A hedonic orexigenic subnetwork within the human hippocampus

Only recently has the rodent hippocampus been implicated in orexigenic appetitive processing 1,2 . This function has been found to be mediated at least in part by lateral hypothalamic inputs involving an orexigenic neuropeptide, melanin-concentrating hormone 3 . This circuit remains elusive in humans. Here, we combine tractography, intracranial electrophysiology, cortico-subcortical evoked potentials, and brain-clearing 3D histology to identify an orexigenic circuit involving the lateral hypothalamus converging in a hippocampal subregion. We found that low-frequency power is modulated by sweet-fat food cues and this modulation was specic to the dorsolateral hippocampus. Lastly, structural and functional analyses of this circuit in a human cohort exhibiting dysregulated eating behavior revealed connectivity that was inversely related to body mass index. Collectively, this multimodal approach describes an orexigenic subnetwork within the human hippocampus implicated in obesity and related eating disorders. Hippocampal segmentation. Tractography used generate a probabilistic map each with voxel’s streamline probability to the LH for all 178 subjects from the normative HCP dataset. Each subjects’ streamline probability map to the LH was transformed to standard MNI 09c space so that they could be averaged and concatenated into a normative weighted average group map of streamline probability between the hippocampal area and the LH across the 178 HCP subjects. We performed this analysis to dene the hippocampal subregions in the normative HCP data and then applied these subregions to the binge-prone cohort. We then used k-means to segment group average hippocampus streamline probability maps. This hypothesis-free method uses successive iterations to assign each voxel to one of two clusters, without the application of external spatial constraints. For the case of large inter-voxel similarities in streamline count, the algorithm fails to identify two distinct clusters. Resulting clusters represented normative hippocampal subregions based on its connectivity to the LH in standard MNI 09c space. Finally, we co-registered the normative clusters to the MRI images from our subjects implanted with depth electrodes as well as the binge eating cohort.

power (~4-6 Hz, condition-speci c pre-stimulus normalized) in the dlHPC was signi cantly higher (p < .05, paired non-parametric cluster-based permutation testing, using null-distribution cluster size to correct for multiple comparisons) during anticipation of the sweet-fat solution compared to a neutral taste (Fig. 1B,   1C top panel). While higher frequencies may re ect more local activity, lower frequencies are thought to be advantageous in routing information across distant areas as their longer period accommodates the temporal demand of conduction velocity across multiple synaptic delays 14 . This pro le was observed immediately following the cue (~ 110 ms) (Fig. 1B) and was localized to contacts within the dlHPC subregion ( Fig. 1C-D). These ndings suggest that dlHPC low-frequency power differentially supports anticipation of appetitive food items.

Evoked potentials support interconnections
Given that tractography cannot assess the potential monosynaptic nature of interactions between two brain regions 15 , we performed direct, single-pulse, electrical stimulation (bipolar; biphasic positive; 0.5 Hz; 6mA; pulse width of 200 s; 49 trials; 120 seconds total) in a human subject with rare, if ever, recordings from both LH and sweet-fat-responsive dlHPC electrodes ( Fig. 2A). Voltage de ections (or evoked potentials) are typically observed within 10-100 ms from stimulation onset when recording from a region directly connected to the stimulation site 16,17 . Following stimulation pulses to the sweet-fat-responsive dlHPC electrodes, we observed sharp and fast (~25ms) antidromic evoked potentials recorded in the single electrode within the LH area ( Fig. 2B and Fig. S5 for raw, common average and bipolar rereferenced from single-trials signal in electrodes). Similar stimulation pulses encompassing the LH area electrode also resulted in fast, orthodromic, evoked potentials in the dlHPC ( Fig. 2C and Fig. S6 for raw, common average and bipolar re-referenced from single-trials signal in electrodes). These bidirectional, fast, evoked potentials recorded in both regions following stimulation of the other are indicative of monosynaptic circuit-interactions between them.

