Central nervous system adaptation mechanism of obesity induced by high-fat high-sucrose diet: Differences in brain function between obese and obese resistant rats based on fMRI Authors: Qi Fang, Shantong Zhao, Qiutong Lin, Xiaoyun Zhu, Changhao Zheng, Yunlong Wang, Tao Lu, Lili Zhu Published: 2026 ABSTRACT ObjectiveThe prevalence of metabolic obesity continues to rise, yet its pathogenesis remains controversial. This study focuses on the central neural regulatory differences underlying phenotypic divergence between diet-induced obese (OB) and obesity-resistant (OR) rats following long-term high-fat and high-sucrose diet (HFHSD) exposure.MethodsForty male Wistar rats were randomized to normal diet group (ND,n = 8) or HFHSD group (n = 32). After 8 weeks, HFHSD rats were split into OB (n = 7) and OR (n = 8) by weight. OB/OR continued HFHSD for 8 weeks with food intake recorded. At week 15, OGTT and functional magnetic resonance imaging (fMRI)-derived amplitude of low-frequency fluctuations (ALFF) were performed. At week 16, metabolic and somatic parameters were measured.ResultsCompared to the OR group, OB rats exhibited persistent hyperphagia, elevated triglycerides, TyG index, and leptin levels, significantly increased liver, heart, and kidney masses, and impaired glucose tolerance (all p < 0.05). fMRI revealed significantly reduced ALFF in the hindbrain (bilateral brainstem and cerebellar molecular layers) and increased ALFF in the right_ dysgranular_retrosplenial cortex (RSC) of OB rats. Metabolic load parameters correlated negatively with hindbrain ALFF and positively with RSC ALFF, with a significant negative correlation between hindbrain and RSC activity.ConclusionLong-term HFHSD induces functional remodeling of the hindbrain–RSC circuit in obese rats, and this neuroplastic alteration is closely associated with systemic metabolic burden, suggesting its involvement in the pathophysiology of obesity. Introduction Although genetic predisposition provides a basis for obesity, it cannot fully explain the rapid global increase in obesity prevalence over recent decades [1]. Primary obesity has become an increasingly severe global public health issue [2]. Fundamentally, obesity results from a chronic imbalance in which energy intake persistently exceeds energy expenditure, causing surplus energy to gradually accumulate as adipose tissue [3]. Overweight and obesity are independent risk factors for arterial hypertension, dyslipidemia, type 2 diabetes mellitus, coronary heart disease, cerebrovascular disease, arthritis, sleep apnea syndrome, and various malignancies, significantly affecting patients’ quality of life and imposing substantial socioeconomic burdens [4]. Notably, under identical high-fat, high-sugar dietary conditions, some individuals exhibit an OR phenotype, whereas others develop obesity [5,6]. Such phenotypic divergence may result from interindividual differences in the responsiveness of central nervous system circuits to dietary signals [7]. The Western diet is widely recognized as a key driver of the global obesity epidemic [8]. HFHSD alters peripheral metabolic status and exerts direct effects on neural circuits that regulate ingestive behavior [7]. Romero et al. showed that the dietary fatty acid margaric acid inhibits Piezo2 function in a dose-dependent manner, thereby reducing the firing frequency of neurons in response to mechanical stimulation [9]. Piezo2 is a mechanosensitive ion channel expressed in vagal afferent terminals and mediates gastrointestinal mechanosensation; functional suppression of this channel may therefore weaken the transmission of mechanical signals from the gut [10,11]. In obese rats induced by HFHSD, vagal nerve activity elicited by gastric distension is reduced by 60–70% [12]. The vagal–brainstem circuit relays visceral sensory information via vagal afferents to the NTS, where signals are integrated and then transmitted through the dorsal motor nucleus of the vagus to modulate gastric function [13]. Accordingly, alterations in the brainstem–vagal pathway, a core network for interoception, may play a significant role in the phenotypic differentiation of obesity [14]. Although obesity is frequently accompanied by dysglycemia, dyslipidemia, and insulin resistance, these metabolic derangements are more plausibly regarded as consequences rather than initiating events of neural circuit dysfunction [2,7]. Given the impact of Western diets on neural systems, delineating the differences in central regulation between distinct obesity phenotypes is of considerable importance for advancing our understanding of obesity pathogenesis. Neural plasticity refers to the inherent