Abstract

   Demyelination in the central nervous system (CNS) disrupts neuronal
   communication and promotes neurodegeneration. Despite the widespread
   use of cuprizone-induced demyelination models to study myelin injury
   and repair, the mechanisms underlying oligodendrocyte apoptosis and
   regeneration are poorly understood. This study investigated the dynamic
   cellular and molecular changes that occur during demyelination and
   remyelination, with a focus on glial cell responses, blood-brain
   barrier (BBB) integrity, and neuroimmune interactions. C57BL/6J mice
   exposed to cuprizone exhibited weight loss, sensorimotor deficits, and
   cognitive decline, which were reversed during remyelination.
   Histological and immunofluorescence analyses revealed reduced myelin
   protein levels, including myelin basic protein (MBP) and
   myelin-associated glycoprotein (MAG), and decreased oligodendrocyte
   populations during demyelination, with recovery during repair. The BBB
   permeability increases during demyelination, is associated with the
   decreased expression of tight junction proteins (ZO-1, Occludin), and
   normalizes during remyelination. Single-cell RNA sequencing revealed
   dynamic shifts in glial cell populations and upregulated Psap-Gpr37l1
   signaling. Neuroimmune activation and oxidative stress peak during
   demyelination, characterized by elevated ROS, MDA, and immune cell
   infiltration, followed by recovery. Transcriptomic profiling revealed
   key inflammatory pathways (JAK-STAT, NF-κB) and hub genes associated
   with demyelination and repair. These findings provide insights into
   myelin repair mechanisms and highlight potential therapeutic targets
   for treating demyelinating diseases.

   Keywords: cuprizone, demyelination, remyelination, blood–brain barrier
   integrity, glial cell, neuroinflammation, snRNA-seq, RNA-seq

1. Introduction

   Myelin within the central nervous system is a structurally intricate,
   multilayered membrane that tightly ensheathes nerve fibers, providing
   electrical insulation and support to ensure the efficient propagation
   of nerve impulses [[34]1]. Primarily synthesized by oligodendrocytes,
   myelin is rich in lipids and specific proteins and is distributed
   across various regions of the brain and spinal cord [[35]2,[36]3]. When
   the myelin integrity is compromised, the resulting loss of axonal
   insulation not only impedes nerve impulse transmission but also
   triggers irreversible neurodegenerative processes, forming the core
   pathological mechanism in neurological diseases such as multiple
   sclerosis [[37]4,[38]5]. Although hereditary disorders of myelin
   development are relatively rare [[39]6], the interplay of environmental
   pathogens [[40]7], metabolic dysregulation [[41]8], and immune
   dysfunction [[42]9] constitutes the major contributing factor to the
   more prevalent acquired demyelinating lesions observed in clinical
   practice.

   Cuprizone is commonly employed in experimental models to induce
   demyelination by disrupting the mitochondrial function within
   oligodendrocytes, leading to energy metabolism imbalance and the
   activation of apoptotic pathways [[43]10,[44]11]. The apoptosis of
   oligodendrocytes results in unavoidable myelin degradation,
   recapitulating clinical demyelinating disease features [[45]12,[46]13].
   Upon the cessation of cuprizone administration, endogenous repair
   mechanisms are activated, triggering remyelination, thus providing an
   ideal model for studying injury and repair processes [[47]14]. However,
   traditional feeding methods are limited in their precise experimental
   designs because of significant fluctuations in intake and potential
   systemic toxic side effects [[48]15,[49]16]. The adoption of
   intragastric administration methods enables accurate dose titration and
   improved reproducibility, thereby advancing the study of the critical
   molecular mechanisms underlying demyelination and remyelination.

   Glial cells play pivotal roles in regulating injury responses and
   repair processes following CNS damage [[50]17]. Microglia are swiftly
   activated to phagocytose myelin debris [[51]18], whereas reactive
   astrocytes proliferate and secrete a range of cytokines and growth
   factors, creating a multifaceted inflammatory milieu [[52]19]. Recent
   studies have revealed profound metabolic reprogramming in glial cells
   under stress conditions, and that microglia shift toward glycolysis to
   meet phagocytosis energetic demands [[53]20], whereas astrocytes
   restructure lipid metabolism to mitigate oxidative stress [[54]21].
   This metabolic reprogramming may critically influence the outcome of
   myelin regeneration. Furthermore, fibrinogen deposition resulting from
   BBB leakage has been identified as a factor that amplifies inflammation
   through the activation of innate immune receptors, potentially
   establishing a detrimental feedback loop that hinders repair processes
   [[55]21,[56]22].

   In this study, CPZ was administered via gavage to achieve precise
   dosing in demyelination and remyelination experimental models and to
   assess the BBB permeability in treated mice. By integrating
   histological examination with RNA sequencing, a comprehensive analysis
   of the transcriptomic characteristics across various stages of
   demyelination and remyelination was conducted, aiming to elucidate the
   key molecular mechanisms underlying oligodendrocyte apoptosis and
   regeneration.

2. Methods

2.1. Mice and Experimental Model

   Male C57BL/6J mice, aged 10 weeks, were procured from the Animal
   Medical Department of Kunming Medical University (SYXK(DIAN)K2020-006).
   The mice were housed under specific pathogen-free (SPF) conditions with
   a 12 h light/dark cycle at a density of four per cage. All experimental
   procedures received approval from the Animal Ethics Committee of
   Kunming Medical University (Protocol Number: kmmu20230296).
   Demyelination was induced by daily oral gavage of 0.5% cuprizone (500
   mg/kg) suspended in corn oil for four consecutive weeks. Control mice
   received the same volume of corn oil on the same schedule. Following
   the 4-week treatment period, the cuprizone administration ceased to
   allow spontaneous remyelination. Throughout the experiment, all the
   animals had ad libitum food and water. Mice from the control,
   demyelination, and remyelination groups were sacrificed concurrently
   ([57]Figure 1A).

Figure 1.

   [58]Figure 1
   [59]Open in a new tab

   Alterations in sensorimotor and cognitive functions across
   demyelination and remyelination in cuprizone-induced mouse model. (A)
   Experimental timeline of cuprizone-induced mouse model. (B) Body weight
   measurements of cuprizone-administered mice at various time points. (C)
   Evaluation of short-term memory changes during demyelination and
   remyelination via novel object recognition test (NOR). (D) Behavioral
   trajectory plots. All p values were calculated via one-way ANOVA,
   followed by Tukey’s multiple comparison test.