MCH+ projections to dlHPC
Given that MCH is an orexigenic neuropeptide produced in the LH area with a well-described role in appetition 3,8,9 , we next tested for MCH+ projections in the dlHPC subregion. To do so, we leveraged another rare opportunity afforded by a post-mortem sample of human tissue for the immunolabelingenabled 3D imaging of solvent cleared organs (iDISCO) procedure (Fig. 3A). This technique allowed for 3D immunostaining and visualization of axonal projections carrying speci c peptides within tissue cuboids, whereas conventional techniques would be limited in visualizing axons intersecting histological slices 18 .
First, we manually identi ed the location of our sample in a corresponding coronal slice in the highresolution MNI 09c template brain (Fig. S7). Second, we extracted a representative dorsolateral section that encompassed the dlHPC subregion in the template brain (Fig. 3A). This section was then processed through the iDISCO brain-clearing procedure, staining for MCH (Phoenix Pharmaceuticals, Inc and Alexa Fluor 647 Thermo Fisher). Third, the cleared and stained section was again manually overlaid to the corresponding coronal slice of the high-resolution MNI 09c template brain with the additional overlay of the tractography-identi ed dlHPC subregion (Fig. 3B). We found that the dlHPC section contained MCH+ orexigenic projections visualized using the brain-clearing 3D histology ( Fig. 3C; Suppl. Video).

LH-dlHPC connectivity implicated in obesity
A human binge eating cohort (n=34, females) was subdivided into overweight/obese (body mass index, BMI ≥ 25 m 2 /kg, n=17) and lean (BMI < 25 m 2 /kg, n=17) groups. We con rmed the dlHPC contained the LH-dlHPC node, previously de ned by LH streamlines, by co-registering our normative hippocampal subregions of interest and the atlas-based LH (Fig. 4A) to images acquired from these human subjects. Akin to the normative cohort described above (Fig. 1A), we also found signi cantly higher normalized counts of LH streamlines in the left (t = -4.585; p < .001) and right (t = -3.609; p < .001) dlHPC voxels of this cohort as compared to the hippocampal voxels outside the dlHPC (i.e., non-dlHPC hippocampal voxels) (Fig. 4B).
We next assessed whether structural and functional connectivity of the LH-dlHPC circuit differ between overweight/obese and lean groups. We found that resting state functional connectivity (rsFC) between the dlHPC and LH area was signi cantly decreased in overweight/obese compared to lean subjects (t = 2.51; p = .018) (Fig. 4C). Probabilistic tractography-based structural connectivity index (CI, as de ned by Tschentscher et al. 19 ) between the dlHPC and the LH area was also signi cantly decreased in obese/overweight compared to lean groups in the left (t = 2.13; p = .042) but not right (t = 1.07; p = .295) hemispheres (Fig. 4D). We con rmed that these connectivity ndings were speci c to the dlHPC subregion by performing a similar analysis between the non-dlHPC hippocampal voxels and the LH area. No differences in LH-non-dlHPC structural CI nor rsFC were observed between overweight/obese and lean groups (Fig. S8).
As we were ultimately interested in the overall multivariate pattern of these functional and structural circuit alterations, we t a multivariate logistic regression model including neuroimaging as well as behavioral variables (detailed in Online Methods) to predict whether a subject belongs to the overweight/obese or lean group. Using backwards elimination, we identi ed LH-dlHPC rsFC (β = -9.886, p = .044) and LH-left dlHPC CI (β = -14.676, p = .037) as the only independent predictors of obesity, with a variance in ation factor (VIF) of 1.32 (VIF < 2.5 suggests negligible collinearity between variables) 20 . Such ndings further implicate this circuit in obesity involving dysregulated eating behavior.

Discussion
As a higher-order processing center involved in integrating external and internal stimuli, the hippocampus is uniquely positioned as an important node for orexigenic appetition 1,2,21 . Here, we characterized an hedonic orexigenic subnetwork within the human hippocampus. Structurally, LH streamlines converge in the dorsolateral aspect of the hippocampus (i.e., dlHPC); interconnections between the LH and dlHPC are further validated via single-pulse stimulation of the dlHPC resulting in sharp and fast voltage de ections in the LH area. This hippocampal subregion contains MCH+ projections presumably derived from LH soma 3 . Functionally, the dlHPC exhibits speci c eld potential responses during anticipation of a highcaloric, sweet-fat solution. Lastly, the LH-dlHPC circuit is perturbed in patients with obesity involving dysregulated eating patterns.