capability of the nervous system to adapt to environmental changes through modulation of synaptic strength and circuit remodeling [3]. Matikainen-Ankney et al. argue that persistent obesity-related neuroplastic changes are a core mechanism underlying the difficulty of weight loss and weight regain [7]. High-fat diets can alter neural circuits within the brainstem even before significant weight gain occurs [15]. The ALFF, measured via fMRI, quantifies the magnitude of blood oxygen level-dependent (BOLD) signal oscillations in the frequency band of 0.01-0.1 Hz, reflecting baseline local neuronal activity [16]. ALFF is a reliable method for detecting inter-phenotypic differences in brain function [17]. Zhou et al. identified significant differences in fractional fALFF in the right insula, brainstem, and cerebellum of patients with postprandial distress syndrome [18]. Similarly, Ling et al. reported a negative correlation between the relative autonomy index of exercise motivation and resting-state ALFF in the right posterior cerebellar lobe, implicating the posterior cerebellum in the regulation of feeding motivation [19]. Enhanced functional synchrony between the dorsal anterior cingulate cortex and RSC has also been observed in individuals with binge-eating disorder [20]. Collectively, these findings highlight the essential roles of the brainstem, cerebellum, and RSC in regulating ingestive behavior and interoceptive processing. This study tests the central hypothesis that long-term HFHSD-induced phenotypic divergence between OB and OR rats is associated with differences in by ALFF within specific brain regions, and that these neural differences partially mediate the development and maintenance of obesity. To test this, OB and OR rat models will be established via a controlled dietary protocol. Resting-state fMRI data will be acquired to calculate ALFF, alongside correlation analyses among ALFF, dynamic feeding behaviors, and metabolic parameters. Materials and methods Experimental animals A total of 40 male Wistar rats (5-6 weeks old) were purchased from SPF (Beijing) Biotechnology Co., Ltd. [License No.: SCXK (Beijing) 2024-0001]. After 2 weeks of acclimatization feeding with a standard diet, their body weight was 283.81 ± 13.95 g (mean ± SD). The animals were housed in a specific pathogen-free (SPF) facility under controlled environmental conditions: ambient temperature of 22 ± 2 °C, relative humidity of 50 ± 10%, and a 12-h light/12-h dark cycle. Rats were housed 2-3 per cage with ad libitum access to water and food. Obesity induction and grouping Following acclimatization, a total of 40 male Wistar rats were randomly assigned to two groups: a ND group (n = 8) and a HFHSD group (n = 32). The ND group was fed a normal diet for 16 weeks, and only body weight was measured to validate the OB/OR model (no further interventions were performed). The HFHSD group was fed an HFHSD ([Composition: 70% basal diet, 15% lard, 15% sucrose; Manufacturer: SPF (Beijing) Biotechnology Co., Ltd., Cat. No. SFD010]) for 8 consecutive weeks to establish models of diet-induced obesity and obesity resistance. At the end of the 8th week, all rats in the HFHSD group were ranked according to body weight: the 7 rats with the highest body weight were assigned to the OB group (n = 7), and the 8 rats with the lowest body weight were assigned to the OR group (n = 8). The remaining 17 rats, whose body weights fell within the intermediate range and did not meet the criteria for extreme phenotype selection, were excluded. Following group allocation, both the OB and OR groups continued to receive the same HFHSD to maintain the obese phenotype for a further 8 weeks. During this period, food consumption was recorded every two days. Body weight was measured weekly using an electronic balance. The obese phenotype was determined based on body weight and the Lee index, calculated using the following formula[21]:Leeindex=Bodyweight(g)3Naso−length(cm)×100 Body weight change was calculated as the final body weight minus the initial body weight at the end of the experiment. Serum collection and organ indices After the 16-week dietary intervention, rats were fasted for 10 h, anesthetized with isoflurane. Blood was collected from the abdominal aorta into 5 mL serum separation tubes. After standing at room temperature for 2 h, the blood was centrifuged at 2500 rpm for 15 min. The collected serum was aliquoted and stored in a −80 °C freezer until analysis. Immediately following dissection, the whole brain, liver, heart, and bilateral kidneys were harvested, rinsed with pre-cooled physiological saline, blotted dry with filter paper, and weighed (balance