GABAB Receptor Signaling in the Mesolimbic System Suppresses Binge-like Consumption of a High-Fat Diet.

Binge eating could contribute to the development of obesity, and previous studies suggest that gamma-aminobutyric acid (GABA) type B receptor (GABABR) signaling is involved in the regulation of binge eating. Here, we show that time-restricted access to a high-fat diet (HFD) induces binge-like eating behavior in wild-type mice. HFD consumption during restricted time was significantly increased in corticostriatal neuron-specific GABABR-deficient mice compared with wild-type mice. Furthermore, the GABABR agonist baclofen suppressed HFD intake during restricted time in wild-type mice but not in corticostriatal or dopaminergic neuron-specific GABABR-deficient mice. In contrast, there were no significant differences in food consumption among genotypes under ad libitum access to HFD. Thus, our data show that the mesolimbic system regulates food consumption under time-restricted but not ad libitum access to HFD and have identified a mechanism by which GABABR signaling suppresses binge-like eating of HFD.

Introduction

Obesity has become a major health concern worldwide, as it is associated with the development of various diseases such as type 2 diabetes, cardiovascular diseases, cancer, and mood-related disorders (Finkelstein et al., 2009). Obesity is caused when energy intake overwhelms energy expenditure, and predisposing factors to obesity include excess in palatable and calorie-rich food intake such a high-fat diet (HFD) (O'Rahilly, 2009), as well as irregular eating such as binge eating (Kessler et al., 2016). Feeding behavior is controlled by both the homeostatic and reward systems (Waterson and Horvath, 2015). The latter is mainly composed of the mesolimbic system, in which dopaminergic neurons in the ventral tegmental area (VTA) project to the nucleus accumbens (NAc) and caudate putamen (CPu) in the striatum (Kenny, 2011).

Gamma-aminobutyric acid (GABA), the inhibitory neurotransmitter that has been implicated in the regulation of the mesolimbic system (Hayes et al., 2014), acts on two types of receptors: ionotropic GABAA and GABAC, and metabotropic GABAB receptors (GABABRs) that are located both pre- and postsynaptically (Bettler et al., 2004, Gassmann and Bettler, 2012). Previous studies suggest that the GABABR agonist baclofen is effective in reducing binge-like eating in rodents (Berner et al., 2009, Buda-levin et al., 2005, Czyzyk et al., 2010, Rao et al., 2008, Wojnicki et al., 2013, Wong et al., 2009) as well as binge eating in humans (De Beaurepaire et al., 2015, Broft et al., 2007, Corwin et al., 2012), although the site of action remains to be elucidated.

The time-restricted access to HFD causes binge-like eating behavior, as shown by increases in motivation to consume (Lardeux et al., 2013) and gradual increases in the consumption (Furlong et al., 2014, Valdivia et al., 2015). These behaviors are accompanied by the activation of neurons in both VTA and NAc (Valdivia et al., 2015) and increases in extracellular dopamine concentrations in the NAc (Liang et al., 2006, Naef et al., 2015, Sahr et al., 2008), suggesting that the time-restricted access to HFD is a good model to investigate the mechanisms underlying binge eating of HFD.

To investigate the role of GABABRs in the mesolimbic system in binge eating, we generated mice that lack GABABRs exclusively in dopaminergic or corticostriatal neurons and compared their binge-like eating behavior induced by the time-restricted access to HFD with wild-type (WT) littermate mice.

Results

Generation of Dopaminergic Neuron-Specific and Corticostriatal Neuron-Specific GABABR-Deficient Mice