The interrogation of a neural circuit underlying appetitive processing in living humans poses unique challenges, and has mostly relied on functional MRI and non-invasive electrophysiology [22][23][24] . For over two decades, however, in-vivo structural investigations of human brain circuits have been made possible by diffusion MRI-based tractography 25,26 . The key limitation of tractography is that it may be prone to false positives and negatives, and it may also not allow distinction between afferent and efferent projections 27,28 . Nonetheless, tractography ndings can be supported by direct interrogation of circuits with (1) stimulation-induced evoked potentials 29 and (2) post-mortem brain clearing 3D histology 15 . Here, we used high-resolution diffusion MRI to de ne the hippocampal subregion where LH streamlines are more densely populated (i.e., the dlHPC). Thereafter, we also applied the two aforementioned modalities uniquely in parallel to further probe and characterize LH connections within the dlHPC.
Stimulation of either dlHPC or the LH caused fast, sharp and reproducible voltage de ections in the other region, potentially indicating monosynaptic connections between them 16 . The shapes of the evoked potentials differed depending on which node received stimulation or recordings, which may re ect cytoarchitecture differences between the two regions and the position of the electrode with regards to the tissue 17 . In a post-mortem hippocampal specimen, we then used iDISCO 3D histology, immunostaining for MCH, a neuropeptide that is involved in feeding and primarily synthesized in the LH and its immediate adjacencies 3,8-10 , to further con rm and characterize orexigenic LH projections within the dlHPC subregion. These histologically-de ned LH projections within the dlHPC subregion shed light on the directionality of the previously described evoked potentials, with the responses recorded in the LH area (following dlHPC stimulation) likely representing antidromic effects of stimulating these projections, as previously described in different subnetworks 30,31 . This dlHPC subregion was selected for functional examination with intracranial electrophysiological recordings during the sweet-fat incentive paradigm.
Comparable to our work, the same sweet-fat paradigm used here has been applied to functional MRI studies that showed increased hippocampal activation in response to sweet-fat compared to neutral stimuli 32 . In addition, hippocampal activation in response to food stimuli has been reported to be decreased following intranasal insulin administration 33 . While these studies place the human hippocampus at the intersection of energy homeostasis and appetitive processing, functional fMRI and non-invasive electrophysiology, respectively, lack the temporal and spatial resolution for uncovering differential hippocampal subregion involvement. Moreover, reports on hippocampal connectivity underlying dysregulated eating and obesity are scarce, and conspicuously absent are studies examining hypothalamic inputs in humans. Subjects undergoing brain mapping with intracranial electrophysiology provide a unique opportunity to overcome these limitations in the interrogation of speci c regions of interest during controlled assays such as a food-incentive paradigm 12,13 .
Intracranial electrophysiology of this dlHPC subregion revealed ~4-6 Hz discriminatory power. This frequency range overlaps with theta ranges, prominent rhythms in the both the rodent and human hippocampus ascribed to mnemonic processes including organizing events in a sequence, spatial navigation and exploration and facilitating transient interactions for memory encoding and retrieval 34,35 . Lower frequencies including theta have also been observed in neocortical areas and linked to both mnemonic and cognitive control processes 36 . The ubiquitous presence of this rhythm across areas and behavioral contexts led to a hypothesis of its more general role 37 , such as mediating information transfer between the recruited regions and at temporal scales associated with a given behavioral context. Given that the hippocampus is a higher order node integrating multimodal information coming from neocortical and subcortical structures, this low frequency pro le may mediate information transfer between the dlHPC and LH as well as other recruited structures recruited during food anticipation. Our report of low frequency power increase in this time window is consistent with power increase in the same frequency range in mice exposed to olfactory sweet-fat cues, re ecting a degree of generality of this signal to appetitive food anticipation irrespective of the sensory modality of the cue [1][2][3]38 .
