Experiment 1Weight change

Relative to vehicle-treated rats, chronic semaglutide treatment reduced weight gain in males and induced weight loss in females (Fig. 2a-b). This is supported by significant main effects of Treatment (F1,60 = 130.24, p < 0.001), Day (F16,960 = 79.41, p < 0.001), and Sex (F1,60 = 196.85, p < 0.001), as well as significant Treatment x Day (F16,960 = 41.41, p < 0.001) and Day x Sex (F16,960 = 97.30, p < 0.001) interactions. After Day 1 of treatment, SEMA-treated rats (regardless of sex) weighed significantly less than VEH-treated rats for the remainder of the experiment (p ≤ 0.006) (Fig. 2a). At the end of the 17-day treatment regimen, both male and female rats treated with semaglutide had reduced weight gain with respect to baseline body weight relative to those treated with vehicle (Fig. 2b; Males: t = 8.038, p < 0.001; Cohen’s d = 2.842; Females: t = 6.668, p < 0.001; Cohen’s d = 2.358). There was a significant effect of Treatment (F1,60 = 109.1, p < 0.001) and Sex (F1,60 = 226.9, p < 0.001), but no Treatment x Sex interaction (F1,60 = 3.7, p = 0.059) for body weight on the final day of treatment. As evident in Fig. 2b, most female rats treated with semaglutide showed a decrease in body weight relative to baseline levels, whereas most male rats showed attenuated weight gain rather than weight loss. For both sexes, however, there was approximately a 10% difference in the change from baseline body weight in rats treated with semaglutide relative to those treated with vehicle, demonstrating the validity of the treatment regimen.

Pavlovian conditioned approach (PavCA)

Chronic semaglutide administration had no effect on the acquisition of PavCA behavior (Fig. 2c). Table 1 presents the LMM analyses of the individual behavioral measures (i.e., lever-cue contacts, probability of lever-cue contact, latency to lever-cue contact, food-cup entries, probability of food-cup entry, and latency to food-cup entry) used to calculate PavCA Scores. For each measure, there was a significant main effect of Session (p < 0.001), but no main effects of Sex (p ≥ 0.292) or Treatment (p ≥ 0.240).

As shown in Figs. 2c-d, by the end of PavCA training, all groups tend to sign-track (i.e., PavCA Score > 0.5) rather than goal-track. There was a significant effect of Session (F4,88 = 29.82, p < 0.001), but no significant main effects of Treatment (F1,63 = 0.01, p = 0.943) or Sex (F1,63 = 1.53, p = 0.220) for the PavCA Score (Fig. 2c). The PavCA Index (i.e., average of PavCA Score on Sessions 4 and 5) also did not differ between treatment groups or sex (Treatment (F1,60 = 0.01, p = 0.907); Sex (F1,60 = 2.75, p = 0.102)), and there were no significant interactions (Treatment x Sex interaction (F1,60 = 1.38, p = 0.246)) (Fig. 2d). Thus, chronic semaglutide treatment does not appear to affect the acquisition of Pavlovian conditioned approach behavior to a food-paired cue.

Conditioned reinforcement

Chronic semaglutide treatment enhanced the conditioned reinforcing properties of the lever-cue (Fig. 2e-g). Although both treatment groups responded more in the active nose port that resulted in lever-cue presentation (Effect of Port (F1,60 = 147.62, p < 0.001)), SEMA-treated rats showed enhanced responding relative to VEH-treated rats (Treatment (F1,60 = 11.68, p = 0.001); Port x Treatment (F1,60 = 13.25, p < 0.001)) (Fig. 2e). While SEMA-treated rats responded significantly more than VEH-treated rats in the active nose port (t = 3.834, p < 0.001; Cohen’s d = 0.959), there was no difference between groups in responding in the inactive nose port (t = 0.455, p = 0.632; Cohen’s d = 0.114). There was not a significant effect of Sex (F1,60 = 0.03, p = 0.864) on nose port responding, but there was a significant Port x Sex interaction (F1,60 = 9.80, p = 0.003). Males responded more in the inactive port than females (p = 0.003), with no sex differences in responding in the active port (p = 0.099). Upon lever presentation, SEMA-treated rats made more contacts with the lever than VEH-treated rats (Effect of Treatment (F1,60 = 4.22, p = 0.044)); post hoc t = 2.085, p = 0.041; Cohen’s d = 0.521) (Fig. 2f). There was not a significant effect of Sex (F1,60 = 0.003, p = 0.956) nor a Treatment x Sex interaction (F1,60 = 0.16, p = 0.688) for lever-cue contacts during the conditioned reinforcement test. In agreement with the above, the incentive value index was greater for rats treated with semaglutide relative to those treated with vehicle (Effect of Treatment (F1,60 = 9.91, p = 0.003)); post hoc t = 3.145, p = 0.003; Cohen’s d = 0.786) (Fig. 2g). There was not a significant effect of Sex (F1,60 = 2.04, p = 0.159) nor a Treatment x Sex interaction (F1,60 = 0.062, p = 0.803) for the incentive value index. Together, these data indicate that chronic semaglutide treatment enhances the incentive motivational value of a cue previously associated with food reward.

