Avoidant/restrictive food intake disorder (ARFID) is a complex eating disorder characterized by persistent avoidance or restriction of food intake that is not driven by concerns about body weight or shape. ARFID is characterized by an extremely low food intake and may be driven by sensory aversions to certain foods, a low interest in food, and/or traumatic experiences related to food intake, leading to avoidance. Although ARFID has a number of specific criteria and symptoms, many aspects of its diagnosis, development, and treatment remain unclear. In this narrative review, we summarize the available evidence on molecular, cellular, and neurobiological mechanisms potentially underlying the core manifestations of ARFID, including food avoidance and restriction, sensory-based food aversion, and fear of aversive consequences. The potential mechanisms discussed include altered sensory processing, dysregulation of hunger and satiety signals, conditioned taste aversion and other forms of associative learning, alterations in the microbiome-gut-brain axis, immune mechanisms, and genetic factors. Because direct evidence from ARFID populations remains limited, we also consider findings from related but diagnostically distinct conditions and phenotypes, including picky or selective eating, food neophobia, and sensory food aversion. Such findings may provide insights into mechanisms relevant to ARFID but cannot be assumed to be specific to the disorder. By integrating evidence across these biological domains, this review aims to identify potential mechanisms underlying the heterogeneous manifestations of ARFID and highlight priorities for future research.
One prospective animal model in regenerative and developmental neuroscience is a precocial spiny mouse. However, the pattern of its central nervous system development remains unexplored. To identify critical periods in the postnatal formation of general and exploration‑related behavior in spiny mice, open field behavior was assessed across the first three postnatal months. Locomotor activity, thigmotaxis, vertical activity, and grooming were analyzed. Based on these findings, two key time points in the spiny mouse postnatal ontogeny were identified: (1) between 1st and 2nd days (transition from minimal movement to active walking, with jumping and unsupported rearing behavior first appearing with sharp increase in all exploratory behaviors with peripheral preference), and (2) between 5th and 10th days (transition to mature complex exploratory patterns characterized by more diverse and frequent behavioral repertoire and broader area coverage). The first stage may correspond to a marker improvement in locomotor abilities, while the second and third stages are associated with a sharp increase in exploratory behavior.
The gut microbiome acts as a primary regulator of host homeostasis, influencing the entire body through bidirectional communication along the gut-brain axis (GBA). Dysbiosis, which is defined as a state of microbial imbalance involving alterations in community composition and function, can disrupt the synthesis of important microbiota-derived metabolites, such as short-chain fatty acids (SCFAs), bile acids and neurotransmitter precursors. This can lead to impaired essential host signalling pathways. There is growing evidence that metabolic alterations associated with dysbiosis contribute to the onset and progression of neurodegenerative disorders (NDDs), including Alzheimer's disease (AD), Parkinson's disease (PD), Huntington's disease (HD), multiple sclerosis (MS) and amyotrophic lateral sclerosis (ALS). In this context, G protein-coupled receptors (GPCRs) act as essential molecular transducers that link microbial metabolites to intracellular signalling networks. Aberrant GPCR activation, driven by altered metabolite profiles, modulates key downstream pathways including cAMP, MAPK, PI3K/Akt, NF-κB and Ca2 + signalling. This promotes neuroinflammation, oxidative stress, mitochondrial dysfunction and pathological protein aggregation - hallmark processes underlying neurodegeneration. By identifying convergent and disease-specific signalling pathways, the review highlights mechanistic nodes of therapeutic relevance and discusses GPCR-centric emerging and other microbiome-targeted strategies aimed at restoring metabolic and signalling homeostasis in neurodegenerative disorders.