Previous studies have also implicated the human hippocampus and hypothalamus in dysregulated eating and obesity. Neuroimaging work reported hippocampal activation in response to food stimuli increased in children with obesity and dysregulated eating (e.g., eating in dissonance with homeostatic requirements) 39 . Other work reported a decrease in hippocampal activation in response to food images that predicts post-task levels of chocolate consumption, and abnormal hippocampal activation during reward processing in subjects with dysregulated eating behaviors 40,41 . In addition to these neuroimaging studies, hippocampal concentration of metabolites (e.g., creatine + phosphocreatine) have also been reported to be increased in obese/overweight subjects, potentially indicating BMI-related alterations in in ammatory cytokines and adipokines within the hippocampus 42 . With regards to the hypothalamus, prior work established an association between MCH overexpression and obesity in animal models 8 and neuroimaging work reported increased hypothalamic activation during a task requiring inhibitory control in subjects with dysregulated eating 43 . Only one study has reported abnormal rsFC between the LH and multiple brain regions in adolescents with excessive weight 44 ; while the hippocampus was identi ed as functionally connected to the LH, this subnetwork's involvement in obesity was not reported. Our discovery of this link, however, is likely due to the identi cation of the dlHPC as the subregion of interest, as ndings of decreased LH-hippocampal connectivity in the obese state were not observed when including the hippocampal voxels outside the dlHPC.
The present investigation provides evidence supporting decreased rsFC and structural connectivity between LH and dlHPC in obese/overweight females. These ndings only emerged after the de nition of the dlHPC as the subregion of interest, as our two cohorts showed no differences in connectivity between the non-dlHPC voxels and the LH. Moreover, the nding that functional and structural connectivity measures were signi cant predictors of overweight/obese vs. lean group assignments supports the notion that the LH-dlHPC appetitive processing node is indeed altered in the obese state. Putting these ndings in context of the reports discussed above, structural and functional abnormalities involving the MCH+ LH-dlHPC node uncovered here may predispose individuals struggling with dysregulated eating behavior to obesity.
A few considerations are noteworthy regarding the interpretations of the present ndings. Firstly, we did not include males or subjects struggling with other forms of dysregulated eating. We chose to use a homogeneous female only binge eating cohort to ensure that our ndings would not be skewed by sex or behavioral differences. Our electrophysiological analysis, however, are not sex speci c. Secondly, directionality cannot be inferred from our cortico-subcortical evoked potential experiment, as the observed LH signal can be explained by either retrograde activity along the LH axons, or anterograde activity along the dlHPC axons. Thirdly, MCH+ projections may be coming from either the LH or its immediately adjacent structures, such as the zona incerta 45,46 . Fourthly, we had only a single case with simultaneous LH and dlHPC intracranial electrodes for the experiment with cortico-subcortical evoked potentials and a single sample for the 3D histology; together, however, these novel, complimentary methods afforded us a unique opportunity to leverage and cross-validate these approaches to interrogate a speci c human orexigenic subnetwork. Lastly, what remains unclear is the extent to which the properties of the appetitive circuit identi ed here generalize to other hedonic behaviors.
Collectively, the convergence of modalities has elucidated a circuit that is perturbed in a disease-relevant state, furthering our understanding of how speci c node interactions within the human brain are involved in obesity and related eating disorders.   Resting-state fMRI scans from the binge-prone cohort were preprocessed using fMRIPrep 1.2.3 4 . In brief, the preprocessing of the functional image involved skull-stripping, co-registration to the T1 reference image, and head motion and susceptibility distortion corrections. After removal of non-steady state volumes and spatial smoothing with a 6mm FWHM isotropic Gaussian kernel, ICA-AROMA was used to identify motion-related noise components in the BOLD signal 5 . Framewise displacement (FD) and root mean square variance over voxels of the temporal derivative of time courses (DVARS) were calculated 6,7 . Global signals were extracted within the cerebrospinal uid (CSF), white matter (WM), gray matter (GM), and whole-brain masks. XCP Engine 1.0 was used to perform denoising of the preprocessed BOLD output from fMRIPrep, utilizing the estimated confound parameters 8,9 . This included demeaning and removal of any linear or quadratic trends and temporal ltering using a rst-order Butterworth bandpass lter (0.01-0.08 Hz). These preliminary preprocessing steps were then followed by confound regression of ICA-AROMA noise components, together with mean WM, CSF, and global signal regressors. All regressors were bandpass ltered to retain the same frequency range as the data to avoid frequency-dependent mismatch 8 . Whereas preprocessing was performed on the diffusion MRI data from the binge-prone cohort to prepare the images for probabilistic tractography using the FSL suite 10,11 , the normative HCP diffusion MRI data had already been preprocessed (with the minimal preprocessing pipeline). The diffusion-weighted images were corrected for motion and geometric distortions using the 'topup' and 'eddy' functions, similar to that applied in HCP's preprocessing pipeline. For each subject, diffusion and T1-weighted images were co-registered using boundary-based registration.