Progressive ratio responding

Chronic semaglutide treatment enhanced progressive ratio responding for the food reward (Fig. 2h-j). Both groups of rats distinguished between the active and inactive ports, with more responding in the active port (Port (F1,28 = 51.99, p < 0.001)). However, relative to those treated with vehicle, rats treated with semaglutide showed more responding into the active port that resulted in reward delivery (Treatment (F1,28 = 7.43, p = 0.011); post hoc SEMA vs. VEH, t = 2.835, p = 0.009; Cohen’s d = 1.002) (Fig. 2h). There were no significant differences in responding in the inactive port (Port x Treatment interaction (F1,28 = 8.36, p = 0.007); posthoc SEMA vs. VEH (t = 1.962, p = 0.067; Cohen’s d = 0.694)), and no effect of Sex (F1,28 = 1.25, p = 0.272). SEMA-treated rats also had significantly higher breakpoints (Treatment (F1,28 = 10.45, p = 0.003)); post hoc t = 3.256, p = 0.003; Cohen’s d = 1.151) (Fig. 2i) and received more rewards in total (Treatment (F1,28 = 9.9, p = 0.004); post hoc t = 3.125, p = 0.004; Cohen’s d = 1.105) (Fig. 2j) relative to VEH-treated rats. Importantly, all SEMA- and VEH-treated rats consumed the banana pellets rewarded during progressive ratio responding. Contrary to our hypothesis, these findings indicate that rats treated with chronic semaglutide are more motivated to respond for small food rewards.

To determine whether differences in learning to respond for a primary reinforcer might have influenced the results of the progressive ratio test, we established an acquisition criterion: rats were required to respond at least 1.5 times more in the active port than the inactive port, and to reach a minimum progressive ratio of 15. Under this criterion, one male rat in the vehicle treatment group from Experiment 1 failed to meet the learning threshold. Exclusion of this rat did not alter the results (see Supplementary Information); therefore, this animal is included in all figures and analyses. Additionally, survival analysis was conducted to further examine group differences in performance on the progressive ratio test, supporting the conclusions drawn above (see Supplementary Information, Figure S1).

When analyzed across the entire population, responding in the active port during the progressive ratio test was significantly correlated with responding in the active port during conditioned reinforcement (r2 = 0.486, p = 0.005, data not shown). However, when analyzed within groups, responding in the active port during the progressive ratio and conditioned reinforcement tests was significantly correlated in VEH-treated rats (r2 = 0.701, p = 0.002), but not in SEMA-treated rats (r2 = 0.281, p = 0.292). Thus, we do not believe that greater responding at the active port during the conditioned reinforcement test impacted responding during the progressive ratio test for semaglutide-treated rats.

Experiment 2Weight change

As in Experiment 1, chronic semaglutide treatment reduced weight gain in male rats and induced weight loss in female rats relative to vehicle-treated counterparts (Fig. 3a-b). This is supported by significant main effects of Treatment (F1,59 = 149.17, p < 0.001), Day (F16,944 = 37.46, p < 0.001), and Sex (F1,59 = 88.61, p < 0.001) (Fig. 3a). Additionally, there were significant Treatment x Day (F16,944 = 29.81, p < 0.001) and Day x Sex interactions (F16,944 = 29.02, p < 0.001). After Day 1 of treatment, SEMA-treated rats weighed significantly less than VEH-treated rats (regardless of sex) for the remainder of the experiment (p ≤ 0.004) (Fig. 3a). At the end of the 17-day treatment regimen, both male and female rats treated with semaglutide had reduced weight gain with respect to baseline body weight relative to those treated with vehicle (Fig. 3b) (Males: t = 9.218, p < 0.001; Cohen’s d = 3.313; Females: t = 6.808, p < 0.001; Cohen’s d = 2.412). There were significant main effects of Treatment (F1,59 = 131.38, p < 0.001) and Sex (F1,59 = 104.17, p < 0.001), and a Treatment x Sex interaction (F1,59 = 7.02, p = 0.010) on the final day of treatment. Similar to Experiment 1, most male rats treated with semaglutide showed attenuated weight gain while most female rats treated with semaglutide showed weight loss relative to baseline levels; but both sexes showed an approximate 10% difference in the change from baseline body weight relative to those treated with vehicle.

