Neural Mechanism of Hunger-gated Food-seeking and Evaluating

The physiological need for energy evokes motivated feeding behaviours that help to ensure survival. However, the neural mechanisms underlying the generation of food motivation remain poorly understood. We investigated these mechanisms by subdividing feeding-related motivated behaviours into food-seeking, evaluating, and swallowing. Micro-endoscopic results indicated that neurons containing leptin receptors (LepRs) in the lateral hypothalamus (LH) are the major food-specific subpopulation of LH neurons. Optogenetic manipulation of LH neurons bidirectionally regulated both food-seeking and evaluating. Furthermore, micro-endoscope data revealed that distinct LH neurons encode seeking and evaluating. Computational modelling analysis demonstrated that LH neurons encode motivation, whereas neurons containing agouti-related peptide and neuropeptide Y (AgRP/NPY) encode the need for food. Additionally, slice studies revealed that NPY decreases inhibitory input to LH neurons via LH interneurons. This mechanism explains the permissive gate role of hunger (food need) in seeking/evaluating motivation. Together, the present study provides a comprehensive neural mechanism of how physiological needs drive distinct motivated behaviours. Introduction To maintain homeostasis, physiological need drives the diverse motivations required to evoke appropriate behaviours 2, . In the context of feeding, motivated behaviours are divided into “seeking” (moving toward the target), “evaluating” (examining the target by tasting, smelling, etc.), and “swallowing” (consuming the target) 5, 6, . These separate food-motivated behaviours are orchestrated by exteroceptive and interoceptive signals in the brain to execute the appropriate feeding behaviours required for survival . To study the precise mechanisms orchestrating these distinct motivations, we developed specific behavioural experimental paradigms to induce and distinguish the three feeding behaviours. Heterogenous GABAergic neurons are widely distributed in the lateral hypothalamus (LH) and are involved in various motivated behaviours, including feeding 10, 11, 12, . Among LH neurons, leptin receptor (LepR) neurons are reported to be associated with feeding behaviour, although the results of previous studies remain controversial 15, 16, 17, . To investigate whether distinct neural populations separately encode motivation for food-seeking and/or evaluating, we focused on LH and LH neurons using these newly developed behavioural paradigms. Next, we investigated how neurons containing agoutirelated peptide and neuropeptide Y (AgRP/NPY), which are well known to regulate feeding, orchestrate motivation for food-seeking/evaluating 20, . Neural dynamics and causal perturbations observed in these experiments revealed that two distinct LH populations separately motivate seeking and evaluating during the state of hunger, and this process is gated by AgRP/NPY neurons that encode the need for food. Results LH neurons are food-specific subpopulation of LH neurons To investigate food-specific neurons within the population of LH neurons, three food context tests and one non-food context test were performed using a micro-endoscope (Fig. 1a, c, e). The results for 218 LH neurons showed that only a minor subpopulation of neurons (8%) was food specific (Fig. 1e, g-j). These findings imply the existence of an isolated food-specific population within the vast population of LH neurons. Since leptin decreases food motivation by inhibiting LH neurons 15, , and LH neurons have been reported to be associated with feeding . Therefore, we hypothesised that LH neurons may represent this food-specific subpopulation. To identify the distribution of LepR neurons in the LH, we performed whole-LH three-dimensional (3D) tissue clearing (Supplementary Video 1) and 2D histological mapping using LepR-tdTomato mice (Extended Data Fig. 1a–k). Our results indicated that the LH neurons were mainly distributed in the middle (-1.5 mm from bregma) and posterior (-2.2 mm from bregma) regions. Based on these distribution data, we targeted LepR neurons in the middle region of the LH. LH neural activity was measured using a micro-endoscope (Fig. 1b, d, f). Most LH neurons (63%) were specific to the food context (Fig. 1h, k,l), and most LH neurons did not respond to water (Extended Data Fig. 2a–e). Based on the available single-cell RNA sequencing data for the LH, we discovered that LH neurons comprise only 4% of LH neurons (Extended Data Fig. 1m). A previous study has also reported that LH neurons comprise <20% of LH neurons. Although LH neurons represented only a minor portion (4–20%) of LH neurons, most LH neurons were food-specific (63%), in contrast to findings for LH neurons (8%). Collectively, these results suggest that LH neurons are the major food-specific neurons among LH neurons (Extended Data Fig. 2k). LH neurons are activated during food-seeking and evaluating behaviors To identify the