Probabilistic tractography. Probabilistic tractography was used to evaluate the connections between LH and hippocampus. The LH mask was de ned on the standard T1 MNI152 09c template adapted from CIT168 Subcortical In Vivo Probabilistic Atlas 12 , while the hippocampi masks were de ned using the Harvard-Oxford brain atlas. This registration was performed using Advanced Normalization Tools (ANTs), which consists of two successive steps of linear and nonlinear registration between the subject's brain and the MNI brain. In a third step, the MNI-de ned ROIs were registered to the subject's space. FSL's Bayesian Estimation of Diffusion Parameters Obtained using Sampling Techniques (BEDPOSTX) was used to conduct Monte Carlo sampling of probability distribution of diffusion parameters at each voxel, accounting for up to three crossing bers directions within a voxel 13 . Fiber tracking was performed with FSL's Probtrackx2, using distance correction and each hippocampal voxel as seed and the LH as target 14 .
A total of 5000 seed points were used to generate streamlines from each seed voxel, and only the streamlines that reached the target were retained for further analysis. The results of Probtrackx are summarized in a map of "streamline probability" and "waytotal", representing the probability for each seed voxel to reach the target and the total number of streamlines from a given seed that reached the target, respectively. The strength of the connections between seed and target was calculated as a tractography connectivity index (tractography-CI), as de ned in a previous study by the formula: log(waytotal)/log(5000 x Vseed) 15 . The waytotal resulting from the tractography was log-transformed and divided by the log-transformed product of the generated sample streamlines in each seed voxel (5000) and the number of voxels in the respective seed mask (Vseed). The log-transformation increased the likelihood of reaching normality, which was tested using the Shapiro-Wilk test 16 .
Hippocampal segmentation. Tractography was used to generate a probabilistic map each with voxel's streamline probability to the LH for all 178 subjects from the normative HCP dataset. Each subjects' streamline probability map to the LH was transformed to standard MNI 09c space so that they could be averaged and concatenated into a normative weighted average group map of streamline probability between the hippocampal area and the LH across the 178 HCP subjects. We performed this analysis to de ne the hippocampal subregions in the normative HCP data and then applied these subregions to the binge-prone cohort. We then used k-means to segment group average hippocampus streamline probability maps. This hypothesis-free method uses successive iterations to assign each voxel to one of two clusters, without the application of external spatial constraints. For the case of large inter-voxel similarities in streamline count, the algorithm fails to identify two distinct clusters. Resulting clusters represented normative hippocampal subregions based on its connectivity to the LH in standard MNI 09c space. Finally, we co-registered the normative clusters to the MRI images from our subjects implanted with depth electrodes as well as the binge eating cohort.
Sweet-fat incentive paradigm. Participants (n=9, Table S1) underwent surgical implantation of depth electrodes for neurosurgical epilepsy monitoring. The location of electrode implantation was determined solely based on clinical needs and thus varied across participants. The inclusion criteria for this study were the presence of at least one hippocampal depth electrode. Consent to participate in this study was obtained according to the Declaration of Helsinki and approved by the institutional ethical committee. We have adapted the sweet-fat incentive paradigm, also known as the Milkshake task 17 for intracranial electrographic recordings during cued anticipation and consumption of a sweet-fat and a taste-neutral solution, as previously published and described in detail by our group 18,19 . Data acquisition and experimental task were previously described 18 . Brie y, neural activity was sampled at 1024 Hz from AdTech electrodes while subjects engaged in a sweet-fat incentive computer-based paradigm (Fig. S1). Each trial in this paradigm began with a 2-sec xation cross presented on a computer screen -this period is referred to as the pre-stimulus period. This was followed by a 1-sec presentation of an image of a glass of either water or of milkshake, which served as a cue for the solution to be subsequently delivered through a mouthpiece to the subject for consumption. Before the solution was delivered, a 2-sec image of a xation cross was viewed. The 1-sec presentation of the solution to be delivered and this 2-sec xation cross period are referred to as an anticipatory period (3-sec). Following the anticipatory period is a 5-sec receipt/consummatory period, consisting of a 3-sec solution delivery period followed by a 2-sec consumption of solution period. Sweet-fat and taste-neutral trials were presented in a randomized order, with a total of 80 to 100 trials evenly split between sweet-fat and taste-neutral conditions. Upon task completion, the participants were asked to rate on a 1-10 Likert scale the quality of the sweet-fat solution (Likert scale, 1-10) and which solution (sweet-fat vs. taste-neutral) they preferred.
