BMSC exosomes promote neurogenesis and alleviate behavioral deficits in chronic traumatic encephalopathy: an animal model-based study.

Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Stem Cell Res Ther . 2026 Mar 5;17:132. doi: 10.1186/s13287-026-04943-4 Search in PMC Search in PubMed View in NLM Catalog Add to search BMSC exosomes promote neurogenesis and alleviate behavioral deficits in chronic traumatic encephalopathy: an animal model-based study Peng Liu Peng Liu 1 Department of Experimental Surgery, Tangdu Hospital, The Fourth Military Medical University, Xi’an, 710038 Shaanxi China 2 Department of Pharmacology and Toxicology, Institute of Basic Medicine Science, Xi’an Medical University, Xi’an, 710021 China Find articles by Peng Liu 1, 2, # , Zhuangzhuang Bai Zhuangzhuang Bai 1 Department of Experimental Surgery, Tangdu Hospital, The Fourth Military Medical University, Xi’an, 710038 Shaanxi China Find articles by Zhuangzhuang Bai 1, # , Yongfei Yang Yongfei Yang 1 Department of Experimental Surgery, Tangdu Hospital, The Fourth Military Medical University, Xi’an, 710038 Shaanxi China Find articles by Yongfei Yang 1, # , Xiao Li Xiao Li 1 Department of Experimental Surgery, Tangdu Hospital, The Fourth Military Medical University, Xi’an, 710038 Shaanxi China Find articles by Xiao Li 1 , Jinghua Xia Jinghua Xia 1 Department of Experimental Surgery, Tangdu Hospital, The Fourth Military Medical University, Xi’an, 710038 Shaanxi China Find articles by Jinghua Xia 1 , Qian Yang Qian Yang 1 Department of Experimental Surgery, Tangdu Hospital, The Fourth Military Medical University, Xi’an, 710038 Shaanxi China Find articles by Qian Yang 1, ✉ Author information Article notes Copyright and License information 1 Department of Experimental Surgery, Tangdu Hospital, The Fourth Military Medical University, Xi’an, 710038 Shaanxi China 2 Department of Pharmacology and Toxicology, Institute of Basic Medicine Science, Xi’an Medical University, Xi’an, 710021 China ✉ Corresponding author. # Contributed equally. Received 2025 Aug 14; Accepted 2026 Feb 16; Collection date 2026. © The Author(s) 2026 Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ . PMC Copyright notice PMCID: PMC13072477  PMID: 41787581 Abstract Chronic traumatic encephalopathy (CTE), a progressive neurodegenerative disorder, poses a significant threat to human health. The lack of validated animal models has impeded mechanistic studies and the development of treatments for CTE. Recent evidence suggests that bone marrow mesenchymal stem cell-derived exosomes (BMSC-exos) represent a promising strategy for treating central nervous system injuries; however, their efficacy and mechanisms of action in CTE remain unexplored. In this study, we developed and optimized a CTE mouse model that recapitulates the core clinical features observed in CTE patients, including the delayed symptom onset. Using this model, we investigated the therapeutic effects of BMSC-exos. Our results indicate that BMSC-exos ameliorated anxiety-like behaviors and cognitive deficits in CTE mice, restoring them to levels comparable to those in noninjured control mice. Mechanistically, analysis of the hippocampal subgranular zone (SGZ) revealed that BMSC-exos restored the chronic CTE-induced reduction in the number of doublecortin (DCX)-positive immature neurons without altering the population of Sox2-Nestin–double-positive neural stem cells, indicating a primary effect on promoting neuronal differentiation efficiency or immature neuron survival rather than stem cell proliferation. Furthermore, BMSC-exos preserved neuronal structural integrity during late-stage CTE, indicating a critical role in maintaining synaptic plasticity and dendritic complexity. Collectively, our study provides promising evidence for the therapeutic potential of BMSC-exos in CTE, offering new insights for future CTE therapeutics. Supplementary Information The online version contains supplementary material available at 10.1186/s13287-026-04943-4. Keywords: Chronic traumatic encephalopathy, Bone marrow mesenchymal stem cell, Neurogenesis, Neuronal structural plasticity, Animal model Introduction Chronic traumatic encephalopathy (CTE) is a neurodegenerative disease that presents a formidable challenge to public health and neuroscience. Triggered by repetitive mild traumatic brain injuries (mTBIs), such as those sustained in contact sports or military service, CTE manifests years or decades later with debilitating cognitive, behavioral, and motor deficits [ 1 ]. Despite the prevalence of repetitive head injuries in modern society, the understanding of CTE and its potential neuropathological implications, as well as treatment methods, has developed very slowly over the past century. On the one hand, the diagnostic methods for CTE rely on postmortem examinations, leading to many cases being misdiagnosed or missed entirely [ 1 ] and obstructing the construction and expansion of the CTE brain bank [ 2 ]. On the other hand, there is currently a lack of widely accepted animal models for CTE, which also limits preclinical research on this condition [ 3 ]. Although researchers have attempted to establish