Both serum serotonin and cardiovascular-risk (CVR) burden are implicated in Alzheimer's disease (AD)-related vascular and neurodegenerative processes, yet their interrelationship across the AD continuum remains unclear. This cross-sectional study examined whether serum serotonin was associated with three complementary measures of CVR burden across the AD continuum and according to CSF amyloid status in 428 participants (138 cognitively normal [CN] individuals, 206 participants with mild cognitive impairment [MCI], and 84 participants with AD), using covariate-adjusted regression models with FDR correction. Following covariate-adjusted analyses, serum serotonin declined progressively from CN to MCI and AD, whereas all CVR measures were higher in MCI and AD than in CN participants. Among participants with MCI and pooled amyloid-positive (Aβ+) participants, greater CVR burden was consistently associated with lower serum serotonin across the Framingham Risk Score (FRS), pooled PCA-derived CVR score, and cohort-specific PCA-derived CVR score. Together, these within-group findings suggest that lower serum serotonin accompanies greater vascular-risk burden in clinically and biologically vulnerable individuals. However, CVR × diagnosis and CVR × amyloid-status interaction tests were not significant after FDR correction. Besides, lower serum serotonin and greater CVR burden were separately associated with poorer cognitive and functional outcomes in both MCI and Aβ+ participants. In short, greater CVR burden was associated with lower serum serotonin, particularly within MCI and Aβ+ groups, and both factors were related to less favorable clinical outcomes, underscoring the need for longitudinal and mechanistic studies to define serotonin's role at the vascular-neurodegenerative interface in AD.
Learning and memory are regulated by physiological processes involved in hippocampal plasticity. Grounding-based cranial coupling to Earth potential has been proposed; however, its effects on cognition remain unclear. This study investigated whether prolonged grounding-based cranial coupling affects learning, memory, and hippocampal biochemistry in male Wistar rats. Forty-eight rats were randomly assigned to Control, Sham, and grounding exposure groups (7, 14, 21, and 28 days; n = 8/group). Grounding was applied continuously (24 h/day) through a cranial electrode connected to an Earth grounding system. Cognitive performance was evaluated using the Morris Water Maze, Novel Object Recognition, Shuttle Box, and Open Field Test. Hippocampal levels of TNF-α, IL-1β, IL-6, malondialdehyde (MDA), superoxide dismutase (SOD), glutathione (GSH), and brain-derived neurotrophic factor (BDNF), together with serum corticosterone, were measured. Prolonged grounding exposure (21-28 days) was associated with improved learning and memory, reflected by enhanced spatial, recognition, and passive avoidance memory, while locomotor activity remained largely unchanged. Behavioral improvements occurred alongside reduced hippocampal IL-1β, IL-6, TNF-α, and MDA levels, increased SOD activity, elevated GSH and BDNF levels, and lower serum corticosterone. These findings provide preliminary evidence of an association between sustained grounding-based cranial coupling and physiological changes related to hippocampal function and cognitive performance. However, this study did not assess cortical or hippocampal electrical activity, membrane potential, or tissue-level current flow; therefore, the observed behavioral and biochemical changes should not be interpreted as evidence of a direct electrical neuromodulatory mechanism. Further studies are required to clarify pathways and determine causality.
Cholesterol 24-hydroxylase (CYP46A1) regulates brain cholesterol homeostasis and synaptic plasticity, playing a crucial role in ischemic stroke. Although previous studies have reported post-ischemic CYP46A1 upregulation, its spatiotemporal dynamics remain poorly defined. To elucidate these dynamics, we investigated the expression of CYP46A1 and other essential cholesterol homeostasis-related genes from 6 h to 3 days after permanent middle cerebral artery occlusion (pMCAO) in CB-17 mice. We utilized single-cell and single-nucleus transcriptomics, regional quantitative PCR, and high-resolution immunohistochemistry. CYP46A1 is predominantly expressed in neurons. Following ischemia, the cholesterol network exhibited a dynamic spatiotemporal divergence. Acutely (6 h post-ischemia), surviving regions transiently upregulated cell-autonomous cholesterol synthesis genes and CYP46A1. Subacutely (3 days), this response shifted toward a widespread upregulation of glia-dependent cholesterol transport genes and general CYP46A1 downregulation. At 24 h, CYP46A1 protein was substantially reduced in the necrotic core and superficial layer II/III of the peri-infarct cortex, but upregulated in deeper layer V, hippocampus, and lateral striatum. Notably, this localized upregulation spatially coincided with reactive microglial hypertrophy. These findings indicate that CYP46A1 is dynamically modulated in viable tissues following ischemic stress. This spatial divergence likely reflects a synergistic interaction between inflammatory propagation and neural circuit-mediated oxidative stress. Resolving these spatiotemporal profiles provides a rigorous foundation for evaluating CYP46A1 functionality and developing stage-specific therapeutic interventions.