Pavlovian conditioned approach (PavCA)

Experiment 2 was designed to classify rats as sign-trackers, intermediate responders, or goal-trackers prior to administration of semaglutide or vehicle to assess potential phenotype-dependent effects. However, the distribution of phenotypes was skewed, with a predominance of sign-trackers and relatively few goal-trackers, limiting our ability to perform sufficiently powered statistical comparisons between groups. Therefore, for subsequent data analyses and illustrations, phenotype groups were collapsed.

Chronic semaglutide treatment had no effect on the expression of PavCA behavior. Table 2 presents the LMM analyses of the individual behavioral measures (i.e., lever-cue contacts, probability of lever-cue contact, latency to lever-cue contact, food-cup entries, probability of food-cup entry, and latency to food-cup entry) used to calculate PavCA Scores. These metrics are presented separately for Sessions 1–5 (pre-treatment, Acquisition) and 6–10 (during treatment, Expression). For behavior across Sessions 1–5, there was a significant main effect of Session (p ≤ 0.012) for each measure except for food-cup entries (F4,72 = 2.29, p = 0.068), indicative of a change in behavior as the cue-reward relationship was learned. There were no significant main effects of Treatment for any of the measures (p ≥ 0.723) across Sessions 1–5, as was expected given that Treatment groups were counterbalanced across these measures. Female rats overall showed a higher probability to contact the lever-cue (Effect of Sex: F1,64 = 6.08, p = 0.016). However, there were no main effects of Sex for any of the other measures (p ≥ 0.076). For Sessions 6–10, there was a significant main effect of Session for each measure (p ≤ 0.007), but no significant main effects of Treatment (p ≥ 0.184) or Sex (p ≥ 0.495).

As shown in Fig. 3c-d, by the end of PavCA training for Sessions 1–5, all groups tended to sign-track (i.e., PavCA Score ≥ 0.5) rather than goal-track. Congruent with the findings described above for individual metrics, there was a significant main effect of Session (F4,73 = 7.94, p < 0.001), but no main effect of Treatment (F1,73 = 0.03, p = 0.870) or Sex (F1,73 = 3.16, p = 0.08), and no significant interactions (p ≥ 0.095) for PavCA Scores across Sessions 1–5 (Fig. 3c). For Sessions 6–10, the tendency to sign-track increased with further training (Effect of Session: F4,84 = 9.28, p < 0.001), but the expression of PavCA behavior was not affected by chronic semaglutide treatment (Treatment (F1,60 = 0.00, p = 0.998)) or Sex (F1,60 = 0.04, p = 0.841). There were no significant differences between males and females across individual sessions (p ≥ 0.391), and no other interactions were statistically significant (p ≥ 0.084).

In agreement with the data above, there were no significant differences between treatment groups in the PavCA Index which was averaged across Sessions 4–5 (Acquisition) and 9–10 (Expression) (Fig. 3d). There was no effect of Treatment (F1,59 = 0.03, p = 0.861), Sex (F1,59 = 0.82, p = 0.370), or Treatment x Sex interaction (F1,59 = 2.29, p = 0.136) for the Acquisition phase, nor were there significant effects for the Expression phase (Treatment (F1,59 = 0.09, p = 0.760), Sex (F1,59 = 0.01, p = 0.909), Treatment x Sex interaction (F1,59 = 3.25, p = 0.077)). Thus, consistent with the findings from Experiment 1, chronic semaglutide treatment does not appear to affect the expression of Pavlovian conditioned approach behavior to a food-paired cue.