Electrode localization. Pre-surgical MRI scan was co-registered to post-surgical CT scan for electrode visualization and localization as described previously 20 . Locations of depth electrodes within the medial temporal lobe were then examined by one rater with expertise in medial temporal lobe anatomy in neuroimaging (D.A.N.B). Electrodes in direct contact with the hippocampal area were selected for further evaluation. Then, we co-registered the normative hippocampal clusters (i.e., dlHPC and non-dlHPC) that we had previously de ned in standard MNI09c template brain to each subject's native space (Fig. 1D). All hippocampal electrodes (n=54) were labelled according to whether they were in direct contact with dlHPC or not (i.e., non-dlHPC). This localization was performed prior to time-frequency analysis.
Task data preprocessing and analyses. Electrophysiological data was downsampled to 1000 Hz, notch ltered for 60 Hz and 2nd-3rd harmonics, and Laplacian re-referenced in FieldTrip as previously described 18,19 . Artifact timepoints were de ned as voltage values greater or less than the mean signal of all 10-sec trials concatenated, recorded from the same channel plus 4 multiples of its standard deviation. Any trial with at least 1 detected artifact timepoint was excluded. Time-Frequency analysis was implemented using the wavelet toolbox in Matlab. There are three input parameters: 1) minimum frequency, set to 3, 2) maximum frequency, set to 250, and 3) NumVoices, set to 32. The toolbox generates 'scales' based on the desired frequency range (de ned by min to max frequencies), which then get mapped into frequencies. The trial vector, scales vector, and 'morl' are inputs to the cwtft function in matlab, which generates the wavelets and power extraction. Wavelets were rst tested on ground truth data with known spectral properties before use on experimental data.
Trial instantaneous power values were normalized to power at the same frequency and channel during the 1-sec prestimulus period across all trials in the same condition (condition speci c pre-stimulus normalization). The prestim duration of any trial with at least 1 detected artifact timepoint was excluded from the normalizing distribution (see previous section). Condition speci c pre-stimulus normalization was used to account for possible differences in baseline power before stimulus presentation. Results were reproduced using an alternative normalization method whereby power values were normalized relative to the distribution of power in the same frequency and channel, during the entire recording.
Spectral analyses were primarily focused on the anticipation period (1-sec cue, and 2-sec post-cue xation). Statistical differences in time-frequency power between conditions were calculated using cluster-based permutation testing 21 . Brie y, this involved calculating a t-statistic in each time-frequency voxel, between the two conditions (sweet fat vs. neutral), thereby generating the observed t-map. The distribution for each voxel was generated by pooling the time-frequency maps from all channels and subjects (trials were averaged to generate a single map per channel). The observed t-map was then compared to a null distribution (shu ed condition labels) of t-maps generated over 1000 paired permutations. A p-value for each voxel was obtained by comparing the observed to the null t-value at the same time-frequency voxel, thereby generating a p-map. Clusters of contiguous voxels with a p < .05 were identi ed and compared to the null-distribution cluster size. Observed clusters with sizes larger than the 95th percentile of those from the null distribution were considered signi cant after correction for multiple comparisons.