To generate dopaminergic neuron-specific GABABR deficient (D-KO) mice, GABAR mice were crossed with dopamine transporter (DAT)-Cre (DAT-Cre) mice. Then, we crossed the GABAR DAT-Cre mice with GABAR or GABAR mice to yield GABAR DAT-Cre mice and littermate controls (hereafter termed WT mice). GPR88 is an orphan G-protein-coupled receptor that is highly expressed in striatal and cortical neurons (Hisatsune et al., 2013, Quintana et al., 2012). To generate corticostriatal neuron-specific GABABR-deficient (CS-KO) mice, GABAR mice were crossed with GPR88-Cre mice. Then, we crossed the GABAR GPR88-Cre mice with GABAR or GABAR mice to yield GABAR GPR88-Cre mice and WT mice. Deletion of the GABA receptor allele in D-KO mice was only detected in DNA extracts from the VTA (Figures 1A and S1A). Similarly, deletions of the GABA receptor allele in CS-KO mice were detected in DNA extracts from the NAc, CPu, medial prefrontal cortex (mPFC), and orbitofrontal cortex (OFC) (Figures 1B and S1B). In contrast, no recombined alleles were detected in WT mice (Figures 1A, 1B, S1A, and S1B). To visualize dopaminergic neuron-specific and corticostriatal neuron-specific Cre-mediated recombination, we crossed D-KO and CS-KO mice to ROSA26 Cre-reporter knockin mice (hereafter termed R26GRR), in which green fluorescence changed to red fluorescence in Cre-recombined cells (Hasegawa et al., 2013). GABAR DAT-Cre R26GRR mice expressed tdsRed-positive cells in the VTA, and the tdsRed co-localized with DAT immunostaining (Figures 1C and S1C). Likewise, GABAR GPR88-Cre R26GRR mice expressed tdsRed-positive cells in NAc and CPu, and the tdsRed was co-localized with GPR88 immunostaining (Figures 1D and S1D). Immunostaining of both DAT and GABABRs revealed that GABABRs were expressed in the dopaminergic neurons of the VTA in WT mice, whereas GABABRs-expressing cells in the VTA were rarely detected in D-KO mice (Figures 1E and 1F). Similarly, immunostaining of both GPR88 and GABABRs revealed that GABABRs were expressed in the GPR88-positive neurons in NAc, CPu, mPFC, and OFC in WT but not in CS-KO mice (Figures 1G–1J and S1E–S1H).

Figure 1

Generation of Dopaminergic Neuron-Specific and Corticostriatal Neuron-Specific GABABR Deficient Mice

(A and B) Detection of deleted GABAR alleles (Δ) in GABARDAT-Cre (D-KO) mice and GABARGPR88-Cre (CS-KO) mice. DNA was extracted from different tissues, and deletion of the floxed allele was detected by PCR. Vta, ventral tegmental area; Sub, substantia nigra; Hyp, hypothalamus; Cor, cerebral cortex; Hip, hippocampus; Cer, cerebellum; BS, brain stem; mPfc, medial prefrontal cortex; Ofc, orbitofrontal cortex; NAc, nucleus accumbens; CPu, caudate putamen; PC, positive control; NC, negative control. PCR reaction with GAPDH was used as an internal control.

(C and D) Double-color imaging of EGFP (green) and tdsRed (magenta) fluorescence to assess DAT-Cre and GPR88-Cre. The VTA of GABARDAT-Cre R26GRR mice as compared with that of GABARR26GRR mice (C). The NAc and CPu of GABARGPR88-Cre R26GRR mice as compared with those of GABARR26GRR mice (D). Scale bar: 100 μm.

(E and F) The representative photographs showing the staining of DAT (green), GABAB1R (magenta), and DAPI (blue) in VTA in WT and D-KO mice. White arrow heads show colocalization of DAT and GABAB1R. Scale bar: 40 μm.

(G–J) The representative photographs showing the staining of GPR88 (green), GABAB1R (magenta), and DAPI (blue) in NAc (G and H) and CPu (I and J) in WT and CS-KO mice. White arrow heads show colocalization of GPR88 and GABAB1R. Scale bar: 20 μm. All data are from male mice.

See also Figure S1.