Conditioned reinforcement

As in Experiment 1, chronic semaglutide enhanced the conditioned reinforcing properties of the lever-cue (Fig. 3e-g). Relative to vehicle, rats treated with semaglutide showed enhanced responding in the active nose port that resulted in lever-cue presentation, but not in the inactive port (Fig. 3e; Effects of Port (F1,59 = 86.25, p < 0.001); Port x Treatment interaction (F1,59 = 4.12, p = 0.047); post hoc SEMA vs. VEH active port: t = 2.124, p = 0.040; Cohen’s d = 0.535; post hoc SEMA vs. VEH inactive port: t = 0.12, p = 0.885; Cohen’s d = 0.03). Overall, female rats made more nosepokes than male rats (Sex (F1,59 = 6.89, p = 0.011)), with responding in the active port primarily driving this effect (Port x Sex interaction (F1,59 = 10.22, p = 0.002); Sex effect at the active port (p = 0.003)). Congruent with greater responding in the active port, rats treated with semaglutide made more lever contacts than those treated with vehicle (Fig. 3f; Effect of Treatment (F1,59 = 5.27, p = 0.025); post hoc t = 2.343, p = 0.022; Cohen’s d = 0.591), and SEMA-treated rats had a significantly higher Incentive Value Index than VEH-treated rats (Fig. 3g; Treatment (F1,59 = 6.65, p = 0.012); post hoc t = 2.607, p = 0.011; Cohen’s d = 0.657). Although there were no sex differences for lever contacts, female rats had a significantly higher Incentive Value Index than males (Sex (F1,59 = 6.43, p = 0.014)), but there was no Treatment x Sex interaction (F1,59 = 0.019, p = 0.890). Together, these data align with those reported in Experiment 1, suggesting that chronic semaglutide treatment enhances the incentive motivational value of a cue previously associated with food reward.

Progressive ratio responding

Chronic semaglutide treatment enhanced progressive ratio responding for the food reward (Fig. 3h-j), similar to Experiment 1. While both groups responded more in the active port than the inactive port (Effect of Port (F1,59 = 83.19, p < 0.001); posthoc (VEH: p = 0.011; SEMA: p < 0.001), SEMA-treated rats responded more in both ports compared to VEH-treated rats (Treatment (F1,59 = 28.11, p < 0.001); Port x Treatment interaction (F1,59 = 28.99, p < 0.001); posthoc active port: t = 4.984, p < 0.001; Cohen’s d = 1.256; post hoc inactive port: t = 3.545, p < 0.001; Cohen’s d = 0.893). Additionally, females made more nosepokes overall than males (Effect of Sex (F1,59 = 13.94, p < 0.001); Port x Sex interaction (F1,59 = 13.35, p < 0.001); posthoc Active: p < 0.001; Inactive: p = 0.004). In agreement with the nose port data, SEMA-treated rats had higher breakpoints relative to VEH-treated rats (Fig. 3i; Effect of Treatment (F1,59 = 34.41, p < 0.001); post hoc t = 5.361, p < 0.001; Cohen’s d = 1.351), and female rats had higher breakpoints relative to male rats (Effect of Sex (F1,59 = 18.53, p < 0.001)); but treatment did not impact the sex effects (Treatment x Sex interaction (F1,59 = 1.16, p = 0.287)). Rats treated with chronic semaglutide also received more rewards than those treated with vehicle (Fig. 3j; Effect of Treatment (F1,59 = 32.79, p < 0.001); t = 5.168, p < 0.001; Cohen’s d = 1.302), and females received more rewards than males (Effect of Sex (F1,59 = 21.97, p < 0.001)), but treatment did not impact the sex effects (Treatment x Sex interaction (F1,59 = 0.01, p = 0.922)). Importantly, all SEMA- and VEH-treated rats consumed the banana pellets rewarded during progressive ratio responding. Consistent with Experiment 1, these findings indicate that chronic semaglutide treatment enhances the motivation to respond for small food rewards.

We applied the same acquisition criterion as in Experiment 1 (i.e., at least 1.5 times more responding in the active versus inactive port and a minimum progressive ratio of 15) to assess whether differences in learning for a primary reinforcer might have influenced progressive ratio test results. Based on this criterion, seven male rats in the VEH-treated group from Experiment 2 failed to meet the learning threshold. Although these learning differences were more prevalent in VEH-treated rats, exclusion of these animals did not alter the results (see Supplementary Information); therefore, they are included in all figures and analyses. As in Experiment 1, survival analysis was also conducted to further examine group differences in progressive ratio test performance, supporting the conclusions drawn above (see Supplementary Information, Figure S1).

When analyzed across the entire population, responding in the active port during the progressive ratio test was significantly correlated with responding in the active port during conditioned reinforcement (r2 = 0.385, p = 0.002). However, when analyzed within groups, responding in the active port during the progressive ratio and conditioned reinforcement tests was not significantly correlated in VEH-treated rats (r2 = 0.301, p = 0.1) or SEMA-treated rats (r2 = 0.349, p = 0.05). Thus, we do not believe that responding during the conditioned reinforcement test impacted responding during the progressive ratio test.

