Neurobiology of Disease最新文献

Neurodegenerative diseases represent a major and growing clinical challenge due to their progressive nature, biological heterogeneity, and limited therapeutic options. Recent advances in artificial intelligence (AI) have introduced new analytical strategies for extracting clinically relevant information from complex biomedical data, offering complementary tools to established diagnostic and research approaches. This review provides a critical and method-comparative synthesis of AI applications in neurodegenerative diseases, with emphasis on studies published between 2022 and 2025. Rather than cataloging algorithms, the review evaluates how specific AI methodologies are selected, implemented, and validated across diverse data modalities, including molecular profiles, neuroimaging, biosensors, speech, gait, and electronic health records. Across Alzheimer's disease, Parkinson's disease, and other neurodegenerative disorders, the reviewed evidence indicates that AI-based models can support early risk stratification, disease characterization, and monitoring when applied within clearly defined analytic and clinical contexts. Importantly, performance gains are shown to depend strongly on data quality, feature representation, validation design, and alignment between model architecture and biological signal, rather than on algorithmic complexity alone. Emerging paradigms, including multimodal integration and next-generation AI frameworks, are discussed in relation to their methodological contributions rather than clinical readiness. By systematically comparing analytical strategies and highlighting sources of variability across studies, this review underscores the importance of methodological transparency, uncertainty-aware evaluation, and biological interpretability. Collectively, the work positions AI as an enabling and adjunctive analytical framework that can enhance neurodegenerative disease research and clinical decision support when deployed with rigor and caution, providing a balanced perspective on current capabilities and future directions.

Myotonic dystrophy type 1 (DM1) arises from toxic CUG-expanded DMPK transcripts that sequester Muscleblind-like (MBNL) proteins, yet how this molecular lesion perturbs brain development in congenital DM1 (CDM) remains unknown. Here, we identify an unanticipated developmental role for MBNL2 in outer radial glial cells, a progenitor population critical for cortical expansion. We demonstrate that MBNL2 is expressed in these cells both in vivo and in forebrain organoids derived from patient-specific human induced pluripotent stem cells (hiPSCs), rendering them particularly sensitive to MBNL2 titration. Using genome editing to excise the CTG repeats in the DMPK gene, we provide evidence that the expanded trinucleotide tract directly contributes to defective neuronal migration and impaired differentiation of late-born cortical neurons in CDM organoids. These findings redefine MBNL2 as a potential regulator of human corticogenesis and uncover a developmental mechanism by which RNA toxicity drives this severe form of DM1. By uncovering a prenatal origin for CDM neuropathology linked to MBNL2 dysfunction, this work opens avenues for therapeutic strategies targeting early developmental windows.

Fragile X-associated tremor/ataxia syndrome (FXTAS), caused by the FMR1 premutation allele, is associated with brain degeneration, yet the mechanisms behind this neurodegeneration still need to be elucidated. Apoε polymorphism has been widely implicated in brain aging in cognitively healthy individuals and brain deterioration in Alzheimer's disease. This study aimed to examine the interaction of Apoε genotypes, FXTAS clinical symptoms, FMR1 molecular measures, and age, towards brain pathophysiology and cognitive functions. This longitudinal study includes MRI data collected from 205 male premutation carriers with and without FXTAS clinical symptoms and compared to 86 healthy male controls aged 40-85 years. The investigation includes FXTAS-related brain volumes, IQ, self-control behaviors, FMR1 molecular measures, and Apoε genotypes. In carriers with FXTAS, the presence of the Apoε2 allele showed a possible association with more favorable neuroimaging markers, such as reduced white matter hyperintensities, and lower incidence of the middle cerebellar peduncle sign, patterns that were not observed in carriers without FXTAS. Specifically, the presence of Apoε2 allele exhibited a potential protective effect on brain degeneration, and cognitive functions among FXTAS patients; on the contrary, the Apoε4 allele was associated with a worsening of brain volume and brain degeneration in carriers with no FXTAS symptoms. The identification of Apoε genotypes in FMR1 premutation carriers before any clinical symptoms of FXTAS are observed may improve symptomatic management leading to better outcomes for these individuals.

