Excessive consumption of high-calorie foods is a key contributor to the development and progression of obesity in humans and animals. Chronic exposure to a HFD profoundly influences eating behaviours, particularly those driven by the pleasurable (that is, hedonic) properties of food. To investigate how a chronic HFD affects these behaviours, we placed C57Bl/6 mice on a chronic HFD. Although these mice consistently preferred high-fat chow over standard chow in their home cages, they paradoxically exhibited a reduced drive to opportunistically consume high-calorie foods in an acute feeding assay, even when no effort was required to obtain the food. One possible explanation is that chronic HFD exposure leads to a reduction in the hedonic value of high-calorie foods, decreasing their pleasurable or rewarding aspects.
a, Body weight of REG mice that are switched to HFD and then returned to REG. Arrows indicate timing of acute feeding assays on REG (green) and HFD (red) (***P< 0.001, 1-way repeated measures ANOVA with Holm–Šídák multiple comparisons test; n = 12 mice). b,c, Mean weekly consumption of regular chow and high-fat (HF) chow (**P = 0.0022) (b) and caloric intake (**P = 0.0036) (c) in home cages while on REG or HFD (n = 3 cages; normalized as grams per mouse per week; 2-sided paired Student’s t-test). d, Mean jelly consumption during acute feeding assays for REG mice and after 4 weeks of HFD (***P< 0.001, 2-sided paired Student’s t-test; n = 12 mice). e, Experimental design. f, Acute feeding assay. g, Timeline: trial 1 (habituation, no food); trials 2 and 3 (food presentation, chow or jelly, counterbalanced); trial 4 (NAcLat→VTA opto-tagging). h, Food consumption for REG and HFD mice (***P< 0.001, 2-way repeated measures ANOVA with Holm–Šídák test; REG: n = 8 mice, HFD: n = 7 mice). i,j, DLC behavioural motifs for REG (i) and HFD (j) mice, with example unit firing rates and piezo activity. k, Top, z-scored average of all recorded action potentials across trials relative to events. Total unit events analysed for each motif to determine whether the unit shows significantly increased response (IR) or decreased response (DR) relative to baseline (average unit waveform in inset). Bottom, sample action potentials during feeding or walking (arrows show event onsets; *P = 0.024, ***P = 0.0002; 2-sided Wilcoxon signed-rank test). l,m, Relative z-score average of individual NAcLat→VTA units for REG (l) and HFD (m) mice during different behavioural motifs. Bar graphs show percentage of IR, DR and non-responsive units in each behavioural motif (***P = 0.0002, 2-sided Chi-squared test for proportions with Bonferroni correction for multiple comparisons; REG: n = 21 units from n = 8 mice, HFD: n = 20 units from n = 7 mice). Data are mean ± s.e.m. (error bars or shading).
The mesolimbic dopamine system, consisting of dopamine-producing cells in the VTA projecting to the nucleus accumbens (NAc), has been implicated in the motivational aspects of feeding behaviour, although other dopaminergic projections are also likely to be involved. Activation of VTA dopamine neurons projecting to the NAc is associated with the rewarding aspects of food consumption. Anticipation of food or liquid rewards enhances dopamine neuron firing, promoting goal-directed behaviours. Conversely, chronic HFD exposure has been shown to reduce dopamine activity in both mice and humans, potentially impairing reward-related processes and contributing to obesity.
Although much research has focused on the role of dopamine neurons in feeding and obesity, less is known about the effects of chronic HFD on inhibitory projections from the NAc to the VTA. The NAc provides substantial GABAergic input to the VTA, directly and indirectly influencing dopamine neurons. We previously demonstrated that optogenetic stimulation of the NAcLat→VTA pathway induces robust reward-related behaviours, such as place preference and intracranial self-stimulation, possibly via disinhibition of dopamine neurons. However, whether increased neural activity in this pathway is associated with hedonic feeding behaviours and how it is affected by diet-induced obesity remain unclear.
To study the effects of chronic HFD on NAcLat→VTA activity during feeding behaviours, we first injected C57Bl6 mice with a retrogradely transported virus carrying Cre recombinase (pseudotyped equine infectious anaemia virus, RG-EIAV-Cre) into the lateral VTA and a Cre-dependent adeno-associated virus (AAV) carrying ChR2 into the NAcLat. The mice were also implanted with a custom-made drivable optoelectrode (optrode) above the NAcLat (Fig. 1e). In these mice, ChR2 expression was largely restricted to the NAcLat and not observed in the NAc core or NAc medial shell (Extended Data Fig. 1a–d). Two weeks after stereotaxic surgery, mice were randomly split into two cohorts, with one cohort remaining on a regular diet (REG mice; 4% fat, standard mouse chow) and the other being placed on a HFD where both regular (4% fat) and high-fat chow (60% fat) were freely available in the home cage. HFD mice rapidly gained weight when compared with REG mice (Extended Data Fig. 1e). After 30 days of diet, we recorded the neural activity of opto-tagged NAcLat→VTA cells (examples of opto-tagging in units from REG and HFD mice are shown in Extended Data Fig. 1i–r) during free exploration of an open-field chamber containing calorie-rich (jelly) and low-calorie (chow) foods (Fig. 1f,g). The behaviour of REG and HFD mice was recorded on video and analysed using DeepLabCut (DLC) to identify discrete behavioural motifs (Extended Data Fig. 1f), which included actions such as feeding, touching an empty food cup, rearing, turning and various forms of locomotion at different velocities (Extended Data Fig. 1g,h). A piezo sensor placed under the food cup was used to detect precise feeding event timestamps, which showed a strong correlation with DLC-detected feeding events (Extended Data Fig. 2a–d). Food consumption was measured by weighing the food cups after each session. As expected, REG mice consumed significantly more jelly than chow, whereas HFD mice consumed less jelly overall (Fig. 1h). To analyse neural activity, we quantified unit firing rates before and during the onset of each behavioural motif and classified responses as unchanged (non-responsive), significantly increased (IR type) or significantly decreased (DR type) (Fig. 1i–k). No significant differences were observed in the average time spent in each DLC-detected motif between REG and HFD mice (Extended Data Fig. 2e). Of note, firing rates were negatively correlated with total time spent in each motif, with higher firing rates during motifs with less time spent (for example, jelly and chow feeding or touching the empty cup) and lower firing rates during longer-duration motifs such as locomotion (Extended Data Fig. 2f). Next, we assessed the proportions of classified response types in REG and HFD mice to determine whether they differed between diets. In REG mice, opto-tagged units showed high firing rates during jelly consumption, with the majority of units exhibiting significantly increased responses, whereas other behavioural motifs frequently showed decreased responses. By contrast, opto-tagged units in HFD mice displayed lower firing rates during jelly consumption, with none of the tagged units reaching statistical significance (Fig. 1l,m). Similar results were observed in piezo-based analyses, with increased firing rates during jelly consumption in REG mice and a marked reduction in HFD mice (Extended Data Fig. 2a,b). Non-tagged units also exhibited higher firing rates during jelly consumption compared to other motifs as well as reduced proportions of IR responses in HFD mice compared with REG mice for both DLC- and piezo-based analyses, although the effect size was smaller (Extended Data Fig. 2g–j). Together, these results suggest that increased activity of NAcLat→VTA cells is associated with hedonic feeding, but chronic HFD disrupts this relationship.
Next, we examined whether increased activity in the NAcLat→VTA pathway is sufficient to induce feeding behaviour. We injected AAV-hSyn-ChR2 or AAV-hSyn-eYFP into the NAcLat of C57Bl6 mice and implan