several animal models to explore the pathogenesis of CTE, these models can only partially simulate the pathological changes and symptoms associated with CTE and may not fully align with the pathology and phenotypes of the condition; thus, long-term validation of these models is lacking [ 4 , 5 ]. Consequently, an animal model tailored for evaluating regenerative therapeutic strategies (such as stem cell-derived exosomes) for CTE should ideally not only replicate core neuropathological features—such as tauopathy, neuroinflammation, and progressive neuronal dysfunction—but also reliably recapitulate the delayed onset and progressive nature of clinical symptoms. The dynamic pattern of behavioral deficits with an initial acute phase followed by a latent period and subsequent chronic symptom reemergence—a hallmark of clinical CTE—remains challenging to consistently model and is critical for gauging the sustained efficacy of therapeutic candidates. This temporal profile is essential for assessing interventions that aim to modify the disease course by promoting long-term neuroplasticity and repair. Therefore, we aimed to develop and optimize a modeling protocol that specifically addresses this need, creating a promising platform for translational therapeutic research in CTE. Adult neurogenesis primarily occurs in the subventricular zone (SVZ) and the subgranular zone (SGZ) of the dentate gyrus [ 6 ]. Neural stem cells (NSCs), which are a class of cells with self-renewal and multipotent differentiation potential, exist in these regions. They can produce progenitor stem cells, as well as neurons and various types of glial cells, through asymmetric cell division. The newly generated neurons undergo further maturation and migration, integrate into neural circuits, and play crucial roles in neuronal activity, learning and memory, and emotional regulation [ 7 , 8 ]. The process through which new neurons are generated from NSCs in the hippocampus is known as adult hippocampal neurogenesis (AHN), which enhances the plasticity of the hippocampus and is crucial for learning, memory, and emotional regulation. Despite the differences in clinical manifestations and pathological mechanisms, progressive neuronal loss/death and structural and functional deficits of the nervous system are common characteristics of neurodegenerative diseases [ 6 ]. Numerous studies have indicated that impaired neurogenesis is among the key factors involved in the progression of neurodegenerative diseases. Bone marrow mesenchymal stem cells (BMSCs), a type of mesenchymal stem cell (MSC), are recognized for their efficacy in treating central nervous system injuries because of their ability to repair tissue, inhibit neuroinflammation, and mediate nerve repair through cytokine secretion [ 9 ]. Exosomes are secreted by most living cells, including MSCs, and have a diameter of approximately 50–180 nm. Compared with stem cells, they offer advantages such as high safety, the ability to freely cross the blood‒brain barrier, and prolonged stability [ 10 , 11 ]. Proteomic and functional studies have also indicated that MSC-derived exosomes harbor multifunctional cargo (e.g., neurotrophic factors and proneurogenic miRNAs), which could enhance neurogenesis and neuronal integrity [ 12 , 13 ]. Although the regenerative and neuroprotective properties of BMSC-exos have been documented in models of ischemic stroke, Alzheimer's disease, and other neurological conditions [ 14 – 17 ], the applicability of this promising 'cell-free' therapy to CTE is unknown. Therefore, it is critical to evaluate whether BMSC-exos can mitigate the specific pathological cascade and functional deficits associated with CTE. In this study, we developed and optimized a CTE modeling protocol and conducted a comprehensive evaluation of its effectiveness as a robust tool for investigating CTE-related mechanisms. Utilizing this model, we then examined the therapeutic effects of BMSC-derived exosomes (BMSC-exos) on CTE mice and explored the underlying mechanisms, with a focus on hippocampal neurogenesis and neuronal structural plasticity. To further elucidate the associated biochemical pathways, we investigated the effects of BMSC-exos on tau phosphorylation, synaptic protein expression, and neuroinflammation. Therefore, we hypothesized that our optimized CTE mouse model, which recapitulates the delayed symptom onset characteristic of human CTE, would provide a promising platform for evaluating the therapeutic potential of BMSC-exos. We further postulated that BMSC-exos would alleviate CTE-associated deficits by enhancing hippocampal neurogenesis and preserving neuronal structural plasticity. Collectively, our findings provide promising evidence for the therapeutic potential of BMSC-exos in CTE, offering new insights for future therapeutic development. Materials and methods Animals Adult male C57BL/6 mice were purchased from the Experimental Animal Center of the Fourth Military Medical University. The primary reason for using only male mice was to eliminate potential confounds associated with the estrous cycle in females. The mice were housed in cages (5 per cage) under a 12-h