Outcome prediction after traumatic spinal cord injury (SCI) remains challenging due to patient heterogeneity, highlighting the need for better prognostic tools. Neural tissue damage and blood-spinal cord barrier disruption expose the immune system to spinal cord proteins, eliciting autoantibody responses that may be beneficial or detrimental. This study aimed to identify the (auto)antibody profile of SCI patients, and examine the prognostic antibody biomarker potential. A healthy and a SCI cDNA phage display library were screened for novel antibodies using SCI samples (n = 12/11). Antibody reactivity was validated using phage ELISA in 291 samples from 190 SCI patients collected at baseline (0-4 days post-injury [dpi]) and follow-up (15-30 dpi; 31-56 dpi). Correlations between antibody reactivity and clinical characteristics including SCI level, and American Spinal Injury Association impairment scale (AIS), were analysed. Immunofluorescent stainings were used to validate expression of two antigenic targets. We identified antibodies against 6 novel autoantigens (University Hasselt [UH.] SCI.104/105/106/108/109/110). A panel of three antigens (UH.SCI.104/109/110) demonstrated increased antibody reactivity in 31.3% of SCI patients with AIS improvement versus 4.8% with no improvement, resulting in a positive likelihood ratio of 6.56. Patients with injuries above thoracic level 4 had significantly lower antibody reactivity against UH.SCI.105/110 compared to patients with lower lesions. Anti-UH.SCI.108/110 antibodies bound to astrocytes, in mouse spinal cord tissue and primary cell cultures, confirming disease-relevant reactivity. Antibodies targeting the novel antigens demonstrated prognostic biomarker potential, supporting their future use in outcome prediction and patient stratification for SCI management and clinical trial design.

Heterogeneity in the severity of Parkinson's disease (PD) inhibits the effective interpretation of clinical trial outcomes. Multi-omics analysis may help explain the pathological mechanisms underlying disease progression and reveal biomarkers of clinical severity. We performed Multi-Omics Factor Analysis (MOFA) on whole blood RNA, miRNA and cerebrospinal fluid (CSF) and blood plasma proteomics from the Parkinson's Progression Marker Initiative (PPMI), to identify molecular factors correlated with motor (MDS-UPDRS3) and cognitive (Semantic Fluency Test, SFT) function. Three molecular factors significantly correlated with the MDS-UPDRS3 score and two with SFT, which remained significant after adjusting for age, sex, and medication dose. We used the identified factors to stratify patients into subgroups with distinct motor and cognitive severity. The severe motor clusters showed deregulation of cytotoxic natural killer cell mechanisms in peripheral blood, and changes to proteins associated with the endoplasmic reticulum and dense core vesicle in CSF. The severe cognitive clusters showed changes in the complement system and synaptic dysfunction. Our analysis capitalizes on multi-omics data integration to enrich our understanding of the mechanisms driving motor and cognitive decline in PD, to support precision medicine.

Anxiety exacerbates symptoms in various psychiatric disorders. In conditions such as obsessive-compulsive disorder (OCD) or Tourette syndrome, anxiety intensifies stereotypic and repetitive behaviors. Rodent self-grooming, a structured, repetitive innate behavior, serves as an effective rodent platform for studying these behaviors in neuropsychiatric research. Anxiety is also linked to altered functioning of the dopamine (DA) system, particularly within the substantia-nigra pars compacta (SNc), the main DA source to the dorsal striatum through the nigro-striatal pathway. Striatal modulation by DA signal also plays a complex role in repetitive behaviors and OCD-like symptoms, suggesting this system as linking anxiety to the induced exacerbation of repetitive behavior. In the present study, we observed several long-term effects of anxiety-inducing foot shock on grooming behavior. Recording single unit neuronal activity in the SNc revealed distinct response patterns related to grooming behavior with changes in the magnitude and timing following the shock treatment. Notably, DA neurons of different nigro-striatal pathways demonstrated different changes in different response pattern type units. DA neurons projecting to the dorsolateral striatum (DLS) showed increase, while those targeting the dorsomedial striatum (DMS) exhibited decrease in transient activity - suggesting a shift in cortico-striatal circuitry of behavioral control. These neural changes were correlated with the observed behavioral alterations following adversity. Furthermore, targeted stimulation of SNc DA neurons projecting to the DLS rescued the anxiety-induced behavioral effects, highlighting the critical role of the nigro-striatal pathway to the DLS in mediating the interaction between anxiety and repetitive behaviors, thus offering future direction for mitigation of relevant psychiatric symptoms.

