Food cue regulation of AgRP hunger neurons guides learning

Introduction:

AgRP neurons are activated by fasting, and this causes hunger^1–4 – an aversive state that motivates the seeking and consumption of food^5,6. Eating returns AgRP neuron activity towards baseline in three distinct timescales: rapidly and transiently following sensory detection of food cues^6–8, slowly and longer lasting in response to nutrients in the gut^9,10, and even more slowly and permanently with restoration of energy balance^9,11. The rapid regulation by food cues is of particular interest as its neurobiological basis and purpose are unknown. Given that AgRP neuron activity is aversive⁶, the sensory cue-linked reductions in activity could function to guide behavior. To evaluate this, we first identified the circuit mediating sensory cue inhibition, and then selectively perturbed it to determine function. Here we show that a lateral hypothalamic (LH) glutamatertic → dorsomedial hypothalamic (DMH) GABAergic¹² → AgRP neuron circuit mediates this regulation. Interference with this circuit impairs food cue inhibition of AgRP neurons, and importantly, greatly impairs learning of a sensory cue-initiated food acquisition task. This is specific for food as learning of an identical water acquisition task is unaffected. We propose that decreases in aversive AgRP neuron activity⁶ mediated by this LH^{Glutamatergic} → DMH^GABAergic → AgRP neuron circuit increases the incentive salience¹³ of food cues, and thus facilitates the learning of food acquisition tasks.

Main:

We hypothesized that the rapid regulation of AgRP neurons by food cues likely arises from an afferent neural pathway that is distinct from those that cause sustained inhibition. One such afferent to AgRP neurons is the GABAergic vDMH projection that expresses leptin receptor (Lepr) and prodynorphin (Pdyn)¹². DMH^Lepr neurons are activated by the sensory detection of food and their activation is proportional to caloric value and palatability¹². To precisely determine the relationship between food cue detection and changes in neuronal activity, we trained mice to associate an arbitrary cue (light) with access to food (Ensure). Food-restricted animals were trained on a two-alternative forced choice task (2AFC) in which the illumination of a light source was randomly presented at one of two active ports, indicating food availability in the illuminated port¹⁴. With training, animals learned to poke the correct port associated with this initially arbitrary visual cue (light) to receive the food reward (Ensure) (Fig. 1a). Using this paradigm in combination with fiber photometry, we found that in trained animals DMH^Lepr neurons are robustly activated and their activation follows visual cue presentation (green line, Fig. 1b, Extended Data Fig. 1a) as opposed to nose poking for ensure (black line). As anticipated, this neural response to the otherwise arbitrary light cue was absent in naïve mice – hence it developed over time as the animal learned to associate the cue-light with food availability (Fig. 1c–1d, Extended Data Fig. 1a–1b).

Figure 1: Regulation of DMH^Lepr neurons by learned food cues.

(a) Illustration of 2AFC behavioral task.

(b) Representative heatmap of a well-trained mouse using DMH^Lepr neuron fiber photometry to observe cue responses during a single behavioral session. Single trials have been sorted by response time. Green line indicates light onset (time = 0), black line indicates time when the mouse poked correctly.

(c) Example traces in one animal when completely naïve to the 2AFC task (top) and once the animal has become proficient at the task (bottom).

(d) Mean peak amplitude quantification of DMH^Lepr neuron fiber photometry responses when naïve and trained in the 2AFC task. Peak amplitude was taken within a 2-second window following cue-light presentation. n = 4 (mice). Unpaired two-tailed t-test, *p=0.0241.

All data represents the mean ± S.E.M.

DMH^Lepr neurons are clearly rapidly activated by the sensory detection of food cues (¹², Fig. 1b and Extended Data Fig. 1a). However, it is unknown if DMH^Lepr neurons are also capable of responding to nutrients in the gastrointestinal tract similar to AgRP neurons^9,10. To address this, we infused both caloric (Ensure and glucose) and non-caloric (saline and saccharin) substances directly into the stomach. DMH^Lepr neurons were activated by infusions of caloric substances, but not by non-caloric substances (Extended Data Fig. 1c–1g). However, compared to sensory detection of food, the DMH^Lepr neuron response to gut-infusions was slower and smaller in magnitude (~5 min, ~5% ΔFn) compared to the pellet drop response (~1 sec, ~20% ΔFn) (Extended Data 1d–1f, 1h). Interestingly, this regulation appears not to be consequential for satiation as inhibition of DMH^Lepr neurons failed to increase the amount of food eaten during a meal (Supplementary Text and Extended Data Fig. 1i).

Since, DMH^Lepr neurons are unnecessary for regulating processes that occur over longer timescales, i.e. satiation (Extended Data Fig. 1i) or satiety¹², we next investigated the basis for and function of the rapid regulation of DMH^Lepr neurons by food cues. To find the monosynaptic afferents to DMH^Lepr neurons, we used EnvA pseudotyped, G-deleted rabies tracing^15,16 and channelrhodopsin (ChR2)-assisted circuit mapping (CRACM). Remarkably, while many sites provide GABAergic input, only one site provides glutamatrgic input – the lateral hypothalamus (LH) (Supplementary Text and Extended Data Fig. 2). Since LH^Vglut2 neurons are excitatory and would activate DMH^Lepr neurons, and since the general LH^Vglut2 population has been shown to be involved in sensory cue detection¹⁷, we characterized the LH^Vglut2→DMH^Lepr circuit in greater depth.

The LH is a well-known, but poorly understood, central regulator of motivational drive, reward learning, and feeding behavior^18,19. Prior studies have demonstrated that LH^Vglut2 neurons, en masse, potently suppress food intake and are aversive^20–23. However, since LH^Vglut2 neurons are extremely diverse, both genetically and in their projection targets^24,25, we sought to test effects of activating the DMH-projecting subset of LH^Vglut2 neurons on feeding and valence (i.e. positive/rewarding versus negative/aversive). Since these LH^Vglut2 neurons activate DMH^Lepr neurons, which in turn inhibit AgRP neurons, we hypothesized that their activation would produce effects consistent with AgRP neuron inhibition – namely, suppress food-intake and be appetitive during times of caloric deficiency (when AgRP neurons are active). To test this, we bilaterally injected cre-dependent AAV-ChR2 into the LH of Vglut2-ires-Cre animals and placed a single optic fiber within the midline above the DMH to selectively stimulate LH^Vglut2→DMH terminals (Fig. 2a). We then measured food intake while optically stimulating LH^Vglut2→DMH terminals beginning at the onset of the dark cycle (when AgRP neuron activity is high, and mice are inclined to eat). Consistent with our proposed circuit, we found that LH^Vglut2→DMH terminal stimulation significantly decreased food intake (Fig. 2b). To evaluate the valence of the LH^Vglut2→DMH circuit, we used a real-time place preference assay (RTPP) in which animals were optically stimulated in one side of a behavioral arena and could roam between the stimulation and non-stimulation side. When ad libitum fed animals were placed in the RTPP arena at the onset of the light cycle (i.e. when AgRP activity is low⁸), the animals were agnostic to LH^Vglut2→DMH stimulation (Fig. 2c, Extended Data Fig. 3a). We then repeated the RTPP assay at the onset of the dark cycle (i.e. when AgRP activity is high⁸). Under these conditions, ad libitum fed animals then showed a robust preference (~80%) for LH^Vglut2→DMH stimulation, which is expected to inhibit AgRP neuron activity (Fig. 2d, Extended Data Fig. 3a). Finally, to drive AgRP activity to a maximal level and to control for the time-of-day, we food-restricted the animals overnight and repeated the RTPP assay at the onset of the light cycle. Remarkably, the animals displayed a robust preference (~80%) for LH^Vglut2→DMH stimulation (Fig. 2e, Extended Data Fig. 3a). Importantly, total locomotor activity was not significantly different between XFP and ChR2 expressing groups (Extended Data Figs. 3b–d). Consistent with these LH^Vglut2→DMH neurons not promoting aversion, we found that, unlike other LH^Vglut2 neurons, they do not send collaterals to sites known to promote aversive behaviors^20–23 (Extended Data Fig. 3e–3g). In total, these findings suggest that the DMH-projecting LH^Vglut2 neurons are functionally and anatomically distinct from previously studied LH^Vglut2 neurons^20–23,26. Furthermore, as AgRP neurons promote hunger and aversion^{1–4,6,27,28}, the observed effects (decreased hunger and being appetitive) are consistent with these LH^Vglut2 neurons being upstream of the DMH^{Lepr(GABAergic)}→AgRP circuit.

Figure 2: Functional characterization of a novel LH^Vglut2→DMH^Lepr neuronal circuit.

(a) Experimental design schematic and representative image. AAV-DIO-ChR2 was bilaterally injected into LH^Vglut2 neurons and a single optic fiber was placed midline above the DMH. Scale bar = 500 μM. n=8 (mice).

(b) Optogenetic stimulation of LH^Vglut2→DMH terminals decreases nighttime food-intake in ad libitum fed mice. n = 8 (mice). Repeated-measures two-way ANOVA; Sidak’s multiple comparison test; **p=0.0016, ***p=0.0002.