model:METTLER TOLEDO ME104/02, precision 0.1 mg). Organ indices were calculated using the formula:Organindex=Organweight(g)Bodyweight(g)×1000 Oral glucose tolerance test (OGTT) and fasting blood glucose measurement OGTT was performed during the 15th week of the experiment. After a 6-h fast, baseline fasting blood glucose was measured via blood sampled from the tail vein. Subsequently, a 50% glucose solution was administered by oral gavage at a dose of 2.5 g/kg body weight. Blood glucose concentrations were measured at 0 min (before gavage), 30 min, 60 min, and 120 min post-gavage using blood sampled from the tail vein with a Roche blood glucose meter (model: Accu-Chek active) and compatible test strips. The area under the glucose curve (AUC) was calculated using the trapezoidal rule. Fasting blood glucose was measured during the 16th week; rats were fasted for 10 h, and blood was collected from the tail vein for baseline measurement. Serum lipid parameters and TyG index Serum levels of total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C) were measured using an enzymatic colorimetric method with a Beckman Coulter AU480 automated clinical chemistry analyzer. The TyG index was calculated using the formula below, with fasting glucose and TG concentrations converted to mg/dL: [22]TyGindex=lnFastingTG(mg/dL)×Fastingglucose(mg/dL)2 Atherosclerotic index calculation Atherosclerosis-related indices were calculated based on lipid parameters to assess cardiovascular disease risk. The atherogenic index (AI) was calculated as follows [23]:AI=CHOL(mmol/L)−HDL(mmol/L)HDL(mmol) Serum leptin measurement Serum leptin was measured by ELISA (Elabscience, Wuhan, China; Cat. E-EL-R0582; sensitivity 0.1 ng/mL; intra-/inter-assay CV<10%).Diluted serum samples were added to antibody-precoated wells, incubated for 90 min at 37 °C, and washed thrice. HRP-conjugated antibody was added and incubated for 1 h at 37 °C. After substrate addition (15 min, dark, room temperature), absorbance was read at 450 nm. Sample concentrations were calculated by reference to a standard curve. fMRI data acquisition fMRI scans were performed during the 15th week of the experiment. Imaging data were acquired using a Bruker 7T magnetic resonance system (PharmaScan 70/16 US) equipped with a rat body coil and surface coil for radiofrequency pulse transmission and reception. Prior to data acquisition, rats were anesthetized with 2.5% isoflurane mixed with oxygen in an induction chamber until no response to toe pinch was observed. During scanning, anesthesia was maintained with 1.5–2.5% isoflurane mixed with oxygen. Rats were placed in a prone position and fixed on an animal bed with their heads immobilized in the scanner adapter. Body temperature was maintained using a SC100 water bath system (Thermo Scientific, USA). A physiological monitoring system (PC-SAM32 software) was used to continuously monitor the respiratory rate, which was maintained at 60–70 breaths per minute by adjusting the isoflurane level accordingly. For structural imaging, a T2-weighted TurboRARE sequence was used with the following parameters: TR/TE = 4720/33 ms, slice thickness = 0.6 mm, field of view = 25 × 20 mm, image matrix = 256 × 256, and number of averages = 3. Functional images were acquired using a T2∗-weighted FID-EPI sequence (T2∗_FID_EPI_sat) with the following parameters: repetition time (TR) = 2000 ms, echo time (TE) = 20 ms, field of view (FOV) = 25 × 20 mm2, matrix = 80 × 64, slice thickness = 0.6 mm, and repetitions = 400. Imaging data analysis Functional image preprocessing was performed using MATLAB 2022b and SPM12. The preprocessing steps included: removal of the first 10 vol to allow for signal stabilization; slice timing correction; and head motion correction using rigid body transformation (subjects with translation >3 mm or rotation >3° in any direction were excluded). Individual T2 structural images were co-registered to the mean functional image after motion correction. Images were then normalized to the MNI standard space using linear and nonlinear transformations, and the resulting deformation fields were applied to the functional images, which were resampled to a voxel size of 3 mm × 3 mm × 3 mm. Spatial smoothing was performed using a Gaussian kernel with a 6 mm full width at half maximum (FWHM). Subsequently, linear detrending was applied to remove signal drift. Nuisance covariates, including white matter signal, cerebrospinal fluid signal, and 24 head motion parameters (Friston 24), were regressed out using a linear regression approach ALFF was calculated by converting the time series of each voxel to the frequency domain using FFT. The square