Evoked potentials. As previously described, we performed single-pulse stimulations at rest using an intracranial electrical waveform generator and switchbox 22,23 (MS-120BK-EEG and PE-210AK, Nihon Kohden, Tokyo, Japan). Electrical stimulation was delivered through adjacent pairs of electrodes in biphasic pulses (6mA; 200 μs/phase, 49 trials) at a frequency of 0.5 Hz for a total of 120 seconds. We measured electrical potentials in response to stimulation with a video EEG monitoring system using a sampling rate of 2000 Hz (version WEE-1200, Nihon Kohden). We analyzed the single-pulse stimulation data using custom Matlab scripts. We rst applied a high-pass butterworth lter (1 Hz) to exclude slow varying effects and segmented evoked responses time series from recording channels were into 2500 ms epochs time-locked to stimulus onset (500 ms pre-stimulus and 2000 ms post-stimulus). Then, we rereferenced the data to the common average signal, excluding stimulated channels, channels with artifacts, and channels with large, evoked responses, as previously described 24 . Finally, to exclude potential effects of pre-stimulus signal uctuations, we applied a baseline correction by subtracting the average signal between 200 ms and 20 ms prior to stimulus onset. To ensure that these preprocessing steps did not introduce a bias, we also provided line traces of bipolar re-referenced and the single-trials raw signal from the recording electrodes (Fig. S5).
Human hippocampal sample. In accordance with the local IRB, a post-mortem sample of the left hippocampal area (Fig. 2C, left panel) was obtained from a whole brain with no known pathologies that had been extracted 24 hours after death and placed in 10% formalin for one day. The sample was perfused and stored in PFA 4%. For the brain-clearing procedure, we extracted a representative dorsolateral hippocampal section (Fig. 2C, mid panel) and transferred the sample to 1M phosphatebuffered saline (PBS). The sample was stored at 4°C until the iDISCO protocol was performed. iDISCO brain-clearing 3D histology. After antibody validation, we selected a representative dorsolateral section for the iDISCO protocol to con rm our hippocampal area segmentation (Fig. 1A). The section was approximately 1.0 x 0.8 x 0.4 cm and was pretreated with methanol following the iDISCO protocol 25 using a modi ed immunostaining protocol. The sample was rinsed with PBS with 2% TritonX-100 two times for 1 hour. We incubated the sample in permeabilization solution, PBS with 0.2% Triton, for 30 min, and blocking solution, PBS with 0.2% TritonX-100, 10% DMSO, and 6% Donkey Serum, for 1 hour. The sample was incubated with 1:500 anti-MCH in PTwH (PBS, 0.2% Tween-20, 1% Heparin 10mg/ml, 0.2% Sodium Azide) for 10 days, nutating at 37° C. After 10 days, the samples were rinsed 3 times for 5 minutes and then every few hours and left nutating at room temperature overnight. On the next day, the sample was incubated in the secondary antibody, Alexa Fluor 647 anti-Rabbit (ThermoFischer scienti c-USA, A-31573), 1:250 in PTwH and 3% Donkey serum and 0.2% Sodium azide at 37°C, nutating for 10 days, covered from light. Following secondary incubation, the sample was again rinsed in PTwH for 2 days, and the iDISCO clearing protocol was followed (https://idisco.info/idisco-protocol/update-history/) Histological assessment of lateral hypothalamic connections in the dorsolateral hippocampus. The iDISCO brain-clearing 3D histology results were used to con rm the hippocampal area segmentation. The anti-MCH was used to identify orexigenic projections within our hippocampal sample. We assessed whether our sample from the dorsolateral hippocampal subregion, de ned based on the higher number of probabilistic tractography streamlines, contained projections expressing MCH, an orexigenic neuropeptide primarily produced in the LH area. First, we manually identi ed the location of our whole sample in a corresponding coronal slice in the high-resolution MNI 09c template brain (Fig. S7) and then extracted a representative dorsolateral section that overlapped with the dlHPC (Fig. 2C). Following immunolabeling and clearing, the nal MCH-stained sample was again manually overlaid to the corresponding coronal slice in the high-resolution MNI template brain and the tractography-based hippocampal probability map of LH area streamlines (Fig. 2D).
Demographics, clinical and behavioral data of binge-prone cohorts. Subjects' consent was obtained according to the Declaration of Helsinki and approved by the institutional ethical committee (IRB-35204).