2.2. Neurobehavioral Tests

   Behavioral assessments were conducted to evaluate neurological function
   changes in the mice during the cuprizone-induced demyelination and
   remyelination phases. Sensorimotor deficits were assessed via the tape
   removal test and foot fault test following established protocols
   [[60]23]. A 4 × 3 mm adhesive tape was applied to the hairless region
   of the forepaw, and the time taken for the mouse to touch and remove
   the tape was recorded. If the mouse failed to touch or remove the tape
   within 120 s, the time was recorded as 120 s. The foot fault test was
   used to evaluate sensorimotor coordination during spontaneous movement.
   The mice were placed on an elevated grid 30 cm above the ground and
   were allowed to walk freely for 1 min. The total number of steps and
   foot faults was recorded.

   The novel object recognition (NOR) test was employed to assess the
   learning and memory in the rodents [[61]24]. The mice were tested via a
   two-phase procedure: (1) Training phase: Mice freely explored two
   identical objects in a square open field for 10 min, followed by a 10
   min rest period. (2) Testing phase: One object was replaced with a
   novel object, and the mice explored both objects for 10 min. The
   preference index was calculated as the ratio of the time spent
   exploring the novel object to the total exploration time [[62]25].

2.3. Luxol Fast Blue Staining

   For the histopathological evaluation, deep anesthesia was induced via
   sodium pentobarbital (0.3%, 40 mg/kg), followed by transcardial
   perfusion in PBS. The brain tissues were carefully harvested and fixed
   in 4% paraformaldehyde before being embedded in paraffin. Serial 5 μm
   thick sections were prepared for subsequent staining. The degree of
   demyelination was assessed via Luxol Fast Blue (LFB) (L0294,
   Sigma-Aldrich, St. Louis, MO, USA) staining. The stained sections were
   examined under a light microscope (Nikon ECLIPSE 80i, Nikon
   Corporation, Tokyo, Japan), and digital images were captured for
   further analysis.

2.4. Evans Blue Permeability Assay

   To evaluate the BBB integrity, a 2% Evans blue (EB) solution was
   injected retro-orbitally and was allowed to circulate for 2 h. Deep
   anesthesia was then induced via sodium pentobarbital, followed by
   transcardial perfusion with PBS for brain tissue extraction.
   Whole-brain EB infiltration was documented via a high-resolution
   digital imaging system. For quantitative analysis, brain tissues were
   homogenized in 500 μL of 50% (w/v) trichloroacetic acid via an
   ultrasonic homogenizer (HD4400, WIGGENS, Straubenhardt, Germany) in an
   ice bath. The homogenate was incubated at 4 °C for 24 h and centrifuged
   at 13,000× g for 30 min. The supernatant was collected, and the
   absorbance at 620 nm was measured via a multimode microplate reader
   (Infinite M200 PRO, Tecan, Tecan, Männedorf, Switzerland). The
   concentration of extravasated dye was determined via a standard curve
   and expressed as μg/g of brain tissue.

2.5. FITC–Dextran Tracer Assay

   A 200 μL solution of FITC-labeled dextran (10 kDa, 1 mg/mouse) was
   administered via retro-orbital venous plexus injection, allowing
   systemic circulation for 2 h. The brain tissues were carefully
   extracted without transcardial perfusion and were subsequently fixed in
   4% paraformaldehyde. The samples were sequentially dehydrated in a 30%
   (w/v) sucrose solution, embedded in OCT compound, and cryosectioned at
   −20 °C. Coronal sections (30 μm) were prepared via a cryostat and
   mounted onto slides pretreated with 0.05% (w/v) poly-L-lysine. The
   sections were sealed with an antifade mounting medium to preserve the
   fluorescence signals. Imaging was performed via a laser scanning
   confocal microscope (LSM880, Zeiss, Oberkochen, Germany), and the
   fluorescence intensity was quantified via the ZEN 3.0 imaging analysis
   system.

2.6. Immunofluorescence Staining

   Following transcardial perfusion in ice-cold PBS, brain tissues were
   carefully extracted and fixed in 4% paraformaldehyde (PFA, pH 7.4) for
   24 h. The samples were cryoprotected in 30% (w/v) sucrose, embedded at
   the optimal cutting temperature (OCT), and frozen at −20 °C. Coronal
   sections (20 μm) were prepared via a cryostat. After being washed in
   PBS, the sections were permeabilized and blocked for 2 h at room
   temperature in PBS containing 5% donkey serum and 0.3% Triton X-100.
   Primary antibody incubation was performed overnight at 4 °C in PBS
   supplemented with 5% donkey serum and 0.3% Triton X-100. The primary
   antibodies used included anti-MAG (1:300, [63]AB277535, Abcam,
   Cambridge, UK), anti-MBP (1:500, AB7349, Abcam, Cambridge, UK),
   anti-PDGFRα (1:100, AF1062, R & D Systems, Minneapolis, MN, USA),
   anti-CC1 (1:500, [64]AB239828, Abcam), anti-Olig2 (1:200, PRB009, Oasis
   Biofarm, Hangzhou, China), anti-ZO-1 (1:100, [65]AB307799, Abcam),
   anti-Occludin (1:100, 80545, Proteintech, Wuhan, China), anti-GFAP
   (1:500, AB32760, Abcam), and anti-IBA1 (1:500, AB32760, Abcam)
   antibodies. After washing, the sections were incubated with secondary
   antibodies at room temperature for 1 h. The secondary antibodies used
   included Alexa Fluor 488-conjugated anti-rabbit IgG (1:500, A-21206,
   Thermo Fisher, Waltham, MA, USA), Alexa Fluor 488-conjugated anti-mouse
   IgG (1:500, A-28175, Thermo Fisher, MA, USA), Alexa Fluor
   594-conjugated anti-rabbit IgG (1:500, A-11037, Thermo Fisher, MA,
   USA), Alexa Fluor 594-conjugated anti-mouse IgG (1:500, A-11032, Thermo
   Fisher, MA, USA), and Alexa Fluor 594-conjugated anti-rat IgG (1:500,
   A-21208, Thermo Fisher, MA, USA). The sections were counterstained and
   mounted onto a DAPI-containing antifade mounting medium. Fluorescence
   imaging was performed via a laser scanning confocal microscope (LSM880,
   Zeiss), and quantitative analysis of the fluorescence intensity was
   conducted via ZEN 3.0 imaging software.