root of the power spectrum within 0.01–0.08 Hz was averaged as the ALFF value, which was then converted to z-scores across the whole brain to obtain standardized zALFF maps. Statistical analysis All data were statistically analyzed using GraphPad Prism 9.5.1 and Python 3.9 (with libraries including pandas, numpy, matplotlib, scipy.stats, warnings, and matplotlib as mpl). Between-group comparisons: Due to the relatively small sample sizes (OB group n = 7, OR group n = 8), data are presented as median (interquartile range). Group differences were compared using the non-parametric Mann-Whitney test (two-tailed). A p-value <0.05 was considered statistically significant. Imaging data comparison: Voxel-wise independent samples t-tests were performed on the zALFF maps between the two groups using SPM12. Statistical significance was determined using False Discovery Rate (FDR) correction, with a significance threshold set at voxel-level P < 0.005 and cluster-level p < 0.05. Correlation analysis: Variables showing significant differences between groups (including imaging features, serum biochemical parameters, metabolic indices, and organ indices) were selected for correlation analysis. Spearman's rank correlation (two-tailed) was used to assess associations between variables. A correlation heatmap was generated using the corrplot package in R, with a significance level set at p < 0.05. Results Dietary intake and weight change Fig. 1Metabolic and physiological parameters in OB and OR rats after a 4-month HFHSD. (A) Weekly food intake (weeks 8-16); (B) Body weight changes; (C) Lee index; (D) Organ weights; (B) Organ indices. OB: obesity group: OR: obesity-resistant group. Data are presented as median (Q1-Q3). Significant difference va. OR: na, not significant, ∗p < 0.05; ∗∗p < 0.01,∗∗∗p < 0.001. Metabolic and physiological parameters in OB and OR rats after a 4-month HFHSD. (A) Weekly food intake (weeks 8-16); (B) Body weight changes; (C) Lee index; (D) Organ weights; (B) Organ indices. OB: obesity group: OR: obesity-resistant group. Data are presented as median (Q1-Q3). Significant difference va. OR: na, not significant, ∗p < 0.05; ∗∗p < 0.01,∗∗∗p < 0.001. The OB group exhibited significantly higher food intake than the OR group during weeks 8–12 and weeks 14–16 (p < 0.001 for both periods), as well as during week 13 (p < 0.01, Fig. 1A).During the 16week HFHSD intervention, the OB group showed significantly greater weight gain than the OR group starting from week 1 (p < 0.001), and the weight difference between these two groups continued to widen over time. Compared with the ND group, the OB group also had significantly higher weight gain from week 1 to week 6 (p < 0.01) and from week 7 to week 16 (p < 0.001,).For the OR group relative to the ND group: no significant difference was observed at week 1 nor from week 9 to week 16 (all p > 0.05).At week 2 and week 5, the OR group showed significantly lower body weight than the ND group (p < 0.01); at weeks 3, 4, and 6–8, the OR group was also significantly lower than the ND group (p < 0.001, Fig. 1B). Organ mass and lee index After 16 weeks of dietary intervention, no significant difference was observed in brain mass between the two groups (OB: 2.19 [2.14–2.37] g vs. OR: 2.26 [2.15–2.33] g, p > 0.05). However, the liver, heart, and kidney masses in the OB group were significantly higher than those in the OR group (liver: 18.43 [17.76–21.82] g vs. 13.16 [11.85–14.23] g; heart: 1.64 [1.54–1.67] g vs. 1.25 [1.23–1.41] g; kidney: 3.75 [3.57–4.04] g vs. 3.01 [2.80–3.20] g; all p < 0.001, Fig. 1D). Additionally, the Lee index in the OB group was significantly higher than that in the OR group (31.70 [30.70–32.00] vs. 29.90 [29.53–30.53], p < 0.05, Fig. 1C). Organ index analysis The brain index (3.02 [2.62–3.09] vs. 4.33 [4.16–4.41], p < 0.001) and cardiac index (2.13 [1.96–2.33] vs. 2.45 [2.42–2.62], p < 0.05) in the OB group were significantly lower than those in the OR group. No significant differences were observed in the liver index or renal index between the two groups (p > 0.05, Fig. 1E). Serum lipid profile and serum leptin levels No significant differences were observed between the two groups in terms of HDL-C, LDL-C, or total cholesterol levels (p > 0.05).The triglyceride level in the OB group was significantly higher than that in the OR group (0.83 [0.69–1.38] g/L vs. 0.43 [0.34–0.59] g/L, p < 0.01, Fig. 2A). The serum leptin concentration in the OB group was significantly higher than that in the OR group (0.80 [0.65–1.16] ng/mL vs. 0.43 [0.40–0.51] ng/mL, p < 0.01, Fig. 2C). The atherosclerotic index did not differ significantly between the OB group (p > 0.05).Fig. 