light/dark cycle (lights on at 7 am) with food and water ad libitum. The room temperature was maintained at 22 ± 2 °C. The mice were handled daily for 1 week to adapt to the conditions before treatment. All the animal procedures were approved by the Institutional Ethics Committee of the Fourth Military Medical University. All efforts were made to minimize the number of animals used and the distress experienced by the animals. The work has been reported in line with the ARRIVE guidelines 2.0. The mice were randomized into experimental groups. Groups included 8–10 animals for behavioral tests and a minimum of 3 animals for histochemical (≥ 2 sections/animal) and morphological (≥ 6 neurons/group) evaluations, ensuring statistical power and ethical compliance. The group size is consistent with the standards in the field for rodent behavioral studies. The experimenters who conducted the behavioral tests and analyzed the behavioral data were blinded to the group assignments of the mice throughout the experiment.​ CTE modeling The modeling method was improved on the basis of the traditional weight drop injury model [ 3 , 18 ]. In brief, the mice (20–25 g) experienced closed head injury after being anesthetized with 3% isoflurane (RWD; Shenzhen, China). The impact procedure was performed immediately upon cessation of anesthetic administration. Specifically, the anesthetized mice were secured in a prone position on a foam board, with their heads placed under a PVC tube with side openings to reduce resistance during the process of weight descent. A 50-g weight (a stainless steel ball with a diameter of approximately 2.3 cm) was released from a height of 70 cm to strike the heads of the mice. The velocity at impact was approximately 3.7 m/s. The head was positioned directly under the center of the PVC tube, ensuring that the impact was delivered to the midline of the skull, avoiding the eyes and brainstem (as shown in Supplementary Fig. 1A). After impact, the weight was intercepted by a net pouch at the bottom of the PVC tube to prevent secondary injury. This modeling device has been granted a utility model patent (Supplementary Fig. 1A, Patent No.: ZL 2022 2 2309056.8). The entire procedure, from anesthesia to impact, was completed within a consistent timeframe for each mouse by a single trained researcher. After the modeling procedure was complete, the mice were returned to their breeding cages. They were then observed for any signs of distress (e.g., labored breathing, seizures) for an additional 5 min. Mice that died because of anesthesia-related complications or exhibited skull fractures or seizures following impact were excluded from the study. The model mice received a single head injury daily for a duration of 28 days, allowing for a ~ 24-h recovery period between injuries (CTE28). The overall survival rate for the 28-day modeling protocol was 97%. This extended regimen was chosen to simulate the cumulative effect of repetitive head impacts encountered in human populations at risk (e.g., athletes engaged in contact sports, such as football and rugby players) within a feasible experimental timeframe. It was designed to induce stable, quantifiable chronic behavioral and pathological deficits, thereby creating a reliable platform for longitudinal therapeutic evaluation. In addition, behavioral tests were conducted on the 3rd, 7th, 14th, and 28th days after modeling was completed. The control group of mice received the same anesthetic treatment, but no striking was performed. We also designed different strike frequencies (1 or 7 times) and observation time windows to optimize a modeling scheme that best simulates the characteristics of CTE disease. Rotarod test The rotarod test was performed using a rotarod apparatus (model LE8505, Panlab). The testing protocol, including both constant and accelerated speed modes, was ​​adapted from​​ established methods in the literature [ 19 ]. The mice were habituated to the task 1 day before testing. For the constant-speed mode, the speed was set to 16 rpm, and the mouse was placed on the rotating drum until it fell, after which the latency time was recorded. Each test was repeated three times for each mouse, and the average was taken. For the accelerated mode, the mice were placed on the rotating drum, which was accelerated from 4 to 40 rpm over a 5 min period. The latency to fall and the speed at the time of falling were recorded. Open-field test The mice were placed individually into an open-field chamber (40 × 40 × 40 cm) under low light (25 lx) for 30 min. A computer imaging video tracking system (SMART 3.0, Panlab) was used to monitor overall distance of activity, which was considered to indicate motor function. During the first 5 min of the test, the number of times the mice entered and the duration of time spent in the central area of the chamber were recorded separately to characterize the anxiety-like behavior of the mice. ​Furthermore, to provide a comprehensive behavioral profile, the total distance traveled, average velocity, and time spent in the side and corner areas were analyzed if needed. Y-maze test The apparatus consisted of three Plexiglas