Naïve Sprague-Dawley rats (F0 male and all females) were purchased from Shanghai Laboratory Animal Center, Chinese Academy of Sciences. F1 and F2 generation of the rats were bred in our own laboratory. Both F0 and their descendants were housed at 23°C on a 12-h reverse dark/light cycle (on 20:00, off 8:00), and room humidity was controlled at 40% ± 10%. Rats were housed in groups (3–4 rats per cage) and allowed free access to food and water unless otherwise specified. Animals used for behavioral tests were male rats 8–12 weeks of age, and were food-restricted and maintained at 85% original body weight. All animal treatments were in strict accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by Animal Care and Use Committee of Shanghai Medical College of Fudan University.

Male rats were randomly assigned to Control group (n = 12) or Sucrose group (n = 49). Sucrose group rats were trained to press the active lever for 45 mg sucrose pellets (Bio-Serv, Flemington, NJ, USA), while those assigned to the Control group received cues without reward. The training underwent fixed ratio1 (FR1, 1 lever press/pellet) for 5 days, FR5 (5 lever presses/pellet) for 2 days, and then switched to progressive-ratio (PR) schedule, in which lever press required for each successive pellet increases by progressive increments. The PR session stopped when the rat takes more than 1 h to achieve the response requirement. Lever presses needed for the last pellet in the PR schedule was defined as the break point. Rats that achieved 50 lever presses were considered to have acquired self-administration behavior, and those that failed in the 5th FR1 session were excluded. Lever presses during FR5 and break point of individual rats were scored according to the equation (X_i − X¯)/s.d. (Student’s t-statistic) and named as intake and motivation score, respectively. In the equation, X_i is the behavior value for each rat (average of lever pressed in the FR5 sessions for intake score, and break point for motivation score), X¯ and s.d. are the mean and standard deviation of the population behavior readout (Le et al., 2017). F0 Sucrose group rats (n = 10) were from each of tenth percentiles of performance score (addition of intake score and motivation score), and Control F0 (n = 6) rats were randomly chosen from Control group. Twenty-four hours after the last session, each rat from Sucrose or Control F0 group was housed with two naïve female rats to generate F1 offspring. Six of 10 Sucrose F0 and four of six Control F0 gave birth to pups within 20–25 days after mating, and the F1 rats were housed in groups after weaning period and used for subsequent behavioral tests or mating (4 and 6 litters from Control and Sucrose F0, respectively). F2 rats were sired by crossing naïve F1 from each litter with two naïve female rats (4 and 6 litters from Control and Sucrose F1, respectively).

Apart from those animals used for breeding experiment, some of the other Sucrose group rats were subjected to surgery. In F1 generation, to facilitate the rats to self-administer cocaine, rats were trained to press the active lever to get 45 mg food pellets (Bio-Serv, Flemington, NJ, USA) in the operant chambers (Med-Associates, St. Albans City, VT, USA) for 7 days. Those met the criteria of obtaining 100 food pellets per FR5 session were subjected to surgery. A silastic catheter was positioned about 3 cm into the right jugular vein, and the other end was attached to a stainless steel pedestal mounted to the rat’s skull by dental cement. The rats were allowed 7 days to recover from surgery. Catheters were flushed daily with 0.1 ml saline containing heparin (30 IU/ml) and gentamicin (0.5 mg/ml).

After recovery, rats were allowed to self-administer cocaine in daily sessions. When the rat pressed the active lever, an injection of cocaine (Qinghai Pharmaceutical Firm) at 0.75 mg/kg/injection over 4 s was delivered and accompanied by a conditioned cue, including the illumination of the stimulus light and an audible tone for 20 s. Presses on the inactive lever had no programmed consequences. Rats were first trained on 4 h fixed-ratio FR1 program for 5 days, then FR3 for 2 days and FR5 for 5 days. Then 6-h PR program was used, in which response requirement for each successive injection increased by progressive increments. The catheter patency was verified after PR schedule by anesthesia with chloral hydrate and data of rats with a catheter problem was excluded. Cocaine performance score for each individual was calculated the same as sucrose performance score.