2Comparison of metabolic parameters between OB and OR groups after a 4-month HFHSD. (A) Serum lipid parameters; (B) OGTT results; (C) Serum leptin concentration; (D) OGTT-AUC (B) TyG index.Significant difference va. OR: ns, not significant; ∗p < 0.05; ∗∗p < 0.01. Comparison of metabolic parameters between OB and OR groups after a 4-month HFHSD. (A) Serum lipid parameters; (B) OGTT results; (C) Serum leptin concentration; (D) OGTT-AUC (B) TyG index.Significant difference va. OR: ns, not significant; ∗p < 0.05; ∗∗p < 0.01. Oral glucose tolerance test (OGTT) and fasting blood glucose During the OGTT, the OB group exhibited significantly higher blood glucose levels at both 0 min (baseline) and 120 min compared to the OR group (0 min: 7.50 [7.40–7.90] mmol/L vs. 7.05 [6.48–7.38] mmol/L (p < 0.05); 120 min: 8.40 [8.00–9.20] mmol/L vs. 7.05 [6.50–8.23] mmol/L (p < 0.05). However, no significant differences were observed at the 30 min and 60 min time points (p > 0.05, Fig. 2B). Fasting blood glucose did not differ significantly between the OB group and the OR group (p > 0.05). Metabolic and insulin resistance parameters The TyG index in the OB group was significantly higher than that in the OR group (8.31 [8.10–8.76] vs. 7.53 [7.26–7.98], p < 0.001, Fig. 2E). The OGTT-AUC for blood glucose was also significantly higher in the OB group compared to the OR group (1098 [1035–1137] vs. 995.3 [940.5–1028], p < 0.001, Fig. 2D). Differences in spontaneous neural activity across brain regions Compared to the OR group, the OB group exhibited significant differences in ALFF values across multiple brain regions (voxel p < 0.005, cluster p < 0.05, FDR corrected): significantly reduced ALFF values were observed in the Brainstem_L (T = −5.45, cluster size = 39), Brainstem_R (T = −5.07, cluster size = 66), Molecular_Layer_of_the_Cerebellum_L (T = −4.53, cluster size = 37), and Molecular_Layer_of_the_Cerebellum_R (T = −4.41, cluster size = 34); whereas significantly increased ALFF values were noted in the Retrosplenial_Dysgranular_Cortex_R (T = 9.55, cluster size = 67) (Fig. 3 and Table 1).Fig. 3Brain regions showing ALFF differences between the OB and OR groups. Significant differences in ALFF were observed in the OB group compared to the OR group in the following regions: Brainstem R (A), Brainstem L (B), Molecular_Layer_of_the_Cerebellum_L(C), Molecular_Layer_of_the_Cerebellum_R(D), and Retrosplenial Dysgranular_Cortex_R (E).Table 1Region-specific differences in the amplitude of low-frequency fluctuations between OB and OR groups.Table 1Brain regionMNI peak coordinatesT valueCluster sizep ValueXYZBrainstem_R1.42−2.87−14.06−5.0766<0.05Brainstem_L−2.18−1.97−11.36−5.4539<0.05Molecular_Layer_of_the_Cerebellum_L−2.781.04−13.76−4.5337<0.05Molecular_Layer_of_the_Cerebellum_R4.721.04−13.16−4.4134<0.05Retrosplenial_Dysgranular_Cortex_R3.824.64−9.569.5567<0.05MNI, Montreal Neurological Institute; T-value, statistical value of peak voxel showing differences in amplitude of low-frequency fluctuations (ALFF) between OB and OR groups. The statistical threshold was set at an uncorrected voxel-level p < 0.005 and a cluster-level p < 0.05 with false discovery rate (FDR) correction for multiple comparisons. Only clusters with a minimum of 20 contiguous voxels are reported. Brain regions showing ALFF differences between the OB and OR groups. Significant differences in ALFF were observed in the OB group compared to the OR group in the following regions: Brainstem R (A), Brainstem L (B), Molecular_Layer_of_the_Cerebellum_L(C), Molecular_Layer_of_the_Cerebellum_R(D), and Retrosplenial Dysgranular_Cortex_R (E). Table 1Region-specific differences in the amplitude of low-frequency fluctuations between OB and OR groups.Table 1Brain regionMNI peak coordinatesT valueCluster sizep ValueXYZBrainstem_R1.42−2.87−14.06−5.0766<0.05Brainstem_L−2.18−1.97−11.36−5.4539<0.05Molecular_Layer_of_the_Cerebellum_L−2.781.04−13.76−4.5337<0.05Molecular_Layer_of_the_Cerebellum_R4.721.04−13.16−4.4134<0.05Retrosplenial_Dysgranular_Cortex_R3.824.64−9.569.5567<0.05MNI, Montreal Neurological Institute; T-value, statistical value of peak voxel showing differences in amplitude of low-frequency fluctuations (ALFF) between OB and OR groups. The statistical threshold was set at an uncorrected voxel-level p < 0.005 and a cluster-level p < 0.05 with false discovery rate (FDR) correction for multiple comparisons. Only clusters with a minimum of 20 contiguous voxels are reported. Region-specific differences in the amplitude of low-frequency fluctuations between OB and OR groups. MNI, Montreal Neurological Institute; T-value, statistical value of peak voxel showing differences in