arms arranged at angles of 120 degrees to each other, converging at a triangular central area. Each arm measured 32 cm × 8 cm × 15 cm. During the test, a mouse was placed in the central area and allowed to explore the maze for 10 min. The total number of entries into the arms and the number of alternations (successive and consecutive entries into all three arms) were recorded by a computer imaging video tracking system (SMART 3.0, Panlab), and the spontaneous alternation percentage (alternation%) was calculated by the following formula: alternation\% = alternation number total arm entries - 2 x 100 % . Three-chamber test The device was an open transparent acrylic box (40 × 60 × 20 cm) divided into three chambers (one central chamber and two side chambers) by two walls with doors (5 × 8 cm) at the center of the bottom. Each side chamber contained a cylindrical cage (diameter 7 cm, height 17 cm) surrounded on its sides by several transparent acrylic columns. The test was divided into two phases (each lasting 10 min). In the first phase (pre-test, as shown in Fig. 2 A), the social preferences of the mice, where the test mice were placed in the device and a stranger mouse was placed in one of the cylindrical cages, were examined. The stranger mice were age- and sex-matched C57BL/6 mice that had no prior contact with the test mice. The time spent by the test mouse exploring the cage with the mouse and the empty cage was recorded. The second phase (post-test, Fig. 2 D) focused on assessing the social memory and cognitive abilities related to the social interactions of the mice. The test mice remained in the device, while another novel stranger mouse was placed in another cylindrical cage, and the time spent exploring the familiar mouse versus the novel mouse was recorded. In accordance with the literature [ 20 ], the preference indices for the two stages are calculated separately. For the pre-test phase, the social preference index was calculated as follows: (T mouse − T empty cage )/(T mouse + T empty cage ). For the post-test social novelty index, the following formula was used: (T novel mouse − T familiar mouse )/(T novel mouse + T familiar mouse ). Fig. 2. Open in a new tab Social preference and social cognitive ability in the three-chamber test. A Schematic of the pre-test phase: mice choose between an empty cage and a stranger mouse-containing cage. B Both groups showed a significant preference for the stranger mouse across all timepoints, confirming intact innate sociability. C ​ Social preference index in the pre-test phase. D Schematic of the post-test phase: choice between familiar (preexposed) and novel stranger mice. E Control mice consistently preferred the novel mouse (p < 0.05), and CTE28 mice exhibited comparable exploration times (p > 0.05), indicating impaired social recognition memory. F Social novelty index in the post-test phase. Compared with control mice, CTE28 mice exhibited a significantly reduced index, indicating impaired social recognition memory. G Representative heatmaps during the three-chamber test (blue: low activity, green: medium, yellow/red: high). Data are presented as the mean ± SEM; n = 8–10/group; *p < 0.05, **p < 0.01, ***p < 0.001 Isolation, culture and characterization of BMSCs BMSCs were obtained from 2-week-old C57 mice. The mice were euthanized (by cervical dislocation) and soaked in 75% alcohol, after which the femurs and tibias of the mice were obtained by dissection. A 5-mL syringe filled with complete medium (α-MEM with GlutaMAX (Gibco™) as the basal medium plus 15% FBS (Gibco™) and 1% penicillin‒streptomycin) was used to flush out the bone marrow as much as possible. The bone marrow was collected and blown evenly with a Pasteurpipette and then placed in a T75 culture flask. The cells were incubated at 37 °C in 5% CO 2 , and half of the media was changed every 3 days. When the cell density reached 80%, the cells were passaged at a ratio of 1:2. BMSCs were characterized prior to exosome extraction to confirm their identity and functional capacity. For phenotypic characterization, the cells were analyzed by flow cytometry using antibodies against positive (CD29, CD44, and Sca-1) and negative (CD117 and CD31) markers. Flow cytometry analysis was performed by Cyagen Biosciences (Suzhou, China) (see Supplementary Materials and Methods). For functional characterization, the trilineage differentiation potential was assessed using standard induction protocols. Briefly, cells were induced toward adipogenic, osteogenic, and chondrogenic lineages using specific differentiation media (Cyagen Biosciences). Subsequent staining with oil red O, alizarin red S, and alcian blue was performed to confirm lipid droplet formation, calcium deposition, and proteoglycan synthesis, respectively (see Supplementary Materials and Methods).