(c-e) Real-time place preference: optogenetic stimulation of LH^Vglut2→DMH terminals is without effect when mice are calorically replete (p=0.6531) (c) and appetitive when mice are calorically deficient at night (p=0.008) (d) and during fasting (p=0.0129) (e). n = 7 (ChR2) and 7 (XFP). Two-tailed, unpaired t-test, *p≤0.05, **p≤0.01.

(f) Experimental design schematic and representative image. AAV-DIO-GCaMP6s was injected into LH^Vglut2 neurons. Optic fiber placed over the DMH to perform fiber photometry in LH^Vglut2 terminals. Scale bar = 500 μM. n=9 (mice).

(g-j) LH^Vglut2→vDMH axons are rapidly activated upon the sensory detection of food (k) and are scalable by caloric value and palatability (l-n). Heat map (k) represents the trial-by-trial response in one representative animal from 15 mg pellet presentation (time=0). Line represents average ± S.E.M.

(k) Quantification of mean peak response within the first five-seconds following pellet presentation. One-way ANOVA; Friedman test ***p=0.0007. n=9 (mice).

All data represents the mean ± S.E.M.

The LH is known to participate in many aspects of feeding behavior^18,19,24 and LH^Vglut2→NAc neurons have been found to be necessary for signaling cue-outcome associations¹⁷. With this in mind, we asked if the DMH-projecting LH^Vglut2 neurons were responsive to food pellet drops. We used axon fiber-photometry to record LH^Vglut2→DMH axonal Ca²⁺-activity in freely-moving mice (Fig. 2f) while presenting food pellets of increasing caloric value and palatability^7,12(Fig. 2g–2k). LH^Vglut2→DMH axons were rapidly activated upon the presentation of a food item, were scalable to caloric content, and the responses were transient (returning to baseline within ~10–20 seconds) (Fig. 2h–2k, Extended Data Fig. 3i–3j). In addition, we found that the response was mostly specific for food in that a much smaller, and much more delayed response was seen when a non-edible object was presented (Extended Data Fig. 3h–3i) or when water was presented to water-deprived mice (Extended Data Fig. 3k–3m). In total, our findings suggest that LH^Vglut2→DMH neurons are activated by sensory cues preceding food ingestion, and that they likely provide this information to DMH^Lepr neurons.

If the LH does indeed provide food cue information to the DMH^Lepr→AgRP pathway, then selectively inhibiting LH^Vglut2→DMH afferents should attenuate the food cue response in DMH^Lepr neurons. We employed two distinct methods to determine if LH→DMH afferents were necessary for the cue-evoked response in DMH^Lepr neurons. First, we used a retrograde AAV-FlpO virus to selectively express FlpO within LH neurons that project to the DMH. To record Ca²⁺-activity while simultaneously inhibiting LH→DMH afferents, we injected flp-dependent AAV-hM4Di within the LH and cre-dependent AAV-GCaMP6s within the DMH of Lepr-ires-Cre mice (Fig. 3a). Flp-dependent AAV-hM4Di efficiently hyperpolarized LH neurons (Extended Data Fig. 4a–4b). Notably, inhibiting LH→DMH afferents attenuated the food cue evoked response in DMH^Lepr neurons by ~70% (Figs. 3b–3c). CNO injections in GFP-control mice was without effect (Extended Data Fig. 4c–4d). To specifically interrogate the role of LH^Vglut2 neurons in mediating DMH^Lepr neuron food cue responses, we used Vglut2-ires2-FlpO::Lepr-ires-Cre mice to express flp-dependent AAV-hM4Di in LH^Vglut2 neurons and cre-dependent AAV-GCaMP6s in DMH^Lepr neurons (Fig. 3d). Selectively inhibiting LH^Vglut2 neurons also attenuated the rapid DMH^Lepr cue-evoked response by ~68% (Figs. 3e–3f). Collectively, these studies show that LH^Vglut2 neurons drive food cue-evoked responses in DMH^Lepr neurons.

Figure 3: LH afferents to the DMH are necessary for rapid food cue-evoked responses in DMH^Lepr and AgRP neurons.

(a) Retrograde AAV-FlpO and AAV-DIO-GCaMP6s were injected into the DMH of Lepr-ires-Cre mice and AAV-fDIO-hM4Di was injected into the LH to inhibit LH→DMH projecting neurons. Bottom: Sample images of hM4Di expression (left), GCaMP6s expression, and fiber placement (right). Scale bar = 200 μM. n=4 (mice).

(b) Example traces of vehicle/CNO recording session within the same animal. Vehicle and CNO recording sessions were split so that treatment comparisons were done within day. Vehicle 1.1 represents the first vehicle recording session and Vehicle 1.2 represents the second recording session. Vehicle 2.1 represents vehicle recording session on the second day, and CNO 2.2 represents the second recording session with CNO injection. Black vertical line represents cue presentation. Purple box represents the analyzed 2s period. Blue horizontal line represents peak of vehicle control from the first recording session.

(c) LH→DMH inhibition attenuates the cue response in DMH^Lepr neurons. Normalized responses are indicated by ΔF/F₀, 2 (R2, second recording session) divided by ΔF/F_0, 1 (R1, first recording session). n = 4 (mice). Two-tailed t-test, **p=0.0035.

(d) AAV-fDIO-hM4Di injected into the LH and AAV-DIO-GCaMP6s within the DMH of Vglut2-ires-FlpO::Lepr-ires-Cre mice. Bottom: Sample images of AAV-fDIO-hM4Di expression (left) and AAV-DIO-GCaMP6s expression and fiber placement (right). Scale bar = 500 μM. n= 4 (mice).

(e) Example traces from one mouse (as explained in (b)).

(f) LH^Vglut2 inhibition attenuates the cue response in DMH^Lepr neurons. n = 4 (mice); Two-tailed t-test, *p=0.0009.

(g) AAV-fDIO-hM4Di injected into the LH and AAV-DIO-GCaMP6s within the ARC of Vglut2-ires-FlpO::AgRP-ires-Cre mice. Bottom: Sample images of expression and fiber placement, AAV-fDIO-hM4Di (right) and AAV-DIO-GCaMP6s (left). Scale bar = 500 μM. n=4 (mice).

(h) Example traces in one mouse during vehicle and CNO (as explained in (b)).

(i) LH^Vglut2 inhibition decreases the cue response in AgRP neurons. n = 4 (mice), Two-tailed t-test, *p=0.027.

All data represents the mean ± S.E.M.

To determine if LH^Vglut2 neurons do indeed cause food cue activation of AgRP neurons, we used Vglut2-ires2-FlpO::AgRP-ires-Cre mice to restrict expression of flp-dependent AAV-hM4Di within LH^Vglut2 neurons and cre-dependent AAV-GCaMP6s within AgRP neurons (Fig. 3g). Strikingly, selective inhibition of LH^Vglut2 neurons robustly attenuated the rapid, cue-evoked response in AgRP neurons (~76% decrease in response magnitude) but did not affect sustained AgRP inhibition during food consumption of large food pellets (Fig. 3h–3i, Extended Data Fig. 4f). Thus, LH^Vglut2 neurons have an essential role in food cue-evoked inhibition of AgRP neurons but appear not to be involved in gut nutrient-mediated regulation of AgRP neurons.

We next explored the function of this food cue-mediated regulation of the LH^Vglut2→DMH^Lepr→AgRP circuit. In mice that learned the task, CNO/hM4Di inhibition of LH→DMH neurons had no effect on task performance (Extended Data Fig. 4e, 4g–4j). Perhaps this lack of effect is due to habit-like performance in such highly trained mice²⁹. To determine if this food cue-regulated circuit plays a role in the learning of this task, we expressed cre-dependent AAV-tetanus toxin (TeNT) within DMH^Lepr neurons to eliminate evoked synaptic release within the circuit (Fig. 4a). We then trained naïve mice on the 2AFC task (Fig. 1a) until they mastered the task (or a maximum of 21 training days). We calculated the number of correct responses in addition to the errors made, including misses (failing to poke within the response window), and false alarms (poking in the incorrect port). We also assessed the response time (how long after cue presentation the poke was performed). TeNT- mediated DMH^Lepr silencing caused a significant delay in task acquisition (Figs. 4b–4c). In addition, the TeNT mice took many more days of training to reduce their misses, false alarms, and their response times (Extended Data Fig. 5a–5c). The higher misses and false alarms, and in particular, the longer response times suggest that in TeNT mice, the cue-light is less effective in motivating food-seeking behavior. Consistent with the view that the DMH^Lepr→AgRP neuron circuit is not involved in either satiety¹² or satiation (Extended Data Fig. 1i), body weight (Extended Data Fig. 5d) and post-fast refeeding (Extended Data Fig. 5e) were normal in TeNT mice.

Figure 4: Afferent modulation of AgRP neurons is required for learning a cue-initiated food acquisition task.

(a) AAV-DIO-TeNT injection into DMH^LepR neurons. Bottom: Example images of TeNT expression. Scale bar = 500 μM. n= 5 (mice).