Time-Restricted Access to HFD Gives Rise to Binge-like Eating

To examine the hedonic regulation of HFD intake, male WT mice were divided into three groups, “control group,” “ad libitum HFD group,” and “restricted HFD group” (Figures 2A and 2B). All mice were fed only a chow diet (CD) from conditioning day 1–5. Then, the mice in control group were given free access to both CD and HFD for 2 days, followed by access to only CD (Figures 2A and 2B). The mice in ad libitum HFD group could access both CD and HFD from the conditioning day 6 to the end of the experiments (Figure 2B). In the restricted HFD group, mice were given free access to both CD and HFD on conditioning days 6 and 7, followed by access to only CD from conditioning day 8–12. The mice were then given the access to HFD for 2 h (zeitgeber time 12 to 14), whereas they could always access CD from experimental day 1–6 (Figures 2A and 2B). The protocols of conditioning and experimental days are determined based on previous studies (Bake et al., 2014, Berner et al., 2008, Cao et al., 2014, Czyzyk et al., 2010, Johnson and Kenny, 2010, King et al., 2016) with some modifications.

Figure 2

Time-Restricted Access to HFD Gives Rise to Binge-like Eating

(A) Photographs of chow feeding (termed as C) and CD and HFD feeding (termed as C + H).

(B) Protocol for the experiment. WT mice were divided into three groups, control group (Control), ad libitum HFD group (Ad-HFD), and restricted HFD group (R-HFD). The mice in R-HFD have ad libitum access to CD and HFD for 2 h (ZT 12–14) and only CD during the rest of the day (for 22 h) in experimental days.

(C) The mice in Ad-HFD showed increases in body weight compared with R-HFD and Control (group: F (2,15) = 9.580, p = 0.002; time: F (17,255) = 25.412, p < 0.001; group × time interaction: F (34,255) = 9.311, p < 0.001, n = 6 per group).

(D) The mice in Ad-HFD showed increases in daily food intake compared with R-HFD and Control (group: F (2,15) = 8.479, p = 0.003; time: F (17,255) = 36.294, p < 0.001; group × time interaction: F (34,255) = 5.309, p < 0.001, n = 6 per group).

(E) The mice in R-HFD consumed more calories during the restricted time (zeitgeber time 12 to 14) than mice in the other 2 groups (group: F (2,16) = 80.096, p < 0.001; time: F (2,32) = 175.227, p < 0.001; group × time interaction: F (4,32) = 26.085, p < 0.001, n = 5–9 per group).

(F) The ratios of HFD intake to daily calorie intake on experimental days 3–6 were significantly increased compared with that on experimental day 1 in R-HFD (time effect: F (5,50) = 9.580, p < 0.001, n = 11). The calorie intake in Ad-HFD group was almost from HFD, but not from CD, throughout experimental days (n = 6).

All values are mean ± SEM. Statistical analysis were performed using two-way ANOVA assessed by repeated measures (C–E) or one-way ANOVA assessed by repeated measures (F) followed by Bonferroni post hoc test. #p < 0.05, ###p < 0.001 versus Control; *p < 0.05, ***p < 0.001 versus R-HFD; †††p < 0.001 versus Ad-HFD; §p < 0.05, §§§p < 0.001 versus the ratio of HFD in R-HFD on experimental day 1. See also Table S1 for the details of statistics.

The mice in the ad libitum HFD group showed increases in both body weight and daily food intake compared with restricted HFD and control groups (Figures 2C and 2D). The findings were supported by a group effect (F (2,15) = 9.580, p = 0.002, for body weight; F (2,15) = 8.479, p = 0.003, for food intake) and an interaction effect between time and group (F (34,255) = 9.311; p < 0.001, for body weight; F (34,255) = 5.309, p < 0.001, for food intake). In both control and restricted HFD groups, the food intake was decreased when the mice were returned to CD from HFD on conditioning day 8, as reported previously (Berner et al., 2008, Czyzyk et al., 2010). The mice in the restricted HFD group consumed only HFD during the restricted time (zeitgeber time 12 to 14), and they consumed more calories during this period than mice from the other two groups (Figure 2E). The mice in the restricted HFD group consumed about 60% of the daily calorie intake from HFD during the restricted time on experimental day 1, and consistent with previous studies (Johnson and Kenny, 2010, Valdivia et al., 2015), the ratios of HFD intake to daily calorie intake on experimental days 3–6 are significantly increased compared with that on experimental day 1 (time effect; F (5,50) = 80.096, p < 0.001, Figure 2F). On the other hand, calorie intake in the ad libitum HFD Ad-HFD group was almost from HFD, but not from CD, throughout experimental days (Figure 2F). These data suggest that time-restricted access to HFD gives rise to a pattern of binge-like eating of HFD.