​ Extraction, identification and injection of BMSC-exos After 3 to 4 passages, when the BMSC density reached approximately 70%, the culture medium was thoroughly aspirated. The cells were then washed with PBS and replenished with fresh culture medium, in which the standard FBS was replaced by exosome-free FBS. After 72 h of cultivation, the culture medium was collected. BMSC-exos were subsequently isolated using the ultracentrifugation method, as reported previously [ 21 , 22 ]. The culture medium was centrifuged at 300× g for 10 min, followed by centrifugation at 2000× g for 10 min and at 10,000× g for 30 min to remove cell debris and large organelles. The supernatant was subsequently centrifuged at 100,000 g (no brake mode) for 90 min, after which the resulting pellet was suspended in PBS and subjected to another round of ultracentrifugation. The resulting pellet, which contained BMSC-exos, was resuspended in PBS for subsequent identification and injection. All centrifugation steps were performed at 4 °C. The final exosome pellet was resuspended in 100 μL of sterile PBS. The isolated BMSC-exos were characterized on the basis of three critical criteria: particle size distribution, specific marker expression, and morphological characteristics. Nanoparticle tracking analysis (NTA):​​ A small amount of BMSC-exos was diluted in 1 mL of PBS to achieve the optimal concentration. The size distribution and particle concentration were analyzed using a Particle Metrix ZetaView device according to the manufacturer's instructions. The expression of specific exosomal markers was confirmed by Western blotting. The protocol was described in our previous publication [ 23 ]. The samples were probed with antibodies against the positive markers: Alix (1:1000; A2215; ABclonal), Tsg101 (1:1000; ab125011; Abcam), and CD9​ (1:1000; A19027; ABclonal), as well as the negative marker GM130​ (1:1000; A11408; ABclonal) to assess potential organelle contamination.​​ The morphology of the exosomes was examined by transmission electron microscopy (TEM) (Hitachi, Japan). Briefly, 10 μL of the exosome suspension was deposited onto a copper grid for 1 min. The sample was negatively stained with 10 μL of 2% uranyl acetate for 1 min, followed by air drying at room temperature. Images were acquired using a Hitachi TEM instrument operated at 100 kV. After the CTE model was established, the mice were randomly assigned to receive either exosome treatment or a PBS placebo. In the treatment group (CTE28 + EXO), BMSC-exos (protein concentration: 0.3 µg/µL, based on previous in vivo studies [ 16 , 24 ]) were injected into mice via the tail vein. The injection was administered once daily at a volume of 100 µL for 7 consecutive days. Animals in the control and untreated CTE28 groups were injected with an equivalent volume of PBS.​ EdU injection and observation In the last three days of the CTE modeling phase, the mice were given intraperitoneal injections of EdU (100 mg/kg) daily. Then, brain slices were stained using a BeyoClick™ EdU-594 Kit (Cat#: C0078S; Beyotime) according to the manufacturer's instructions, and the number of EdU-positive cells in the SGZ was recorded. After EdU staining, immunofluorescence staining for NeuN (antibody: mouse anti-NeuN; 1:200; MAB377; Sigma‒Aldrich) was performed to accurately assess adult neurogenesis on the 28th day post-modeling. Immunofluorescence staining On the corresponding day, the mice were deeply anesthetized by inhalation of 1.5–3% isoflurane and then intracardioventricularly perfused with saline (approximately 50 mL) followed by 4% paraformaldehyde (approximately 100 mL). The brain was removed and postfixed in 4% paraformaldehyde for 48 h, followed by sequential cryoprotection in 20% and 30% sucrose solutions. The brain tissue was then embedded in OCT and coronally sectioned at a thickness of 10 μm (Leica CM1950) through the hippocampus in accordance with the standardized mouse brain atlas [ 25 ]. Immunofluorescence staining was used to assess the expression of Sox2, Nestin and DCX in the SGZ regions of the hippocampus using a standard procedure [ 24 ]. After being washed with PBS, the brain sections were permeabilized by incubation in 0.2% Triton X-100 in PBS for 1 h at room temperature (RT). The sections were then incubated in blocking solution (3% bovine serum albumin (BSA) and 0.2% Triton X-100 in PBS) for 1 h at RT to reduce nonspecific binding. After being washed with PBS, the samples were incubated overnight at 4 °C with a cocktail of primary antibodies diluted in blocking solution, including rabbit anti-Sox2 (1:200, GTX101507, GeneTex), mouse anti-Nestin (1:500, MAB353, Sigma‒Aldrich), and guinea pig anti-DCX (1:800, AB2253, Sigma‒Aldrich) primary antibodies. Following three washes with PBS, the sections were incubated at RT for 1 h with the corresponding fluorophore-conjugated secondary antibodies (1:500, Invitrogen). Finally, the sections were covered with drops of SlowFade Diamond Antifade mounting medium with DAPI ( S36968 ; Invitrogen). Using an Olympus SLIDEVIEW VS200 slide scanner, sections were imaged to quantify Sox2/Nestin double-positive cells (NSCs) and DCX-positive cells (immature neurons) per unit length in the SGZ region. Sparse neuronal labeling strategies and imaging methods A dual AAV vector system was employed for the and bright labeling of sparse neurons to observe changes in the structural plasticity of mice. Briefly, equal volumes of rAAV-EF1α-DIO-EYFP-WPREs (AAV2/9 type, titer: 2.53 × 10 12 vg/mL, PT-0899, BrainVTA) and diluted rAAV-CMV-Cre-hGH pA (AAV2/9 type, titer: 5.8 × 10 12 vg/mL, PT-0025, BrainVTA, diluted 1:50,000 in PBS) were mixed and injected bilaterally into the hippocampal CA1 (AP − 2.0 mm, ML ± 1.6 mm, DV − 1.5 mm; 150 nL/side at 20 nL/min). After surgery, the mice were allowed to recover for 2 weeks before they underwent CTE modeling. At the designated time points, mouse brain tissues were obtained using the method described above, and vibratome sectioning​​ of the hippocampal region was performed (Leica VT1200S; thickness, 150 μm). Confocal Z-stacks (step size, 0.8 μm step size; laser, 488 nm) were acquired with an Olympus FV3000 microscope. Maximum intensity projections were generated for quantitative analysis of neuronal morphology. The inclusion criterion for morphological analysis was the successful 3D reconstruction of at least 10 neurons per mouse in the target hippocampal CA1 region. Morphological analysis was performed with a minimum of 3 animals and no fewer than 6 neurons per group​ for key results. Total dendritic length and branch numbers were analyzed from 20 × magnification micrographs using ImageJ. Sholl analysis was performed as follows: A series of concentric circles were overlaid at 10-μm radial increments from the soma centroid. Branch complexity was quantified by counting the number of dendritic intersections with each circle. The parameters were set as follows: starting radius = soma diameter (10 μm), ending radius = maximum dendritic extension (200 μm), and step size = 10 μm. The basal dendritic spines were analyzed from 100 × images, with segments sampled 30–50 μm from the soma and extending ≥ 20 μm. Western blot analysis The protocol was described in our previous publication [ 23 ]. Hippocampal tissues were homogenized in RIPA lysis buffer containing protease and phosphatase inhibitors. Protein concentrations were determined using a BCA assay. Equal amounts of protein were separated by SDS‒PAGE and transferred to PVDF membranes. After blocking, the membranes were incubated overnight at 4 °C with primary antibodies against p-Tau (Thr181) (1:500, CST, D9F4G), PSD-95 (1:1000, Millipore, MAB1596), synaptophysin (1:5000, Proteintech, 17785–1-AP), GFAP (1:1000, ABclonal, A19058), and Iba1 (1:1000, Abcam, 178847), with anti-GAPDH (1:5000, ABclonal, AC033) as a loading control. After the membranes were incubated with HRP-conjugated secondary antibodies, the bands were visualized using a chemiluminescence detection system (Bio-Rad) and quantified with Image Lab software. Statistical analysis An observer who was blinded to the experimental conditions and treatments evaluated the data. All data are expressed as the mean ± standard error of the mean (SEM), and SPSS 18.0 software was used for data processing and analysis. The normality of the data distribution was assessed using the Shapiro‒Wilk test, and the homogeneity of variance was verified using Levene's test. Data that met these assumptions were analyzed using parametric tests. Unpaired Student's t tests were used for comparisons between two groups. One-way ANOVA or two-way ANOVA followed by multiple-group comparison was performed using post hoc Tukey’s test. A critical value for significance of p < 0.05 was used throughout the study. All figures were generated using GraphPad Prism version 8.0. Results Behavioral phenotype changes in CTE model mice at varying post-modeling intervals The overall modeling and behavioral testing process is illustrated in the upper section of Fig. 1 . Model (CTE28) and control mice showed comparable rotarod performance metrics in both the constant-speed mode (latency to fall, ​​Fig. 1 A) and the accelerated mode (latency to fall and falling speed, Fig. 1 B, C) throughout the observation period. Notably, the comparable performance between CTE28 and control mice on the rotarod test indicate that the modeling procedure ​​did not induce general motor coordination or balance deficits. In the open-field test, there were no significant differences in the total distance traveled between the two groups of mice across all observation time points post-modeling (Fig. 1 D). These findings indicate that the modeling approach is relatively mild and does not adversely affect the basic motor abilities of the mice. These findings from the ​​rotarod and open-field tests​​ serve as crucial internal controls, suggesting that subsequent behavioral impairments observed in other tests are specifically related to cognitive and affective dysfunctions rather than confounding motor disabilities. Fig. 1. Open in a new tab Longitudinal assessment of motor, affective, and cognitive functions in CTE28 mice. Upper part: Experimental timeline. A Rotarod latency (constant speed of 16 rpm) showing comparable motor coordination between the control (green) and CTE28 (red) groups across post-modeling intervals (3rd, 7th, 14th, and 28th days). B , C Accelerated rotarod latency ( B ) and falling speed ( C ) demonstrating preserved motor function in both groups. D Total distance traveled in the open-field test confirming unimpaired locomotor activity. E , F Anxiety-like behaviors quantified by