Dopaminergic neurons (DANs) exhibit subtype-specific vulnerability in Parkinson's disease (PD), but the molecular basis of selective resilience remains poorly understood. Here, we investigated the role of neuropeptide receptor (NPR) signaling in DAN survival using single-nucleus RNA sequencing of 8065 DANs from postmortem substantia nigra of individuals with PD and matched controls. Despite pronounced neuronal loss in PD, surviving DANs showed a higher NPR transcript burden and increased NPR gene co-expression per cell. Using a 25-gene NPR score, we stratified DANs into high-, mid-, and low-NPR tiers and identified distinct molecular signatures. High-NPR DANs exhibited increased expression of the resilience marker CALB1 and an inverse correlation with PD genetic-risk signals. Low-NPR DANs preferentially expressed vulnerability markers (SOX6, ALDH1A1, AGTR1) and displayed a positive association with PD genetic risk. Notably, Mitochondrial Complex I subunits (NDUFS2, NDUFB10) were relatively enriched in low-NPR DANs at baseline and were further reduced in PD specifically within this tier. Among dopaminergic subtypes, the PD-susceptible SOX6_AGTR1 neurons displayed minimal NPR activity, while resilient CALB1⁺ subtypes showed elevated NPR signaling. Moreover, SOX6_AGTR1 neurons preferentially expressed the transcription factors PGR and CLOCK, but the expression of both factors was significantly reduced in PD within this subtype, with no significant change in CALB1⁺ subtypes. Integrating these findings with genome-wide association study (GWAS) enrichment and external datasets, we identified PRLR and CRHR1 as key mediators of dopaminergic resilience, highlighting their potential as targets for neuroprotective therapy. Together, our data implicate NPR signaling as a molecular correlate of dopaminergic resilience in PD and highlight specific receptor pathways for therapeutic development.

Charcot-Marie-Tooth disease (CMT) is one of the most prevalent inherited peripheral neuropathies. CMT type X1 (CMTX1), caused by mutations in the GJB1 gene, represents the most common X-linked subtype with central nervous system (CNS) involvement. Here, we report the identification and functional characterization of a novel GJB1 variant (c.554C > T, p.Thr185Ile) in a CMTX1-affected family and its pathogenic impact using patient-derived induced pluripotent stem cells (iPSCs) and three-dimensional (3D) neural organoid models. The GJB1 gene encodes connexin 32 (Cx32), a gap junction protein. Immunofluorescent analysis revealed aberrant intracellular reduction and aggregation of the mutant Cx32 protein, suggesting impaired gap junction function. iPSC-derived neural organoids carrying the GJB1 mutation exhibited significant delay in neural differentiation and disrupted neural rosette organization. These findings underscore the critical role of Cx32 in neural development and provide a physiologically relevant platform for underlying CMTX1 pathological mechanisms on central nervous system. The established GJB1-variant organoid model holds promise for investigating genotype-phenotype correlations and facilitating the development of targeted therapeutic strategies for CMTX1.

Fear-related disorders, characterized by inappropriate learned fear and resistance to extinction, are among the most prevalent psychiatric conditions. Potential modifications in neurogenesis and brain activity were studied as possible individual factors associated with the development of these pathologies. By using Pavlovian fear conditioning and extinction paradigm, male and female mice were categorized in resilient and susceptible phenotypes based on their individual fear extinction behavior. Increased neurogenesis, as revealed by higher expression of the early neuronal marker doublecortin in the subgranular zone of the dentate gyrus, was observed in susceptible male and female mice. This result suggests the existence of a compensatory mechanism given that the DNA-alkylating agent temozolomide induced an impairment of fear extinction and a reduction of neurogenesis in male mice. The use of c-Fos immunofluorescence revealed several brain regions that were differently activated in susceptible animals, although these differences were less evident in female mice. Categorization by k-means clustering based on c-Fos labelling was significantly associated with phenotype of extinction in male, but not female, animals. Pairwise Pearson correlations between brain regions showed that resilience and susceptibility to fear extinction are related to divergent circuit-level reorganizations. These findings reveal new individual factors involved in the variability of fear extinction response which could be of interest for the development of future therapeutic strategies.