{"paper_id":"22df99b3-4df0-4f5a-9bf6-8035edc874d9","body_text":"Title \nObesity dysregulates feeding-evoked response dynamics in hypothalamic satiety neurons \n \nAuthors \nMarta Porniece1, Jessica Baker1, Charlotte D. Ausfahl1, Stephen X. Zhang1*, Mark L. Andermann1* \n \n \n1Division of Endocrinology, Diabetes and Metabolism, Department of Medicine, Beth Israel \nDeaconess Medical Center, Harvard Medical School, 02115, Boston, MA, USA \n \n*Correspondence: xz5184@nyu.edu and manderma@bidmc.harvard.edu \n \n \n  \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nAbstract \nMelanocortin-4 receptor -expressing neurons in the paraventricular nucleus of the \nhypothalamus (PVHMC4R) integrate hunger-promoting and hunger-suppressing signals to regulate \nsatiety. Food consumption-evoked responses in PVHMC4R neurons increase gradually during meal \nconsumption to promote satiety, and disrupting this process drives massive obesity. These critical \nsatiety neurons are strongly affected by a high-fat diet, yet the impact on their functional properties \nremains unknown.  We used fiber photometry to track PVH MC4R neurons’ responses to the \nconsumption of drops of milkshake in animals fed a chow diet or a high-fat diet (HFD), both after \nobesity was established and after its reversal. PVHMC4R neurons in HFD-fed animals showed \ngreater consumption-evoked responses than chow -fed animals  at the early stages of meal \nconsumption, and these responses did not increase further during the meal . HFD-fed animals \nalso showed reduced licking vigor and motivation to consume Ensure. Switching HFD-fed obese \nanimals to a normal chow diet (NCD) re-engaged the motivation to consume Ensure, partially \nrestoring early-meal neural responses  to a lower level, but did not restore the increase in  \nconsumption-evoked response magnitude across the mea l. These findings highlight functional \nalterations in hypothalamic satiety -promoting neurons in obesity  and provide insight into the  \npathological neural consequences of an obesogenic environment. \n  \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nIntroduction \nNeurons in the paraventricular nucleus of the hypothalamus (PVH)  that express the \nmelanocortin-4 receptor (PVHMC4R) are critical for regulating food intake and maintaining energy \nbalance1–4. Recent research has advanced our understanding of how PVHMC4R neurons process \nthese signals under normal physiological conditions5,6. In lean animals, PVHMC4R neuron activation \nis regulated by  integration of  inputs from two antagonisti c upstream neuronal populations : 1) \nhunger-promoting Agouti-related peptide (AgRP) neurons , which selectively engage and inhibit \nPVHMC4R neurons by releasing neuropeptide Y ( NPY) that acts on  NPY1 receptors7–10 (among \nother mechanisms 11–14), and 2) the satiety-promoting pro -opiomelanocortin (POMC) neurons , \nwhich release α-MSH peptide from synaptic terminals onto MC4Rs in PVH2,5,6,15,16. This dynamic \ninterplay is further modulated by circulating levels of metabolic hormones such as leptin  and \ninsulin, which reflect the body’s energy state and act on AgRP and PO MC neurons to fine-tune \nthe downstream response in PVHMC4R neurons17–19. During a meal, increased α-MSH and reduced \nNPY release together elevate the intracellular second messenger cAMP in PVHMC4R neurons to \npromote satiety 5,6,20–22. This mechanism ensures that caloric intake aligns with the body’s energy \nneeds, preventing overeating and promoting energy balance . Notably, animals with AAV-\nmediated expression of the cAMP-degrading phosphodiesterase PDE4D3 -Cat in PVH MC4R \nneurons exhibit hyperphagia and rapid weight gain  along with  altered intrinsic excitability of \nPVHMC4R neurons and impaired sensitivity to feeding-related excitatory inputs2,5,23. \nThe functionality of this finely tuned hypothalamic circuit becomes compromised in \nobesity24–27. For example, rare genetic variants that decrease α-MSH release by POMC neurons \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nlead to early onset, severe, and rapid weight gain 27–32. These conditions  involving reduced \nstimulation of the MC4R are effectively treated by administering MC4R agonist s, such as \nsetmelanotide. These agonists bind to and activate MC4R, mimicking the natural signaling that \nwould normally suppress appetite and promote energy expenditure32–35. In contrast, diet-induced \nobesity (DIO) may be associated with a saturation of MC4R signaling  as well as other neural \ncircuit dysfunction (see below) . Accordingly, while  setmelanotide treatment acutely improve s \nmultiple metabolic parameters in DIO, chronic setmelanotide-induced PVHMC4R neuron activation \nis not effective in reducing food intake and body weight in DIO36,37. \nDIO results from the excessive consumption of calorie-dense foods and has been linked \nto hypothalamic inflammation, gliosis, and other forms of hypothalamic injury24. DIO dysregulates \nfeeding circuit s by desensitizing AgRP and POMC neuron responses  to food  and altering \nexcitability and neuropeptide signaling 25,38,39. In DIO, AgRP/NPY neurons show increased \nspontaneous firing due to altered intrinsic excitability40,41, while  POMC neurons exhibit a decrease \nin spontaneous activity due to a hyperpolarized membrane potential42. Furthermore, in vivo fiber \nphotometry recordings reveal obesity -driven reductions in intragastric nutrient- or hormon e-\ninduced modulation of AgRP neurons, which may either promote or reduce food intake (e.g., via \ndesensitization of AgRP responses to intragastric infusion of fat or blunting of ghrelin-induced \nAgRP neuron activation, respectively)25,43. Obesity also attenuates the rapid responses of AgRP \nneurons to sensory food cues  and food consumption25,38,44,45. Additionally, obesity blunts the \nresponses of AgRP and POMC neurons to a variety of hormonal inputs that vary between fasted \nand fed states , such as ghrelin, CCK, leptin , and insulin25,40,41,46,47. These disruptions impair \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nPVHMC4R neuron sensitivity to upstream inputs and predict weaker elevations in cAMP during \nfeeding and a weaker meal -related increase in PVH MC4R excitation, thereby compromising the \ncritical role of these neurons in energy balance, meal size, and weight regulation5,37,47,48. In the \nPVH, long-term HFD exposure induces the loss of MC4R protein  abundance and mitochondrial \ncontent in PVHMC4R neurons, even though the number of PVHMC4R neurons remains the same48,49. \nThis loss is accompanied by diminished a-MSH expression in the hypothalamic arcuate nucleus, \nfurther suggesting that exposure to dietary fat induces alterations in α-MSH-MC4R signaling48. In \nsummary, excessive dietary fat consumption disrupts melanocortin signaling by impairing \nupstream inputs to PVHMC4R neurons and how PVHMC4R neurons integrate these inputs to regulate \nbehavior. \nHere, we investigate  whether and how obesity-related disruptions in melanocortin  \nsignaling manifest in functional changes in PVHMC4R neuron responses during feeding. We aimed \nto understand the maladaptive plasticity in PVHMC4R neurons that may contribute to overeating in \nDIO. We employed fiber photometry to track the real-time activity of PVHMC4R neurons during meal \nconsumption in both lean and obese states, as well as following the transition from HFD back to \na normal chow diet . Our results provide new insights into the plasticity of hypothalamic satiety \nmechanisms in response to changes in diet and highlight the potential for targeted interventions \nto restore energy balance in obesity. \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nResults \nFeeding-related responses in PVHMC4R neurons are elevated early in a meal in HFD-fed \nanimals \nTo assess the state-dependent integration of satiety signals in PVH MC4R neurons in lean \nand obese states (Fig. 1a), we provided animals with ad libitum normal chow diet (NCD) or calorie-\ndense high-fat diet (HFD , 60% of calories from fat)  from 5 weeks of age.  We then selectively \nexpressed the calcium sensor GCaMP6s ( AAV-Syn-Flex-GCaMP6s) in PVHMC4R neurons in \nMC4R-Cre mice, and placed an optic fiber above the PVH for fiber photometry recordings (Fig. \n1b). After recovery from surgery, the two cohorts of animals were fed limited, calorie-matched \namounts of NCD or HFD  daily (~9.5 kcal/day; Fig. 1c, d) . This food restriction motivated the \nanimals to perform a simple operant tone-conditioned feeding task for 4-5 weeks.  \nFor both experimental groups , we monitored body weight and food intake in the home \ncage (Fig. 1c, d). During the fiber photometry recordings, we tracked Ensure consumption, licking \nvigor, and locomotion (Fig . 1e, f , Supplementary Fig. 1 a, b). Upon food restriction, HFD -fed \nanimals required a longer time to reach a steady -state level of reduced  body weight \n(Supplementary Fig. 1c, d) and to engage in and learn the operant task (not shown). HFD-fed \nmice were ~30% heavier than the NCD-fed mice, consistent with their elevated weight before the \nstart of food restriction  (Fig. 1c). Despite calorie matching to ensure comparable acute hunger \nstates (Fig . 