We explored processes reflecting movement stability during holding a pointing posture involving the trunk and lower limbs. In particular, we focused on characteristics of drifts and random walk (RW) in joint configuration spaces involved in shoulder movement, endpoint-to-shoulder movement, and their sum - the endpoint movement. Young, healthy participants performed a pointing movement from a natural standing posture that required motion of the arm, trunk, and leg joints. They held this posture for 60 s with open and closed eyes and with and without an elastic band applying a force in the backward direction. Motion analysis in a two-dimensional action space was used. Drift and RW were quantified within and orthogonal to the uncontrolled manifold (UCM, solution space) within the respective spaces of elemental kinematic variables. Drifts along and orthogonal to the UCM with characteristic times of 2-8 s were observed, primarily along the UCM, where the drifts were slower. RW was marginally anti-persistent (stabilizing) when quantified in the joint configuration spaces. There were no effects of vision on drift and RW characteristics, in contrast to earlier force production tasks. We observed no major effects of the external load, indicating that stability-related behavior was robust to the tested change in external forces. Overall, the findings suggest differences in the organization of stability between force-production and movement tasks, while the underlying neural mechanisms remain to be established. Single-trial analysis of drifts and RW is a promising tool for studies of disordered movement stability.
Ketogenic diet (KGD) combined with ketamine (KET) suppresses food restriction-evoked hyperactivity in activity-based anorexia (ABA) mouse model of anorexia nervosa (AN) for more than 28 days, thereby improving weight retention and reducing anxiety, even after mice return to standard diet (SD) and are re-challenged with ABA (KGD + KET → SD, "KET + KGD group"). Suppression of these AN-like behavior is not sustained in mice fed KGD without KET (KGD → SD, "KGD-only group"). We hypothesized that the sustained efficacy of KET + KGD is associated with altered prevalence of certain synapses in the dorsal hippocampus that influence behavior. We compared the prevalence of excitatory and inhibitory synapses between KET + KGD (N = 7) versus KGD-only (N = 7) groups, using electron microscopy and GAD immunolabeling to distinguish GABAergic inhibitory synapses from non-GABAergic excitatory synapses. Pyramidal cell layer (PCL) was analyzed to assess the contribution by parvalbumin + subtype of GABAergic neurons known to reside there, while stratum lacunosum-molecular (SLM) was analyzed to assess inhibition by non-PV + GABAergic neurons known to reside there. In both PCL and SLM, the KET + KGD group exhibited greater correlation between GABAergic inhibition of PNs and weight retention. In SLM of the KET + KGD group, the areal density of excitatory synapses was less. Inhibitory synapses onto GABAergic dendrites (disinhibition which may also augment excitability) correlated negatively with hyperactivity for the KET + KGD group only. KET + KGD, but not KGD-only, may sustain ABA resilience by enhancing plasticity of both glutamatergic and GABAergic synapses. This jointly sustained plasticity underlying hippocampal excitability may be the cellular bases for KET + KGD treatment's suppression of AN relapses.
Traumatic brain injury (TBI) is a major cause of persistent cognitive and functional impairment. Citicoline has been investigated as a neuroprotective therapy after TBI, but its clinical efficacy remains uncertain. This single-center, triple-blind, randomized, placebo-controlled pilot trial evaluated the effects of citicoline on 90-day cognitive and functional outcomes in adults with mild-to-moderate TBI. Sixty patients were randomly assigned in a 1:1 ratio to receive citicoline 1000 mg/day or matched placebo for 90 days. The primary outcome was the Mini-Mental State Examination (MMSE) score at 90 days. Secondary outcomes were the Barthel Index and Glasgow Outcome Scale-Extended (GOS-E) scores at 90 days and safety outcomes. At 90 days, MMSE scores were higher in the citicoline group than in the placebo group (25.2 ± 2.1 vs. 21.5 ± 3.4; P < 0.001). The citicoline group also had higher Barthel Index scores (85.0 ± 0.5 vs. 70.0 ± 0.8; P < 0.001) and GOS-E scores (5.2 ± 1.3 vs. 3.6 ± 0.8; P < 0.001). New or exacerbated intracerebral hemorrhage occurred in two patients receiving citicoline and three receiving placebo (P = 0.65). These preliminary findings suggest that citicoline may be associated with better 90-day cognitive and functional outcomes after mild-to-moderate TBI. Confirmation in adequately powered multicenter randomized trials is required.