reduced central area time ( E ) and number of entries ( F ) in CTE28 mice (*p < 0.05, **p < 0.01). G Representative movement trajectories (left) and activity heatmaps (right, blue: low activity, green: medium, yellow/red: high) during open-field testing. CTE28 mice exhibit restricted central exploration acutely (3rd day) and chronically (28th day). H Comparison of Y-maze arm entries indicating intact locomotion. I , J Working memory deficits evidenced by reduced spontaneous alternation numbers ( I *p < 0.05 28th day) and percentages ( J *p < 0.05 3rd and 28th day) in CTE28 mice, with transient recovery at interim timepoints. K Representative movement trajectories. Data are presented as the mean ± SEM; n = 8–10/group With respect to anxiety-like behavior, as shown in Fig. 1 E, F, on the 3rd day after modeling, compared with mice in the control group, the mice in the CTE28 group exhibited a significant reduction in the time spent in the central area during the first 5 min of the test (p = 0.015). A ​​trend toward decreased​​ entries into the central area was also observed in the mice in the CTE28 group. These findings suggest that CTE modeling induces anxiety-like behavior in mice during the acute phase. This symptom disappeared on the 7th day after modeling and partially reemerged on the 14th day. On the 28th day post-modeling, compared with control mice, the CTE28 mice ​​exhibited a recurrence of anxiety-like behavior​​, characterized by a significantly shorter time spent in the central area and fewer entries (p = 0.017). Representative movement trajectories​​ and ​​corresponding heatmaps​ for the first 5 min of the open-field test in both groups of mice at the different post-modeling time points are presented in ​​Fig. 1 G. In the Y-maze test, the total number of arm entries did not differ significantly between the two groups at the different time points (Fig. 1 H), ​​further indicating that the general locomotor activity of the mice was not impaired.​ During the acute phase post-modeling (3rd day), compared with control mice, CTE28 mice exhibited a significant reduction in the spontaneous alternation percentage​​ (alternation%) (p = 0.022), ​​indicating cognitive impairments affecting working memory.​​ ​​This cognitive deficit​​ in CTE28 mice ​​subsequently recovered (7th and 14th day).​​ However, ​​upon longer-term assessment (28th day), a reemergence of cognitive impairment in the CTE28 group mice was observed, characterized by a significantly lower alternation number (Fig. 1 I) and alternation%​​ (Fig. 1 J) than in control mice (p = 0.045). As shown in Fig. 2 B, mice in both groups preferentially explored the cage containing the stranger mouse over the empty chamber during the pre-test phase across all observation time points (3rd, 7th, 14th and 28th days). No significant difference in the preference index was observed between the two groups of mice (Fig. 2 C). These findings indicate that the development of CTE did not alter the innate social preference of the mice. In the post-test phase, control mice consistently demonstrated a preference for the novel mouse over the familiar one at all time points, a behavior that was significantly impaired in CTE28 mice (Fig. 2 E). Furthermore, the preference index was significantly greater in control mice than in CTE28 mice across all time points (Fig. 2 F). These results indicate a persistent impairment in social cognitive function in the CTE28 mice. Representative heatmaps​​ for the three-chamber test in both groups of mice at the different post-modeling time points are presented in ​​Fig. 2 G. To further comprehensively characterize the phenotypes of our CTE model, we conducted additional assessments, including the novel object recognition test (NORT) and the forced swim test (FST), in an independent cohort of mice at 28 days post-modeling (see Supplementary Materials and Methods). The CTE mice exhibited cognitive deficits in the NORT and exhibited depressive-like behavior in the FST (Supplementary Fig. 1B-D). Consistent with CTE pathology, hippocampal p-Tau (Thr181) levels were significantly elevated in the CTE model group compared with those in the control group (Supplementary Fig. 1E). We also performed immunohistochemical (IHC) analysis for phosphorylated tau at the Ser202/Thr205 epitopes (see Supplementary Materials and Methods). Consistent with the Western blot results, CTE28 mice exhibited a significant increase in p-Tau (Ser202/Thr205) immunoreactivity compared to controls (Supplementary Fig. 1F–I). In addition to the aforementioned continuous 28-day CTE modeling method (CTE28), we also established additional murine models with varying impact frequencies, including single-impact exposure (CTE1) and 7-day repeated impacts (CTE7). Behavioral assessments were conducted at multiple post-modeling intervals (see Supplementary Fig. 2).