(b) DMH^Lepr neuron silencing attenuates correct responses. n = 5 (GFP), 5 (TeNT). Red line = line of best fit (***p<0.0001, plateau = 97.49 (GFP), 132.4 (TeNT); tau = 6.928 (GFP), 20.80 (TeNT)). Two-way repeated measures ANOVA, main effect of days = ****p<0.0001, main effect of group = ****p<0.0001.

(c) Silencing DMH^Lepr neurons increases time to reach learning criterion (>80% correct across three consecutive days). 5/5 GFP and 4/5 TeNT mice learned. n = 5 (GFP), 5 (TeNT). Two-tailed, unpaired t-test, *p=0.0238.

(d) AAV-DIO-TeNT injection into DMH^LepR neurons. Bottom: Example images of TeNT expression. Scale bar = 500 μM. n=6 (mice).

(e) Silencing DMH^Lepr neurons does not affect behavioral performance for learning to obtain water. n = 4 (GFP), 6 (TeNT). Red line = line of best fit (plateau = 129.3, tau = 11.53). Two-way repeated measures ANOVA, main effect of days = ****p<0.0001, main effect of group = 0.7203.

(f) Silencing DMH^Lepr neurons does not affect learning (>80% correct across three consecutive days) when water-deprived mice are trained to receive water rewards. n = 4 (GFP), 6 (TeNT). Two-tailed, unpaired t-test, p=0.4375.

All data represents the mean ± S.E.M.

While AgRP neuron-regulating DMH^Lepr neurons project only to the arcuate nucleus¹², the DMH does contain other Lepr-expressing neurons that project elsewhere¹². Given this, the impairment in learning caused by TeNT inhibition could be due to the silencing of DMH^Lepr neurons that project elsewhere (Extended Data Fig. 5f). To address this, we used Pdyn-ires-Cre mice to express TeNT selectively in DMH^Pdyn neurons. Note, we previously established that DMH^Pdyn neurons send long range projections exclusively to the ARC where they inhibit AgRP neurons¹² (and Extended Data Fig. 6). These mice were then trained on the 2AFC task. Of interest, silencing DMH^Pdyn neurons modestly increased bodyweight in TeNT-expressing animals (Extended Data Fig. 5l) but did not affect post-fast refeeding (Extended Data Fig. 5m). As bodyweight was not increased when DMH^Lepr neurons were silenced (Extended Data Fig. 5d), this bodyweight effect may be due to the silencing of Lepr-negative DMH^Pdyn neurons. Importantly, silencing DMH^Pdyn neurons, like silencing DMH^Lepr neurons, significantly altered task performance in that it delayed task acquisition, increased mistakes made, and increased response time (Extended Data Fig. 5g–5k). Given that DMH^Pdyn neurons project selectively to AgRP neurons, these findings along with those from the DMH^Lepr silencing study (Fig. 4a–4c, Extended Data Fig. 5a–5c), strongly supports the view that the DMH → AgRP neuron circuit plays a key role in the learning of a sensory cue – food acquisition task.

To specifically test the role of the LH→DMH segment of the circuit in learning the 2AFC task, we selectively silenced DMH-projecting LH neurons by bilaterally injecting a retrograde AAV-Cre virus into the DMH and a cre-dependent TeNT AAV into the LH (Extended Data Fig. 5n). Indeed, silencing these neurons significantly impaired the rate of task acquisition and significantly increased misses and response times (Extended Data Fig. 5o–5s). Importantly, TeNT-mediated silencing of LH→DMH neurons did not affect body weight or post-fast refeeding (Extended Data Fig. 5t–5u) suggesting this LH→DMH neuron projection is not necessary for satiety or satiation as similarly seen when silencing DMH^Lepr neurons (Extended Data Fig. 5d–5e). Collectively, these data (Fig. 4a–4c, Extended data Fig. 5a–5u) show that the LH^Vglut2→DMH^Lepr/Pdyn→AgRP circuit plays an important role in promoting the mastery of this caloric deficiency-driven, sensory cue-initiated food acquisition task.

To assess the possibility that TeNT expression in DMH^Lepr neurons somehow nonspecifically interferes with deprivation state-motivated learning of an operant task, we generated a cohort of TeNT-expressing water-deprived mice (Fig. 4d) and determined their ability to learn the identical task with water, instead of food, as the reward. We hypothesized that the principles of learning in these two tasks would be identical, with the sole exception that water cue regulation of aversive thirst neurons^6,30–32, instead of food cue regulation of hunger neurons, is what promotes learning. Since the DMH → AgRP circuit is involved in the latter, but not the former, this experiment controls for nonspecific effects. As shown in Fig. 4e–4f, Extended Data Fig. 5v–5x, TeNT expression had no effect on learning in the water-oriented version of the task. Thus, the DMH^Lepr→AgRP neuronal circuit is specific for learning food cue-initiated food acquisition tasks, and is congruous with the very specific role of AgRP neurons in regulating food but not water intake.

Discussion:

These findings lead us to propose the following model: caloric deficiency activates AgRP neurons (reviewed in ref.¹¹) and causes the aversive feeling of hunger⁶. Environmental cues instructive for food acquisition engage the LH^Vglut2→DMH^Lepr→AgRP neuron circuit, and transiently reduce AgRP neuron activity. Related to the proposal that AgRP neurons transmit a negative-valence teaching signal⁶, these “appetitive” falls in aversive AgRP neuron activity, over time, increase the incentive salience¹³ of food cues, thereby facilitating the learning of food acquisition tasks. If true, this model implies a very important general neurobiological mechanism for how homeostatic deficiency states promote the learning of tasks directed at acquiring the cognate goals (i.e. caloric deficiency/food, dehydration/water, etc.) – by providing the “substrate”, which are the deficiency-activated aversive drive neurons^6,30–32, upon which the inhibitory sensory cue-regulated afferent neurons can operate.

Methods

Experimental subjects

Vglut2-IRES-Cre (JAX #: 016963)³³, Vglut2-IRES2-FlpO (JAX #: 030212) (Unpublished, donating investigator Hongkui Zeng, Allen Institute for Brain Science), AgRP-IRES-Cre (JAX #: 012899)³⁴, Pdyn-GFP³⁵ and wildtype mice (JAX#:101045) were obtained through Jackson Laboratories (Bar Harbor, ME) or in house. Lepr-IRES-Cre³⁶ mice were maintained as previously described¹². All mice were maintained on a mixed genetic background unless otherwise noted. Pdyn-IRES-Cre²⁷ mice were maintained as previously described and kept on a congenic C57 BL/6J background. The National Institute of Health and Beth Israel Deaconess Medical Center Institutional Animal Care and Use Committee approved all animal care and experimental procedures. Mice were housed at 22–24°C, 20–30% humidity with a 12:12 light:dark cycle with standard mouse chow (Teklad F6 Rodent Diet 8664) and water was provided ad libitum, unless otherwise stated. All diets were provided as pellets. For all behavioral studies, we used male mice between 8 and 20 weeks of age. For electrophysiogical recordings, we used male mice between 8 and 12 weeks of age.

Brain tissue preparation

Mice were terminally anesthetized with chloral hydrate (Sigma-Aldrich C8383) and transcardially perfused with phosphate-buffered saline (PBS) followed by 10% neutral buffered formalin (Fisher Scientific SF100). Brains were extracted then cryoprotected in 20% sucrose. All brains were sectioned coronally on a freezing sliding microtome (Leica Biosystems) at 40 μM and collected in four equal series.

Immunohistochemistry

Tissue sections were washed with 0.1 M phosphate-buffered saline (pH 7.4) then blocked in 5% normal donkey serum/0.2% Triton X-100 in PBS for 1 h at room temperature. Sections were then incubated overnight at room temperature in blocking solution containing: rat anti-mCherry (1:3000, Invitrogen M11217) and chicken anti-GFP (1:1000, Invitrogen A10262). Secondary detection was performed with Alexa Fluor 488 or 594 conjugated donkey anti-chicken or donkey anti-rat (1:1000, Invitrogen) for 1 h at room temperature. After secondary incubation, sections were washed and mounted onto gelatin-coated slides and fluorescent images were obtained with an Olympus VS120 slide scanner microscope. Our inferred Bregma coordinates on all histological images were adopted from a stereotaxic atlas (³⁷, Second Edition).