1d), the HFD animals consumed less Ensure volume during the experimental \nrecordings (Supplementary Fig. 1a, b). Both NCD- and HFD-fed animals steadily reduced the ir \nlicking rates from early to late phases of the session (Supplementary Fig. 1e, f). Compared to \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nNCD-fed animals, food-seeking-related locomotion was negligible in HFD-fed animals (Fig. 1e, f \ntop panels, Supplementary Fig. 1g), and it was further reduced to a minimum level late near the \nend of the meal (Supplementary Fig. 1g). \nUsing a head-fixed fiber photometry setup, we tracked calcium signals in real -time in \nparallel with licking behavior and locomotion (Fig. 1e, f, Supplementary Fig. 1h, i). Of all the trials \nin a session, we only analyzed the trials  in which tone-evoked licking triggered Ensure delivery  \n(“triggered trials”), as only these trials contributed to the gradual increase in satiety. NCD-fed \nanimals consumed the palatable drops of Ensure from the beginning of the 60-minute meal, with \nlicking vigor reduc ing throughout the meal as animals got satiated (Fig. 1e). Accordingly, after \naveraging all recordings (n = 39 sessions from 6 mice) , we  observed weak feeding-evoked \nresponses early in the meal  (first 15 trials) (Fig. 1e, g, h, Supplementary Fig. 1j). Similar to our \nprior study5, as the animals progressed through the meal , feeding-evoked responses gradually \nincreased (Fig. 1e (top left panel)). This gradual increase in MC4R neuron  activity and parallel \nreduction of licking vigor tracks the approach to the sated state as the animal continues to eat, \nwhich is likely to curb food intake once sufficient caloric intake is achieved5. \nIn contrast, HFD-fed animals (n = 33 recording sessions / 5 mice) stopped eating earlier \nduring the experiment compared to the NCD -fed animals (Fig . 1e, f , lower panel s; see also  \nSupplementary Fig. 1g, h below). As discussed above, we matched levels of total daily calorie \nintake between NCD and HFD cohorts from the beginning of the food restriction  and training \nphase of our protocol ( to attempt to match acute hunger states  across mice , Fig. 1 c, d). \nNevertheless, at the start of the meal, HFD-fed animals exhibited stronger feeding-evoked neural \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nresponses than NCD -fed animals, and these acute responses persisted until the end of the \nfeeding bout (Supplementary Fig. 1j, k, left panels). \nWe next assessed the magnitude of PVH MC4R neuron responses  during the food \nconsumption phase (0 – 10 s post-tone, restricted to trials with Ensure delivery). Specifically, we \ncompared responses early in the meal (first 15 triggered trials) and late in the meal (last 15 \ntriggered trials) in each recording. On average, these neural responses were minimal in NCD-fed \nanimals in early trials (Z-score (NCD early) = 0.12 ± 0.11), but exhibited a robust increase in late \ntrials (Fig. 1e (top panel); Fig. 1i) (Z-score (NCD late) = 0. 86 ± 0.09), consistent with our prior \nstudy5. While some variation in early trial activity was observed among individual mice,  all \nindividual NCD-fed animals exhibited a n overall  increase in consumption-related neural \nresponses in the late meal phase (Z-score (NCD early) = 0.19 ± 0.24, Z-score (NCD late) = 0.91 \n± 0.13)) (Fig. 1g, h, j). \nIn contrast , PVHMC4R neurons in HFD-fed animals were already highly driven during \nfeeding in the early phase of the meal (Fig. 1f (top panel); Fig. 1g (lower panel), Fig. 1i) (Z-score \n(HFD early) = 0.72 ± 0.09), with no further increase observed throughout the meal (Z-score (HFD \nlate) = 0.79 ± 0.09) (Fig. 1i). This pattern was consistent across all subjects (Z-score (HFD early) \n= 0.73 ± 0.19, Z-score (HFD late) = 0.80 ± 0.20) (Fig. 1h (lower panel), Fig 1j). Additionally, HFD-\nfed animals did not show a change in response from early to late in the meal when considering a \nlonger window of 0-20 seconds post tone (Supplementary Fig. 1k). These experiments confirm a \nprior study5 showing that PVHMC4R neurons are increasingly excited by food consumption as NCD-\nfed animals transition from hunger to satiety, and provide the first evidence that this within-meal \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nchange in neuronal responses is dysfunctional in HFD-fed animals. In particular, stronger-than-\nexpected feeding-related responses in PVH MC4R neurons at the start of a meal  may explain the \ndiminished motivation to consume Ensure in these obese animals (see Discussion). \n \nHFD alters neural and behavioral responses during the progression from hunger to satiety \nWe observed considerable inter-session variability in the number of trials with Ensure \nconsumption required for the mice to become sated. Thus, we sought to subdivide the sessions \nby the number of trials required for satiation. In this way, we could compare neural and behavioral \nresponses in  the two groups of mice when matching for this  index of motivation. However, \nbecause NCD-fed mice often consumed Ensure in all 60 trials in the experiments in the above \nexperiments (Figure 1), we ran additional experiments in which we extended the session from 60 \nto 90 trials to allow for voluntary meal termination in both NCD-fed (n = 59 recordings / 7 mice) \nand HFD-fed (n = 54 recordings / 7 mice) animals.  \nDuring these extended recording sessions, l ean animals continued to show progressive \nincreases in feeding-related PVHMC4R responses (Fig. 2a) throughout a larger number of  trials \nthan was possible in the earlier experiments . I n these longer experiments , NCD -fed mice \nconsumed Ensure during 70.1 ± 2.1 out of 90 possible trials, whereas HFD-fed animals only \nconsumed Ensure during 52.9 ± 1.98 out of 90 trials (Fig. 2b). Consistent with our findings in \nFigure 1, in NCD-fed animals, excitatory responses increased over the meal, with a positive mean \nchange in Z-score (ΔZ-score) between early and late in the meal of 0.52 ± 0.09. However, in HFD-\nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nfed animals, the increase was  minimal (ΔZ-score = 0.12 ± 0.08; Fig. 2c). Similar results were \nobserved across individual animals (ΔZ-score: NCD = 0.73 ± 0.07, HFD = 0.33 ± 0.10) (Fig. 2d). \nIn these extended sessions, we first attempted to examine the across-trial changes in the \nanimal’s motivation to consume food by measuring lick vigor during cue presentation. Throughout \nthe meal, NCD-fed animals exhibited higher lick vigor than HFD -fed animals (Fig. 2e). In NCD-\nfed animals, lick vigor began to decline steadily after 50 trials, coinciding with the increase in \nPVHMC4R responses as mice approached satiety. In contrast, in HFD-fed animals, lick vigor began \nto decline much earlier, after only 20 trials (Fig. 2e). \nWe hypothesized that on different recording days, animals’ hunger levels may have varied, \npotentially leading to different response magnitudes in PVHMC4R neurons even when considering \nresponses at the same time point in each session (Fig. 2b). Benefitting from the extended session \nduration, we could now stratify experiments by meal size before satiation (i.e. by the number of \ntriggered trials where Ensure was delivered, in sessions with at least 15  triggered trials) and \nexamine effects on  the motivation to consume the food  (licking vigor during cue) (Fig. 2f – j, \nSupplementary Fig. 2a) and the development of neural responses over phases of the meal (Fig. \n2k – o, Supplementary Fig. 2b). First, we found that both HFD-fed and NCD-fed animals exhibited \nvery variable lick rates early in the meal. Sessions with smaller meal sizes (fewer total triggered \ntrials and lower caloric intake) exhibited lower lick vigor (i.e., the initial lick rate during the cue was \n4-5.5 licks/s for NCD mice versus 2-5.7 licks/s for HFD mice) (Fig. 2f – j, p, Supplementary Fig. \n2a). \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nWe then analyzed the strengthening of PVHMC4R responses throughout the meal across \ndifferent meal sizes. In HFD-fed animals, sessions with short meals exhibited the highest neural \nresponses early in the meal, and these responses reached their peak (Fig. 2k) as the lick rate fell \nto its minimum level (Fig. 2f), re flecting lower engagement and peak satiety. For short meal \nsessions with (30–44 triggered trials ), HFD-fed animals exhibited elevated PVH MC4R responses \ndespite the  reduced lick rate  early in the meal . In longer meal sessions (45–59 and 60 –74 \ntriggered trials), lick ing vigor  early in the meal  was comparable between NCD - and HFD -fed \ngroups, yet PVHMC4R responses were initially larger and  remained elevated in HFD -fed animals \n(Fig. 2g – h, l – m; Supplementary Fig. 2h – i). For the longest meal size (75 – 90 triggered trials), \nNCD-fed mice showed gradually increas ing PVHMC4R responses and consistently robust \nmotivation to consume the food  across the entire meal . In contrast, HFD -fed mice, despite \ndemonstrating comparable lick vigor early in the meal, became satiated earlier, and their PVHMC4R \nresponse magnitude did not increase throughout the session (Fig. 2I, j, n, o, Supplementary Fig. \n2a – b). Overall, this analysis revealed that PVHMC4R response magnitudes exhibited distinct \ndynamics throughout the meal in NCD-fed and HFD-fed animals, and also varied across sessions \nand mice within each group depending on meal size. Critically, despite both HFD- and NCD-fed \nmice having similar lick vigor early during a session (e.g., Fig. 2h), HFD -fed mice still showed \nelevated PVHMC4R responses in the early trials (e.g., Fig. 2m). This indicates that the reduced \nmotivation to consume food is likely not the primary driver of the elevated early session PVHMC4R \nresponses observed in HFD -fed animals. This finding supports the idea that obesity alters the \nunderlying satiety-related signaling mechanism in PVHMC4R neurons. \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nFurther analyses revealed that the PVHMC4R responses in HFD -fed animals negatively \ncorrelated with meal size (HFD: r = -0.46, p = 0.001) (Fig. 2q), whereas lick rate early in the meal \npositively correlated with the number of trials triggered in the session for both diet groups  (Fig. \n2p). This indicates that in HFD -fed animals, early PVH MC4R responses and initial motivation to \nconsume food early in a session predict subsequent meal size and caloric intake. \nSimilar to cue -evoked licking, licking during the reward period (1  – 9 s) post-tone in HFD-fed \nanimals progressively declined with meal size (Supplementary Fig. 2c, d). We also assessed \nlocomotion both before cue onset  and after receiving the reward . Although HFD -fed animals \nexhibited reduced locomotion, this did not correlate with meal size (Supplementary Fig. 2e, f).  \nWe next explored the relationship between within-meal increases in PVHMC4R responses \nand overall meal engagement across sessions (Supplementary Fig. 2g) . To this end, we \ncorrelated session-by-session differences in the number of triggered trials (a measure of  meal \nsize) with the magnitude of increase in PVHMC4R response magnitude across the session. This \nanalysis revealed a positive correlation in the NCD-fed group and a negative correlation in the \nHFD-fed group (NCD: r = -0.34, p = 0.012, HFD: r = 0.26, p = 0.048 ). This further highlights the \ndecoupling of satiety circuit activity and meal size in HFD -fed mice . Interestingly, when we \nassessed sex-specific differences, we found that NCD-fed females exhibited slightly larger neural \nresponses than males early in the meal  – a trend that was even more pronounced in HFD -fed \nanimals (Supplementary Fig. 2h). \nLastly, we also considered the possibility that the observed reduction in licking vigor and \nincreased early-meal PVHMC4R excitation in HFD-fed mice was due to insufficient caloric restriction \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nin the context of the animals’  recent energy surplus. We therefore recorded sessions with \novernight fasting in addition to the chronic food restriction protocol, to increase their hunger drive \n(Supplementary Fig. 2i – m). The additional fasting created an additional ~9,5 kCal/day and ~9 \nkCal/day calorie deficit for NCD- and HFD-fed mice, respectively. However, additional fasting did \nnot significantly alter licking or PVHMC4R response dynamics for either group (Supplementary Fig. \n2i – m). Together, t hese results suggest that diet -induced obesity results in  tonically high \nexcitation of PVHMC4R neurons that is not simply a consequence of caloric surplus but may reflect \na disruption in the adaptive signaling required for proper meal termination and satiety regulation, \npotentially contributing to obesity maintenance. \n \nEffects of Diet Switching on MC4R Neuron Function and Feeding Behavior \nTo investigate whether PVHMC4R neuron dysfunction in diet-induced obesity is reversible, \nwe implemented a diet-switching protocol (Fig. 3a). A subset of animals previously fed a HFD (30 \nrecordings / 7 animals) were switched to a NCD (42 recordings / 7 animals), while other NCD (52 \nrecordings / 6 animals) and HFD (25 recordings / 4 animals) groups remained on their initial diets \n(NCD: 43 recordings / 6 animals; HFD: 23 recordings / 4 animals). This design enabled cross -\ncomparisons among three groups: NCD -to-NCD, HFD-to-NCD, and HFD -to-HFD. Prior to this  \nsecond phase of neural recordings ( calorie-matched across groups), all animals were provided \nad libitum  access to their respective diets, followed by food restriction  and re-habituation to \noperant conditioning. \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nIn this second phase of recordings, the NCD-to-NCD group again exhibited the previously \nobserved PVHMC4R neuron response potentiation across the meal, with the strongest transients \noccurring near the end of the 90-trial session (Fig. 3b, e, h). This activity correlated with vigorous \nlicking during cue presentation throughout the experiment (Fig. 3b, e; Supplementary Fig. 3a). In \ncontrast, during the initial phase of HFD consumption, the HFD-to-NCD group showed greater \nPVHMC4R responses from the first bite but progressively reduced licking and ceased Ensure \nconsumption by trial 80 (Fig. 3c, left panel, Supplementary Fig. 3c). After at least six weeks on \nNCD, these same animals displayed an increase in lick vigor during cue presentation (Fig. 3c, \nbottom right panel; Supplementary Fig. 3b), indicating a reinstatement of motivation to consume \nfood. PVHMC4R neuron responses remained attenuated relative to prior responses while on HFD \nfor most of the experiment  (Fig. 3f, h), but they did not show a noticeable increase in response \nmagnitude across the meal, similar to the HFD -to-HFD group (Fig. 3i, Supplementary Fig. 3d). \nThe HFD-to-HFD group maintained high early-meal PVHMC4R responses (Fig. 3d, g, h) and ceased \nto engage in Ensure consumption even earlier in the session, never exceeding 60 trials. Licking \nvigor steadily declined  across trials, reflecting a persistent drop in motivation to consume food \n(Supplementary Fig. 3c), as observed during their initial HFD exposure. \nFurther analysis revealed that only the NCD-to-NCD group exhibited a significant increase \nin PVHMC4R responses from early to late meal stages  (Fig. 3i). Interestingly, the HFD-to-NCD \ngroup displayed an intermediate pattern between the HFD-to-HFD and NCD-to-NCD, where there \nwas a slight, but nonsignificant increase in PVH MC4R responses across the meal  (Fig. 3 i; \nSupplementary Fig. 3d). When evaluating the PVHMC4R response elevation across the meal (Fig. \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\n3j), we confirmed a significantly smaller increase in response across trial s in the HFD -to-HFD \ngroup (Fig. 3j). Notably, six weeks of NCD exposure in formerly HFD -fed animals failed to fully \nrestore this range to NCD levels (ΔZ -score: NCD-to-NCD = 0.37 ± 0.03, HFD -to-NCD = 0.14 ± \n0.05, HFD-to-HFD = -0.05 ± 0.07)  (Fig. 3j). Despite this, the HFD -to-NCD group consumed a \nvolume of Ensure comparable to the NCD-to-NCD group and significantly higher than the HFD -\nto-HFD group during individual recording sessions (Fig. 3k). \nIn summary, switching HFD -fed obese animals to NCD partially restored both neural \nactivity (Fig. 3b, h, f) and feeding behaviors (Supplementary Fig. 3a). Specifically, in the HFD-to-\nNCD group, PVHMC4R responses were closer to expected levels in the early meal stage (Fig. 3h) \nbut did not show the same increase in response magnitude over the meal  as the NCD-to-NCD \ngroup (Fig. 3i). This suggests a partial recovery of these neurons' ability to integrate satiety signals \n(Fig. 3h) . Moreover, the motivation to consume the Ensure, which had been significantly \ndiminished in obese animals, was reinstated following the diet switch. Licking and food  intake \nimproved (Fig. 3k, Supplementary Fig. 3b), supporting the idea that a lean diet can rescue some \naspects of both behavioral and neural dysfunction induced by obesity. These findings underscore \nthe dynamic nature of the MC4R system in regulating feeding and energy balance. While diet -\ninduced obesity impairs PVH MC4R neuron function, transitioning to a healthier diet only partially \nrestores the neurons' ability to integrate hunger and satiety signals.  \n  \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nDiscussion \nThe present study sheds light on the role of PVHMC4R neurons in integrating satiety signals \nduring food consumption in lean and diet -induced obese mice, highlighting the alterations in \nneuronal activity and feeding behavior that occur with prolonged exposure to a high-fat diet (HFD). \nOur findings also confirm that in lean NCD -fed animals, PVH MC4R neurons exhibit increase d \nresponses during feeding that are closely linked to meal progression and satiety5. Conversely, in \nHFD-fed animals, th e response dynamics were impaired - PVHMC4R neurons exhibit increased \nresponse magnitude early in a meal, consistent with the mice beginning the meal in a more sated \nstate. This elevated response magnitude then stayed roughly constant throughout the meal (in \ncontrast to the progressive increase in response magnitude in lean mice). This suggests a failure \nto properly integrate satiety signals as the meal progresses , or that a ceiling effect prevent s a \nfurther rise in neural responses. \nHow do our findings in PVHMC4R neurons relate to prior studies of DIO in upstream neural \ncircuits involved in feeding behavior? Prior studies have shown that DIO mice exhibit increased \nspontaneous firing in AgRP/NPY neurons due to altered intrinsic excitability40,41. Moreover, these \nneurons are insensitive to leptin41,50,51. In contrast, POMC neurons exhibit a decrease in \nspontaneous activity due to a hyperpolarized membrane potential 42. Moreover, long-term HFD \nexposure induces a loss of MC4R protein abundance in the PVH MC4R neurons, even though the \nnumber of cells remains the same48,49. This loss is accompanied by diminished α-MSH expression \nin the arcuate nucleus, further supporting the notion that dietary fat exposure alters α-MSH-MC4R \nsignaling48. The associated eleva tion in inhibitory NPY signaling (via Gi-coupled NPY1Rs) and \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nreduction in excitatory α-MSH signaling (via Gs-coupled MC4Rs) would thus be expected to result \nin a decrease in cAMP in PVHMC4R neurons at baseline (when the mouse is not consuming food). \nWhen we suppressed elevations in cAMP in PVHMC4R neurons in a previous study using selective \nexpression of a modified phosphodiesterase (PDE), mice developed massive obesity \naccompanied by  a lower frequency of spontaneous activity and spontaneous postsynaptic \npotentials in PVHMC4R neurons in vitro5. Thus, we initially predicted that PVHMC4R neurons in HFD \nmice would be silent at baseline and more weakly activated during consumption, as seen in PDE-\noverexpression animals, but instead we observed the opposite effects in our experiments, where \nend-of-meal responses were similar to those in NCD mice. \nWe also considered other prior studies using in vivo fiber photometry recordings of calcium \nactivity in DIO, which paint a more complex picture25,43. Acute delivery of various foods established \nthat obesity attenuates the rapid sensory inhibition of AgRP neurons during food \nconsumption25,38,44,45. Additionally, DIO blunts the responses  of AgRP neurons  to a variety of \nhormonal inputs, such as ghrelin, CCK, leptin , and insulin25,40,41,46,47. These disruptions likely \nprevent changes in cAMP in PVHMC4R neurons in response to fast (consumption-related) and to \nmore persistent (satiety-related) drops in AgRP neuron activity25,37,45,48. While these findings may \nhelp explain the lack of a gradual elevation in response magnitude over the meal in our HFD mice, \nthey do not explain why response magnitude is higher in HFD mice at the start of the meal.  \nIn lean animals, the gradual increase in PVH MC4R neuron activity was accompanied by a \nreduction in licking vigor, a behavior closely linked to the development of satiety. Th e observed \nincrease in consumption-evoked neural responses across trials was faster than in our previous \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nstudy5, consistent with the fact that the animals satiated faster in the current study. This difference \ncould be due to batch differences in body  weight. In contrast , in HFD -fed animals, PVH MC4R \nneurons showed unexpectedly strong responses early in the meal and remained similarly \nresponsive throughout the meal . Given the higher body weight and metabolic demand  in HFD \nanimals, we would have expected them to engage in longer feeding bouts. Surprisingly, despite \nbeing food-restricted and calorie-matched to NCD-fed mice, HFD-fed mice displayed premature \nsatiation. Further, the degree of this premature satiation was predictable from neural response  \nmagnitudes early in the meal. \nInterestingly, the results from the diet -switching experiment provide insight into the \npotential reversibility of these impairments , as previous studies show recovery of expression of \nMC4R and α-MSH after returning to a low-fat diet for four weeks48. While the HFD-to-NCD animals \nin our study exhibited some recovery in motivation to consume food, as evidenced by an increase \nin licking vigor and larger meal size, the PVHMC4R neuronal responses did not fully return to the \nlevels observed in the NCD-to-NCD group. Notably, some drawbacks must be considered when \ninterpreting the results of the diet-switching experiments in Figure 3. The prolonged expression \nof the calcium indicator and potential overtraining of the animals to perform the task may have \ncontributed to the slightly less pronounced PVH MC4R neuron response potentiation observed \nduring the second recording after the diet switch to NCD. This is reflected in the lower ΔZ-scores \ncompared to those in Fig. 2c. \nAdditionally, HFD -fed mice devalue NCD as a food resource, consistent with blunted \nAgRP neuron sensitivity to NCD and reduced intake  of NCD25,38. A two -week withdrawal from \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nlong-term HFD exposure is insufficient to restore the AgRP neuron responses to sensory food \ndetection and consumption38. In our study, although we returned mice to NCD for over  6 weeks \nbefore recordings, only a partial recovery of responses was observed. These findings suggest \nthat the neuronal dysfunction in HFD -fed animals may not be entirely reversible with short -term \ndiet intervention, indicating a potential long-term impact of HFD exposure on the plasticity of the \nPVHMC4R neurons.  \nAlthough DIO increases the spontaneous firing of AgRP neurons39,51 and inhibits POMC \nneurons42, and AgRP neurons become less sensitive to leptin 39, administration of a leptin \nantagonist in DIO mice promotes additional food intake52, Hence, our finding of elevated food \nconsumption response early in the meal may be due to the persistently high level of leptin in DIO \nmice, both before and after the meal. In both cases, leptin is likely to saturate its receptor, leading \nto consistently elevated excitability of POMC neurons and reduced excitability of AgRP neurons. \nUnder these conditions, cAMP levels in PVHMC4R neurons may not return to a low level in between \nmeals, potentially preventing a downward resetting of the strength of consumptio n-evoked \nsynaptic inputs to PVHMC4R neurons. Concurrently, we observed that calorie-restricted HFD mice, \ndespite being calorie-matched to the NCD group and theoretically having the same acute caloric \ndeficit while exhibiting increased adiposity and leptin levels , had small meal sizes during the \nrecordings, as expected based on their higher early-session PVHMC4R neuron responses. \nThis hypothesis is consistent with the counterintuitive finding from Sherrer and colleagues \nshowing that clamping leptin at a lower level in DIO mice actually reduces body weight, despite \nthe anorexigenic actions of leptin in healthy mice 53. This may allow consumption -evoked \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nresponses to drop to a lower magnitude early in the meal, and the associated restoration in leptin \nsensitivity may also be involved in restoring the normal increase in response magnitude \nthroughout the meal.  \nOur findings, together with the aforementioned effects of leptin antagonist administration \nin DIO mice, suggest that the mechanisms responsible for suppressing homeostatic feeding, such \nas those mediated by leptin , are at least partially intact in HFD mice and are engaged from the \nbeginning of the meal. This indicates that the physiological \"brakes\" that typically signal satiety \nand reduce the drive to eat are functional to some extent. In addition, the persistence of feeding \ndespite these satiety-promoting signals could also  imply that a second, parallel motivational \nsystem may be at play. Specifically, circuits involved in processing the hedonic or rewarding \nproperties of high-fat diet (HFD) food might be overriding or competing with homeostatic signals54. \nThese reward-related pathways could be contributing to the decision to initiate and sustain eating, \nindependent of true caloric need, particularly in the context of energy-dense, palatable food.  \nFuture experiments, such as  assessment of peptide transmission from inputs onto  \nPVHMC4R neurons in DIO, will be necessary to further investigate the mechanisms underlying \nthese changes and to better guide therapeutic interventions. \n  \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nMethods \nAnimal care \nAll animal procedures were approved and completed in compliance with the Institutional \nAnimal Care and Use Committee at Beth Israel Deaconess Medical Center (BIDMC). All mice \nwere single-housed in Innocage® plastic ventilated cages (Innovive) and kept in a room on a 12h \nlight/12h dark cycle. Room temperature (18-22°C) and humidity were controlled within the rodent \nhousing room. From five weeks of age to approximately 12 weeks of age mice were kept on an \nad libitum diet of either standard mouse chow diet (LabDiet®, Formulab Diet Irradiated, 5008) or \nblue dye high-fat diet (Research Diets, D12492i, Rodent Diet ) with 60 % of total kcal from f at, \n20% from carbohydrates, 20%from protein. Both male and female MC4R-Cretg/wt (B6.129S4-\nMc4rtm1Lowl/J) mice were included in the experiments . The n umber of mice used for each \nexperiment is shown in each Figure. Experiments were not conducted in an experimenter-blinded \nmodel, but animals were randomly assigned to either a lean diet or a high-fat diet group. \n \nChronic food restriction \nAfter recovering from stereotaxic surgery, the mice were chronically food restricted either \non normal chow pellets (Bio-Serv, Product # F0171, Dustless Precision Pellets, 500 mg, Rodent \nGrain-Based Diet, Lot # 295488.00) or blue dye high -fat diet for the remainder of the first round \nof photometry recordings  (Figure 1 – 2). For the second recording phase (Figure 3) , we food-\nrestricted the animals on the same or on a different diet. Mouse body weight and food intake were \nrecorded to keep both groups of mice at ~80% of their original body weight (Supplementary Fig. \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\n1c, d). Some recordings (Supplementary Fig. 2c – e) involved acute overnight fasting of the food-\nrestricted animals the day before (16h), to increase motivation to eat. After the end of a recording \nthat was preceded by acute fasting, the animal was refed with about twice the amount of food  \ngiven during the standard food-restriction protocol to maintain body weight at ~80% of the original \nbody weight. \n  \nStereotaxic surgery \nWe performed stereotaxic viral injections and headpost implantation following a protocol \nthat was described previously5. Briefly, AAV1-Syn-FLEX-GCaMP6s (Addgene, 100845 -AAV1) \nwas injected unilaterally into the PVH of MC4R-Cretg/wt mice (150 nL, Bregma: AP -0.6 mm, ML -\n0.2 mm, DV -4.75 mm) at a titer of 1.05 x 10 13 gc/ml. Single optic fibers with a metal ferrule \n(MFP_400/430/LWMJ-0.57_1m_FC-ZF1.25 (F)_LAF ; Doric Fibers) were implanted above the \nPVH (Bregma: AP -0.6 mm, ML -0.25 mm, DV -4.6 mm) in each mouse. C&B Metabond (Parkell) \nwas used to cement the titanium metal headposts onto the cranium following surgery. Mice \nrecovered for 2 weeks post-surgery before any other intervention. Viral expression and accurate \nfiber positions were verified using the post-hoc histology.  \n  \nBehavioral training on wheels \nTwo weeks after surgery, animals started a 3 -4 week  protocol of habituation followed by \nbehavioral training. Head-fixed mice were placed on a three-dimensional (3D)-printed standalone \ncircular treadmill for three days at increasing intervals of 20, 30, and 45 minutes, respectively. On \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nDays 4 through 8 (D4-8), animals were introduced to Ensure (Ensure® Plus nutrition shake, 1.47 \nkcal/mL) via manual administration through a syringe three times during the one -hour training \nsession. Starting at D9, mice were head-fixed and placed on the circular treadmill with access to \na lickspout for 25 minutes. Ensure was manually dispensed from the lickspout, and the mice were \nconsidered acclimatized when they licked regularly from this lickspout. Training then continued \non the circular treadmill for another 30 minutes. Mice were further acclimated to the recording set-\nup by letting them freely run on the treadmill and self -administer Ensure from the lickspout. \nAnimals then underwent classical conditioning to associate an audible cue with the delivery of \nEnsure, followed by operant conditioning in which mice would lick  to trigger the Ensure delivery \nvia a solenoid pump after an audible cue. Milkshake delivery speed was controlled by gravity and \nregulated by a solenoid pulse (5 x 150 ms pulses, 150 ms between the pulses, 3 μL per pulse, \n~25 μL total per trial)  during all training sessions . For unconditional training, usually, 3  – 4 45-\nminute-long sessions with 75 Ensure deliveries after a cue (released at random intervals between \n45 s and 70 s ) were carried out per animal. For operant conditioning training, at least 3 training \nsessions were carried out to ensure that animals were robustly triggering the Ensure release after \neach audible cue.  Operant conditioning training sessions were initially short and were then \ngradually extended across days as  the behavior became more reliable.  If animals (especially \nHFD-fed mice) were not progressing well through initial learning of this task, they underwent \nadditional classical conditioning training sessions, and then restarted the operant conditioning.  \n \nHead-fixation and food delivery \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nMice ran on a circular treadmill with running speed measured by an IR beam while their head \nwere fixed. Head-fixation reduces motion artifacts while recording neural activity. In all recordings, \nEnsure was delivered through a lickspout, with the following protocol : 10 x 150 ms pulses, 150 \nms between the pulses, 3 μL per pulse, ~50 μL total per trial. Milkshake delivery during recording \nwas controlled using the Nanosec (https://github.com/xzhang03/NidaqGUI) photometry-\nbehavioral system. The total consumed Ensure throughout the experiment was noted.  \n \nHead-fixed photometry \nHead-fixed photometry experiments were conducted as described previously 5,23. For GCaMP6s \nrecordings, excitation light from 465 nm (PlexBright LED 465nm, Plexon) and 405 nm (PlexBright \nLED 405nm, Plexon) LEDs were modulated by LED drivers and combined in a three -port mini-\ncube (FMC4_IE(400-410)_E(460-490)_F(500-550)_S, Doric) to transmit to the implanted optical \nfiber via a patch cord (NA 0.57, MFP_400/430/LWMJ-0.57_1m_FC-ZF1.25(F)_LAF, Doric). The \nemitted light was collected in a femtowatt photoreceiver (2151, Newport).  Photometry excitation \nlight was controlled by the Nanosec photometry-behavioral system5,23. Briefly, excitation light was \nmodulated as interleaved pulses (465 LED ON/405nm LED OFF for 6 ms; 465 LED OFF/405nm \nLED ON for 14 ms) . The respective datapoints of  the 6ms pulse were extracted to quantify the \nphotometry signal. The recording was conducted in the dark. Three to five recordings per mouse \nwere performed in two conditions: 1) food  restriction (FR) and 2) food restriction plus acute \novernight fasting of FR mice prior to the experimental recording. Acute overnight fasting  (16 h) \nwas implemented to increase the hunger and motivation to work for food, particularly in the HFD \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\ngroup. Each recording session consisted of a 5-minute baseline period, followed by structured \ntrials (60 s inter-trial interval) for 60 minutes (Fig. 1, 60 trials) or 90 minutes (Fig. 2, 90 trials). An \naudible cue (tone, 2 kHz) was presented on each trial, at which point the mouse had 1 second to \nlick to trigger the lickspout for Ensure delivery via the lickspout.  If no  lick trigger (no tongue \npressure exerted on the spout) occurred during the cue window , the mouse would not get any \nEnsure and must wait till the next audible cue (1 min later) to try again. An IR beam was used to \nmeasure the speed of free locomotion on the circular treadmill. \n \nHistology \nMice were terminally anesthetized using tribromoethanol ( i.p. 250 mg/kg ) and tran scardially \nperfused with PBS and then formalin (10%). The tissue was fixed in 20% sucrose in PBS before \nextraction and microtome slicing  (40 µm). Slides were stained with antifade mounting medium \nwith DAPI (VECTASHIELD® HardSet ™) before capturing the images with an Olympus VS120 \nslide scanner microscope. \n \nAnalysis of head-fixed feeding assays \nData analysis was performed as described previously using custom software written in \nMATLAB (MathWorks)  and Python (https://github.com/xzhang03/Photometry_analysis)5,23. \nPhotometry signals in the photodetector trace were identified from LED pulses associated with \nillumination of the 465 nm LED and the median of the respective datapoints were extracted. The \nresulting trace (50 Hz sampling) was further filtered with a 10 Hz low-pass filter. The data were \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nsplit into 60-s long traces beginning 10 s prior to the cue. We calculated baseline fluorescence \n(mean of 10 s pre-cue, denoted Fo ) and fractional changes in fluorescence from the baseline \n(ΔF/Fo). Given that animals could voluntarily  trigger the Ensure release  repeatedly, with the \nnumber of total triggered trial s depending on their hunger state and motivation , we considered \nonly those recordings where at least 30 trials were triggered, unless otherwise stated, and for the \nanalysis, we only selected the triggered trials for each session. The recordings from each animal \nfrom different days were concatenated to determine a single Z-score per animal, which was then \nused to determine the Z-score across the recordings. \n Heatmaps were calculated by averaging across all usable trials across sessions from each \nmouse, and then a mean heatmap was computed across mice. Responses to Ensure during food \nconsumption were estimated as the mean response during the peri-licking phase, 1-10 s post \ntone. For some analyses, as indicated in the Figure Legends, mean responses from 1-20 s post \ntone were considered. For the early meal analyses, we calculated the mean of the first 15 \ntriggered trials, whereas the late meal was defined using the last 15 triggered trials. Lick rate \nduring the cue was defined as the number of licks per second during the tone presentation. The \nlick rate during the reward window was calculated as the mean lick rate from 1-10s post tone. ΔZ-\nscore (Fig. 2c, d, 3j) was calculated as a difference between the mean response during the peri-\nlicking phase (1 – 10 s post tone) of the last 5 triggered trials and the first 5 triggered trials.  \n \nStatistical analysis \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nStatistical analyses were performed in Python using the scipy.stats package. The numbers of \nrecordings, animals , and statistics used are indicated in the figure legends. In summary , the \nstatistical significance was determined with 1) two -way ANOVA and multiple comparisons \nperformed using the Tukey HSD post hoc test in Fig 3h, k,, Supplementary Fig. 1h; 2) paired, two-\ntailed t-test between early and late meal trials Fig 1j Supplementary Fig. 1j, 3d; 3) unpaired, two-\ntailed Student’s t -test in Fig. 2b  – d; Supplementary Fig. 1e, f; 4) one-way ANOVA with Tukey \nHSD post hoc test in Figure 3j; 5) linear mixed effects (LME) models (mixedlm of statsmodels \nPython module) to account for dependencies originating from repeated  recording sessions of \nindividual animals in Fig. 1i, 3i. \nWe fit a linear mixed -effects model to analyze the effects of diet and meal phase on the \nmeasured values while accounting for repeated measures (i.e., repeated sessions) within \nindividual mice. The model included diet as a categorical fixed effect with \"NCD\" as the reference \nlevel and meal phase as an additional fixed effect. An interaction term between diet and meal \nphase was also included to assess whether the relationship between these factors influenced the \noutcome variable. To account for variability a cross individual mice, we specified \"Mouse\" as a \nrandom effect. The model was implemented using the formula:  Value∼Diet×Meal \nphase+(1∣Mouse), where \"Value\" represents the dependent variable, \"Diet\" (NCD or HFD) and \n\"Meal phase\" (Early meal or Late meal) are fixed effects, and \"Mouse\" is the grouping factor for \nrandom intercepts. This approach allowed us to assess both the main effects and their interaction \nwhile controlling for inter -mouse variability. We used this model to analyze data presented in \nFigure 1i. \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nFor diet-switched experiments, we fit a linear mixed -effects model to analyze the effects \nof Condition (a combination of start diet and second diet) and Meal Type (Early meal vs. Late \nmeal) on the measured values while accounting for repeated measures within individual mice. \nThe model included Condition as a categorical fixed effect with ‘ NCD to NCD’ as the reference \nlevel, and Meal phase as an additional fixed effect. An interaction term between Condition and \nMeal phase was also included to assess whether the relationship between these factors \ninfluenced the outcome variable. To account for variability across individual mice, we specified \n\"Mouse\" as a random effect . The model was implemented using the formula:  \nValue∼Condition×Meal phase +(1∣Mouse). Where \"Value\" represents the dependent variable,  \n\"Condition\" (NCD -> NCD, HFD -> NCD, or HFD -> HFD) and \"Meal phase\" (Early meal or Late \nmeal) are fixed effects, and  \"Mouse\" serves as the grouping factor for random intercepts.  