amplitude of low-frequency fluctuations (ALFF) between OB and OR groups. The statistical threshold was set at an uncorrected voxel-level p < 0.005 and a cluster-level p < 0.05 with false discovery rate (FDR) correction for multiple comparisons. Only clusters with a minimum of 20 contiguous voxels are reported. Multivariate correlation analysis A correlation analysis was conducted on all variables showing significant inter-group differences. The ALFF values of the bilateral brainstem and bilateral cerebellar molecular layers were strongly positively correlated with each other (r > 0.6), and negatively correlated with the ALFF value of the RSC, as well as with TG, TyG index, liver mass, heart mass, kidney mass, and body weight change (∣r∣ = 0.52–0.75, p < 0.05). The ALFF value of the RSC was positively correlated with the above metabolic indicators (r = 0.60–0.69, p < 0.05), but negatively correlated with the cardiac index (r = −0.53, p < 0.05). In terms of metabolic indicators, TG, TyG index, leptin, Lee index, and organ masses showed significant positive correlations with each other (p < 0.05), whereas the cardiac index exhibited negative correlations with body weight change and other metabolic indicators (Fig. 4).Fig. 4Imiegrated cocrelation wodysis of multiple indicatora. Spearman correlation analysis was conducted on the neuroimaging, metabolic, and physical indicators with significant Iner-group dimmon."∗p < 0.05; ∗∗p < 0.01 ∗∗∗p < 0. 001. Imiegrated cocrelation wodysis of multiple indicatora. Spearman correlation analysis was conducted on the neuroimaging, metabolic, and physical indicators with significant Iner-group dimmon."∗p < 0.05; ∗∗p < 0.01 ∗∗∗p < 0. 001. Discussion Behavioral data demonstrated that, from the 8th week post-grouping through the 16th week, OB rats exhibited persistent and marked hyperphagia. Prolonged 16-week exposure to the HFHSD induced systemic metabolic dysregulation in OB rats, as evidenced by elevated serum triglyceride levels, an increased triglyceride-glucose (TyG) index, a higher Lee index, and increased weights of major visceral organs. Correlation analysis revealed that these metabolic burden indicators were all strongly and positively correlated with liver weight. These findings faithfully recapitulate the classical trajectory of obesity progression, evolving from an initial imbalance in energy intake to systemic metabolic dysfunction and pathological remodeling of target organs. The TyG index is widely recognized as a surrogate marker for insulin resistance [22]. Recent studies have further validated its predictive value for metabolic dysfunction-associated fatty liver disease and visceral adiposity [24,25]. Notably, our findings suggest that the TyG index may serve as a more sensitive early predictor of metabolic dysfunction-associated fatty liver disease in diet-induced obesity models. Interoception refers to the brain's sensing, integration, and interpretation of internal physiological state signals, forming a core psychophysiological mechanism for maintaining homeostasis, shaping emotional experiences, and guiding cognitive decision-making [14,26]. As a key relay station for visceral sensory information, the brainstem receives mechanical, chemical, and immune signals via vagal afferent pathways and participates in mediating the formation of aversive memories associated with discomfort [13,27,28]. Low et al. found that activation of the anterior deep cerebellar nuclei (aDCN) reduces the anticipated reward value of food by increasing striatal baseline dopamine levels while simultaneously attenuating food-evoked phasic dopamine release [29]. This mechanism promotes satiety by shortening meal duration and reducing meal size without affecting meal frequency. Moreover, the cerebellum can perceive and integrate visceral and limbic information through cerebello-hypothalamic-circuits, sending modulatory outputs to hypothalamic nuclei that regulate feeding and gastrointestinal function [30]. The hindbrain serves as the primary central integrative hub for vagal afferent signals and circulating metabolic cues, playing a decisive role in the efficiency of negative feedback regulation of energy homeostasis [27,[31], [32], [33]]. Our study showed that OB rats exhibited a significant reduction in the ALFF of BOLD signals in hindbrain-associated regions, and that these values were significantly negatively correlated with all measured peripheral metabolic burden parameters. This finding suggests that the severity of peripheral metabolic derangement is closely linked to the degree of functional suppression within hindbrain interoceptive integration circuits. Persistent lipotoxicity