GABAB Receptor Signaling in Corticostriatal Neurons Suppresses HFD Consumption under Time-Restricted Access to HFD

To clarify the role of GABABR signaling in the hedonic regulation of HFD intake, we placed male D-KO and CS-KO mice on time-restricted HFD access. The HFD intake during the 2 h was significantly increased in male CS-KO, but not in D-KO, compared with WT mice in experimental days 1 and 2 (Figures 3A and 3B). This was supported by a genotype effect between WT and CS-KO mice (F (1,14) = 7.475, p = 0.003) and post hoc tests between WT and CS-KO mice on experimental days 1 (F (1,14) = 7.661, p = 0.015) and 2 (F (1,14) = 6.603, p = 0.022). On the other hand, the daily intake of CD was similar among groups (Figures S2A and S2B). We also found similar results in female mice (Figures 3C, 3D, S2C, and S2D). There were no differences in body weight among WT, D-KO, and CS-KO mice during these experiments (data not shown). These data suggest that endogenous GABABR signaling in corticostriatal neurons, but not in dopaminergic neurons, suppresses HFD intake under time-restricted access to HFD.

Figure 3

GABAB Receptor Signaling in Corticostriatal Neurons Suppresses HFD Consumption under Time-Restricted Access to HFD

(A and C) HFD intake of male (A) and female (C) GABARDAT-Cre (D-KO) and GABAR (WT) mice during 2 h (ZT 12–14) in R-HFD (male and female: WT, n = 11; D-KO, n = 5).

(B and D) HFD intake of male (B) and female (D) GABARGPR88-Cre (CS-KO) and WT mice during 2 h (ZT 12–14) in R-HFD. The HFD intake during the 2 h was significantly increased in CS-KO compared with WT mice on experimental days 1 and 2 (male: genotype: F (1,14) = 7.475, p = 0.003; time: F (5,70) = 31.215, p < 0.001; genotype × time interaction: F (5,70) = 1.074, not significant, WT, n = 11; CS-KO, n = 5; female: genotype: F (1,15) = 5.025, p = 0.041; time: F (5,75) = 20.126, p < 0.001; genotype × time interaction: F (5,75) = 0.811, not significant, WT, n = 10; CS-KO, n = 7).

All values are mean ± SEM. Statistical analyses were performed using two-way ANOVA assessed by repeated measures followed by Bonferroni post hoc test. *p < 0.05 versus WT. See also Table S2 for the details of statistics.

Baclofen Suppresses HFD Consumption under Time-Restricted Access to HFD via GABABR Signaling in Dopaminergic and Corticostriatal Neurons

To evaluate the effects of GABABR agonists on the hedonic regulation of HFD intake, we injected male mice in the restricted HFD group with baclofen interperitoneally 30 min before the beginning of dark cycle (ZT 12) on experimental days 1 and 5. Baclofen at a dose of 3 μg/g body weight reduced HFD intake for 10 min, 30 min, and 2 h in male WT mice compared with vehicle on both days (treatment effect; day 1: F (1,12) = 39.602, p < 0.001; day 5: F (1,12) = 21.484, p = 0.001), whereas it had no effect on the daily intake of CD (Figures 4A and 4D). In contrast, the effect of baclofen on HFD intake was absent in male D-KO (Figures 4B and 4E) and CS-KO mice (Figures 4C and 4F). Similar results were found in female mice on experimental day 5 (Figures 4G–4I). Baclofen at a dose of 0.3 μg/g body weight also suppressed HFD intake for 2 h compared with vehicle on day 5 in male WT mice but not in D-KO or CS-KO mice (Figures S3B–S3D). Baclofen had no effect on the locomotor activity in WT (Figure S3A), D-KO, and CS-KO mice (data not shown). Thus, baclofen suppresses HFD intake via GABABR signaling in the mesolimbic system under time-restricted access to HFD.