temporal dynamics of LH neurons in food-seeking and evaluating behaviours, we performed various food-related tests using fibre photometry (Fig. 2a-b). LH neural activity significantly increased at each feeding bout with specific time-locked temporal dynamics (Fig. 2c-g, Extended Data Fig. 2f–j). Interestingly, LH neural activity increased even before physical contact with food, implying that LH neurons may be involved in seeking or evaluating. To provide sufficient temporal distinction between seeking and evaluating, we designed tests specific to each behaviour (Fig. 2h, k). Before conditioning, since mice were not aware of the food location, they explored the whole maze (non-goaldirected locomotion) (Fig. 2i). LH neural activity did not increase during this non-goal-directed locomotion (Fig. 2l). LH neural activity started to increase when mice evaluated food at the end of the corridor. However, after conditioning, the mice moved directly to the food at the end of the corridor (goaldirected seeking; significantly shorter time to food contact) (Fig. 2j). When compared with that observed before conditioning, LH neural activity started to increase significantly when mice initiated food-seeking, and this increase was sustained during the evaluating stage (Fig. 2m). Additional tests revealed that LH neural activity decreased when mice voluntarily terminated both seeking and evaluating, suggesting that LH neural activity is significantly correlated with these behaviours (Extended Data Fig. 3a-j). To precisely quantify the temporal onset of LH neural activity and compare it with the voluntary onset of food-seeking, we allowed mice to make a voluntary decision regarding when to initiate foodseeking behaviour. The mice were trained in the doorless shelter with an appropriate amount of electrical shock and food reward, such that they intermittently made a voluntary decision to move out of the shelter to seek the food at the end of the corridor (Fig. 2n). LH neural activity began to increase significantly before the mice voluntarily initiated seeking (Fig. 2o-r, Supplementary Video 2). Furthermore, the onset of LH neural activity significantly preceded the onset of seeking by an average of approximately 6 s (Fig. 2s-u). These results indicate that LH neurons exhibit a causal temporal relationship with food-seeking behaviour, suggesting that activity in LH neurons represents the motivation for seeking, not the consequence of seeking behaviour. Activation of LH neurons evoke food-seeking and evaluating To investigate whether the activation of LH neurons increases food-seeking or evaluating behaviour (Fig. 3a, b), we displayed two foods and two non-food objects with similar appearance in the arena in an alternative order and measured behavioural changes during LH neural activation (Extended Data Fig. 4a). Activation of LH neurons significantly increased the distance moved, velocity in the seeking zone, the frequency of visiting the evaluating zones, and the time spent in the evaluating zones (Extended Data Fig. 4b-g). To scrutinise the role of LH neurons in the regulation of food-seeking, mice were conditioned to food hidden in the four corners of an open-field chamber filled with bedding (Fig. 3c, Supplementary Video 3). On the photostimulation day, the mice were placed in the same area covered with bedding without food to evoke seeking behaviour only, without any evaluating. Activation of LH neurons significantly increased seeking behaviours (digging with nose, digging with paw, and digging after floor exposure), entry into the seeking zones, and locomotion (Fig. 3d-i). These results indicate that activation of LH neurons evokes seeking behaviour. Consistent with the activation of middle LH neurons, chemogenetic activation of posterior LH neurons also significantly evoked seeking behaviours (Extended Data Fig. 4p-r). To precisely investigate evaluating behaviour and distinguish it from swallowing, the mice were confined to a small chamber to minimise seeking (Fig. 3j). To accurately distinguish evaluating behavior from swallowing behavior, we performed two independent tests using different behavior analysis methods. First, we used a deep-learning based animal pose estimation method (DeepLabCut). The results of this analysis revealed that stimulating LH neurons significantly evoked only evaluating behaviors and not swallowing behavior (Fig. 3k-n, Supplementary Video 4). Second, by using a manual behavior analysis method, we revealed that stimulating LH neurons significantly evoked evaluating behaviors (Fig. 3o, p, Extended Data Fig.4h-l). These results demonstrate that the stimulation of LH neurons evokes only foodevaluating but not swallowing behavior. Additionally, to investigate the valence effect of LH neurons, real-time place preference was performed (Extended Data Fig. 4m). The results showed that mice significantly changed their preference to the laser stimulation side (Extended Data Fig. 4n middle), in accordance with previous results . When t