2.7. Flow Cytometry

   Brain mononuclear cells were isolated via Percoll density gradient
   centrifugation (17-0891, GE Healthcare, Chicago, IL, USA). Following Fc
   receptor blockade, the cell suspensions were incubated with
   fluorescence-conjugated monoclonal antibodies at 4 °C for 30 min in the
   dark. The antibodies used included anti-mouse CD45-PerCP (103130,
   BioLegend, San Diego, CA, USA), anti-CD3-FITC (100204, BioLegend, CA,
   USA), and anti-F4/80-APC (123115, BioLegend, CA, USA). After staining,
   the cells were washed twice in PBS containing 2% fetal bovine serum
   (FBS) and were resuspended in 300 μL of PBS. Flow cytometry was
   performed using a FACSCanto II (BD Biosciences, CA, USA) instrument
   equipped with 488 nm and 640 nm lasers. The data were analyzed via
   FlowJo software (version 10.1), and the representative gate strategies
   are presented in [66]Supplementary Figure S1A.

2.8. Western Blot

   Total protein was extracted from tissue samples via RIPA lysis buffer
   (P0013B, Beyotime, Shanghai, China), while membrane proteins were
   isolated via a dedicated extraction kit (P0033, Beyotime, Shanghai,
   China), with protease inhibitors added to prevent protein degradation.
   The protein concentration was quantified via a BCA protein assay kit
   (P0011, Beyotime, Shanghai, China). Equal amounts of protein samples
   were subjected to sodium dodecyl sulfate-polyacrylamide gel
   electrophoresis (SDS-PAGE) and transferred to polyvinylidene fluoride
   (PVDF) membranes (Millipore, Burlington, MA, USA). The membranes were
   incubated with primary antibodies, including anti-MAG (1:2000,
   [67]AB277535, Abcam), anti-MBP (1:2000, AB7349, Abcam), anti-ZO-1
   (1:1000, [68]AB307799, Abcam), anti-Occludin (1:1000, 80545,
   Proteintech, Wuhan, China), anti- NF-κB p65 (1:2000, 8242T, Cell
   Signaling Technology), and anti- Phospho-NF-κB p65 (1:2000, 3033T, Cell
   Signaling Technology, Danvers, MA, USA) antibodies. HRP-conjugated
   secondary antibodies were subsequently applied, and protein bands were
   visualized via a chemiluminescence imaging system. Band intensities
   were quantified with ImageJ software, version 1.54.

2.9. Quantitative Real-Time PCR (qRT-PCR)

   Total RNA was extracted from the cortical and corpus callosum tissues
   using the TRIzol reagent (15596018, Thermo Fisher Scientific, Waltham,
   MA, USA). The RNA was then reverse-transcribed into cDNA via the
   PrimeScript RT reagent kit (TK1621, Thermo Fisher Scientific), with 1
   μg of total RNA used as input. Quantitative real-time PCR (qRT-PCR) was
   performed on an ABI Q6 Real-Time PCR System (Applied Biosystems, Foster
   City, CA, USA) with SYBR Green Master Mix (Roche) and specific primers
   (sequences are provided in [69]Supplementary Table S1). The reaction
   conditions were set as follows: 1 min of predenaturation at 95 °C,
   followed by 40 cycles of 5 s of denaturation at 95 °C and 30 s of
   annealing/extension at 60 °C. The gene expression was quantified by the
   2-ΔΔCq method.

2.10. Oxidative Stress-Related Factor Detection

   The brain tissue was minced and filtered through 70 μm mesh to obtain a
   single-cell suspension. The levels of intracellular reactive oxygen
   species (ROS) were quantified via a DCFH-DA Flow Cytometry Kit (S0033S,
   Beyotime, Shanghai, China) via flow cytometry. Brain tissue was
   homogenized in PBS, and supernatant was collected. The malondialdehyde
   (MDA) content and superoxide dismutase (SOD) activity were determined
   via the thiobarbituric acid method (S0131S, Beyotime, Shanghai, China)
   and the WST-1 assay (A001-3-2, Nanjing Jiancheng Bioengineering
   Institute, Nanjing, China) kits, respectively, according to the
   manufacturers’ instructions. The data were normalized to the total
   protein concentration measured via the BCA method.

2.11. scRNA-seq and Data Mining

   scRNA-seq count data from the cortex and corpus callosum of normal and
   cuprizone-induced demyelinated mice were downloaded from the GEO
   database (GEO accession number: [70]GSE204769). A standard Seurat
   analysis pipeline was employed to construct Seurat objects, with a more
   stringent mitochondrial content threshold (5%) applied during data
   filtering. The data were log-normalized via the normalizeData function,
   and the mitochondrial gene percentages were regressed during data
   scaling. Principal component analysis (PCA) identified the top 3000
   most variable genes, whereas UMAP dimensionality reduction utilized the
   first 20 principal components. Clustering analysis was performed with a
   resolution of 0.3. Manual inspection and detailed annotations were
   subsequently performed on the basis of widely accepted cell type
   markers and annotations from the original publication’s authors
   [[71]26]. Cell-cell communication networks were inferred via the
   CellChat and CellPhoneDB tools on the basis of single-cell
   transcriptomics data. Briefly, expression data were loaded, and a
   CellChat object was created using the Seurat object. Ligand-receptor
   databases were then loaded and filtered for information related to
   “secretory signaling,” and overexpressed genes and their interactions
   in each cell type were identified. Intercellular communication
   probabilities were calculated, and communication signals involving
   fewer than 10 cells were excluded. Pathway-level communication networks
   were then inferred. Differentially expressed genes (DEGs) in
   subclusters from different experimental groups were visualized via
   volcano plots. Pseudobulk analysis was applied to the DEG analyses. The
   thresholds for DEG selection were an absolute log fold change (FC)
   greater than 0.585 and a p value less than 0.05. The resulting gene
   list was used as input for the subsequent pathway analysis. A Venn
   diagram was generated via InteractiVenn
   ([72]http://www.interactivenn.net/ (accessed on 5 October 2024)) to
   visualize the overlaps and relationships between the DEGs. To further
   explore the functional significance of the overlapping DEGs, KEGG
   pathway enrichment analysis was performed via the clusterProfiler
   package, and significantly enriched pathways (p < 0.05) were visualized
   via a bubble plot.

