolytic-like effects under these experimental conditions in mice. Indeed, previous studies from our laboratory have compared the effect of saline and the vehicle of THC (ethanol, cremophor, and distilled water) in the mouse elevated plus-maze. Both groups of mice spent the same percentage of time in the open arms revealing the absence of effect on anxiety-like responses of the vehicle used to dissolve THC (Balerio et al. 2005, 2006). THC induces anxiolytic responses when administered at low doses (0.3 mg/kg; Valjent et al. 2002). However, THC did not modify the operant performance of mice trained with standard pellets in the present study, indicating a lack of anxiolytic-like effects at the dose of THC used (1 mg/kg), in agreement with previous studies (Valjent et al. 2002). Therefore, the modification of the operant performance of mice trained with chocolate-flavored and fat-enriched pellets was not due to a possible effect of THC on the anxiety-like responses.

Cannabinoid agonists such as THC produce antinociceptive effects in several animal models. In mice, doses of THC higher than 1 mg/kg are required to induce antinociceptive effects (Martin and Lichtman 1998). In agreement, other studies have revealed that the dose of THC used in the present work (1 mg/kg, i.p.) is devoid of antinociceptive effects in a mouse model of thermal nociception (Monory et al. 2007). Moreover, the selective effect of THC in the operant behavior of mice trained with chocolate and fat-enriched pellets, but not in mice obtaining normal pellets, demonstrates that these effects are not due to any possible antinociceptive action of THC.

Mice significantly increased the number of responses on the non-reinforced lever when foot-shock was administered. This is a well-characterized consequence of the contingency between the instrumental response and an aversive stimulus such as the electric shock (Dardano 1968; Tullis and Walters 1968; Yoshino and Kimura 1991). In agreement with previous studies (Yoshino and Kimura 1991), a negative relationship between the intensity of the electric shock administered and the number of lever presses on the reinforced lever was also revealed in the present work. This effect has been previously demonstrated in fixed ratio and progressive ratio schedules of reinforcement. Thus, pigeons trained on a progressive ratio schedule of reinforcement learned to rotate the response in a second manipulandum in order to decrease the effort required to obtain the reward. This behavior was severely impaired when a punishment was associated to the second manipulandum leading the animal to perform an alternative persisting response on the first manipulandum in order to obtain the reward (Dardano 1968).

The main effect observed after THC administration was a significant decrease in the number of non-reinforced lever presses in the case of chocolate-flavored and fat-enriched pellets, leading to an improved level of discrimination between levers. This effect was not observed in the case of standard pellets. These responses are unlikely to be due to the sedative effects of THC (Arizzi et al. 2004; Carriero et al. 1998; McLaughlin et al. 2005) since: (1) the same dose of THC has been proved ineffective to inhibit locomotor activity (Wiley et al. 2005), (2) the operant performance of mice trained with standard pellets was not modified by THC, and (3) the number of reinforced lever presses was not decreased after THC administration in any of the experimental conditions assessed (standard, chocolate-flavored, and fat-enriched pellets). As mentioned, animals exposed to a punishment engage in alternative behavioral responses in an operant paradigm (Dardano 1968). This alternative behavior was also revealed in the present study by an enhancement in the number of presses on the non-reinforced lever leading to a loss of discrimination between levers. THC made this alternative behavior less favorable for the animal, which recovered the discrimination after the foot-shock exposure. This improved discrimination could be due to a habituation to the aversive situation generated by the foot-shock since CB1 cannabinoid receptors play a crucial role in the extinction of aversive memories (Marsicano et al. 2002; Kamprath et al. 2006). However, this habituation and the resultant enhanced discrimination observed under THC treatment specifically occurred when highly palatable or caloric foods were employed.

A modulation of the hedonic properties of food has been proposed to explain the ability of cannabinoids to regulate feeding behavior (Arnone et al. 1997; De Vry et al. 2004; Harrold et al. 2002; Miller et al. 2004; Rowland et al. 2001). Interestingly, THC effects were only revealed in the case of chocolate-flavored and fat-enriched pellets. We hypothesize that the effects of THC were only revealed when the reinforcing properties of the pellets were enhanced by high palatability or high caloric content. In agreement, several studies have demonstrated that the cannabinoid system is involved in the regulation of palatable food consumption (Harrold et al. 2002; Miller et al. 2004; Rademacher and Hillard 2007; Rowland et al. 2001). Thus, THC administration increased the consumption of chocolate containing food but had no effects on standard food intake (Koch and Matthews 2001), while rimonabant selectively decreased the intake of a highly palatable chocolate-flavored beverage (Maccioni et al. 2008). The operant performance of mice during training with fat-enriched pellets was similar to that of standard pellets. It is important to underline that the high caloric value of this kind of food could induce a rapid satiety, which could also account for the final operant performance of these animals, in spite of the possible high hedonic properties of fat-enriched pellets. Thus, THC may directly modulate the operant performance to obtain high caloric food, independently of the palatability of the fat-enriched pellets. In line with this hypothesis, THC preferentially increased the intake of a fat-enriched diet over standard diet or the same fat-enriched diet sweetened with saccharin (Koch 2001). Moreover, rimonabant decreased and the cannabinoid agonist CP55940 increased the break point to obtain a high caloric food (corn oil) in mice (Ward and Dykstra 2005).