Figure 4

Baclofen Suppresses HFD Consumption under Time-Restricted Access to HFD via GABABR Signaling in Dopaminergic and Corticostriatal Neurons

(A, D, and G) Intake of HFD during ZT12-14 and daily intake of CD in male (A and D) and female (G) GABAR (WT) mice in R-HFD under treatment of baclofen on experimental days 1 (A) and 5 (D and G). Baclofen at a dose of 3 μg/g body weight reduced HFD intake for 10 min, 30 min, and 2 h in male WT mice compared with vehicle on days 1 (treatment: F (1,12) = 39.602, p < 0.001; time: F (2,24) = 626.053, p < 0.001; treatment × time interaction: F (2,24) = 20.949, p < 0.001, n = 7 per group) and 5 (treatment: F (1,12) = 21.484, p = 0.001; time: F (2,24) = 165.583, p < 0.001; treatment × time interaction: F (2,24) = 4.286, p = 0.026, n = 7 per group) as well as in female WT mice on day 5 (treatment: F (1,10) = 46.609, p < 0.001; time: F (2,20) = 167.084, p < 0.001; treatment × time interaction: F (2,20) = 4.465, p = 0.025, n = 6 per group).

(B, E, and H) Intake of HFD during ZT12–14 and daily intake of CD in male (B and E) and female (H) GABARDAT-Cre (D-KO) and WT mice in R-HFD under treatment of baclofen on experimental days 1 (B) and 5 (E and H). The inhibitory effect of baclofen on HFD intake was absent in male D-KO on days 1 (treatment: F (1,35) = 22.669, p < 0.001; genotype: F (1,35) = 0.568, not significant; treatment × genotype interaction: F (1,35) = 9.658, p = 0.004, n = 9–10 per group) and 5 (treatment: F (1,32) = 23.803, p < 0.001; genotype: F (1,32) = 2.969, not significant; treatment × genotype interaction: F (1,32) = 6.546, p = 0.015, n = 7–10 per group) as well as in female D-KO mice on day 5 (treatment: F (1,32) = 20.867, p < 0.001; genotype: F (1,32) = 4.326, p = 0.025; treatment × genotype interaction: F (1,32) = 6.026, p = 0.02, n = 7–10 per group).

(C, F, and I) Intake of HFD during ZT12–14 and daily intake of CD in male (C and F) and female (I) GABARGPR88-Cre (CS-KO) and WT mice in R-HFD under treatment of baclofen on experimental days 1 (C) and 5 (F and I). The inhibitory effect of baclofen on HFD intake was absent in male CS-KO on days 1 (treatment: F (1,31) = 16.937, p < 0.001; genotype: F (1,31) = 12.671, p < 0.001; treatment × genotype interaction: F (1,31) = 4.503, p = 0.021, n = 7–10 per group) and 5 (treatment: F (1,36) = 9.661, p = 0.004; genotype: F (1,36) = 6.320, p = 0.017; treatment × genotype interaction: F (1,36) = 9.770, p = 0.003, n = 9–11 per group) as well as in female CS-KO on day 5 (treatment: F (1,38) = 18.808, p < 0.001; genotype: F (1,38) = 5.077, p = 0.03; treatment × genotype interaction: F (1,38) = 10.005, p = 0.003, n = 9–11 per group).

All values are mean ± SEM. Statistical analysis were performed using two-way ANOVA assessed by repeated measures (A, D, and G) or two-way factorial ANOVA (B, C, E, F, H, and I) followed by Bonferroni post hoc test. *p < 0.05, ***p < 0.001 versus vehicle in WT. Ns, not significant. See also Table S3 for the details of statistics.