2.12. Bulk Transcriptome Analysis

   Total RNA was extracted from cortical and whole-brain tissue samples
   via TRIzol reagent (15596018, Thermo Fisher Scientific, MA, USA).
   Sequencing libraries were constructed with the Illumina TruSeq Stranded
   Total RNA Sample Prep Kit, ensuring that the input RNA quantity met the
   standard quality control threshold of ≥1 μg. Paired-end 150 bp
   sequencing was performed on the Illumina NovaSeq 6000 platform.
   Following quality assessment, the raw sequencing reads were aligned to
   the C57BL/6J mouse reference genome (Ensembl release 100) via HISAT2
   (v2.2.1). Differentially expressed genes (DEGs) were identified across
   groups and samples via the DESeq2 package, with a significance
   threshold set at |log2FC| > 1 and p < 0.05.

   To further elucidate the gene coexpression network characteristics, a
   scale-free topology network was constructed via weighted gene
   coexpression network analysis (WGCNA). Coexpression modules were
   identified through a dynamic tree-cut algorithm, and correlations
   between module eigengenes and phenotypic traits were computed. Gene
   Ontology (GO) functional annotation and KEGG pathway enrichment
   analyses were performed on genes within significant modules and DEGs.
   Overlapping genes from the DEGs and key module genes were imported into
   the STRING database to construct a protein-protein interaction (PPI)
   network, applying a confidence threshold of >0.7. The network topology
   was visualized via Cytoscape, version 3.10.1, and the hub genes were
   identified via the cytoHubba algorithm (Degree).

2.13. Statistical Analysis

   All the experiments were performed at least three times, and the data
   are presented as the means ± standard deviations (SDs). Statistical
   analysis of the validation experiments was conducted via Prism 8.0
   software. One-way analysis of variance (ANOVA) followed by Tukey’s
   multiple comparison test was applied to assess the statistical
   significance, with a predefined significance threshold of α = 0.05. For
   omics data, the Benjamini–Hochberg method was applied for corrections
   to ensure the stability of the differentially expressed gene
   identification and pathway enrichment results.

3. Results

3.1. Neurological and Motor Functions in Cuprizone-Induced Mice

   To evaluate the progressive effects of cuprizone-induced demyelination
   and remyelination on cognitive and sensorimotor functions, behavioral
   assessments were conducted at multiple time points: control, 2-week and
   4-week demyelination, and 1-week and 2-week remyelination ([73]Figure
   1A). The mice exhibited significant weight loss during the 4-week
   demyelination phase, which persisted into the remyelination phase but
   showed partial recovery ([74]Figure 1B). Sensorimotor deficits emerged
   from the second week of demyelination, as evidenced by the impaired
   performance in the adhesive removal and foot fault tests, followed by a
   marked improvement after 2 weeks of remyelination ([75]Supplementary
   Figure S1B,C), suggesting partial functional restoration. In the novel
   object recognition test, mice subjected to 2-week and 4-week
   demyelination demonstrated reduced interaction with novel objects,
   indicating cognitive impairment. During remyelination, cognitive
   deficits were partially ameliorated, as reflected by an increase in the
   exploration time ([76]Figure 1C,D).

3.2. Dynamic Reconstruction of Oligodendrocytes During Cuprizone-Induced
Demyelination and Endogenous Remyelination

   The observed deficits in cognitive and sensorimotor functions following
   cuprizone exposure suggest dynamic regulation by the processes of
   demyelination and remyelination. In the cuprizone-treated mouse model,
   the stage-dependent remodeling of corpus callosum myelin was evident.
   Quantitative Luxol fast blue (LFB) histochemical analysis of the
   compact myelin deposition revealed a progressive decrease in the
   optical density at 2 and 4 weeks of exposure, followed by significant
   remyelination at 1 and 2 weeks of recovery ([77]Supplementary Figure
   S1D,E). A similar temporal pattern was observed for the MBP and MAG,
   both exhibiting reduced expression during demyelination and gradual
   restoration during remyelination ([78]Figure 2A–C). These findings
   indicate a strong association between the dynamic regulation of
   myelin-associated proteins and the distinct phases of demyelination and
   remyelination ([79]Figure 2D–F). Immunofluorescence analysis further
   demonstrated a decrease in mature oligodendrocytes (CC1^+OLIG2^+) and
   oligodendrocyte precursor cells (OPCs, PDGFRα^+OLIG2^+) at 2 and 4
   weeks of demyelination. Notably, these populations markedly increased
   at 1 and 2 weeks of remyelination ([80]Figure 2G–J). This dynamic
   cellular response suggests that OPCs are activated and differentiate
   into mature oligodendrocytes during remyelination, providing
   mechanistic insights into myelin repair.

Figure 2.

   [81]Figure 2
   [82]Open in a new tab

   Cuprizone-induced demyelination. (A) Representative immunofluorescence
   images of MBP (red) and MAG (green) in corpus callosum at different
   time points during demyelination and remyelination (scale bar: 100 µm).
   (B,C) Quantitative analysis of MBP and MAG relative intensities. (D)
   Western blot showing changes in MBP and MAG protein expression. (E,F)
   Quantitative analysis of the ratios of MBP/β-actin and MAG/β-actin from
   Western blot images. (G) Representative immunofluorescence images of
   CC1 (red) and Olig2 (green) in the corpus callosum at different time
   points during demyelination and remyelination (scale bar: 20 µm). (H)
   Representative immunofluorescence images of PDGFRα (red) and Olig2
   (green) in the corpus callosum at different time points during
   demyelination and remyelination (scale bar: 20 µm). (I) Normalized
   percentage of CC1^+Olig2^+ cells relative to the total Olig2^+
   population. (J) Normalized percentage of PDGFRα^+Olig2^+ cells relative
   to the total Olig2^+ population. All p values were calculated via a
   one-way ANOVA, followed by Tukey’s multiple comparison test.