Stereotaxic surgeries and viral injections

For viral injections, six-to-eight week old male mice were anesthetized with a ketamine (100 mg kg⁻¹) and xylazine (10 mg kg⁻¹) cocktail diluted in 0.9% saline and placed into a stereotaxic apparatus (Kopf model 940). Subcutaneous injection of sustained release Meloxicam (4 mg kg⁻¹) was provided as postoperative care. A pulled glass micropipette (20–40 μM diameter tip) was used for stereotaxic injections of adeno-associated virus (AAV). For electrophysiological experiments, bilateral injections (25 nL) of purified adeno-associated virus (6.24 × 10¹² viral genomes ml⁻¹) were injected into the NAc (from bregma: +1.3 AP, ±0.5 ML, −4.25 DV), BNST (from bregma: +0.14 AP, ±0.75ML, −4.9 DV), MPO (from bregma: +0.4 AP, ±0.25 ML, −4.75 DV), LH (from bregma: −1.3 AP, ±0.9 ML, −5.3 DV), VTA (from bregma: −3.15 AP, ±0.6 ML, −4.74 DV), IPN (from bregma: −3.4 AP, ±0.0 ML, −4.8 DV), and PAG (from bregma: −4.16 AP, ±0.3 ML, −2.5 DV). For optogenetic experiments, bilateral injections (15 nL) of AAV9-EF1α-DIO-ChR2(H134R)-eYFP purchased from the University of Pennsylvania School of Medicine Vector Core (donating investigator, Dr. Karl Deisseroth Cat # AV-9–20298P; Addgene: 20298; 2.7 × 10¹³ viral genomes ml⁻¹) were injected into the LH (coordinates as above). For in vivo fiber photometry experiments, AAV1-hSyn-DIO-GCaMP6s (University of Pennsylvania Vector core; Addgene Cat. # 100845-AAV1; 1.6 × 10¹³ viral genomes ml⁻¹) was injected unilaterally into either the LH (25 nL, coordinates as above), DMH (25 nL, from bregma: −1.80 AP, ±0.3 ML, −5.2 DV), or ARC (150 nL, from bregma: −1.45 AP, ±0.25 ML, −5.85 DV). For retrograde chemogenetic silencing studies, AAV6-CAG-FlpO (Boston Children’s Hospital Viral Vector Core, modified from Addgene 67829; 2.88 × 10¹⁴ viral genomes ml⁻¹) was bilaterally injected into the DMH (15 nL, coordinates as above) and AAV8-nEF-fDIO-hM4Di-mCherry (Boston Children’s Hospital Viral Vector Core, modified from Addgene 44362 and 55644; 9.98 × 10¹³ viral genomes ml⁻¹) was bilaterally injected into the LH (25 nL, coordinates as above). For Flp-dependent chemogenetic studies in combination with fiber photometry, AAV8-nEF-fDIO-hM4Di-mCherry was bilaterally injected into the LH (25 nL, coordinates as above) or, the DMH (25 nL, coordinates as above). For tetanus toxin mediated silencing studies, AAVDJ-CMV-DIO-eGFP-2A-TeNT (Stanford; Cat. # GVVC-AAV-71; 3.6 × 10¹² viral genomes ml⁻¹) was injected bilaterally into the DMH (15 nL, coordinates as above). Finally, for projection specific tetanus toxin mediated silencing studies, AAVDJ-CMV-DIO-eGFP-2A-TeNT was injected bilaterally into the LH (20 nL, coordinates as above) and rAAV2-hSyn-Cre (Addgene Cat.#: 105553; 1.2 × 10¹³ viral genomes ml⁻¹) was injected bilaterally into the DMH (15 nL, coordinates as above). Animals were allowed to recover for a minimum of three weeks prior to the initiation of any experiments. Following each experimental procedure, accuracy of AAV injections were confirmed via post hoc histological analysis of mCherry, YFP, or GFP fluorescent protein reporters, viral expression of each individual surgery is cataloged in detail within Extended Data Figs. 7–11. All subjects determined to be surgical “misses” were those with absent or low reporter expression and were removed from the experimental data set. In addition, animals were excluded from the dataset if reporter expression was primarily outside of the area of interest. Anatomical boundaries were drawn by using the DAPI signal so that we could clearly discern landmarks within a histological section. Based on the landmarks present, we then inferred the A/P coordinate in the atlas³⁷ and traced the outline of different nuclei and superimposed the outline on our histological images.

Optic fiber implantation

Optic fiber implantations were performed during the same surgery as viral injections (above). For optogenetic photostimulation of LH→DMH terminals, ceramic ferrule (Precision Fiber Products) optical fibers (200 μM diameter core, 0.39 NA, multimode; Thorlabs) were implanted within the midline over the DMH (from bregma: −1.8 AP, ±0 ML, −4.7 DV). For LH^Vglut2→DMH axon fiber photometry, a metal ferrule (Precision Fiber Products) optic fiber (400 μM diameter core, 0.5 NA, multimode; Thorlabs) was implanted unilaterally over the DMH (from bregma: −1.8 AP, ±0.3 ML, −5.1 DV). For DMH^Lepr and ARC^AgRP cell body fiber photometry, a stainless steel ferrule optic fiber (same as above) was implanted unilaterally over the DMH (coordinates as above) or the ARC (from bregma: −1.45 AP, ±0.25 ML, −5.8 DV). Fibers were fixed to the skull using dental acrylic and mice were allowed to recover for three weeks prior to the start of acclimation to behavioral testing.

Monosynaptic rabies mapping

Lepr-IRES-Cre mice were unilaterally injected with a 1:1 mixture of AAV8-EF1α-DIO-TVA-mCherry (University of North Carolina Vector Core, donating investigator Naoshige Uchida) and AAV8-CAG-FLEX-Rabies G (Stanford; Cat. # GVVC-AAV-59) into the DMH (15 nL, coordinates as above) (Extended Data Fig. 2a). Animals recovered for three weeks post TVA/RG transduction to ensure adequate levels of TVA and RG viral expression. Following recovery, animals underwent a second surgery in which animals were injected with SADΔG-EGFP (EnvA) rabies (Salk Gene Transfer Targeting and Therapeutics Core) into the DMH (15 nL, coordinates as above). Animals recovered for 7 days to allow for the retrograde transport of rabies virus and EGFP expression before perfusion and histological processing. Sites of afferent input to DMH^Lepr neurons were assessed by the presence of EnvA-EGFP positive neurons.

Rabies collateral mapping

Vglut2-IRES-Cre mice were unilaterally injected with AAV8-EF1α-DIO-TVA-mCherry into the LH (15 nL) and allowed to recover for three weeks (Extended Data Fig. 3e–3g). Then, SADΔG-EGFP (EnvA) rabies was unilaterally injected into the DMH (15 nL). Animals were allowed to recover for seven days to allow for the retrograde transport of rabies virus and EGFP transgene expression before perfusion and histological processing. Comprehensive examination of SADΔG-EGFP (EnvA) axonal and retrograde transduction was assayed using immunohistochemistry followed by imaging the entire brain for the presence of EGFP expression.

Electrophysiology

Animals were deeply anesthetized, and intracardially perfused with ice-cold dissection buffer (in mM: 2.5 KCl, 1.25 NaH₂PO₄, 20 HEPES, 10 MgSO₄•7H₂O, 0.5 CaCl₂•2H₂O, 92 choline chloride, 25 glucose, 2 thiourea, 5 sodium ascorbate, 3 sodium pyruvate, and 20 NaHCO₃) bubbled with 95% O₂—5% CO₂. Brains were then rapidly removed and immersed in ice-cold dissection buffer. DMH sections were dissected and 300 μM thick coronal slices were prepared using a vibrating microtome (Campden 7000smz 2). Slices recovered for 10 min in a 35 °C submersion chamber filled with oxygenated dissection buffer. Slices were then transferred to a secondary 35 °C submersion chamber filled with oxygenated artificial cerebrospinal fluid (ACSF; in mM: 125 NaCl, 2.5 KCl, 25 NaHCO₃, 2 CaCl₂, 1 MgCl₂, 1.25 NaH₂PO₄, 25 glucose) and allowed to recover for an additional 15 minutes. Slices were then kept at room temperature in oxygenated ACSF for ≥30 min until use (Extended Data Fig. 2b–2m).

Channelrhodopsin-2 assisted circuit mapping (CRACM)

To isolate oIPSCs and oEPSCs (optically-evoked Inhibitory/Excitatory Post Synaptic Currents), slices were placed in a submersion chamber and perfused at 4 mL min⁻¹ with oxygenated ACSF. Cells were visualized with a Scientifica SliceScope Pro 1000 microscope equipped with infrared differential interference contrast optics. DMH^Pdyn GABAergic neurons were identified by GFP fluorescence ventral to the DMC (dorsomedial hypothalamic nucleus compacta). Open-tip resistances for patch pipettes were between 2–4 MΩ and were backfilled with a Cs-based internal solution containing (in mM): 135 CsMeSO₃, 10 HEPES, 1 EGTA, 3.3 QX-314 (Cl⁻ salt), 4 Mg₂-ATP, 0.3 Na₂-GTP, and 8 Na₂-phosphocreatine with pH adjusted to 7.3 with CsOH and osmolarity adjusted to ~295 mOsM by the addition of sucrose. oEPSCs were isolated with membrane potential clamped at V_h = −70 mV and oIPSCs were isolated with membrane potential clamped at V_h = 0 mV. Bath solutions for pharmacological isolation of excitatory or inhibitory currents in whole-cell voltage clamp recordings contained SR95531 (10 μM, gabazine), kynurenic acid (1 mM), tetrodotoxin (TTX, 1 μM), and 4-Aminopyridine (4-AP, 500 μM). To photostimulate ChR2-positive fibers, an LED light source was used (470 nM, Cool LED pE-100). The blue light was focused onto the back aperture of the microscope objective (40X) producing wide-field exposure around the recorded cell of 10–15 mW per mm² as measured using an optical power meter (PM100D, Thorlabs). A programmable pulse stimulator, Master-8 (A.M.P.I.) and pClamp 10.2 and 10.6 software (Molecular Devices, Axon Instruments) controlled the photostimulation output. The oIPSC/oEPSC detection protocol consisted of one blue-light pulse (5 ms pulse length) at 30 s intervals for at least six consecutive sweeps. Changes in series and input resistance were monitored throughout the experiment by giving a test pulse every 30 s and measuring the amplitude of the capacitive current. Cells were discarded if series resistance rose above 25 MΩ (Extended Data Fig. 2b–2m). All electrophysiology data was analyzed using Clampfit 10.2 and 10.6.