driven by chronic high-fat exposure, potential neuroinflammatory processes, and reduced vagal afferent sensitivity may underlie the observed decrease in hindbrain ALFF [2,9,12,34,35]. Reduced spontaneous neural activity in the hindbrain might imply impaired relay and integration of satiety signals [27]. This central desensitization state may shift the hypothalamic energy balance set-point toward a positive energy balance, thereby providing a neurophysiological basis for persistent hyperphagia [36]. In addition, attenuated hindbrain functional activity may further worsen peripheral organ hypertrophy and structural remodeling by impairing vagal-autonomic efferent function [10,11,37]. In striking contrast to the suppressed hindbrain activity, OB rats displayed significantly enhanced ALFF values in the RSC, and the magnitude of this enhancement was positively correlated with various metabolic burden indices. As a key node within the default mode network (DMN), the RSC is involved not only in episodic memory and spatial navigation but also in the associative encoding of environmental cues and reward value [38]. Increased DMN activity during rest has been associated with higher-order cognitive functions, including interoceptive awareness and self-referential processing, which are important for regulating homeostatic feeding [39]. From a neuroanatomical circuitry perspective, cerebellar signals can be relayed to the hippocampal formation via cerebello-thalamo-retrosplenial pathways, thereby contributing to the modulation of cognitive functions such as spatial cognition, cognitive flexibility, and associative learning [[40], [41], [42]]. fMRI studies have also reported co-activation of the RSC and cerebellum during autobiographical memory retrieval tasks, supporting their joint involvement in brain networks underlying episodic memory [43]. Furthermore, the hypothalamus–RSC–hindbrain pathway represents a core element of the rodent recognition memory system: the supramammillary nucleus (SuM) of the hypothalamus regulates hippocampal-dependent spatial memory retrieval, while the RSC serves as a cortical integration hub that converges spatial and contextual information and sends descending projections to the hindbrain to coordinate the encoding, consolidation, and retrieval of spatial navigation and episodic memories [[44], [45], [46], [47]]. However, this study used resting-state fMRI without food-cue tasks or ingestive behavior assessments. Therefore, the enhanced ALFF in the RSC cannot be directly interpreted as reflecting increased encoding capacity for food cues or as driving hyperphagia. One tentative hypothesis is that increased spontaneous activity in the RSC of OB rats may alter excitability in episodic memory circuits, indirectly modulating how environmental context affects feeding under free-feeding conditions. Testing this causal hypothesis will require task-based fMRI combined with behavioral paradigms. Given the inverse covariation pattern observed between hindbrain and RSC activities, their strong associations with peripheral metabolic parameters, the presence of hypothalamic leptin resistance, and previously documented epigenetic modifications, we cautiously propose a preliminary framework termed the “Feeding Adaptation Hypothesis” [48]. This hypothesis suggests an evolutionarily conserved hindbrain-hypothalamus-RSC neural circuit in the mammalian brain [[49], [50], [51], [52], [53]]. Within this circuit, the hindbrain integrates interoceptive signals conveyed via vagal afferents and circulating factors, the hypothalamus acts as the central hub for homeostatic regulation, and the RSC, characterized by high neuroplasticity, may exhibit heightened sensitivity to external environmental cues, although the resting-state design of this study precludes direct validation of this sensitivity [27,53]. Under ecological conditions involving limited availability of calorie-rich foods, transient and reversible suppression of hindbrain activity accompanied by enhanced RSC activity could serve as an adaptive mechanism, facilitating short-term energy storage. This mechanism is phenotypically analogous to hyperphagia observed in animals preparing for hibernation [54,55]. In the modern context of sustained food abundance, however, this evolutionarily adaptive mechanism may become maladaptively entrenched due to chronic high-calorie intake [3,27,31]. Peripheral metabolic dysregulation and the accumulation of lipotoxic substances arising from persistent consumption of HFHSD exert sustained detrimental effects on hindbrain neuronal function, progressively impairing efficient satiety signal transmission. Such attenuation