Intellectual disability (ID) is one of the most serious developmental disorders of children, characterized by significant limitations in intellectual function and adaptive behaviors. However, topological alterations of functional brain networks in children with ID remain unclear. This study combined resting-state functional MRI (rs-fMRI) data and graph theory analysis to quantify the topological properties of functional brain networks in children with ID. Thirty-four children diagnosed with ID and twenty-eight sex-, age-, and education-matched healthy controls (HC) were included. Global, modular and nodal characteristics were calculated to investigate ID-related alterations in functional brain networks. Compared with HCs, decreased clustering coefficient and local efficiency were revealed in children with ID, indicating reduced functional segregation. At the nodal level, children with ID exhibited nodal alterations concentrated in the ventral attention network (VAN), particularly in the left middle cingulate gyrus and bilateral insula, together with nodes in the dorsal attention network (DAN) and the subcortical network (SubN). At an uncorrected threshold, modular connectivity analysis revealed exploratory decreases of subcortical-cortical connectivity between SubN and the somato-motor network (SMN), between the SubN and VAN, between the SubN and frontoparietal network (FPN), and between the SMN and VAN, as well as altered connector and provincial hub indices across the default mode, limbic, frontoparietal, visual and subcortical networks in the ID group. Together, our findings provided preliminary multi-level evidence of functional network reorganization spanning subcortical, sensory, attention, and control systems in children with ID, providing new insights into the topological organization of the cognitive dysfunction in ID.
Aberrant dendritic changes in epilepsy, morphologically measured by the complexity and length of dendrite, as well as the density of dendritic spines, are dynamic and various considering disease course, developmental stages, and various neuronal type across different brain regions. Correspondingly, imbalanced excitatory and inhibitory synapses in epilepsy, quantitatively measured by the expression level and location of synaptic proteins, still vary considering etiology, disease course, and brain regions. And, various mechanisms involved in synaptic pruning add to the complexity of synaptic alterations in epilepsy. Still, sharing mechanisms underlying aberrant dendritic and synaptic plasticity in epilepsy exist, and cytoskeletal alterations can serve as a tool to decipher alterations of dendritic morphology and synaptic plasticity in epilepsy. Evidences supporting aberrant microtubule dynamics and post-translational modifications of tubulin in epilepsy emerge, combined with their involvement in dendritic morphology and synaptic plasticity. In conclusion, this review respectively summarizes the characteristics of dendritic pathology, synaptic alterations, microtubule dynamics, and tubulin's post-translational modifications in epilepsy, and discusses their interplay and correlation in epilepsy, providing insight into how cytoskeletal organization contributes to macroscopic changes in epilepsy.