Optogenetic behavioral experiments

In vivo photostimulation of LH^Vglut2→DMH terminals was conducted by firmly attaching a fiber optic cable (1.25 m long, 200 μM core diameter, 0.63 NA; Doric Lenses) with ceramic split sleeves (Precision Fiber Products) (Fig. 2, Extended Data Fig. 3a–3d). Animals were acclimated by connecting them to a ‘dummy’ fiber optic cable three days before the initiation of the experiment. Animals were stimulated with blue light (465 nM LED; Plexon) at 20 Hz, 5 ms pulses for 1 sec with a 3 sec recovery period (LED off) during stimulation trains to avoid ChR2 desensitization, neuronal transmitter depletion, and tissue heating. Light pulse trains were programmed using a waveform generator (National Instruments) that provided TTL input to the blue light LED. The light power exiting the fiber optic cable measured by an optical power meter (Thorlabs) was 7–8 mW in all experiments. After completion of photostimulation experiments, mice were perfused for assessment of surgical accuracy of both ChR2-expression and optic fiber tip location via histological analysis as described in “Stereotaxic surgeries and viral injections”.

Food intake studies

To test the sufficiency of LH^Vglut2→DMH neurons for satiety, animals were tested under conditions of physiological hunger at the onset of the dark cycle (Fig. 2b). For dark cycle feeding, mice with ad libitum access to food were photostimulated for 5 min prior to the onset of the dark cycle (a time when mice often eat) and photostimulation continued throughout the duration of the study (three hours). For post-fast refeeding assays (Extended Data Figs. 1 and 5), mice were food-restricted for 24 hours then, were given ad libitum access to food. Food was then weighed each hour to determine the amount consumed during the experimental manipulation.

Real-time place preference assays

Animals were placed in a custom-made behavioral arena (transparent acrylic, 25 × 25 × 25 cm) for 20 min (Fig. 2, Extended Data Fig. 3a–3d). One counterbalanced side of the arena was designated as the photostimulation side. The animal was placed in the stimulation side at the onset of the experiment and each time the mouse crossed to the non-stimulation side of the arena, the photostimulation immediately stopped until the animal crossed back into the stimulation side. Behavioral data was recorded with Ethovision software (Noldus Information Technologies). To test photostimulation preference during different hunger states, ad libitum fed animals were placed within the arena immediately prior to the onset of the dark cycle. Following a one-week rest period, ad libitum fed animals were then placed in the same arena and tested for photostimulation preference at the onset of the light cycle. Following a one-week rest period, animals were then fasted overnight then, placed in the arena and tested for photostimulation preference at the onset of the light cycle. Each real-time place preference assay was counterbalanced within animal and within day.

In vivo fiber photometry

Fiber photometry was performed on a rig constructed as follows: A 465 nm LED (PlexBright LED, Plexon) was used as the excitation source which was passed through a fluorescence mini cube (excitation: 460–490 nm, detection: 500–550 nm; Doric Lenses), and transmitted onto the sample via a fiber optic cable (1 m long, 400 μM diameter, 0.48 NA; Doric Lenses). The optic fiber was coupled to the implanted optic fiber with a ceramic mating sleeve (Precision Fiber Products). Light intensity was measured as 100–200 μW at the end of the patch cord and was kept constant across sessions for each animal. Emitted light was collected by a photodetector (2151; Newport). The signal was digitized at 1 kHz with a data acquisition card (National Instruments) and collected with a custom MATLAB (MATLAB2016a; MathWorks) script (Figs. 1–3, Extended Data Figs. 1, 3–4).

For LH^Vglut2→DMH axon fiber photometry recordings (Fig. 2, Extended Data Fig. 3h–3j), animals underwent a 25 minute recording session within their home cage that consisted of: eight ‘small’ chow (15 mg; Bio-Serv) trials, two ‘large’ chow (500 mg; Bio-Serv) trials, and one peanut butter (Reese’s peanut butter chips; Hershey) trial which were dropped into a pyrex petri dish. Each animal underwent only one session per day and were food restricted (85% of ad libitum bodyweight) for at least one week prior to beginning the experiment in which they were habituated to the pellet drops within the petri dish. All trials were pooled to calculate mean peak response (0–5 sec following food presentation) to each food presentation. For object drop experiments, animals underwent a 12 minute recording session within their home cage that consisted of 10 total trials of non-edible object drops. Objects consisted of uniform, white plastic marbles (BC Percision, Hungry Hungry Hippos Marbles). After an object drop session, a 500 mg chow pellet was dropped as a positive control. Each animal underwent a single session per day. All trials were pooled to calculate mean z-score response to each object drop.

For water presentation experiments (Extended Data Figs. 3k–3m), animals were water restricted (85% of original bodyweight) for at least one week prior to beginning the experiment. Animals were then habituated to receiving water in a ceramic dish within their home cage and were given free access for five minutes. All sessions were pooled to calculate the mean peak response (0–30 seconds) and the time to peak response.

Data were analyzed using a custom MATLAB (MATLAB2016b; MathWorks) script (Figs. 1–3, Extended Data Figs. 1, 3–4). Fluorescence traces were down-sampled from 1kHz to 100 Hz and smoothed using a 1 s running average. The fractional change in fluorescence was calculated as ΔF/F = (F-F₀)/F₀ where, F₀ was the mean of all data points from the baseline prior to each trial. In home cage pellet drop/water presentation experiments, F₀ was the average of five seconds before the food drop/water presentation. In operant chamber silencing experiments, F₀ was the average of one second before cue presentation. All trials in a single session were averaged then the mean peak amplitude was taken for quantification.

Two-alternative forced choice task (2AFC)

To determine if LH^Vglut2→DMH neurons and DMH^Lepr neurons are necessary for food cue responses (Figs. 1–3, Extended Data Figs. 1, 4), animals underwent training in a three nosepoke operant chamber (Bpod r2; Sanworks) controlled by a custom MATLAB (MATLAB2016a; Mathworks) script. Briefly, food restricted animals (85% of ad libitum body weight) were trained to associate a light presentation with Ensure (10 μL; Ensure PLUS, vanilla) and were required to nose poke and hold their snouts within the port for 200 ms before ensure was delivered. Light delivery was randomized between the left and right nose poke and, animals had a 10 sec response window with a 10 sec inter-trial interval. The required learning criterion was a success rate of ≥80% across three consecutive days. After animals learned the task, animals were then attached to a fiber optic cable as described above. To determine the necessity of LH afferents on food-cue responses (Fig. 3 and Extended Data Fig. 4), animals underwent two, 11 min recording sessions on the same day separated by a 10 min ‘break’ period. The first session was always a saline run and animals were injected with saline 10 min prior to the onset of the recording session. The second session was a clozapine-n-oxide (CNO; 1 mg kg⁻¹; 0.5% body weight volume) or saline injection 10 min prior to the onset of each recording session. Comparisons between vehicle and CNO recordings were made within day; therefore, an animal received two saline injections or, one saline and one CNO injection in a single recording session. A TTL pulse triggered at the onset of each trial determined cue onset. All trials were pooled to calculate the mean peak response (0–2 sec following cue presentation) and were normalized to the first recording session within day.

Intragastric Catheter Surgery

Mice with DMH^Lepr photometry neural activity signal larger than 10% ΔF/F to chow during fast re-feed were implanted with intragastric catheters³⁸ (Extended Data Fig. 1). During surgery, mice were anaesthetized with isoflurane (1.5–3%) and treated post-operation with buprenorphine (1 mg/kg SQ) analgesia. A midline incision on the abdomen was made through skin and muscle layers. Micro-Renathane catheter tubing 6–7cm in length (Braintree Scientific, MRE-033, 0.033 × 0.014 in) was anchored with epoxy spheres on each end (Devcon Clear Epoxy Adhesive, 92926, Lowes). The catheter was inserted into the fundus of the stomach through a puncture hole and secured with surgical mesh (5 mm diameter piece, Bard, 0112660). The other end of the catheter was directed out of an intrascapular incision. A metal cap made out of 27G blunt needle was placed in the exposed end for seal. The gastric catheter was flushed with sterile water immediately, and daily, after surgery to prevent blockage. Mice were fed with gel chow diet and given at least 1–1.5 week for recovery prior to experimentation. Daily body weight was monitored until stable pre-surgical weight was regained.