This \napproach allowed us to assess both the main effects of  Condition and Meal phase, as well as \ntheir interaction, while controlling for inter -mouse variability. The results from this model are \npresented in  Figure 3i. Data representation with a scatter dot plot graph includes values of \nindividual data points with mean values as a bar and standard error of the mean (s.e.m.) as error \nbars. Line plot graphs indicate the mean and show s.e.m. as a shaded region. \nSignificance was measured against an alpha value of 0.05 unless otherwise stated. * p < 0.05, \n**p < 0.01, ***p< 0.001, ****p<0.0001. \n  \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nReferences \n1. Andermann, M. L. & Lowell, B. B. Toward a Wiring Diagram Understanding of Appetite \nControl. Neuron 95, 757–778 (2017). \n2. Garfield, A. S. et al. A neural basis for melanocortin-4 receptor–regulated appetite. Nat. \nNeurosci. 18, 863–871 (2015). \n3. Sweeney, P., Gimenez, L. E., Hernandez, C. C. & Cone, R. D. Targeting the central \nmelanocortin system for the treatment of metabolic disorders. Nat. Rev. Endocrinol. \n19, 507–519 (2023). \n4. Balthasar, N. et al. Divergence of melanocortin pathways in the control of food intake \nand energy expenditure. Cell 123, 493–505 (2005). \n5. Zhang, S. X. et al. Stochastic neuropeptide signals compete to calibrate the rate of \nsatiation. Nature 637, 137–144 (2025). \n6. De Solis, A. J. et al. Reciprocal activity of AgRP and POMC neurons governs coordinated \ncontrol of feeding and metabolism. Nat. Metab. 6, 473–493 (2024). \n7. Krashes, M. J. et al. Rapid, reversible activation of AgRP  neurons drives feeding \nbehavior in mice. J. Clin. Invest. 121, 1424–1428 (2011). \n8. van den Pol, A. N. Neuropeptide Transmission in Brain Circuits. Neuron 76, 98–115 \n(2012). \n9. Steculorum, S. M. et al. AgRP Neurons Control Systemic Insulin Sensitivity via \nMyostatin Expression in Brown Adipose Tissue. Cell 165, 125–138 (2016). \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\n10. Engström Ruud, L., Pereira, M. M. A., de Solis, A. J., Fenselau, H. & Brüning, J. C. NPY \nmediates the rapid feeding and glucose metabolism regulatory functions of AgRP \nneurons. Nat. Commun. 11, 442 (2020). \n11. Tong, Q., Ye, C.-P., Jones, J. E., Elmquist, J. K. & Lowell, B. B. Synaptic release of GABA \nby AgRP neurons is required for normal regulation of energy balance. Nat. Neurosci. 11, \n998–1000 (2008). \n12. Wu, Q., Boyle, M. P. & Palmiter, R. D. Loss of GABAergic signaling by AgRP neurons to \nthe parabrachial nucleus leads to starvation. Cell 137, 1225–1234 (2009). \n13. Gropp, E. et al. Agouti-related peptide–expressing neurons are mandatory for feeding. \nNat. Neurosci. 8, 1289–1291 (2005). \n14. Luquet, S., Perez, F. A., Hnasko, T. S. & Palmiter, R. D. NPY/AgRP neurons are essential \nfor feeding in adult mice but can be ablated in neonates. Science 310, 683–685 (2005). \n15. Mountjoy, K. G., Robbins, L. S., Mortrud, M. T. & Cone, R. D. The Cloning of a Family of \nGenes That Encode the Melanocortin Receptors. Science 257, 1248–1251 (1992). \n16. Brüning, J. C. & Fenselau, H. Integrative neurocircuits that control metabolism and \nfood intake. Science 381, eabl7398 (2023). \n17. Friedman, J. M. Leptin and the endocrine control of energy balance. Nat. Metab. 1, 754–\n764 (2019). \n18. Ruud, J., Steculorum, S. M. & Brüning, J. C. Neuronal control of peripheral insulin \nsensitivity and glucose metabolism. Nat. Commun. 8, 15259 (2017). \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\n19. Cowley, M. A. et al. The Distribution and Mechanism of Action of Ghrelin in the CNS \nDemonstrates a Novel Hypothalamic Circuit Regulating Energy Homeostasis. Neuron \n37, 649–661 (2003). \n20. D’Agostino, G. & Diano, S. alpha-Melanocyte stimulating hormone: production and \ndegradation. J. Mol. Med. 88, 1195–1201 (2010). \n21. Xu, S. et al. Behavioral state coding by molecularly defined paraventricular \nhypothalamic cell type ensembles. Science 370, (2020). \n22. Molden, B. M., Cooney, K. A., West, K., Van Der Ploeg, L. H. T. & Baldini, G. Temporal \ncAMP Signaling Selectivity by Natural and Synthetic MC4R Agonists. Mol. Endocrinol. \n29, 1619–1633 (2015). \n23. Zhang, G.-W. et al. Medial Preoptic Area Antagonistically Mediates Stress-induced \nAnxiety and Parental Behavior. Nat. Neurosci. 24, 516–528 (2021). \n24. Jais, A. & Brüning, J. C. Arcuate Nucleus-Dependent Regulation of Metabolism-\nPathways to Obesity and Diabetes Mellitus. Endocr. Rev. 43, 314–328 (2022). \n25. Beutler, L. R. et al. Obesity causes selective and long-lasting desensitization of AgRP \nneurons to dietary fat. eLife 9,. \n26. Loos, R. J. F. & Yeo, G. S. H. The genetics of obesity: from discovery to biology. Nat. Rev. \nGenet. 23, 120–133 (2022). \n27. Huszar, D. et al. Targeted disruption of the melanocortin-4 receptor results in obesity in \nmice. Cell 88, 131–141 (1997). \n28. Krude, H. et al. Severe early-onset obesity, adrenal insufficiency and red hair \npigmentation caused by POMC mutations in humans. Nat. Genet. 19, 155–157 (1998). \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\n29. Yeo, G. S. H. et al. A frameshift mutation in MC4R associated with dominantly inherited \nhuman obesity. Nat. Genet. 20, 111–112 (1998). \n30. Vaisse, C., Clement, K., Guy-Grand, B. & Froguel, P. A frameshift mutation in human \nMC4R is associated with a dominant form of obesity. Nat. Genet. 20, 113–114 (1998). \n31. Jackson, R. S. et al. Obesity and impaired prohormone processing associated with \nmutations in the human prohormone convertase 1 gene. Nat. Genet. 16, 303–306 \n(1997). \n32. Montague, C. T. et al. Congenital leptin deficiency is associated with severe early-onset \nobesity in humans. Nature 387, 903–908 (1997). \n33. Kühnen, P. et al. Proopiomelanocortin Deficiency Treated with a Melanocortin-4 \nReceptor Agonist. N. Engl. J. Med. 375, 240–246 (2016). \n34. Ryan, D. H. Setmelanotide: what does it mean for clinical care of patients with obesity? \nLancet Diabetes Endocrinol. 8, 933–935 (2020). \n35. Farooqi, I. S. & O’Rahilly, S. Mutations in ligands and receptors of the leptin–\nmelanocortin pathway that lead to obesity. Nat. Clin. Pract. Endocrinol. Metab. 4, 569–\n577 (2008). \n36. Kievit, P. et al. Chronic Treatment With a Melanocortin-4 Receptor Agonist Causes \nWeight Loss, Reduces Insulin Resistance, and Improves Cardiovascular Function in \nDiet-Induced Obese Rhesus Macaques. Diabetes 62, 490–497 (2013). \n37. Li, H. et al. The melanocortin action is biased toward protection from weight loss in \nmice. Nat. Commun. 14, 2200 (2023). \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\n38. Mazzone, C. M. et al. High-fat food biases hypothalamic and mesolimbic expression of \nconsummatory drives. Nat. Neurosci. 23, 1253–1266 (2020). \n39. Enriori, P. J. et al. Diet-Induced Obesity Causes Severe but Reversible Leptin \nResistance in Arcuate Melanocortin Neurons. Cell Metab. 5, 181–194 (2007). \n40. Briggs, D. I., Enriori, P. J., Lemus, M. B., Cowley, M. A. & Andrews, Z. B. Diet-Induced \nObesity Causes Ghrelin Resistance in Arcuate NPY/AgRP Neurons. Endocrinology 151, \n4745–4755 (2010). \n41. Briggs, D. I. et al. Evidence that diet-induced hyperleptinemia, but not hypothalamic \ngliosis, causes ghrelin resistance in NPY/AgRP neurons of male mice. Endocrinology \n155, 2411–2422 (2014). \n42. Paeger, L. et al. Energy imbalance alters Ca2+ handling and excitability of POMC \nneurons. eLife 6, e25641 (2017). \n43. Beutler, L. R. et al. Dynamics of Gut-Brain Communication Underlying Hunger. Neuron \n96, 461-475.e5 (2017). \n44. Chen, Y., Lin, Y.-C., Kuo, T.-W. & Knight, Z. A. Sensory Detection of Food Rapidly \nModulates Arcuate Feeding Circuits. Cell 160, 829–841 (2015). \n45. Chen, Y. et al. Sustained NPY signaling enables AgRP neurons to drive feeding. eLife 8, \ne46348 (2019). \n46. Dodd, G. T. et al. Intranasal Targeting of Hypothalamic PTP1B and TCPTP Reinstates \nLeptin and Insulin Sensitivity and Promotes Weight Loss in Obesity. Cell Rep. 28, 2905-\n2922.e5 (2019). \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\n47. Dodd, G. T. et al. Insulin signaling in AgRP neurons regulates meal size to limit glucose \nexcursions and insulin resistance. Sci. Adv. 7, eabf4100 (2021). \n48. Nyamugenda, E. et al. Selective Survival of Sim1/MC4R Neurons in Diet-Induced \nObesity. iScience 23, 101114 (2020). \n49. Cooney, K. A., Molden, B. M., Kowalczyk, N. S., Russell, S. & Baldini, G. Lipid stress \ninhibits endocytosis of melanocortin-4 receptor from modified clathrin-enriched sites \nand impairs receptor desensitization. J. Biol. Chem. 292, 17731–17745 (2017). \n50. Briggs, D. I. et al. Calorie-Restricted Weight Loss Reverses High-Fat Diet-Induced \nGhrelin Resistance, Which Contributes to Rebound Weight Gain in a Ghrelin-\nDependent Manner. Endocrinology 154, 709–717 (2013). \n51. Baver, S. B. et al. Leptin Modulates the Intrinsic Excitability of AgRP/NPY Neurons in the \nArcuate Nucleus of the Hypothalamus. J. Neurosci. 