Locomotor activity was measured using commercial open field activity chambers for rats (43.2 cm × 43.2 cm × 30.5 cm, Med-Associates, St. Albans City, VT, USA). Each rat was allowed to freely explore in the chamber for 30 min. Data collection and analysis were done using Med Associates Activity Monitor program. The center area was defined as 30.5 cm × 30.5 cm, and distance traveled, time spent in the center, as were as entries to the center were recorded. In cocaine sensitivity test, rats were habituated to the open field activity chamber for 1 h, and were then administered hourly with saline or cocaine (0.25 mg/kg, i.p.). Locomotor activity was measured during each hour of the testing period.

Rats were subjected to habituation for 30 min in the laboratory, and then placed in the center of the elevated plus maze (Med Associates, St. Albans City, VT, USA; 70 cm above the floor, arms 50 cm × 10 cm × 40 cm) facing the closed arm and allowed to freely explore the maze for 5 min. Time spent in the open arms, and entries to the open arms during the test were recorded.

Active lever presses in cocaine self-administration, sucrose self-administration during FR program was analyzed with mixed linear model with repeated measurements (MMRM). Inactive lever presses and cocaine dose-induced locomotion activity was analyzed using two-way repeated-measures analysis of variance (ANOVA) followed by Bonferroni post hoc tests. Break point was analyzed by Mann-Whitney rank sum test. The open field test and plus maze data were analyzed with Student’s t test. Sample size estimation was conducted on alpha value of 0.05 and desired power of 0.80. Equal variance estimation and sample distribution were examined. P < 0.05 was considered statistically significant. Data are presented as mean ± SEM.

We first evaluated sucrose seeking behavior in a cohort of naïve SD rats. Rats were randomly assigned to sucrose or control group and subjected to an eight-session self-administration test (Figure 1A). Compared with Control group that received no reward during the training process, rats from Sucrose group exhibited significantly higher active lever presses in FR sessions (MMRM, group × FR, χ(1)2 = 508.77, P < 0.001; Group, P < 0.001; FR, P < 0.001), while there was no difference in inactive lever presses (Two-way RM ANOVA, F_{group × session} (6,354) = 0.980, P = 0.439). Furthermore, rats from sucrose group were motivated to seek sucrose reward under PR test (Mann-Whitney rank sum test, U = 0, P < 0.001), and displayed higher inactive lever presses compared with Control animals (Mann-Whitney rank sum test, U = 34.5, P < 0.001), indicating compulsive binge-like sucrose reward seeking behavior. Although sucrose is a rewarding substance, we observed that rats exhibited variable lever pressing for sucrose during FR5 sessions (197–990, median 685) and break point (32–603, median 219). Next, we evaluated intake score, derived from normalized FR5 lever presses, and motivation score, from normalized PR break point for each rat individually. As there was no correlation between intake and motivation score of the cohort (Figure 1B, linear regression, R = 0.118, P = 0.419), they were considered as two distinct properties of sucrose seeking behavior, and thus the addition of intake and motivation scores were used as sucrose performance score.

Epidemiology data indicated high comorbidity rates between binge eating disorder and substance use disorders (Schreiber et al., 2013). To explore if incentive salience to sucrose is reinforcer-specific, some of sucrose-experienced rats were subjected to cocaine SA test subsequently (Figure 2A, FR, MMRM, χ(2)2 = 170.86, P < 0.001; break point, Mann-Whitney rank sum test, U = 141.5, P = 0.007). Analysis of additive scores for each individual rat indicated that the number of lever press for cocaine SA was positively correlated with that of sucrose SA (Figure 2B, linear regression, R = 0.424, P = 0.044), suggesting that sucrose-seeking behavior may predict individual’s performance in response to cocaine-induced reinforcement.

In parallel with the above experiment, some other sucrose-SA experienced male rats were bred with naïve females to see if there is an inter- or trans-generational effect, and Control rats were also bred. Six litters of Sucrose F1 and four litters of Control F1 were generated, and one naïve male from each F1 litter was used to generate F2 offspring. As shown in Figure 3, Sucrose F1 exhibited no difference from Control F1 rats in number of sessions required to achieve >50 lever presses per session in FR1 schedule (Figure 3, Mann-Whitney rank sum test, U = 34, P = 0.231), indicating comparable reward learning of these two cohorts. However, in FR5 sessions, Sucrose F1 exhibited slightly lower lever presses than the control rats (Figure 3, MMRM, group × FR, χ(1)2 = 6.00, P = 0.0143; group, P = 0.224; FR1, P = 0.413; FR5, P = 0.031). There was no significant difference between the two groups in sucrose responding at PR schedule (break point, Mann-Whitney rank sum test, U = 48.5, P = 1.000) or inactive lever presses (FR, two-way RM ANOVA, F_{group × session} (6,114) = 0.280, P = 0.945; PR, Mann-Whitney rank sum test, U = 32.5, P = 0.232).