Gastric Infusion

Upon recovery, gastric infusions of liquid substances listed below were performed in a counterbalanced experimental design, under both overnight fasted and sated conditions (Extended Data Fig. 1). The intrascapularly exposed end of the gastric catheters were connected to tubing and a syringe driven by an infusion pump (Harvard Apparatus, 70–3007). At a rate of 0.1 ml/min, 1 ml infusions were performed over the course of 10 min³⁸. Fiber photometry recordings were collected via a lock-in amplifier (TDT) and the software Synapse (TDT). Each trial was approximately 27 min (>7 min baseline recording, followed by a 10min gastric infusion and a 10min chow refeed). Each mouse underwent infusions of the following infusates: 0.9% isotonic saline, Ensure Original Nutrition Shake (Vanilla), 1% saccharine, 25% D-glucose (equal caloric content as Ensure).

Each photometry recording data point was normalized against the average of the last 5 min of baseline period to produce the normalized traces. For normalized and delta (ΔFn/Fn, %) quantified comparisons, we use 1 min averages at the end of baseline period (t=−1–0 min), 5 min into infusion (t=4–5min), and 10 min into infusion (t=9–10min). Due to the transient nature of neural response to food, comparison of delta (ΔFn/Fn, %) maximum magnitude between ensure infusion and food presentation were made using 30 sec averages at the end of ensure infusion (t=9.5–10 min) and at the beginning of food presentation after saline infusion (t=10–10.5 min).

TeNT-mediated silencing

AAV-DIO-TeNT was injected into the DMH of Lepr-IRES-Cre mice, Pdyn-IRES-Cre mice, or wildtype mice (Fig. 4 and Extended Data Fig. 5). Littermate controls were used for AAV-DIO-TeNT and AAV-DIO-GFP behavioral groups. Following three weeks, animals were then placed within the Bpod arena and trained everyday as described above. Mice were either food or water restricted and maintained ≥85% ad libitum bodyweight. Food and water restriction was performed in separate cohorts of mice. For food-learning assays, mice were trained for a total of 21 days. For water-learning assays, mice were trained for a total of 14 days. Mice were excluded if they were non-learners meaning that they did not increase their performance rate for five consecutive days (One GFP-expressing mouse (in water-deprived group) and one TeNT-expressing mouse (in food-deprived group) were removed based on this criterion). Mice were placed in the Bpod for a total of 20 minutes and allowed to perform as many trials as possible with an ITI of five seconds and a response window of 10 seconds. Mice were given either 10 μL of Ensure or 5 μL of water for food and water-learning assays, respectively. Each day, performance in the task was quantified with a custom MATLAB script (Fig. 4 and Extended Data Fig. 5). It should be noted that Pdyn-ires-Cre and wildtype behavioral cohorts (Extended Data Fig. 5f–5u) were different strains than our Lepr-ires-Cre cohort; this was due to the limited availability of mice during the COVID-19 pandemic. As such, they performed slightly differently on the 2AFC task. Therefore, their learning criterion was lowered to >70% correct responses across three consecutive days.

Quantification and Statistical Analysis

Statistical analyses were performed using Prism 5 and Prism 8(GraphPad) software and are described in the figure legends in all cases. No statistical method was used to predetermine sample size, nor were randomization and blinding methods used. Statistical significance was defined as p<0.05. All data presented met the assumptions of the statistical test employed. As mentioned in sections above, experimental animals were excluded if histological validation revealed poor or absent reporter expression or poor fiber optic placement in the region of interest. These criterion were established prior to data collection. N values reflect the final number of validated animals per group.

Extended Data

Extended Data Figure 1: DMH^Lepr neuron photometry responses are greater with pellet drops than with gut-infusions of caloric substances and inhibiting DMH^Lepr neurons does not affect food intake following an overnight fast.

(a) Heat maps of additional well-trained mice included in Fig. 1d of DMH^Lepr neuron fiber photometry responses to cue (light) onset. Each heat map represents a single behavioral session. Single trials have been sorted by response time. Green line indicates light onset (time = 0), black line indicates time when the mouse poked within the correct port.

(b) Heat maps of all naïve mice included in Fig. 1d of DMH^Lepr neuron fiber photometry responses to cue (light) onset. Each heat map represents a single behavioral session. Single trials have been sorted by response time. Green line indicates light onset (time = 0), black line indicates time when the mouse poked within the correct port. It should be noted that naïve mice perform significantly fewer trials than experienced mice.

(c) Gastric infusion experimental design.

(d-f) Plot of fiber photometry responses from DMH^Lepr neurons aligned to gut infusions of Ensure (d and e), saline (d and e), or caloric and non-caloric sweeteners (f). These were also aligned to food pellet drops post-infusion. Mice were either food deprived or fed ad libitum for sated Ensure infusions. Light color tones indicate S.E.M.

(g) Quantification of fluorescence changes ten minutes post-gut infusions of saline, Ensure, and Ensure when animals were fed ad libitum (sated Ensure). n = 9 (mice); Repeated measures one-way ANOVA, ***p=0.0007.

(h) Comparison between peak fiber photometry responses during the last 30 seconds of Ensure infusion (red) and first 30 seconds of pellet drop response (beige) when animals were given saline infusions. Peak response is significantly larger with pellet drops vs. Ensure infusions. n = 9 (animals); Two-tailed, paired t-test, *p=0.0241.

(i) hM4Di-mediated inhibition of DMH^Lepr neurons does not affect food intake in a post-fast refeeding assay. n = 12 (mice). Two-way ANOVA, p = 0.9992.

All data represents the mean ± S.E.M.

Extended Data Figure 2: LH^Vglut2→DMH neurons preferentially synapse onto DMH^Pdyn-GFP neurons.

(a) Sample image of rabies labeling of afferents to DMH^Lepr neurons. (Top left) Sample of injection site of AAV-DIO-TVA, AAV-DIO-RG, and EnvA-GFP within the DMH. (Right and bottom row) Sample images of rabies labeled afferents. Scale bar = 200 μM. A/P levels are inferred based on histological landmarks. n=3 (animals).

(b) Monosynaptic connection probability of optically evoked IPSCs and EPSCs to vDMH^Pdyn-GFP neurons. X-axis = afferent source.

(c-d) Quantification of oIPSC (a) and oEPSC (b) amplitude from all candidate afferents to DMH^Lepr neurons in the presence and absence of TTX and 4AP. Two-tailed, paired t-test, *p≤0.05. (b) p = 1.664 (NAc), p = 0.0285 (BNST), p = 0.6413 (LH), p = 0.0165 (VTA) (c) p = 0.9366 (LH).

(e) Experimental design schematic. AAV-DIO-ChR2 injected in LH^Vglut2 neurons and oEPSCs were measured in vDMH^Pdyn-GFP neurons. Bottom: Example current-clamp recordings in vDMH^Pdyn-GFP neuron during optogenetic activation of LH^Vglut2 neurons.

(f) LH^Vglut2 neurons preferentially connect to ~80% of DMH^Pdyn-GFP neurons and have sparse connectivity to DMH^Pdyn-GFP negative neurons. n = 15 (cells) GFP+, 17 (cells) GFP-.

(g) oEPSC amplitude is unchanged when TTX and 4AP are added. n = 9 (ACSF) and 11 (TTX+4AP). Two-tailed unpaired t-test, p=0.1666

(h) AAV-DIO-ChR2 was injected into the LH of Vglut2-ires-Cre mice. Whole-cell recordings were performed in vDMH^Pdyn-GFP negative neurons while stimulating LH^Vglut2 terminals. Bottom: representative CRACM traces from vDMH^Pdyn-GFP negative neurons in the presence and absence of TTX and 4AP.

(i) Summarized postsynaptic current amplitude in DMH^Pdyn-GFP negative neurons in the presence and absence of TTX and 4AP. n = 9 (ACSF) and 8 (TTX, 4AP). Two-tailed, unpaired t-test, p = 0.26.

(j) Summarized input resistance in both Pdyn-GFP positive and negative neurons. n = 20 (GFP+) and 17 (GFP–). Two-tailed, unpaired t-test, p = 0.14.

(k) Summarized series resistance in both Pdyn-GFP positive and negative neurons. n = 20 (GFP+) and 17 (GFP–). Two-tailed, unpaired t-test, p = 0.24.

(l-m) Averaged current onset latency (l) and jitter (m) in Pdyn-GFP positive neurons in the absence of TTX and 4AP. n = 8 (cells).

All data represents the mean ± S.E.M.

Extended Data Figure 3: LH^Vglut2→DMH neurons are appetitive, do not send collaterals to the LHb or PAG, have negligible responses to non-edible object presentations, and respond differently to water when compared to food.

(a) Representative wire plots of an XFP and ChR2 expressing animals throughout the entire 20-minute RTPP session across three different conditions: lights-on ad libitum, lights-off ad libitum, and lights-on fasted. Blue line represents stimulation side of the RTPP chamber and black represents the non-stimulation side.

(b-d) Total distance traveled is unchanged between XFP and ChR2 expressing animals throughout the RTPP session across all three conditions (lights-on ad libitum, lights-off ad libitum, and lights-on fasted). n = 7 (animals) for both XFP and ChR2-expressing groups. Unpaired t-test, (b) p = 0.7498, (c) p = 0.5277, (d) p = 0.3482.