There Were No Significant Differences in Energy Balance or the Effects of Baclofen on Feeding Behavior among Genotypes under ad libitum Access to HFD or CD

Finally, we examined the role of GABABR signaling in the mesolimbic system in the regulation of food consumption and body weight under ad libitum access to HFD or CD. There were no significant differences in daily food intake, body weight, feed efficiency (Δ body weight/Δ food intake), or fat pad weight between WT and D-KO mice (Figures 5A–5H) or CS-KO mice on HFD (Figures 5I–5P) or CD (data not shown). There were no significant differences between genotypes in glucose metabolism estimated by fasted serum glucose, intraperitoneal glucose tolerance test, and insulin tolerance test (Figures S4A–S4F). The administration of baclofen (3 μg/g body weight) reduced daily HFD intake and body weight in WT mice, as reported previously (Sato et al., 2007) (Figures 5Q–5T), and it also reduced daily HFD intake and body weight in D-KO (Figures 5Q and 5R) and CS-KO mice (Figures 5S and 5T). These data suggest that the GABABR signaling in the mesolimbic system does not affect the feeding behavior under ad libitum access to HFD.

Figure 5

There Were No Significant Differences in Energy Balance or the Effects of Baclofen on Feeding Behavior among Genotypes under ad libitum Access to HFD

(A, E, I, and M) Body weight of male (A and I) and female (E and M) GABARDAT-Cre (D-KO) (A and E), GABARGPR88-Cre (CS-KO) (I and M) and GABAR+/+ (WT) (A, E, I, and M) mice under ad libitum access to HFD (n = 8–10 per group).

(B, F, J, and N) Epididymal (B and J) and perigonadal (F and N) fat pad weight of male D-KO (B and F), CS-KO (J and N), and WT (B, F, J, and N) mice at the age of 16 weeks (n = 8–9 per group).

(C, G, K, and O) Daily food intake of male (C and K) and female (G and O) D-KO (C and G), CS-KO (K and O), and WT (C, G, K, and O) mice at the age of 8 weeks (n = 6–10 per group).

(D, H, L, and P) Feed efficiency of male (D and L) and female (H and P) D-KO (D and H), CS-KO (L and P) and WT (D, H, L, and P) mice at the age of 8 weeks (n = 6–10 per group).

(Q–T) Body weight changes (Q and S) and daily food intake (R and T) of male D-KO (Q and R), CS-KO (S and T), and WT (Q–T) mice treated with baclofen (3 μg/g body weight every 6 h) or vehicle for 2 days under ad libitum access to HFD (Q: treatment: F (1,20) = 15.961, p < 0.001; genotype: F (1,20) = 3.289, not significant; treatment × genotype interaction: F (1,20) = 2.932, not significant; R: treatment: F (1,20) = 79.515, p < 0.001; genotype: F (1,20) = 2.411, not significant; treatment × genotype interaction: F (1,20) = 3.147, not significant; S: treatment: F (1,20) = 40.887, p < 0.001; genotype: F (1,20) = 0.497, not significant; treatment × genotype interaction: F (1,20) = 0.157, not significant; T: treatment: F (1,20) = 44.684, p < 0.001; genotype: F (1,20) = 0.575, not significant; treatment × genotype interaction: F (1,20) = 0.026, not significant, n = 6 per group).

BW, body weight; FI, food intake. All values are mean ± SEM. Statistical analysis were performed using two-way ANOVA assessed by repeated measures (A, E, I, and M), unpaired t test (B–D, F–H, J–L, and N–P) or two-way factorial ANOVA (Q–T) followed by Bonferroni post hoc test. *p < 0.05, ***p < 0.001 versus vehicle in the same genotype. See also Table S4 for the details of statistics.

Discussion

In the present study, we generated dopaminergic neuron- and corticostriatal neuron-specific GABABR-deficient mice and demonstrated that HFD intake during the time-restricted access was significantly increased in corticostriatal neuron-specific KO mice compared with WT mice. Furthermore, the suppressing effect of baclofen on HFD intake during the restricted time observed in WT mice was absent in both KO mice. On the other hand, there were no significant differences in daily food consumption or body weight under ad libitum access to HFD among genotypes. Thus, our data indicate that the GABABR signaling in the mesolimbic system suppresses food consumption under time-restricted access to HFD.