3.3. BBB Permeability Dynamics During Cuprizone-Induced Demyelination and
Remyelination

   Cuprizone-induced demyelination and remyelination represent a dynamic
   process in which alterations in BBB function may play a critical role.
   To assess the BBB permeability at different time points, EB dye and
   FITC-labeled dextran were administered at defined stages ([83]Figure
   3A). The EB extravasation increased significantly at 2 and 4 weeks
   post-cuprizone but was reduced during the first and second weeks of
   remyelination, indicating BBB restoration ([84]Figure 3B,C). Parallel
   fluorescence measurements of the FITC–dextran confirmed time-dependent
   increases in the macromolecule permeability ([85]Supplementary Figure
   S2A,B).

Figure 3.

   [86]Figure 3
   [87]Open in a new tab

   Effects of cuprizone on BBB permeability in mice. (A) Schematic of
   Evans blue and FITC–dextran injection in deeply anesthetized mice at
   different time points. (B) Representative images of EB leakage in brain
   tissues during demyelination and remyelination phases. (C) Quantitative
   analysis of EB leakage. (D) Representative immunofluorescence images of
   ZO-1 (red) and Occludin (green) in the corpus callosum at different
   time points during demyelination and remyelination (scale bar: 20 µm).
   (E,F) Normalized percentages of ZO-1^+ and Occludin^+ cells relative to
   the total cell population. (G) Western blot analysis of ZO-1 and
   Occludin protein levels in total cell lysates. (H) Western blot
   analysis of ZO-1 and Occludin protein levels in membrane fractions.
   (I,J) Quantitative analysis of ZO-1/β-actin and Occludin/β-actin ratios
   from Western blot images. (K,L) Quantitative analysis of ZO-1/α-tubulin
   and Occludin/α-tubulin ratios from Western blot images. Schematic
   illustrations (A) were created via BioRender
   ([88]https://app.biorender.com/illustrations (accessed on 15 October
   2024)). All p values were calculated via a one-way ANOVA, followed by
   Tukey’s multiple comparison test.

   Immunofluorescence analysis revealed a reduction in ZO1- and
   Occludin-positive cells at 2 w and 4 w of demyelination, followed by a
   marked increase during remyelination (1 w and 2 w) ([89]Figure 3D–F).
   These fluctuations correspond to BBB disruption during demyelination
   and repair during remyelination. Western blot analysis further revealed
   that the ZO-1 and Occludin expression in the membrane fraction
   decreased during demyelination and was restored during remyelination
   ([90]Figure 3H,K,L). However, the total protein levels of ZO-1 and
   Occludin did not show significant changes between the groups
   ([91]Figure 3G,I,J). In addition, the quantitative PCR (qPCR) analysis
   showed that the transcript levels of TJP1 (encoding ZO-1) and Occludin
   were not significantly different among the groups ([92]Supplementary
   Figure S2C,D), suggesting that increased BBB permeability may be
   attributed to protein mislocalization rather than transcriptional
   alterations.

3.4. A Single-Cell Atlas Revealed Dynamic Cellular Remodeling and Signaling
Reprogramming During Demyelination and Remyelination

   In a cuprizone-induced mouse model, the demyelination and remyelination
   processes are closely associated with cellular diversity changes. To
   systematically dissect the dynamic alterations in glial cells, a
   comprehensive cell atlas was established by reanalyzing datasets from
   the GEO database, focusing on the specific heterogeneity in the cortex
   and corpus callosum (CC) regions. After stringent quality control,
   109752 high-quality nuclei were retained. Specific marker genes were
   used to annotate each cell type. For example, microglia are marked by
   C1qa and Hexb; astrocytes by Aqp4 and Gja1; OPCs by Pdgfra and Cspg4;
   oligodendrocytes by Cldn11 and Mbp; vascular cells by Atp13a5 and
   Cald1; and excitatory and inhibitory neurons by Grin, Syt1, Sv2b,
   Slc17a7, Gad1, and Gad2 ([93]Figure 4A) ([94]Supplementary Figure S3A).
   Unsupervised UMAP clustering identified seven transcriptionally
   distinct neural cell populations: excitatory neurons, inhibitory
   neurons, OPCs, oligodendrocytes, astrocytes, microglia, and vascular
   cells ([95]Figure 4B). Further analysis revealed that the cuprizone
   treatment altered the cell subpopulation proportions, with reductions
   in the oligodendrocyte, OPC, and astrocyte populations during
   demyelination and an increase in microglia. These proportions partially
   recovered after cuprizone withdrawal ([96]Figure 4C,D). The coordinated
   depletion-recovery dynamics suggest that endogenous regulatory
   mechanisms maintain cellular diversity despite severe myelin loss.

Figure 4.

   [97]Figure 4
   [98]Open in a new tab

   scRNA-seq analysis of the corpus callosum (CC) and cortex. (A) Scatter
   plot showing the distribution of the marker gene expression levels in
   microglia, astrocytes, OPCs, and oligodendrocytes. (B) UMAP plot of the
   scRNA-seq data. (C,D) Proportions of each cell type during
   demyelination and remyelination. (E) Receptor–ligand pair counts
   between different cell populations in normal mice. (F) Receptor–ligand
   pair counts between different cell populations in demyelinated mice.
   (G) Receptor–ligand pair counts between different cell populations in
   remyelinated mice. (I,J) Volcano plot displaying DEGs in astrocytes and
   oligodendrocytes. p values were calculated via the Wilcoxon rank-sum
   test. (H) Venn diagram showing overlapping DEGs between normal vs.
   demyelinated and demyelinated vs. remyelinated astrocytes, along with
   (J) a bar plot of significantly enriched biological pathways for
   overlapping genes. (I) Venn diagram showing overlapping DEGs between
   OLs (normal vs. demyelinated) and oligodendrocytes (demyelinated vs.
   remyelinated), along with (K) a bar plot of the significantly enriched
   biological pathways for the overlapping genes.

   Cell-cell communication analysis via CellChat in the control,
   demyelination, and remyelination groups revealed no significant
   differences in the number or strength of interactions ([99]Figure 4E–G)
   ([100]Supplementary Figure S3B–D). However, during demyelination, the
   Psap-Gpr37l1 signaling axis between oligodendrocytes and OPCs was
   increased, as was the Psap-Gpr37l1 axis from microglia to OPCs, whereas
   the Psap-Gpr37 axis from astrocytes to oligodendrocytes was reduced
   ([101]Supplementary Figure S3E–G).