(e) DIO-TVA-mCherry was injected into the LH of Vglut2-ires-Cre mice and pseudotyped EnvA-GFP was injected into the DMH to map collaterals of LH^Vglut2 → DMH neurons.

(f) Injection sites within the LH (left) and DMH (right). Scale bar = 500 μM (LH) and 200 μM (DMH)

(g) Collaterals were found within the BNST, MPO/LPO, and AHA and were absent in LH targets known to promote aversion such as the LHb and PAG. n = 3 (animals). Scale bar = 500 μM.

(h) Averaged axonal fiber-photometry response in LH^Vglut2→vDMH terminals across all animals when presented with a non-edible object. Vertical line represents object presentation. n=7 (animals).

(i) Summarized Z-score of LH^Vglut2→vDMH axonal fiber photometry response to pellet (n=9) and object (n=7) drops in fasted mice. One-way ANOVA; Kruskal-Wallis, ****p<0.0001.

(j) Summarized mean peak response in LH^Vglut2→vDMH axonal fiber photometry response to pellet drops when animals were fed ad libitum. n = 9; Two-tailed, unpaired t-test, *p=0.0129.

(k) Averaged axonal fiber-photometry response in LH^Vglut2→vDMH terminals across all animals when given a 500 mg food pellet when food restricted (green) or, a bowl of water when water restricted (blue). Vertical line represents food/water presentation.

(l-m) Quantification of mean peak response of food (in food-deprived mice) and water presentations (in water-deprived mice) (l, p=0.0206). n = 9 (food) and 8 (water). Peak water response occurred at a much later time point than food responses suggesting that the neurons respond differently to water cues when compared to food (m, p=0.0391). n = 9 (food) and 8 (water); Two-tailed, unpaired t-test, *p<0.05.

A/P levels are inferred based on histological landmarks from the referenced mouse atlas.

All data represents the mean ± S.E.M.

Extended Data Figure 4: AAV-fDIO-hM4Di efficiently inhibits LH^Vglut2 neurons, CNO does not affect DMH^Lepr activation, and hM4Di inhibition does not affect sustained AgRP inhibition.

(a) Validation of AAV-fDIO-hM4Di construct using whole-cell patch clamp electrophysiology. Upon CNO wash-on (red), the rheobase increase is indicative of a successfully hyperpolarized LH^Vglut2 neuron given that a higher injection of current is needed to fire an action potential during CNO-wash on.

(b) CNO wash-on also eliminated spontaneous action potential firing in LH^Vglut2 neurons in fDIO-hM4Di expressing neurons.

(c-d) XFP controls were used to determine if CNO would affect DMH^Lepr activation in the absence of hM4Di. CNO did not affect DMH^Lepr neuron photometry responses (c, p=0.7616) or behavioral performance (d). n = 3 (mice).

(e) Inhibiting LH→DMH afferents does not affect behavioral response time in well-trained mice. n = 4 (mice). Relates to Fig. 3a–3c. Two-tailed t-test, p=0.7068.

(f) CNO treatment does not affect the later consummatory response in AgRP neurons when mice are given a large (250 mg) food pellet. n = 4 (mice); Two-tailed t-test, p = 0.61.

(g) Inhibiting LH→DMH afferents in well-trained mice does not affect behavioral performance. n = 4 (mice), Two-tailed t-test, p=0.3819. Relates to Fig. 3a–3c.

(h) Inhibition of LH^Vglut2 neurons does not affect behavioral performance. n = 4 (mice), Two-tailed t-test p=0.3661. Relates to Fig. 3d–3f.

(i-j) Inhibition of LH^Vglut2 neurons does not affect behavioral accuracy (i) or response time (j). n = 4 (mice), Two-tailed t-test, p=0.7107. Relates to Fig. 3g–3i.

All data represents the mean ± S.E.M.

Extended Data Figure 5: TeNT-mediated silencing of DMH^Lepr, DMH^Pdyn, and LH^→DMH neurons significantly attenuates learning a cue-initiated food acquisition task by increasing mistakes made during the task but TeNT-mediated silencing of DMH^Lepr neurons does not affect learning a cue-initiated water acquisition task.

(a-b) DMH^Lepr neuron silencing leads to an increase in response errors in the 2AFC task. TeNT-expressing mice largely miss the response window (a) in the initial phases of training and have an increase in false alarms (b). n = 5 (GFP), 5 (TeNT). (a) Red dotted line indicates line of best fit (***p<0.0001, plateau = 5.735 (GFP), −2.910 (TeNT); tau = 2.923 (GFP), 12.64 (TeNT)). Two-way repeated measures ANOVA, main effect of days = ****p<0.0001, main effect of group = 0.05. (b) Red dotted line indicates line of best fit (***p<0.0001, plateau = 0.1324 (GFP), −162.8 (TeNT); tau = 6.679 (GFP), 117.6 (TeNT)). Two-way repeated measures ANOVA, main effect of days = ****p<0.0001, main effect of group = 0.6.

(c) TeNT-expressing Lepr-Cre mice take significantly longer to nose poke following cue-light presentation. n = 5 (GFP), 5 (TeNT). ). Red dotted line indicates line of best fit (***p<0.0001, plateau = 1.319 (GFP), 0.633 (TeNT); tau = 4.272 (GFP), 9.860 (TeNT)). Two-way repeated measures ANOVA, main effect of days = **p<0.01, main effect of group = 0.08.

(d) Body weight is unaffected upon DMH^Lepr silencing with TeNT (7 weeks post-surgery). n = 5 (GFP), 5 (TeNT). Two-tailed, unpaired t-test, p=0.7043.

(e) Food-intake does not significantly differ with DMH^Lepr silencing following an overnight fast. n = 5 (GFP), 6 (TeNT). Two-tailed, unpaired t-test, p=0.9495.

(f) AAV-DIO-TeNT injection into DMH^Pdyn neurons. Bottom: Example images of TeNT expression. Scale bar = 500 μM. n=5 (mice).

(g-h) Similar to (Fig. 2b) in DMH^Pdyn neurons. n = 7 (GFP), 5 (TeNT). Red line indicates = line of best fit (***p<0.0001, plateau = 82.83 (GFP), 65.52 (TeNT); tau = 5.619 (GFP), 7.337 (TeNT)). Two-way repeated measures ANOVA, main effect of days = ****p<0.0001, main effect of group = **p<0.01. (h) Similar to (Fig. 2c) in DMH^Pdyn neurons (learning criterion >70% correct across three consecutive days). >20 on the y-axis is to indicate that by day 20, mice had not learned. 7/7 GFP and 1/6 TeNT mice learned. n = 7 (GFP), 5 (TeNT). Two-tailed, unpaired t-test, **p=0.0094.

(i-j) DMH^Pdyn neuron silencing leads to an increase in response errors in the 2AFC task. TeNT-expressing mice largely miss the response window (i) in the initial phases of training and do not have a difference in false alarm rate (j). n = 7 (GFP), 5 (TeNT). (i) Red dotted line indicates line of best fit (***p<0.0001, plateau = 7.142 (GFP), 20.89 (TeNT); tau = 0.8597 (GFP), 1.96 (TeNT)). Two-way repeated measures ANOVA, main effect of days = ***p<0.001, main effect of group = *p<0.05. (j) Line of best fit could not be calculated. Two-way repeated measures ANOVA, main effect of days = ***p<0.001, main effect of group = 0.8267.

(k) TeNT-expressing Pdyn-Cre mice take significantly longer to nose poke following cue-light presentation. n = 7 (GFP), 5 (TeNT). ). Red dotted line indicates line of best fit (***p<0.0001, plateau = 1.185 (GFP), 2.748 (TeNT); tau = 2.122 (GFP), 3.939 (TeNT)). Two-way repeated measures ANOVA, main effect of days = ****p<0.0001, main effect of group = **** p<0.0001.

(l) Body weight increases upon long term DMH^Pdyn silencing with TeNT (20 weeks post-surgery). n = 7 (GFP), 5 (TeNT). Two-tailed, unpaired t-test, **p=0.0086.

(m) Food-intake does not significantly differ with DMH^Pdyn silencing following an overnight fast. n = 7 (GFP), 5 (TeNT). Two-tailed, unpaired t-test, p=0.2668.

(n) AAV-DIO-TeNT injection into the LH and rAAV-Cre injection into the DMH in wildtype mice. Bottom: Example images of TeNT expression. Scale bar = 500 μM. n=6 (mice).

(o-p) Similar to (Fig. 2b) in LH^→DMH neurons. n = 5 (GFP), 6 (TeNT). Red line = line of best fit (***p<0.0001, slope = 2.033 (GFP), 1.488 (TeNT); Y-intercept = 44.50 (GFP), 30.02 (TeNT)). Two-way repeated measures ANOVA, main effect of days = ****p<0.0001, main effect of group = p=0.0548. (p) Similar to (Fig. 2c) in LH^→DMH neurons (learning criterion >70% correct across three consecutive days). 5/5 GFP and 2/6 TeNT-expressing mice learned. n = 5 (GFP), 6 (TeNT). Two-tailed, unpaired t-test, p=0.0789.