We analyzed the available clinical and behavioral data from 34 binge-prone females, de ned by at least one weekly episode of eating large amounts of food in short periods accompanied by the feeling of loss of control eating over the prior 6 months (i.e., binge-prone cohort; mean age = 26 ± 5.6 years; BMI = 27.9 ± 8.5; binge frequency = 2.7 ± 1.4 episodes/week) 26 . The number of binge-eating episodes per week was assessed with the Eating Disorder Examination, a standardized diagnostic interview 27 . The Beck Depression Inventory (BDI) and the Beck Anxiety Inventory (BAI) were used to screen for depression and anxiety, respectively 28,29 . The Di culties in Emotion Regulation Scale was used to assess impairment in emotion regulation 30 . The binge-eating-cohort was divided into two subgroups: (1) lean (n=17): BMI < 25 (referred to as lean group); (2) overweight/obese (n=17): BMI > 25 (referred to as obese/overweight group).
Resting-state functional connectivity analysis. Resting-state functional connectivity analysis was performed on the binge-prone cohort's preprocessed resting-state fMRI data using DPABI 4.3/DPARSF which is based on Statistical Parametric Mapping (SPM12, https://www. l.ion.ucl.ac.uk/spm) 31 . Three subjects were excluded due to excessive movement as measured by 1) mean FD > 0.2mm, 2) more than 20% of FD over 0.2mm, or 3) any FD > 5mm 9 . A seed-based approach was performed to examine rsFC in the binge-prone cohort (n=34) by calculating the rsFC between the LH mask as de ned above and each tractography-identi ed hippocampal subregion. Functional connectivity values were extracted for all subjects and utilized in further correlational analyses.
Statistical analyses. Statistical analyses were performed using the RStudio Version 1.2.5042 (RStudio, Inc.). Given the sensitivity of metrics derived from resting-state fMRI and diffusion MRI proneness to numerical distortions related to data acquisition or analytical pipeline, we employed the Tukey's method to remove outliers for each connectivity metric. After checking for normality, we then used the Student's Ttest to compare rsFC as well as tractography-CI between hippocampal subregions and LH in overweight/obese and lean group. One outlier was identi ed and removed from the obese binge eating group and two outliers were identi ed and removed from the lean group in both the rsFC and structural CI analyses in Figures 3C and 3D (left panel), respectively. Additionally, we identi ed and removed one outlier from the lean group in the CI analysis in Figure 3D (right panel). Mann-Whitney U Test was used to compare the corrected number of streamlines between the LH and hippocampal subregions in the binge-prone cohort. To assess potential effect of confounders in the identi ed connectivity differences between the obese and lean group, we t a multivariate logistic regression model to predict whether a subject belongs to the overweight/obese or lean group including the available demographic and behavioral variables in addition to the LH-dlHPC connectivity measurements. The comprehensive list of variables included: age, depression (BDI), anxiety (BAI), binge frequency, restrained eating, emotional eating and externally driven eating scores (from DEBQ), LH-dlHPC-LH rsFC, LH-non-dlHPC rsFC, LH-dlHPC structural connectivity, and LH-non-dlHPC structural connectivity. We then used backwards elimination to identify which combination of variables provided highest predictive power with the lowest total number of explanatory variables to avoid over-tting (Akaike Information Criterion). Finally, a variance in ation factor was calculated to assess potential correlations between explanatory variables 32 . Signi cance was de ned by p < 0.05 for all tests.
Data availability. Anonymized data that support the ndings of this study as well as the code and materials used for analyses are available from the corresponding author upon reasonable request.
Online References

Supplemental Video
The Supplemental Video is not available with this version    This circuit is associated with the obese state involving dysregulated eating behavior in humans. A) Regions of interest co-registered to native space of an exemplary subject: dlHPC (yellow, hotspot), non-dlHPC (red) and LH (blue, adapted from CIT168 Subcortical In Vivo Probabilistic Atlas). (B) Signi cantly higher normalized streamline counts observed between the LH and left dlHPC (t = -4.585; p < .001) and right dlHPC (t = -3.609; p < .001) compared to the non-dlHPC in the overall cohort. (C) rsFC between the