Levodopa is the gold-standard treatment for Parkinson's disease, but chronic use results in levodopa-induced dyskinesia (LID) in approximately 80% of patients. Previous research has identified critical network changes in the cortico-basal ganglia pathway associated with LID; however, no study has extensively recorded motor thalamus single-cell activity in a LID model. The motor thalamus is a key node in the movement pathway, receiving basal ganglia and cerebellar inputs and projecting an integrated signal to the motor cortex, therefore could be a critical site for LID pathophysiology. The current study investigated single-cell and population motor thalamus activity following LID using extracellular electrophysiology. Activity was compared across control and parkinsonian rats, and then Parkinsonian rats that had or had not developed LID following chronic levodopa treatment. Baseline Mthal activity in the LID rat was hyperactive, bursty, and synchronised. Similar but weaker trends were found in Parkinsonian and non-dyskinetic rats. Following acute levodopa administration, motor thalamus activity changed greatly in the LID rat, showing increased firing rate and bursts, while decreasing oscillation strength, LTS burst number, and synchronisation. The non-dyskinetic rat showed opposing changes, hinting at a critical switch in motor thalamus activity associated with LID onset and not long-term levodopa treatment. These results show critical changes in Mthal activity following LID onset, whereby the motor thalamus becomes hyperactive and disorganised in LID. This study expands on previous observations of network changes in the corticobasal ganglia pathway, showing that motor thalamus activity contributes to LID pathophysiology.
Substance use disorder develops in a subset of individuals after repeated drug exposure and is characterized by escalating intake and loss of control over use. Rodent self-administration protocols with varying drug access durations are used to study controlled and compulsive drug use and its neurobiology. Drug-associated stimuli are major triggers for relapse. In humans, the craving response elicited by these stimuli progressively increases during abstinence, contributing to high relapse rates. In animals, an analogous process is observed, termed incubation of cocaine craving, which is characterized by a progressive increase in context-induced drug-seeking behavior over the course of abstinence. Associative learning between environmental stimuli and drug use is encoded by neuronal ensembles reactivated by drug-related stimuli, driving drug seeking. In the NAcc, these ensembles regulate cue-driven behaviors, while PV interneurons provide inhibitory control over network activity. We examined overall NAcc and PV-specific recruitment by measuring Fos expression in response to the cocaine-associated context after restricted or extended access self-administration and assessed whether this recruitment is modulated by forced abstinence. The magnitude of context-induced Fos expression in the NAcc core and shell before abstinence was influenced by the amount of cocaine intake during training, with greater expression in extended access rats. NAcc core and shell Fos expression increased after 30 days of abstinence in both groups but was consistently higher in extended access rats. PV Fos expression decreased in extended access rats following forced abstinence. These findings suggest that the Incubated cocaine craving alters NAcc and PV interneuron recruitment.
As traveling waves mediate neural information processing that depends on arousal states, we hypothesized that traveling waves would represent arousal states. To investigate this, we aimed to analyze brain-wide patterns of traveling waves during resting states using functional magnetic resonance imaging (rsfMRI). Specifically we focused on the temporal dynamics of the brain-wide patterns of rsfMRI traveling waves in association with arousal states. Using a publicly available dataset (N = 27), we extracted infra-slow (0.01-0.1 Hz) activity from blood oxygen level dependent (BOLD) signals. From infra-slow activity, we measured traveling waves via local phase gradient (LPG). Then, we clustered brain-wide LPGs through the k-means algorithm to extract brain-wide patterns of traveling waves, where each cluster represented each pattern. The optimal number of clusters was found to be 2. We analyzed the temporal dynamics of these clusters in relation to arousal states. During low-arousal states, brain-wide patterns of traveling waves featured in Cluster 1 appeared more often. During high-arousal states, those in Cluster 2 did so and lasted longer. As arousal increased, cluster dwell times shortened and transitions between clusters became more frequent, with the opposite pattern observed as arousal decreased. These findings suggest that brain-wide patterns of rsfMRI traveling waves reflect arousal states through their temporal dynamics, forming a syntax underlying arousal.