Experiment 3Weight change

As with Experiments 1 and 2, chronic semaglutide treatment reduced weight gain in males and induced weight loss in females relative to vehicle-treated controls (Fig. 4a-b). This is supported by significant main effects of Treatment (F1,28 = 86.1, p < 0.001), Day (F14,392 = 84.4, p < 0.001), and Sex (F1,28 = 119.61, p < 0.001). Although there was not a significant Treatment x Sex interaction (F1,28 = 0.18, p = 0.678), the majority of female rats treated with semaglutide showed a decrease in body weight relative to baseline levels, whereas all male rats showed attenuated weight gain rather than weight loss. In support, at the end of the 17-day treatment regimen, both male and female rats treated with semaglutide had reduced weight gain with respect to baseline body weight relative to those treated with vehicle (Fig. 4b) (Males: t = 5.815, p < 0.001; Cohen’s d = 2.907; Females: t = 3.491, p = 0.004; Cohen’s d = 1.746). There were significant main effects of Treatment (F1,28 = 43.33, p < 0.001) and Sex (F1,28 = 84.57, p < 0.001), but not a significant interaction on the final day of treatment. These results are consistent with Experiments 1 and 2, with chronic semaglutide causing a ≈ 10% or greater difference in the change from baseline body weight relative to those treated with vehicle.

Fig. 4

The effects of chronic semaglutide treatment on free consumption of chow and banana-flavored reward pellets. Data are shown for vehicle- (VEH, n = 16; 8 male and 8 female) and semaglutide- (SEMA, (n = 16; 8 male and 8 female) treated rats. (a) Mean ± SEM percent change in weight from baseline for VEH- and SEMA-treated rats across 15 days of treatment. Relative to vehicle, chronic semaglutide treatment reduced weight gain in males and induced weight loss in females. (b) Black bars show mean ± SEM weight change from baseline following the final treatment on Day 15. Each data point represents an individual rat. Semaglutide treatment reduced weight gain in both males and females. (c) Mean ± SEM chow consumed per cage of VEH- and SEMA-treated rats presented in 2-day bins over the course of Experiment 3. Chronic semaglutide treatment reduced homecage chow consumption in males and females. (d) Mean ± SEM amount of chow or banana pellets consumed divided by body weight for VEH- and SEMA-treated rats during feeding tests. Each data point represents an individual rat. Relative to vehicle, chronic semaglutide treatment reduced the amount of banana pellets, but not chow, consumed during the free-feeding consumption test. *p < 0.05

Chow consumption

Homecage chow consumption was analyzed in 2-day bins over the course of the 14-day treatment period. Grouping the data in this way reduces the impact of day-to-day variability and provides a more stable estimate of intake, facilitating clearer visualization and interpretation of overall intake patterns. (Fig. 4c). Chronic semaglutide treatment reduced homecage chow consumption across a 14-day treatment period (Effects of Treatment (F1,12 = 50.14, p < 0.001); Sex (F1,12 = 161.53, p < 0.001); Bin (F6,72 = 8.37, p < 0.001)). There was a significant Treatment x Bin interaction (F6,72 = 5.26, p < 0.001) and post hoc comparisons showed that SEMA-treated rats ate less chow than VEH-treated rats during each 2-day bin (p ≤ 0.012). No other interactions were statistically significant (p ≥ 0.087). These data further highlight the validity of our chronic semaglutide treatment regimen, showing that it attenuates homecage chow consumption.

Free-feeding consumption tests

During the free-feeding consumption tests, chronic semaglutide treatment reduced the amount of banana-flavored reward pellets consumed relative to vehicle but had no effect on chow consumption (Fig. 4d). Overall, rats ate more banana-flavored pellets than chow (main effect of Food Type (F1,28 = 32.27, p < 0.001; Treatment x Food Type interaction (F1,28 = 5.29, p = 0.029)). Post hoc comparisons showed that SEMA-treated rats ate significantly fewer banana pellets than VEH-treated rats (t = 2.167, p = 0.038; Cohen’s d = 0.766), with no difference between groups with respect to chow consumption (t = 0.17, p = 0.865; Cohen’s d = 0.060). No other interactions were statistically significant (p ≥ 0.058) for the free-feeding consumption test. Thus, chronic semaglutide treatment appears to decrease the consumption of salient food rewards when such rewards are unlimited.