(q-r) LH^→DMH neuron silencing leads to an increase in response errors in the 2AFC task. TeNT-expressing mice largely miss the response window (q) in the initial phases of training and do not have a difference in false alarm rate (r). n = 5 (GFP), 6 (TeNT). (q) Red dotted line indicates line of best fit (***p<0.0001, slope = −0.299 (GFP), −0.4169 (TeNT); Y-intercept = 16.75 (GFP), 39.91 (TeNT)). Two-way repeated measures ANOVA, main effect of days = *p=0.0416, main effect of group = p=0.1896. (r) Line of best fit is not significantly different for TeNT and GFP-expressing animals (p=0.1724, plateau = −22.34 (GFP), −74.42 (TeNT); Tau = 23.91 (GFP), 83.89 (TeNT)). Two-way repeated measures ANOVA, main effect of days = *p=0.0101, main effect of group = 0.9094.

(s) TeNT-expression in LH^→DMH neurons significantly increases nose poke response following cue-light presentation. n = 5 (GFP), 6 (TeNT). ). Red dotted line indicates line of best fit (***p<0.0001, plateau = 1.790 (GFP), 3.314 (TeNT); tau = 1.818 (GFP), 8.042 (TeNT)). Two-way repeated measures ANOVA, main effect of days = ***p = 0.0004, main effect of group = ****p<0.0001.

(t) Body weight is unaffected upon LH^→DMH silencing with TeNT (8 weeks post-surgery). n = 5 (GFP), 6 (TeNT). Two-tailed, unpaired t-test, p=0.2820.

(u) Food-intake does not significantly differ with LH^→DMH silencing following an overnight fast. n = 5 (GFP), 6 (TeNT). Two-tailed, unpaired t-test, p=0.8666.

(v-x) DMH^Lepr neuron silencing does not affect response errors or response times in the 2AFC task when dehydrated mice receive water rewards. n = 4 (GFP), 6 (TeNT). (v) Red dotted line indicates line of best fit (plateau =57.74, tau = 4.477). Two-way repeated measures ANOVA, main effect of days = ****p<0.0001, main effect of group = 0.942. (w) Red dotted line indicates line of best fit (plateau = −15.32, tau = 11.67). Two-way repeated measures ANOVA, main effect of days = *p<0.05, main effect of group = 0.13. (x) Red dotted line indicates line of best fit (plateau = 0.7882, tau = 2.929). Two-way repeated measures ANOVA, main effect of days = ****p<0.0001, main effect of group = 0.82.

All data represents the mean ± S.E.M.

Extended Data Figure 6: Pdyn-expressing DMH neurons exclusively project to the ARC.

AAV-DIO-ChR2-mCherry was injected into the DMH of Pdyn-Cre mice to search for long range projections of DMH^Pdyn neurons. The ARC was the only area of the brain that received DMH^Pdyn innervation. A portion of this data was previously published in Garfield et. al. 2016¹².

Extended Data Figure 7: Schematic representation of viral spread and fiber placement.

Schematic representing the extent of viral spread (transparent shaded regions) and fiber placement (solid ovals) for every animal related to studies presented in (a) Figs. 1b–1d, Extended Data Fig. 1a, (b) Figure 2a–2e, Extended Data Figure 3a–3d and, (c) Figure 2f–2k, Extended Data Fig. 3h–3m. Please note that different animals are represented by different colors and that “hits” and “misses” are represented on different hemispheres (miss is on the left). GENERAL CRITERION FOR HITS AND MISSES: Animals were deemed a “hit” if viral expression was within the area of interest and fiber placement was either within or immediately dorsal to the area of interest (i.e. the DMH or the LH). Excluded animals are specified by the “miss” label.

REASON EACH ANIMAL WAS CONSIDERED A “MISS”
Figure	Animal	Viral expression/spread	Fiber placement
a	5	-	Dorso-lateral to the DMH
b	10	Unilateral expression	Lateral to midline/DMH
c	10	Ventro-medial to LH	Lateral to DMH

Extended Data Figure 8: Schematic representation of viral spread and fiber placement within Lepr-Cre animals.

Schematic representing the extent of viral spread (transparent shaded regions) and fiber placement (solid ovals) for every animal related to studies presented in Figure 3a–3c and Extended Data Fig. 4e–4g. Please note that individual animals are represented by different colors. Bilateral hM4Di injections were targeted to the LH and bilateral AAV6-FlpO injections were targeted to the DMH. GCaMP6s injections into the DMH were unilateral. CRITERION FOR HITS AND MISSES: Animals were deemed a “hit” if viral expression was within the area of interest (i.e. hM4Di bilateral in the LH, GCaMP unilateral in the DMH) and fiber placement was either within or immediately dorsal to the area of interest (i.e. the DMH). For clarity, hits are shown on the top and misses on the bottom.

REASON EACH ANIMAL WAS CONSIDERED A “MISS”
Animal	Viral expression/spread	Fiber placement
1	(hM4Di) Medial to DMH	Ventral to DMH
2	(hM4Di) unilateral	Ventral to DMH
3	(hM4Di) unilateral	Dorsal to DMH
4	(hM4Di) no expression	Ventral to DMH

Extended Data Figure 9: Schematic representation of viral spread and fiber placement within Vglut2-FlpO::Lepr-Cre animals.

Schematic representing the extent of viral spread (transparent shaded regions) and fiber placement (solid ovals) for every animal related to studies presented in Figure 3d–3f and Extended Data Fig. 4h. Please note that individual animals are represented by different colors. Bilateral hM4Di injections were targeted to the LH and GCaMP6s injections were unilateral into the DMH. GENERAL CRITERION FOR HITS AND MISSES: Animals were deemed a “hit” if viral expression was within the area of interest (i.e. hM4Di bilateral in the LH, GCaMP unilateral in the DMH) and fiber placement was either within or immediately dorsal to the area of interest (i.e. the DMH). For clarity, hits are shown on the top and misses on the bottom.

REASON EACH ANIMAL WAS CONSIDERED A “MISS”
Animal	Viral expression/spread
1	(hM4Di) dorsal to LH
2	(hM4Di) unilateral and dorsal (GCaMP6s) minimal expression and lateral

Extended Data Figure 10: Schematic representation of viral spread and fiber placement within Vglut2-FlpO::AgRP-Cre animals.

Schematic representing the extent of viral spread (transparent shaded regions) and fiber placement (solid ovals) for every animal related to studies presented in Figure 3g–3i and Extended Data Figs. 4i–4j. Please note that different animals are represented by different colors. Bilateral hM4Di injections were targeted to the LH and GCaMP6s injections were unilateral into the ARC. Purple regions indicate areas with no viral expression. GENERAL CRITERION FOR HITS AND MISSES: Animals were deemed a “hit” if viral expression was within the area of interest (i.e. hM4Di bilateral in the LH, GCaMP unilateral in the ARC) and fiber placement was either within or immediately dorsal to the area of interest (i.e. the ARC). For clarity, hits are shown on the top and misses on the bottom.

REASON EACH ANIMAL WAS CONSIDERED A “MISS”
Animal	Viral expression/spread
1	(hM4Di) Unilateral and dorsal to LH (GCaMP6s) no expression
2	(hM4Di) unilateral and dorsal (GCaMP6s) no expression

Extended Data Figure 11: Schematic representation of viral spread.

Schematic representing the extent of viral spread (transparent shaded regions) for every animal related to studies presented in (a) Figure 4a–4c, Extended Data Fig. 5a–5e, (b) Extended Data Fig. 5f–5m, (c) Extended Data Fig. 5n–5u, and (d) Figure 4d–4f and Extended Data Fig. 5v–5x. Please note that different animals are represented by different colors. GENERAL CRITERION FOR HITS AND MISSES: Animals were deemed a “hit” if viral expression was within the area of interest (i.e. the DMH or the LH). Excluded animals are specified by the “miss” label.

REASON EACH ANIMAL WAS CONSIDERED A “MISS”
Figure	Animal	Viral expression/spread
a	7	DMH and VMH expression
b	6	Dorsal to DMH
b	7	Barely detectable levels of expression
c	7	Dorsal to LH
c	8	Diffuse, widespread expression dorsal to LH
c	9	Dorsal to LH (Zona Incerta expression)
c	10	Dorsal to LH (Zona Incerta expression)
d	7	Dorsal to DMH
d	8	Barely detectable in a small region in anterior DMH

Extended Data Figure 12: Example of method used to draw anatomical boundaries in histological images.

DAPI-stained images were taken and contrast was enhanced (left). Images were matched with the appropriate A/P coordinate within the histological atlas (middle) then, boundaries were drawn using the axis as a template (right). All anatomical boundaries represented within the figures were drawn using this method. The atlas used for these studies was The Mouse Brain in Stereotaxic Coordinates by Paxinos and Franklin, Second Edition.

Data and Software Availability

The datasets and custom MATLAB scripts generated and/or analyzed during the current study are available from the corresponding author upon request. MATLAB scripts to run and analyze 2AFC behavior are freely available on Github. For more details on installation please visit: https://sites.google.com/site/bpoddocumentation/installing-bpod.