Classical psychedelics exert therapeutic effects on affective disorders, with serotonin 5-HT2A receptor activation thought to play a central role. The cellular mechanisms by which 5-HT2A signaling modulates neural circuits remain incompletely understood and may differ across mood-regulating brain regions. While psychedelic actions have been extensively studied in the medial prefrontal cortex, the hippocampus which is critical for mood, and memory has received less attention, and effects of selective 5-HT2A agonists on hippocampal neurons remain poorly characterized. Here, we examined the effects of the selective 5-HT2A agonist 25CN-NBOH on synaptic transmission, intrinsic excitability, and intracellular calcium in mouse hippocampal CA1 pyramidal neurons using whole-cell patch-clamp electrophysiology and Fura-2 AM imaging. 25CN-NBOH (10 μM) increased both spontaneous excitatory and inhibitory synaptic transmission, as indicated by elevated sEPSC and sIPSC frequency and amplitude, without affecting miniature events, suggesting action potential-dependent mechanisms. These synaptic effects persisted in the presence of the 5-HT2A antagonist MDL-100907, indicating 5-HT2A -independent synaptic facilitation. Despite increased synaptic drive, neuronal firing and action potential properties were unchanged, consistent with balanced excitation and inhibition. In contrast, 25CN-NBOH induced robust intracellular calcium elevations in CA1 neurons that were significantly reduced by MDL-100907, as well as by TTX and AMPA/NMDA receptor blockade, indicating dependence on both 5-HT2A receptor activation and glutamatergic presynaptic activity. Together, these findings reveal separable mechanisms of action: 5-HT2A-independent, presynaptic facilitation, and 5-HT2A-dependent calcium signaling. These results highlight distinct modes of hippocampal modulation by a selective 5-HT2A receptor agonist and suggest mechanisms through which psychedelics may promote plasticity-related processes.
Auditory attenuation of self-generated sounds is often attributed to motor prediction based on motor commands. Our previous study (Endo et al., 2025) demonstrated that somatosensory inputs also interact with auditory processing during a rhythmic finger tapping sound generation task. The current study further investigated the mechanism underlying this auditory-somatosensory interaction. We recorded auditory event-related potentials (ERPs) during a loudness discrimination task involving self-generated sounds, which were generated by discrete finger tapping movements pressing on a virtual object rendered by a haptic robotic device. Somatosensory input was manipulated by unpredictably varying the stiffness of the virtual object between low and high levels. In each trial, participants judged whether the comparison sound stimulus was louder than the standard stimulus, which was always associated with a discrete finger tapping movement. Also, ERPs elicited by the standard stimulus were compared across somatosensory conditions. At the behavioral level, no significant modulation of loudness perception was observed across somatosensory conditions. In the ERP responses, we did not observe the typical auditory attenuation effects in the N1 and P2 components that are commonly reported for self-generated sounds. In contrast, ERP activity after 300 ms was significantly modulated by somatosensory condition, consistent with our previous findings. Unlike our previous rhythmic finger tapping study, no somatosensory modulation was found in the auditory N1 component. These results indicate that somatosensory inputs consistently interact with the auditory processing of self-generated sounds, while the characteristics of this interaction were found to depend on task context.
Autism Spectrum Disorder (ASD) is a neurodevelopmental condition characterized by persistent challenges in social communication and the presence of restricted, repetitive behaviors. Neuroimaging and post-mortem research have consistently shown that ASD is associated with atypical brain connectivity and microstructural alterations. However, no singular neurobiological marker has been reliably established for diagnosis, and autism diagnoses depend primarily on behavioral evaluations, which may result in delayed or imprecise identification. The biological complexity and heterogeneity of ASD suggest that no single measure is likely to be sufficient, and instead calls for a multidimensional approach that integrates microstructural, neurochemical, and metabolic information. Advanced neuroimaging techniques, particularly ultra-high-field 7 Tesla MRI and phosphorus-31 magnetic resonance spectroscopy (31P MRS), offer promising avenues for detecting subtle neuroanatomical, neurochemical, and metabolic changes in autism. Because studies using 7 T scanners remain sporadic, this review combines a systematic search with a narrative synthesis to assess the current evidence. It brings together post-mortem histological findings with in-vivo ultra-high-field MR evidence across three domains: cortical layer connectivity, excitation-inhibition balance, and bioenergetics. Post-mortem studies establish layer-specific and E/I alterations, while emerging 7 T MRI and MRS findings begin to capture comparable cortical-layer, neurochemical, and bioenergetic differences in living individuals. Because these domains are interdependent, examining them together yields more complete insight than any one in isolation. By showing where these lines of evidence converge, this narrative review identifies directions toward a combination of biologically grounded markers of autism and improved understanding of its neurobiology.