Consistent with previous studies (Berner et al., 2009, Buda-levin et al., 2005, Czyzyk et al., 2010, Johnson and Kenny, 2010, Rao et al., 2008, Valdivia et al., 2015, Wong et al., 2009), WT mice under the time-restricted access to HFD consumed a substantial amount of calories from the HFD, and baclofen suppressed the HFD intake in WT mice. Although the site of action of baclofen has not been investigated so far, we now clearly demonstrate that baclofen suppresses HFD intake via GABABRs in dopaminergic and corticostriatal neurons. The effects of baclofen observed in the present study might be mediated via a decreased dopaminergic tone in the mesolimbic system, as it is reported that (1) dopaminergic neurons in the mesolimbic system receive GABAergic inputs from various neurons (Filip et al., 2015) and (2) GABA inhibits mesolimbic dopamine signaling in the VTA and striatum (Ferrario et al., 2016, Pierce and Kumaresan, 2006).

Our data also showed differences between CS-KO and D-KO mice: HFD intake under the restricted time was significantly increased in CS-KO, but not in D-KO, compared with WT mice. These data suggest that the suppressive effects of endogenous GABABR signaling on hedonic HFD intake during the restricted time is more dominant in cortocostriatal than dopaminergic neurons. Because the striatal neurons receive neural projections from not only the VTA but also other areas such as hippocampus, amygdala, prefrontal cortex, and hypothalamus(Ferrario et al., 2016), it is possible that the lack of GABABRs in the striatal neurons enhances the activity of these neurons. The detailed mechanisms by which the absence of GABABRs in the corticostriatal neurons enhances hedonic consumption of HFD needs to be clarified in future experiments.

Our data clearly demonstrate that there are no significant differences in daily food consumption or body weight between WT, CS-KO, and D-KO mice under ad libitum access to HFD and that baclofen suppressed HFD intake in all genotypes. We previously reported that baclofen decreased orexigenic neuropeptide Y mRNA expression while increasing anorexic proopiomelanocortin mRNA expression in the hypothalamic arcuate nuclei, leading to reduced food intake and body weight in WT mice fed HFD (Sato et al., 2007). Furthermore, proopiomelanocortin neuron-specific GABABR KO mice showed increased body weight under ad libitum access to HFD (Ito et al., 2013). Taken together with the data presented herein, it is suggested that GABABR signaling in the mesolimbic system regulates hedonic food consumption, whereas that in hypothalamic neurons plays an important role in homeostatic regulation of energy balance. These results contrast with previous studies showing that food intake and body weight are increased in mice with a genetic lack of insulin or leptin receptors in the mesolimbic system (Brönneke et al., 2011, Georgescu et al., 2006) and further highlight a role of the GABABR signal in the mesolimbic system in binge-like eating of HFD.

As shown in Figure 2F, intake of HFD during 2 h under the time-restricted access increased only in the first 2 days. These results are consistent with previous studies in which the duration of time-restricted HFD was set for 30 days or longer (Bake et al., 2014, Berner et al., 2008, Johnson and Kenny, 2010, King et al., 2016) and suggest that, although time-restricted access to HFD induces binge-like eating behavior, HFD intake reaches a plateau in the first few days. The finding that CD consumption was decreased when the mice were returned to CD from HFD on day 8 (Figure 2D) is also consistent with previous studies (Berner et al., 2008, Corwin et al., 1998, Czyzyk et al., 2010, King et al., 2016). A possible interpretation is that mice subjected to the time-restricted access to HFD learned to wait for HFD.

In conclusion, our data show that the mesolimbic system regulates binge-like eating of HFD and provide a mechanism by which the GABABR signaling suppresses palatable food consumption.

Limitations of the Study

GPR88-positive neurons include not only striatal neurons but also mPFC and OFC. Indeed, we showed that GABABRs were knocked out in mPFC and OFC in CS-KO mice. As these areas have a crucial role in decision making (Rangel et al., 2008) and have been implicated in reward-guided behavior (Miller and Cohen, 2001), we cannot exclude the possibility that phenotypes observed in CS-KO mice were due to deficiency of GABABRs in mPFC or OFC. Furthermore, both dopamine D1 and D2 receptors are expressed in GPR88-positive neurons (Massart et al., 2009); it remains to be established which receptor is critical for GABABR signaling to suppress binge-eating behavior.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.