As we observed significant reduced FR5 responding to sucrose in Sucrose F1 vs. Control F1, we tested the responding to cocaine to see if the effect is specific to sucrose or a general one. Significant lower lever presses were observed in Sucrose F1 as compared with Control F1 in cocaine self-administration tests (Figure 4A, MMRM, group × FR, χ(2)2 = 20.35, P < 0.001; Group, P = 0.0285; FR1, P = 0.556; FR3, P = 0.007; FR5, P = 0.006), while there was no significant difference in break point (Mann-Whitney rank sum test, U = 48.5, P = 1.000) or in inactive lever presses (FR, two-way RM ANOVA, F_{group × session} (11,384) = 1.587, P = 0.100; PR, Mann-Whitney rank sum test, U = 245.5, P = 0.975). As there was no significant difference in break point between these two groups, we then asked if sensitivity to cocaine was different. We recorded locomotor responses to ascending doses of cocaine injections in a new batch of Sucrose and Control F1 rats, and found that lower locomotion response to cocaine was observed in Sucrose F1 (Figure 4B, two-way RM ANOVA, F_{group × dose} (1,22) = 1.770, P = 0.197; Control F1 vs. Sucrose F1, 25 mg/kg, P = 0.042), indicating desensitization to the drug.

Besides cocaine sensitivity, altered mobility and anxiety level may also affect lever presses in self-administration. A new batch of F1 littermates were subjected to open field activity and elevated plus maze tests. In the open field test, no difference in basal distance traveled, or time spent in the center was observed between Sucrose and Control F1 rats (Figure 5A Student’s t-test, total distance, t₍₄₃₎ = 1.706, P = 0.095; time in the center, t₍₄₃₎ = 0.191, P = 0.849). However, compared with Control F1, Sucrose F1 showed fewer entries to the center zone (Figure 5A right, Student’s t-test, t₍₄₃₎ = 2.109, P = 0.041), and in elevated plus maze model, the Sucrose F1 rats spent less time (Student’s t-test, t₍₃₃₎ = 3.271, P = 0.0025) and exhibited fewer entries to the open arm (Figure 5B, Student’s t-test, t₍₃₃₎ = 3.293, P = 0.0024), indicating higher anxiety level, which likely contribute to the reduction of cocaine- and sucrose-responding in F1 offspring.

We also performed sucrose self-administration experiment in F2 generation. In sucrose self-administration test, there were no significant differences in lever press learning in FR1 sessions (Figure 6, Mann-Whitney rank sum test, U = 37, P = 0.429), or sucrose seeking behavior between Sucrose and Control F2 offspring (FR, MMRM, group × FR, χ(1)2 = 4.29, P = 0.0384, group, P = 0.380; break point, U = 44.5, P = 0.968). Taken together, sucrose binge experience could decrease both sucrose and cocaine intake in F1, but not F2 generation, indicating an inter-generational decrease in reward responding in general.

We then performed correlation analysis to see if there were any relevance in the reward seeking behavior in F0 and their descendants. We first compared the sucrose performance score of Sucrose F1 rats with that of their fathers. A significant positive correlation in additive sucrose scores between F0 and F1 was observed (Figure 7A, linear regression, R = 0.744, P = 0.002). However, no significant correlation between F2 and F0 in sucrose self-administration performance was observed (Figure 7B, linear regression, R = 0.344, P = 0.250). And although there was significant positive correlation between cocaine and sucrose performance score in F0 of each tested individual, no correlation between sucrose score in F0 rats and cocaine score in F1 was observed (Figure 7C, linear regression, R = 0.0797, P = 0.711). Taken together, the maintenance of paternal incentive salience to reward was inter-generational and reinforcer-specific, as we could only observe correlated score of sucrose-SA in sucrose-experienced F0 and its F1, but not F0 sucrose score and F1 cocaine score.