Poststroke depression (PSD) is a prevalent emotional disorder following stroke and can emerge immediately or several months later. It represents the complex interaction between early brain injury-induced damage (immediate depression) and bodily and psychological stress related to neurofunctional deficits (delayed depression), potentially representing two distinct but consecutive pathological processes. Microglia, the brain's resident immune cells, undergo structural and functional changes during ischemic injury and social stress, triggering various pathophysiological processes. In ischemic brain injury, microglia contribute to blood‒brain barrier disruption and neuronal damage via excessive phagocytosis, synaptic pruning, and cytokine release. In the depression-only model, microglial dysfunction is characterized by stress-induced dyshomeostasis, pro-inflammatory responses, neuronal phagocytosis, and even microglial decline, some of which are linked to gut microbiota dysbiosis. Ischemic brain injury may induce persistent alterations in microglial homeostasis, potentially increasing vulnerability to subsequent stress exposure. This proposed microglial sensitization model provides a conceptual framework for understanding delayed PSD, although direct causal evidence remains insufficient. Thus, PSD aligns more with "microgliopathy" than with depression-only. This review provides an introductory overview of "microglial and depression-only," as it may partially align with the characteristics of delayed PSD. The review also highlights potential antidepressant strategies targeting microglia. We propose that multitarget, optimized combination therapies targeting microglial states via various pathological pathways may represent promising approaches to overcome the limitations of traditional PSD treatments. Nonetheless, current challenges include the complexity of PSD modeling and the lack of multiomics studies targeting microglial function, which hinders a precise understanding of microglial roles.
The involvement of microglia (MG) in neuroinflammatory responses has been demonstrated extensively. Recent studies have suggested that dietary restriction suppresses microglial activity; however, the relationship between food deprivation (FD)-induced hypothermia and microglial alterations remains unclear. In the present study, we examined the effects of FD on body temperature (Tb) and Iba1-immunoreactive (Iba1-ir) microglial cells in the hippocampus, thalamus, hypothalamus, and cerebral cortex of rats. Tb was monitored throughout the experimental period, and microglial morphology, cell number, and their correlation with Tb were evaluated following up to 3 days of FD. The results demonstrated that (1) FD significantly reduced Tb during the dark phase, but not during the light phase; (2) FD also decreased microglial density and induced morphological changes consistent with reduced microglial activation in all brain regions examined; (3) there was a positive correlation between Tb and microglial densities in brain regions, except for the cerebral cortex, which showed no significant correlation. These findings demonstrate that FD suppresses microglial activation in multiple brain regions in association with reduced Tb, suggesting that metabolic adaptation to FD modulates neuroimmune homeostasis and may contribute to mechanisms that attenuate neuroinflammation.
Alzheimer's disease (AD) is a common neurodegenerative disease characterized by severe cognitive dysfunctions and brain disorders. The emergence of deep learning methods provides a feasible way to develop effective representations for clinical diagnosis of brain diseases. In this study, we proposed a deep manifold representation learning network to characterize the alteration of functional brain connectivity for AD detection. The proposed framework used a convolutional autoencoder (CAE) to infer low-dimensional manifold representations of functional brain networks reconstructed by phase correlation analysis of electroencephalogram (EEG) signals. To enhance the global associations among latent features, a Transformer-based manifold regularization module was designed to optimize the structure of the learned low-dimensional manifold representations. Experimental results with EEG data showed that CAE-Transformer model could map high-dimensional functional brain networks into low-dimensional representations on Grassmann manifold and improved intra-group similarity and inter-group diversity. Compared with other advanced methods, CAE-Transformer demonstrated better performance in AD detection, achieving an average accuracy of 98.50% and a sensitivity of 99.49% under subject-wise five-fold cross-validation on a single-center EEG dataset containing 2800 samples derived from segmented recordings of 51 participants. The low-dimensional latent representations learned by CAE-Transformer preserved AD-related brain connectivity patterns. Moreover, functional brain networks in the alpha frequency band showed the highest diagnostic performance for AD detection. Our findings showed the variation of brain networks in AD and provided an efficient method for identifying AD patients.