A recent work has assessed the involvement of the endocannabinoid system in the decreased sensitivity to natural rewards induced by restraint stress (Rademacher and Hillard 2007). Thus, chronic restraint stress decreased the intake of sucrose and saccharin. This effect was reversed by the cannabinoid agonist, CP55940, and the inhibitor of anandamide catabolism, URB597, whereas rimonabant enhanced the response suggesting an involvement of the endocannabinoid system in the processing of natural rewards under stressful conditions. These observations are in agreement with the present data since the improvement of the discrimination to obtain chocolate and fat-enriched pellets was observed under a conflict situation, where the exposure to the foot-shocks impaired the operant behavior to obtain these kinds of food. On the other hand, THC administration did not fully restore the instrumental performance for chocolate-flavored and fat-enriched pellets when compared to baseline conditions, suggesting that other neurotransmitters different from the endocannabinoid system are likely involved in these behavioral responses. In support of this hypothesis, it has been reported that the endocannabinoid system closely interacts with other neurotransmitters to modulate reward (such as dopamine), palatability (such as opioids), and metabolic status (such as leptin) (Di Marzo et al. 2001; Solinas and Goldberg 2005; Verty et al. 2004).

When mice were allowed to recover from foot-shock administration and tested without drugs, all the experimental groups previously treated with THC showed a trend for a better recovery. This result was observed even in the case of mice trained with standard pellets, which operant performance was not modified by THC. This result is consistent with the involvement of the endogenous cannabinoid system in the extinction of aversive memories. Thus, a pharmacological or genetic impairment of the cannabinoid system leads to the persistence of fear responses associated with foot-shock exposure (Chhatwal et al. 2005; Marsicano et al. 2002), while an increase in the bioavailability of endocannabinoids facilitates conditioned fear responses extinction (Chhatwal et al. 2005). These responses seem to occur through a habituation-like process (Kamprath et al. 2006). In the light of these findings, we may hypothesize that THC contributed to the process of foot-shock habituation, allowing the mice to switch their attention in order to be focused on the goal-directed behavior of lever pressing for food, even in a drug-free state. Interestingly, the behavioral responses induced by THC were observed in both food-restricted and sated animals. The generalization of these behavioral effects further emphasizes the involvement of the cannabinoid system in maintaining food-directed behaviors.

In conclusion, THC improved lever discrimination of mice trained to obtain highly palatable and highly caloric foods, whereas THC did not modify the performance for standard pellets. These responses were probably due to an increase in either the hedonic or the motivational properties of the tested foods. However, other effects of THC can also account for this improved discrimination, such as a relief of stress effects on reward processes or an habituation to aversive events allowing a switch of attention to food-directed behaviors. The behavioral effects exerted by the cannabinoid system appear to be long-lasting since the recovery of the operant performance in a drug-free state was ameliorated in mice previously treated with THC. The present results highlight the specific role played by cannabinoids in regulating food reinforcement and provide a better understanding of the neurobiological mechanisms underlying the behavioral alterations that contribute to obesity and feeding disorders, such as the preferential attention directed towards relevant palatable or caloric foods. The elucidation of the pathophysiological mechanisms involved in these disorders will help to the future design of more rational therapeutic approaches.

Acknowledgments

This work was supported by the US National Institutes of Health–National Institute of Drug Abuse (NIH–NIDA) (no. 5R01-DA016768), the Spanish “Ministerio de Educación y Ciencia” (no. SAF2007-64062), the DG Research of the European Commission (GEN-ADDICT, no. LSHM-CT-2004-05166; and PHECOMP, no. LSHM-CT-2007-037669), the “Generalitat de Catalunya-DURSI” (# 2005SGR00131 and ICREA Academia) and the Spanish “Instituto de Salud Carlos III” (no. RD06/001/001). M.F.B. was supported by a post-doctoral fellowship from Fyssen Foundation. E.M.G. was supported by a post-doctoral fellowship from the Spanish “Instituto de Salud Carlos III”. We thank Dr. Patricia Robledo for stylistic revision of the manuscript.

Conflict of interest statement

R. Maldonado has received research grants from Sanofi-Aventis, Esteve, and Ferrer. Neither of the other authors have relevant financial interests to disclose, nor a conflict of interest of any kind.

Copyright information

© Springer-Verlag 2009