   A comparative analysis of the DEGs between the control and
   demyelination groups, as well as between the demyelination and
   remyelination groups, revealed changes induced by cuprizone
   ([102]Figure 4H,I) ([103]Supplementary Figure S3H,I). KEGG pathway
   analysis further elucidated the biological processes involved in these
   stages. Astrocytes are involved primarily in the cAMP signaling, Ras
   signaling, and HIF-1 signaling pathways, which are critical for cell
   survival and stress responses ([104]Figure 4J). Microglia were enriched
   in the PI3K-Akt signaling pathway and cytokine-cytokine receptor
   interaction pathways, highlighting their roles in immune responses and
   neuroinflammation ([105]Supplementary Figure S3J). Oligodendrocytes
   were enriched in the PI3K-Akt signaling, MAPK signaling, and axon
   guidance pathways, indicating their importance in cell proliferation
   and survival and axonal regeneration ([106]Figure 4K). Finally, OPCs
   are associated with axon guidance and cell adhesion molecule signaling
   pathways, further emphasizing their roles in remyelination and cell
   migration ([107]Supplementary Figure S3K).

3.5. Neuroimmune Activation and Oxidative Stress Dynamics During the
Demyelination-Remyelination Cycle

   Cuprizone-induced demyelination and subsequent remyelination are
   accompanied by significant alterations in the BBB integrity and
   cellular heterogeneity. These processes are likely influenced by
   neuroimmune responses and oxidative stress, which play pivotal roles in
   driving disease pathology. The dual immunolabeling of microglia and
   astrocytes revealed a time-dependent increase in the IBA1^+ microglial
   density during demyelination, whereas the GFAP^+ astrocytic reactivity
   was significantly suppressed, indicating functional divergence in the
   glial responses ([108]Figure 5A–C). This trend reversed during
   remyelination, suggesting a temporal coordinated transition from
   microglia-driven inflammation to astrocyte-mediated repair ([109]Figure
   5A–C).

Figure 5.

   [110]Figure 5
   [111]Open in a new tab

   Cuprizone-induced neuroinflammation and immune infiltration. (A)
   Representative immunofluorescence images of IBA1 (red) and GFAP (green)
   in the cortical region at different time points during demyelination
   and remyelination (scale bar: 20 µm). (B,C) Normalized percentages of
   IBA1+ and GFAP+ cells relative to the total cell population. (D)
   Representative flow cytometry gate strategy for quantifying CD45 (E),
   CD3 (F), and F4/80 (G) cell populations in the cortex and corpus
   callosum at different stages of demyelination and remyelination. (H,I)
   Flow cytometry analysis of ROS levels in the cortex and corpus
   callosum. The results of the quantitative analysis of the ROS
   generation are displayed as bar graphs. (J,K) Kit-based detection of
   SOD and MDA activity in the cortex and corpus callosum. All p values
   were calculated via a one-way ANOVA, followed by Tukey’s multiple
   comparison test.

   Flow cytometry further revealed significant shifts in the immune cell
   populations within the cortex and corpus callosum ([112]Figure 4D). At
   weeks 2 and 4 of demyelination, the proportions of CD45^+ immune cells,
   CD3^+ T cells, and F4/80^+ macrophages markedly increased, indicating a
   sustained immune response ([113]Figure 5E–G). During remyelination, the
   proportions of these immune cells decreased significantly at weeks 1
   and 2, suggesting the resolution of inflammation and the initiation of
   tissue repair ([114]Figure 4E–G).

   Oxidative stress markers further highlighted cuprizone’s impact on
   cellular damage. At weeks 2 and 4 of demyelination, the reactive oxygen
   species (ROS) and malondialdehyde (MDA) levels were significantly
   elevated, whereas the superoxide dismutase (SOD) activity was
   decreased, reflecting increased oxidative stress and exacerbating
   cellular injury ([115]Figure 5H–K). Consistently, the qPCR analysis
   revealed that the SOD1 transcription was downregulated during
   demyelination and subsequently restored during the remyelination phase,
   further supporting the involvement of antioxidant gene regulation in
   the recovery process ([116]Supplementary Figure S4A). Upon cuprizone
   withdrawal, these oxidative stress markers improved during the first
   two weeks of remyelination ([117]Figure 5H–K), indicating the
   reactivation of endogenous antioxidant defense mechanisms during the
   repair process.

3.6. Transcriptomic Analysis and Inflammatory Regulatory Networks in
Cuprizone-Induced Demyelination and Remyelination

   The critical role of neuroinflammation in cuprizone-induced
   demyelination and subsequent remyelination underscores the molecular
   mechanisms governing these processes. The bulk RNA sequencing of
   cortical and corpus callosum tissues systematically revealed dynamic
   transcriptional reprogramming associated with cuprizone-induced
   demyelination and repair ([118]Figure 6A). Principal component analysis
   revealed a distinct separation in the gene expression profiles between
   the demyelination group and both the control and remyelination groups,
   highlighting the extensive transcriptional alterations following myelin
   loss ([119]Figure 6B). Weighted gene coexpression network analysis,
   with the soft threshold set at 10, identified eight distinct expression
   modules ([120]Supplementary Figure S3B,C). Clustering analysis and
   correlation heatmaps revealed a significant negative correlation
   between demyelination and MEsalmon, whereas remyelination was inversely
   associated with MEbrown, providing insights into module-specific
   functions during pathological damage and repair ([121]Figure 6C).
   Differential gene expression analysis between the normal and
   demyelination groups, as well as between the demyelination and
   remyelination groups, confirmed a marked shift in transcriptional
   regulation following cuprizone exposure ([122]Figure 6D–E). A KEGG
   pathway mapping of the overlapping DEGs revealed cytokine-receptor
   interactions, JAK-STAT signaling, and NF-κB signaling as key regulators
   of these transcriptional changes ([123]Figure 6F-G). Western blot
   analysis further demonstrated that the phosphorylation level of the
   NF-κB pathway component p65 was markedly increased during the
   demyelination phase and decreased during the remyelination phase
   ([124]Figure 2H–I), supporting the key role of NF-κB signaling in
   neuroinflammation dynamic regulation.

Figure 6.