in interoceptive signaling may disinhibit the hypothalamus-RSC circuit, thereby increasing the responsiveness of the RSC to external environmental stimuli, including food-associated cues. This chain of events may ultimately establish a self-perpetuating vicious cycle, in which excessive feeding further suppresses hindbrain function, subsequently promoting RSC hyperactivation and driving additional food intake. Consequently, we hypothesize that the primary mechanism underlying ingestive behavior could shift from being predominantly regulated by internal physiological homeostatic states toward a pattern characterized by elevated spontaneous activity in the RSC and heightened responsiveness to external environmental cues [56]. Additionally, the finding of significantly lower relative organ coefficients for the brain and heart in the OB group compared with OR group suggests possible adaptive trade-offs in energy allocation strategies at the whole-body level. Conclusion Rats rendered obese by long-term HFHSD consumption display functional remodeling of the hindbrain–RSC circuit. These neural alterations are closely associated with indices of systemic metabolic burden. On this basis, we propose the “feeding adaptation hypothesis”: an imbalance between diminished interoceptive sensitivity and heightened sensitivity linked to episodic memory may represent a key neural mechanism that drives and sustains the obese phenotype. Limitations Several limitations of the present study should be acknowledged. First, a single-time-point observational design was employed, with resting-state fMRI acquired only at week 15 of HFHSD exposure. Because the body weight divergence between the OB and OR groups emerged early during HFHSD exposure and the OB group exhibited significantly higher food intake throughout the entire period, the observed between-group differences in ALFF may reflect innate obesity susceptibility, arise secondary to chronic hyperphagia and the obese state, or result from an interaction of these factors. The current design therefore cannot establish a causal relationship between the altered neural activity and the obesity phenotype. Second, fMRI data were collected under anesthesia maintained with 1.5%–2.5% isoflurane. Although real-time physiological monitoring was used to keep the respiratory rate within 60–70 breaths per minute, potential confounding effects of anesthesia on spontaneous brain activity and ALFF estimates cannot be completely excluded. Future studies should adopt anesthesia protocols that more closely approximate the physiological state to validate these findings. Third, we proposed the “feeding adaptation hypothesis” based on the HFHSD-induced obesity and resistance phenotypes. Given that HFHSD exerts unique effects on neural plasticity, the applicability of this hypothesis to other dietary contexts remains to be determined. Future work should incorporate multiple dietary formulations to evaluate the generalizability of the observed neural circuit alterations. Fourth, the relatively small sample size is another limitation. Although the standardized HFHSD minimized confounding variables, the small sample size may have reduced statistical power, increasing the risk of false-negative results. Therefore, these findings require validation in larger studies. CRediT authorship contribution statement Qi Fang: Writing – original draft, Visualization, Validation, Investigation, Formal analysis, Data curation. Shantong Zhao: Validation, Investigation, Data curation. Qiutong Lin: Investigation, Data curation. Xiaoyun Zhu: Project administration, Methodology, Formal analysis. Changhao Zheng: Investigation. Yunlong Wang: Visualization, Formal analysis. Tao Lu: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition. Lili Zhu: Writing – review & editing, Validation, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Ethical statement All animal experiments were conducted in compliance with the guidelines of the Institutional Animal Care and Use Committee of Beijing University of Chinese Medicine (BUCM-IACUC) and the national regulations for the welfare of laboratory animals. The study protocol was reviewed and approved by the BUCM-IACUC (Approval No. [BUCM-2024070103-3003]). Data access statement Access to the data that support the findings of this study is restricted. However, data may be available from the corresponding authors upon reasonable request. Funding statement This work was supported by the 10.13039/501100001809National Natural Science Foundation of China (Grant Nos. [82305036] and [82474488]) Declaration of competing interest The authors declare that they have no competing interests.