34, 5486–5496 (2014). \n52. Ottaway, N. et al. Diet-induced obese mice retain endogenous leptin action. Cell \nMetab. 21, 877–882 (2015). \n53. Zhao, S. et al. Partial Leptin Reduction as an Insulin Sensitization and Weight Loss \nStrategy. Cell Metab. 30, 706-719.e6 (2019). \n54. Berthoud, H.-R., Münzberg, H. & Morrison, C. D. Blaming the Brain for Obesity: \nIntegration of Hedonic and Homeostatic Mechanisms. Gastroenterology 152, 1728–\n1738 (2017). \n \n  \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nAcknowledgments \nThis research was supported by a Walter Benjamin Fellowship from Deutsches \nForschungsgemeinschaft PO 2925/1 -1 (M.P), a Lefler Fellowship, a Charles A. King Trust \nFellowship and NIH K99 DK134853 (S.X.Z.); R01 DK109930, DP1 AT010971, DP1 DK139958, \na McKnight Scholar Award, and grants from the Boston Nutrition and Obesity Research Center \n(P30 DK046200), the Klarman Family Foundation, the Pew Innovation Fund and a Charles Robert \nBroderick III Phytocannabinoid Research Grant (M.L.A.). We thank all members of the Lehtinen \nand Andermann labs for fruitful discussions. We thank M. F. Hammell, J. Edelhaus,  and L. \nMeschisen, for help ing with animal care and histology, P. S. Sunkavalli, Z. A. Stolberg for \nbehavioral experiments, H. Kucukdereli for technical support, P. Soden, P. Kalugin for further lab \nsupport, J. DeBolt, J. Shapiro, A. Pinilla, M. K. Lehtinen for administrative support. \n \nAuthor contributions \nMP and SXZ designed the experiments  with help from MLA . MP , SXZ , and MLA wrote the \nmanuscript. MP conducted all surgeries. MP, JB, and CA performed the behavioral training and \nfiber photometry recordings. JB and CA performed all histology experiments. MP performed data \nanalysis with guidance from SXZ and MLA. \n \nCompeting interests statement \nThe authors declare no competing interests. \n  \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nFigure 1. PVHMC4R neuronal responses are elevated early in the meal in HFD mice \na, Schematic timeline of the experimental paradigm. \nb, Representative images showing viral expression of AAV-Syn-Flex-GCaMP6s and optical fiber \nplacement in the PVH of MC4RCre/wt mice. Scale bar: 200 µm. \nc, Post-surgical bodyweight dynamics, showing initial ad libitum feeding  and the onset of food \nrestriction for each group (normal chow diet, food restricted: NCD FR; high fat diet, food restricted: \nHFD FR). NCD: 6 mice; HFD: 5 mice. \nd, Cumulative calorie consumption across both home cage food and Ensure during experimental \nrecordings for NCD- and HFD-fed mice during food restriction. NCD: 6 mice; HFD: 5 mice.  \ne, f, Mean heatmaps summarizing GCaMP6s photometry signals (top panel) and licking (bottom \npanel) from PVHMC4R neurons of mice in NCD-fed (e) and HFD-fed (f) mice. Mean running speed \nand licking rates across all trials are shown above the heatmaps. Only recordings with a minimum \nof 30 successfully triggered trials were included. T rial structure: 10 s baseline followed by tone \ncue onset (t = 0 s) and Ensure delivery (t = 1 s) conditional on licking in the 1-s interval after tone \nonset. Early (first 15 trials) and late (last 15 trials) meal phases (quantified in i, j) are marked with \ngreen and magenta square insets, respectively. NCD: 39 recordings / 6 mice; HFD: 33 recordings \n/ 5 mice.  \nEarly meal Late meal\nHFD NCD\nNCD HFD\nEnsure response\nby mice\nEnsure response\nby session\nd\nf hg\ni j\nAd libitum access to:\nNormal Chow Diet (NCD)\n or High Fat Diet (HFD)\nFood restriction (FR) on:\nNormal Chow Diet (NCD)\nor High Fat Diet (HFD)\n~5 w 16 - 17w\na b\nDAPI GCaMP6s\ne\nAAV-GCaMP6s injection + optic\nfiber implant in the PVH\n10 - 12 w\nHabituation and \nconditional training \nFiber photometry\nrecording\nc\nFIGURE 1\n*\n**** *\nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\ng, h, Mean heatmaps depicting trial-averaged PVHMC4R neuron activity during early (g) and late \n(h) meal phases for individual NCD-fed (top panel) and HFD-fed (lower panel) mice by combining \nthe sessions for each mouse. NCD: 6 mice, HFD: 5 mice. \ni. Z-scored PVHMC4R neuronal responses during early and late meal phases for each recording in \nNCD-fed and HFD-fed mice. NCD: 39 recordings; HFD: 33 recordings. Linear mixed model (LME), \np(NCD early vs. HFD early) < 0.001, p(Diet x Meal phase) < 0.001. \nj, Z-scored PVHMC4R neuronal responses of each animal during early and late meal phases for all \nrecordings in NCD -fed and HFD -fed mice. NCD : 6 mice, HFD: 5 mice. Paired, two-tailed test \nbetween early and late meal phases. \nc – f, i – j, Data are represented as the mean ± s.e.m. * - P <0.05, ** - P <0.01, **** - P <0.0001. \n  \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\n \nFigure 2. Mid-meal PVHMC4R responses predict meal size in HFD mice \na, Single trial mean peri-licking PVHMC4R neuron responses (1 – 9 s post cue) across 90 recorded \ntrials from NCD-fed and HFD-fed groups (minimum 30 successfully Ensure triggered trials). \nb, Number of triggered trials during a 90-trial session for NCD-fed and HFD-fed groups.  \nc, Difference in late-trial vs. early-trial neuronal responses (ΔZ-score) per recording for NCD-fed \nand HFD-fed groups. \nd, ΔZ-score averaged across sessions from individual NCD-fed and HFD-fed animals. \ne, Mean lick rate during audible cue presentation (t = 0 – 1 s) using the triggered trials as in a. \na – e, NCD: 59 recordings / 7 mice, HFD: 54 recordings / 7 mice. \nf – o, Licking rate during the 1 s cue window (f – j), and single-trial average neuronal responses \nfrom the 1 – 9 s window post cue (peri-licking) (k – o) shown in each panel as the average across \nall sessions with meal size (i.e. # of triggered trials) within a certain range (15–29, 30–44, 45–59, \na f k p\nh\nq\nc\nd\ne\nlg\ni\nj\nm\nn\no\nSessions containing 45 - 59 trials with Ensure\nSessions containing 60 - 74 trials with Ensure\nSessions containing 15 - 29 trials with Ensure\nSessions containing 30 - 44 trials with Ensure\nSessions containing 75 - 90 trials with Ensure\nb\n****\n**\n***\nMouse specific means\nMeal size prediction\nFIGURE 2\nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\n60–74, or 75–90 trials) in NCD -fed (blue shade) and HFD -fed (red shade) mice. Shaded error \nbars are ± s.e.m. across sessions with a given range of meal sizes.  \np, Correlation between meal size (number of triggered trials, i.e. trials with Ensure) and mid-meal \nneuronal responses (averaged across the 15th to the 29th triggered trial) for individual recordings, \nwith trial range represented by light-to-dark shading (blue: NCD; red: HFD). Dashed lines indicate \nlinear fits for NCD and HFD groups. \nq, Correlation between meal size (number of triggered trials) and mid-meal licking rate (averaged \nacross the 15th to the 29th triggered trial) for individual recordings in NCD -fed and HFD -fed \nanimals. Dashed lines indicate linear fits for NCD and HFD groups.  \nf – q, Recordings for each meal size range: NCD (15 –29): 0; HFD (15–29): 6; NCD (30–44): 6; \nHFD (30–44): 13; NCD (45–59): 10; HFD (45–59): 23; NCD (60–74): 13; HFD (60–74): 7; NCD \n(75–90): 28; HFD (75–90): 4. \nb – d, unpaired, two-tailed Student’s t-test. \np – q, Pearson correlation.  \na – o, Data are represented as the mean ± s.e.m across sessions (b, c, f – o) or across mice (a, \nd, e). **- P <0.01, *** - P <0.001, **** - P <0.0001. \n  \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nFigure 3. Partial reversal of PVHMC4R neuron dysfunction and feeding behavior after diet-\ninduced obesity \na, Schematic timeline of the experimental paradigm.  Animals initially maintained on HFD were \neither switched to NCD (HFD to NCD) or remained on HFD (HFD to HFD) at ~21 weeks of age. \nControl animals were maintained on NCD throughout (NCD to NCD). All groups underwent two \nrecording sessions: one on their initial diet and another after diet switching (left and right panels, \nrespectively, in b – d). \ni\nCT \nj\n12 - 14 w\nNCD to NCD  to NCDHFD to NCD\nf g\na\n to HFD\nh\nc\ned\nHFD FR\n5 w\nAAV-GCaMP6s injection + optic\nfiber implant in the PVH\n10 - 12 w\nHabituation and \nconditional training \n16 - 17 w\n1st recording\nNCD ad lib. NCD FR\n2nd recording\nDiet switch\n~21w ~25w ~27w\nHFD ad lib. HFD ad lib. HFD FR\nb\nNCD ad lib. NCD FR NCD ad lib. NCD FR\nNCD to NCD\nHFD to \nk\nEarly meal response Responses across the meal\nEnsure consumption\n*\n*\n****\n****\n***\n*\n****\n**\n********\n**** ****\n**** *\nFIGURE 3\nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nb, Mean heatmaps summarizing GCaMP6s photometry signals (middle panel) and licking events \n(bottom panel) in the same NCD to NCD animals during the first phase of NCD diet (left, NCD to \nNCD) and second phase (right, NCD to NCD). Top panel: mean running speed and licking rate \nacross all triggered trials. Only recordings with a minimum of 30 successfully triggered trials were \nincluded in the averaged heatmap analyses, while untriggered trials were excluded from the \ncalculation of average photometry and behavior signals. \nc, Same as b, but for the HFD to NCD group of mice. \nd, Same as b, but for the HFD to HFD group of mice. \ne – g, Single-trial mean PVHMC4R neuron responses during food consumption (1 – 9 s post cue) \nacross triggered trials from all recordings in which at least 30/90 trials were triggered, for NCD to \nNCD (e), HFD to NCD (f), and HFD to HFD (g) groups, comparing pre-diet-switch (solid line) and \npost-diet-switch (dashed line) conditions. \nh, Z-scored PVHMC4R responses during trials early in the meal (first 15 triggered trials) in NCD to \nNCD (blue), HFD to NCD ( green), and HFD to HFD ( red) groups, comparing pre -diet-switch \n(diamonds) and post-diet switch (circles) conditions. Each dot is a session. \ni, Z-scored PVHMC4R responses post-diet switch during early vs. late parts of the meal (first 15 \n[blue] or last 15 trials [red]) for NCD to NCD, HFD to NCD, HFD to HFD post-diet switch.  \nj, ΔZ-score for late vs. early trials aggregated for individual animals post-diet switch, categorized \nby groups of NCD to NCD, HFD to NCD, HFD to HFD. \nk, Ensure consumption during the individual photometry recording sessions , categorized by \ngroups of NCD to NCD (blue), HFD to NCD (green), and HFD to HFD (red) groups, comparing \npre-diet-switch (diamonds) and post-diet-switch (circles) conditions. \na – j, Recording and animal numbers of the first diet exposure: NCD to NCD: 44 recordings / 6 \nmice; NCD to NCD: 43 recordings / 6 mice; HFD to NCD: 52 recordings / 7 mice; HF D to NCD: \n42 recordings / 7 mice; HFD to HFD: 35 recordings / 4 mice); HFD to HFD: 23 recordings / 4 mice. \nh, k, Two-way Anova with Tukey post hoc test for before vs. after the diet switch  within each \ncondition and comparing first diet phase responses with each other and second diet phase \nresponses with each other. \ni, Linear mixed model (LME), Value ~ C(Condition, Treatment('NCD | NCD')) * Meal phase. \nj, one-way Anova, with Tukey post-hoc test. \nb – k, Data are represented as the mean ± s.e.m. either as error bars (g – k) or as a shaded area \n(b – h). * - P <0.05, ** - P <0.01, *** - P <0.001, **** - P <0.0001. \n  \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\n \nSupplementary Figure 1. Validation of GCaMP6s photometry signals and behavioral \ndynamics during food restriction in NCD- and HFD-fed mice \na, b, Total E nsure consumption during the individual photometry recording sessions ( a) and \nacross individual mice (b) for NCD-fed and HFD-fed mice. \nc, d, Post-surgical bodyweight dynamics during ad libitum feeding  (light gray), food restriction \n(FR, no shading ), and continued food restriction maintenance during the photometry recording \n(FR + recording, dark shading) of NCD-fed (c) and HFD-fed (d) mice. \ne, Mean lick rate during early and late meal phases for NCD-fed (top panel) and HFD-fed (bottom \npanel) mice. \nLocomotion\nSUP. FIGURE 1\n*******\n*\n****\n***\na\ne\ni\n465 nm\n405 nm\nLicking\nc d\nNCDHFD\nNCDHFD\ng\nb\nf\nNCDHFD\nh\nLicking AUC\nEnsure by mice Ensure by session \nj k\nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nf, Area under the curve (AUC) quantification of post-cue lick-rate, normalized to baseline pre-tone \nlick rate, for early and late meal phases of individual NCD -fed (top panel) and HFD-fed (bottom \npanel) mice. \ng, Mean running during the early and late meal phase of NCD -fed (top panel) and HFD -fed \n(bottom panel) mice. \nh, i, Representative heatmap of normalized ΔF/F GCaMP6s fluorescent responses detected \nusing a 465 nm LED (h) and a 405 nm LED, reflecting movement artifacts and auto-fluorescence \n(i). Responses in each trial are normalized to the 10 s baseline before the cue onset. Note the \nrange in i is smaller than in a. \nj, Early and late meal responses to Ensure for NCD-fed and HFD-fed animals, expressed as ΔF/F. \nk, Z-scored PVHMC4R neuronal responses of each animal during early and late meal phases for \nall recordings in NCD-fed and HFD-fed mice, averaged from 1 – 20 s post-cue presentation. NCD: \n6 mice; HFD: 5 mice.  \na, b, f, – unpaired, two-tailed Student t-test. \nk, Two-way Anova with Tukey post hoc test, for Early vs. Late meal within each condition and \nearly phases between conditions. \na – g, j – k, Data are represented as the mean ± s.e.m either as error bars (a, b, f, k) or as shaded \narea (c, d, e, g, j).  \na, b, k, * - P <0.05, *** - P <0.001, **** - P <0.0001. \n  \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\n \nSupplementary Figure 2. Diet- and fasting -dependent modulation of PVH MC4R neuronal \nactivity, licking behavior, and trial engagement in NCD- and HFD-fed mice \na, Lick rate during the 1 s cue window grouped by the number of triggered trials (ranges: 15–29, \n30–44, 45–59, 60–74, 75–90) in NCD-fed (blue shade) and HFD-fed (red shade) mice. \nb, Single trial mean neuronal responses 1 -9 s post cue (peri -licking) grouped by the number of \ntriggered trials (ranges: 15–29, 30–44, 45–59, 60–74, 75–90) in NCD-fed (blue shade) and HFD-\nfed (red shade) mice.  \nFood restricted Food restricted + acutely fasted\nLicking at reward\nLocomotion after rewardLocomotion during cue\nFood restricted + acutely fasted\nLicking during cue\nLicking at reward\nLicking during cue\nSUP. FIGURE 2\nh\nc\ne f\nd\nb j k\nl m\nNCDHFD\ni\ng\na\nFood restricted\nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\nc, Single trial mean licking rate during the reward period (1  – 9 s post-cue) across 90 recorded \ntrials from NCD-fed and HFD-fed groups (minimum 30 successfully triggered trials). \nd, Single trial mean licking rate during the reward (1  – 9 s post cue) across 90 recorded trails \ngrouped by the number of triggered trials (ranges: 15–29, 30–44, 45–59, 60–74, 75–90) for NCD-\nfed (blue shading) and HFD-fed (red shading) mice. \ne, Single trial mean running speed across 90 recorded trails from NCD -fed and HFD -fed mice \n(minimum 30 successfully triggered trials). \nf, Single trial mean running speed during the pre -cue period across 90 recorded trails, grouped \nby the number of triggered trials (ranges: 15–29, 30–44, 45–59, 60–74, 75–90) for NCD-fed (blue \nshade) and HFD-fed (red shade) mice.  \ng, Number of triggered trials out of the 90 possible trials recording session versus ΔZ-score \nbetween computed by subtracting the mean early meal response from the mean late meal \nresponse. The dashed line represents the linear relationship between meal size (triggered trials \nwith Ensure) and the ΔZ-score (i.e. the change in  response to Ensure from early to late in the \nmeal). \nh, Single trial mean peri-licking PVHMC4R responses (1 – 9 s post-cue) across 90 recorded trials \nfrom female (F) and male (M) NCD -fed and HFD -fed groups. Only recordings with at least 30 \nsuccessfully triggered trials were included. M-NCD: 26 recordings / 3 mice; M-HFD: 31 recordings \n/ 4 mice; F-NCD: 33 recordings / 4 mice; F-HFD: 23 recordings / 3 mice. \ni, Single-trial mean peri-licking PVHMC4R responses (1 – 9 s post-cue) across 90 recorded tr ials \nfor food restricted (nonfasted) and food restricted plus acute fasting (fasted) NCD-fed and HFD-\nfed groups. Only recordings with at least 30 successfully triggered trials were included. \nj – m, Heatmaps summarizing GCaMP6s photometry signals (top panel) and licking behavior \n(bottom panel) from PVHMC4R neurons in food-restricted-only session (j,l) and in food restricted-\nplus acutely-fasted sessions (k, m) for NCD-fed (j, k) and HFD-fed (l, m) animals. Mean running \nspeed and licking rates across all triggered trials are shown above the heatmaps. Only recordings \nwith a minimum of 30 successfully triggered trials were included. Trial structure: 10 s baseline \nbefore cue (tone) onset (t = 0 s), followed by Ensure delivery at t = 1 s. \na – e, Nonfasted NCD: 31 recordings / 7 mice (i, j); Food restricted plus acutely fasted NCD: 28 \nrecordings / 7 mice (i, k); Nonfasted HFD: 31 recordings / 7 mice (i, l); Food restricted plus acutely \nfasted HFD: 23 recordings / 7 mice (i, m). \ne – m, Data are represented as the mean ± s.e.m. * - P <0.05, *** - P <0.001, **** - P <0.0001. \nc, e, g, NCD: 59 recordings / 7 mice; HFD: 54 recordings / 7 mice. \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\na, b, d, f, Recordings for each meal size range: NCD (15–29): 0; HFD (15–29): 6; NCD (30–44): \n6; HFD (30–44): 13; NCD (45–59): 10; HFD (45–59): 23; NCD (60–74): 13; HFD (60–74): 7; NCD \n(75–90): 28; HFD (75–90): 4. \n  \nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint \n\n \nSupplementary Figure 3. Impact of diet switch on cue-evoked licking behavior and PVHMC4R \ndynamics during early and late meal phases \na – c, Single-trial mean lick rate during cue presentation (0 – 1 s post-cue onset) across 90 trials \naveraged across recordings with at least 30 successfully triggered trials in the NCD to NCD (a), \nHFD to NCD ( b), and HFD to HFD ( c) groups, comparing pre -diet-switch (solid line) and post -\ndiet-switch (dashed line) conditions. \na – c, Recording and animal numbers during exposure to the first diet: NCD to NCD: 44 recordings \n/ 6 mice; HFD to NCD: 52 recordings / 7 mice; HFD to HFD: 35 recordings / 4 mice. Recording \nand animal numbers during exposure to the second diet: NCD to NCD: 43 recordings / 6 mice; \nHFD to NCD: 42 recordings / 7 mice; HFD to HFD: 23 recordings / 4 mice. \nd, Mean Z-scored PVHMC4R neuronal responses during early (first 15 trials) and late meal (last 15 \ntrials) phases of the meal, for individual mice (averaged across sessions per mouse) across the \nfollowing conditions post-diet switch: NCD to NCD, HFD to NCD, HFD to HFD. Paired, two-tailed \nt-test between early-meal and late-meal trials. \na – d, Data are represented as the mean ± s.e.m across sessions (a – c) or across mice (d). * - P \n<0.05, ** - P <0.01, *** - P <0.001. \nd\na b c\nSUP. FIGURE 3\n*\nMouse specific means\n*\nLicking across trials\nremix, or adapt this material for any purpose without crediting the original authors. \npreprint (which was not certified by peer review) in the Public Domain. It is no longer restricted by copyright. Anyone can legally share, reuse, \nThe copyright holder has placed thisthis version posted May 27, 2025. ; https://doi.org/10.1101/2025.05.22.655553doi: bioRxiv preprint","source_license":"Public-Domain","license_restricted":false}