   [125]Figure 6
   [126]Open in a new tab

   Bioinformatics analysis of transcriptomes in the cortex and corpus
   callosum. (A) Schematic of the experimental design for transcriptome
   sequencing. (B) Principal component analysis (PCA) plot illustrating
   the variability among biological replicates. (C) WGCNA identifies gene
   modules associated with demyelination and remyelination. The heatmap
   displays correlations and p values between modules and traits. (D) The
   volcano plot shows differentially expressed genes (DEGs) between the
   normal and deceased groups. p values were calculated via the Wilcoxon
   rank-sum test. (E) Volcano plot showing DEGs between the demyelinated
   and remyelinated groups. p values were calculated via the Wilcoxon
   rank-sum test. (F,G) Venn diagram illustrating overlapping DEGs between
   normal and demyelinated groups and demyelinated and remyelinated
   groups, with bar plots displaying significantly enriched biological
   pathways for overlapping genes. (H) Western blot analysis of p65 and
   p-p65 protein levels in total cell lysates. (I) Quantitative analysis
   of p65/p-p65 from Western blot images. (J) A protein-protein
   interaction (PPI) network was constructed via the STRING database, with
   hub genes identified via Cytoscape. (K–P) qPCR validation of the mRNA
   expression levels of inflammation-related hub genes normalized to those
   of β-actin. (Q–S) qPCR validation of inflammatory cytokine mRNA
   expression levels normalized to those of β-actin. Schematic
   illustrations (A) were created via BioRender. All p values were
   calculated via a one-way ANOVA, followed by Tukey’s multiple comparison
   test.

   Further protein-protein interaction analyses via the STRING database
   ([127]Supplementary Figure S3D) revealed 20 hub genes ([128]Figure 6J),
   whose dynamic transcriptional patterns in the cortex and corpus
   callosum were visualized via heatmaps ([129]Supplementary Figure S3E).
   The functional annotation suggested strong associations between these
   core genes and tumorigenesis, cardiovascular dysfunction, apoptosis
   regulation, endocrine–metabolic disorders, and immune-related diseases
   ([130]Supplementary Figure S3F). Finally, the qPCR validation revealed
   the significant upregulation of inflammation-associated hub genes and
   key inflammatory mediators, including IL-6, IL-1β, and TNF-α, during
   demyelination, with a marked decline during remyelination ([131]Figure
   6K–S). These findings not only validate the transcriptomic analyses but
   also highlight the dynamic regulation of inflammatory signaling as a
   potential bidirectional modulator of myelin injury and repair.

4. Discussion

   The primary pathological hallmark of demyelinated diseases is the
   widespread loss of myelin, leading to reduced nerve conduction
   efficiency and subsequent sensory-motor and cognitive dysfunction
   [[132]27,[133]28,[134]29]. Compared with other models, the oral
   cuprizone administration model offers advantages in terms of its rapid
   induction, reproducibility, and minimal trauma [[135]30,[136]31]. This
   study demonstrated that cuprizone-induced demyelination progressively
   impairs sensory–motor and cognitive functions, whereas remyelination
   partially reverses these deficits. The sensory-motor impairments
   observed at two weeks post-demyelination correlate with a decrease in
   the axonal conduction efficiency and neuronal hyperexcitability, which
   is consistent with previous findings [[137]32,[138]33] on delayed
   action potential propagation in the motor cortex following
   demyelination. Behavioral recovery temporal dynamics reveal critical
   functional repair windows. Improvements in sensory-motor functions
   during remyelination are synchronized with OPC differentiation and de
   novo myelin formation, a process that can be accelerated by
   activity-dependent mechanisms [[139]34]. The limited recovery of
   cognitive function in the novel object recognition task reflects the
   partial restoration of the myelin integrity, which aligns with
   mechanisms by which adaptive myelination supports synaptic plasticity
   and memory consolidation [[140]35,[141]36].

   The dynamic interplay between demyelination and endogenous
   remyelination constitutes a critical yet incompletely understood axis
   in CNS repair mechanisms. This study elucidates the spatiotemporal
   patterns of oligodendrocyte reconstruction during cuprizone-induced
   myelin remodeling, revealing stage-dependent structural and cellular
   adaptive changes. Quantitative LFB staining revealed progressive
   demyelination during weeks 2–4 of cuprizone exposure, followed by
   significant remyelination during weeks 1–2 of recovery, a trajectory
   paralleled by fluctuations in the MBP/MAG protein expression. The
   temporal alignment of the myelin structural and molecular changes
   highlights myelin plasticity during degeneration and regeneration. The
   number of oligodendrocyte precursor cells (OPCs, PDGFRα^+OLIG2^+) and
   mature oligodendrocytes (CC1^+OLIG2^+) decreased during demyelination
   but increased significantly during remyelination, indicating that OPC
   activation and differentiation are central drivers of endogenous
   repair. This cellular plasticity resembles the differentiation cascade
   observed in zebrafish CNS regeneration models, where radial glial cells
   transiently acquire OPC-like properties to facilitate repair [[142]37].
   Additionally, the activation of endogenous neural stem cells (NSCs) in
   multiple sclerosis to replenish the oligodendrocyte lineage further
   supports OPC mobilization as a conserved repair pathway
   [[143]38,[144]39].

   The BBB serves as a critical interface to regulate central nervous
   system homeostasis, and its dysfunction is closely associated with
   demyelinating diseases such as multiple sclerosis [[145]40,[146]41].
   BBB permeability disruption may exacerbate neuroinflammatory cascades
   and hinder myelin regeneration [[147]42,[148]43]. This study
   demonstrated that dynamic changes in the BBB permeability are linked to
   cuprizone-induced demyelination and subsequent myelin regeneration.
   Both small (EB) and large (FITC–dextran) molecular tracer
   extravasations exhibited synchronous changes, suggesting compromised
   endothelial junction stability rather than irreversible structural
   damage. A transient reduction in the membrane localization of ZO-1 and
   Occludin (with unchanged total protein levels) indicates that
   posttranslational redistribution or cytoskeletal disruption during the
   demyelination phase may lead to decreased junction stability. Further
   supporting this notion, the qPCR analysis showed no significant changes
   in the mRNA expression of TJP1 and Occludin, suggesting that the
   observed junctional defects are likely resulting from intracellular
   trafficking alterations, protein anchorage failure, or interaction with
   inflammatory signaling pathways rather than transcriptional regulation.
   In the regeneration phase, these proteins may be reorganized by
   anti-inflammatory signals or OPC recruitment [[149]44,[150]45,[151]46].
   The restoration of the BBB integrity occurs in parallel with the
   reorganization of tight junction structures, which may provide a
   favorable microenvironment for OPC recruitment and differentiation.
   During the demyelination phase, microglia exacerbate the damage by
   releasing proinflammatory cytokines (e.g., TNF-α and IL-1β)
   [[152]47,[153]48], whereas in the regeneration phase, they switch to a
   phagocytic and anti-inflammatory phenotype that aids in myelin debris
   clearance and neurotrophic factor secretion [[154]49]. For example,
   Gharibani et al. report that kinase C modulators could reduce
   microglial activation and accelerate myelin regeneration, validating
   the interplay between neuroinflammation and barrier repair [[155]50].
   Transient BBB disruption during demyelination may allow peripheral
   immune mediators to infiltrate, thereby exacerbating inflammation and
   oligodendrocyte damage, whereas barrier repair during regeneration may
   limit further injury and promote local repair signaling.

   scRNA-seq has played a pivotal role in elucidating the cellular
   heterogeneity and dynamic remodeling mechanisms underlying
   neurodegenerative processes [[156]51,[157]52]. In the cuprizone-induced
   demyelination and remyelination model, significant changes in the
   cellular composition and signaling dynamics were observed. During the
   demyelination phase, the proportion of oligodendrocytes and astrocytes
   decreased, whereas the proportion of microglia increased, indicating
   neuroinflammation activation. In the remyelination phase, these
   proportions partially recovered, reflecting the activation of
   endogenous repair mechanisms. An analysis of the intercellular
   signaling revealed that during demyelination, the Psap-Gpr37l1
   signaling axis between oligodendrocytes and OPCs was upregulated, and
   microglia also presented increased Psap-Gpr37l1 signaling to OPCs.
   Conversely, astrocytes exhibited a reduction in the Psap-Gpr37
   signaling axis with oligodendrocytes. These signaling alterations may
   play crucial roles in myelin injury and repair [[158]53,[159]54]. KEGG
   pathway analysis revealed that astrocytes specifically activate the
   cAMP and HIF-1 pathways to respond to metabolic stress, which is
   consistent with their core role in maintaining redox homeostasis
   [[160]55,[161]56]. Microglia are enriched in the cytokine-receptor
   interaction pathway, which aligns with the transcriptional signature of
   disease-associated microglia in neurodegenerative disorders [[162]57].
   The coactivation of the PI3K-Akt and MAPK pathways in oligodendrocyte
   lineage cells reflects their classical role in regulating the
   survival-differentiation balance during myelin repair
   [[163]58,[164]59,[165]60]. Compared with the original study on
   [166]GSE204769, which focused on defining DOL and MAFB^hi microglia
   [[167]26], this study systematically quantified regional changes in the
   cellular composition and transcriptional features and integrated
   intercellular communication networks. These analyses further revealed a
   potential role for the Psap–Gpr37l1 signaling axis in regulating glial
   interactions. Future experiments may consider using the conditional
   knockout or pharmacological blockade of the Psap–Gpr37l1 pathway to
   verify its functional necessity in OPC replenishment and blood–brain
   barrier repair, offering translational insights into demyelinating
   diseases.

   Neuroinflammation plays a pivotal role in neurodegenerative disease
   onset and progression. Microglia and astrocyte activation is a hallmark
   of neuroinflammation [[168]61,[169]62], with excessive inflammatory
   responses potentially leading to neuronal damage and disease
   progression. This study reveals the spatiotemporal dynamics of
   neuroimmune activation and oxidative stress during the
   cuprizone-induced demyelination-remyelination cycle. The inverse
   dynamics of microglial proliferation and astrocytic quiescence during
   acute demyelination, along with their functional reverse during the
   repair phase, suggests the existence of a “functional division”
   paradigm in glial cell pathological responses [[170]14]. This biphasic
   regulation correlates with the phenotypic switch in astrocytes—when
   inflammation subsides, the proinflammatory A1 phenotype converts to a
   neuroprotective A2 state [[171]63], whereas microglia shift from
   phagocytic activation to immune regulatory functions [[172]64,[173]65].
   The persistent elevation of CD45^+ leukocytes and tissue-resident
   macrophages during the demyelination phase, coupled with progressive
   oxidative damage marked by ROS/MDA accumulation and SOD depletion,
   confirms that chronic neuroinflammation is both a consequence and a
   driver of myelin toxicity. The transcriptomic analysis further revealed
   the global reprogramming of inflammatory and stress pathways.
   Specifically, Tlr4-mediated NF-κB activation drives the upregulation of
   proinflammatory cytokines (IL6, IL1β, and TNFα), which is consistent
   with its role in amplifying microglial reactivity in experimental
   autoimmune encephalomyelitis [[174]65,[175]66]. Conversely, the
   downregulation of Itgax (CD11c) during the remyelination phase may
   reflect reduced antigen-presenting functions, suggesting the
   attenuation of immune surveillance during the repair phase. The
   sustained high expression of Ptprc (CD45) in demyelinated tissue is
   correlated with T-cell infiltration in human demyelinating plaques
   [[176]67], whereas the downregulation of Itgb2 may reduce leukocyte
   adhesion [[177]68,[178]69]. These hub genes collectively form a
   molecular scaffold linking neuroinflammation with metabolic apoptosis
   pathways and may regulate oligodendrocyte precursor cell survival.

   In summary, the cuprizone-induced demyelination model reveals a dynamic
   relationship between endogenous myelin regeneration and functional
   recovery ([179]Figure 7). Sensorimotor and cognitive deficits correlate
   temporally with myelin loss, whereas the regeneration process partially
   reverses neurological dysfunction. Oligodendrocyte population
   reconstruction, particularly precursor cell differentiation, aligns
   with BBB integrity restoration and neuroinflammation resolution.
   Single-cell profiling revealed a coordinated cellular remodeling
   process, whereas transcriptomic analysis emphasized the dominance of
   inflammatory regulatory networks in driving the demyelination-repair
   cycle. The simultaneous alleviation of oxidative stress and immune
   infiltration during the remyelination phase further highlights the
   critical role of the microenvironment in regulating repair. These
   findings suggest that modulating neuroimmune interactions and enhancing
   endogenous repair mechanisms may provide novel therapeutic strategies
   for treating demyelinating diseases.

Figure 7.

   [180]Figure 7
   [181]Open in a new tab

   Schematic overview of cuprizone-induced demyelination and remyelination
   mechanisms.

Acknowledgments