​ Compared with control mice, CTE1 mice showed virtually no behavioral alterations​​, indicating that an isolated mild head impact is insufficient to trigger persistent neurobehavioral deficits. The CTE7 group exhibited an intermediate and unstable phenotype, suggesting a transitional pathological state. ​​CTE7 mice exhibited significant anxiety-like phenotypes​​ on the 3rd and 14th days, characterized by decreased time spent in the central zone of the open-field test (Supplementary Fig. 2C). ​​Cognitive deficits in the CTE7 mice were temporally unstable​​, with no abnormalities in the Y-maze test but persistent impaired social cognitive function​​ in the three-chamber test (Supplementary Fig. 2D, G, H). The CTE28 group displayed the most severe, pronounced, and persistent behavioral deficits, which closely recapitulated the core features of CTE. Therefore, in subsequent work, we selected the CTE28 modeling paradigm for subsequent investigations exploring exosome-based therapeutics and underlying mechanisms. Therapeutic efficacy of BMSC-exos on behavioral phenotypes in CTE model mice The mouse BMSCs used in this study were characterized to ensure their quality and stemness (Supplementary Fig. 3A, B). Following at least three passages after primary extraction to ensure cellular purity, BMSC culture supernatant was collected for exosome extraction.​​ The obtained BMSC-exos were identified prior to their use in treating CTE model mice. The particle size was analyzed by NTA (Supplementary Fig. 3C). The expression of specific markers was analyzed by Western blotting. The data demonstrated that the isolated exosomes were positive for the exosomal markers Alix (100 kDa), Tsg101 (44 kDa), and CD9 (25 kDa), but negative for GM130 (140 kDa) (Supplementary Fig. 3D). Transmission electron microscopy (TEM) imaging (Supplementary Fig. 3E) revealed the typical cup-shaped or spherical morphology of the purified vesicles, which is a hallmark of exosomes. To track the fate of intravenously administered exosomes, we performed an additional experiment using PKH-67-labeled BMSC-exos (see Supplementary Materials and Methods). Confocal imaging of brain sections revealed PKH-67 signals in the hippocampus of CTE mice that received labeled exosomes, indicating that intravenously injected exosomes reached the hippocampal tissue. Furthermore, colocalization studies demonstrated that the PKH-67 signal was associated with both neurons (NeuN-positive cells) and microglia (Iba1-positive cells), suggesting that BMSC-exos can be taken up by these key cell types in the brain parenchyma (Supplementary Fig. 12). We assessed the therapeutic effects of BMSC-exos on a mouse model of CTE, specifically on the 28th day post-modeling, corresponding to peak symptom manifestation. The overall experimental workflow is illustrated in the upper section of Fig. 3 . Fig. 3. Open in a new tab BMSC-derived exosomes (BMSC-exos) ameliorate anxiety-like behaviors and cognitive deficits in CTE model mice on the 28th day after modeling. Upper part: Experimental timeline. A Baseline locomotor activity assessed by the open-field test. BMSC-exos did not affect the total distance traveled. B Time spent in the central area of the open field, reflecting anxiety-like behavior. Compared with the untreated CTE mice (CTE28), the treated CTE mice (CTE28 + EXO) exhibited a significantly increased center time, which was comparable to that of the noninjured control mice. Compared with control mice, CTE28 mice showed significantly shorter center times (p < 0.05). C Spontaneous alternation percentage (alternation %) in the Y-maze test, indicating working memory. Compared with control mice, PBS-injected CTE28 mice exhibited a significantly reduced alternation% (p < 0.05). Compared with CTE28 mice, BMSC-exo-treated mice (CTE28 + EXO) showed a significantly greater alternation% (p < 0.001) that was comparable to that in the control mice. D Pre-test of the three-chamber test. All groups spent significantly more time investigating the cage containing the stranger mouse than the empty cage (p < 0.05 vs. empty cage), indicating intact social motivation. E ​ Social preference index in the pre-test phase. F Post-test of the three-chamber test. CTE28 mice showed impaired discrimination (p > 0.05 for novel mouse vs. familiar mouse), while CTE28 + EXO mice spent significantly more time investigating the novel vs. familiar mouse (p < 0.05), comparable to the time observed in the control mice. (G)​ Social preference index in the post-test phase. Data represent the mean ± SEM; n = 8–10/group; *p < 0.05, **p < 0.01, ***p < 0.001 BMSC-exos did not affect baseline locomotor activity in mice (Fig. 3 A). However, treated CTE model mice (CTE28-EXO) exhibited significantly increased residence time in the central area of the open field, reaching times comparable to those in the control group (Fig. 3 B). In contrast, compared with control mice, untreated mice (CTE28) spent significantly less time in the center area (p = 0.016), suggesting that BMSC-exo treatment ameliorated anxiety-like behaviors in the CTE model mice.​ Consistent with the center time data, compared with control mice, CTE28 mice exhibited a significant decrease in the number of entries into the center zone and a marked increase in thigmotaxis, as evidenced by increased time spent in the side and corner areas (Supplementary Fig. 4A, C, D). BMSC-exos effectively ameliorated these anomalies, restoring them to levels comparable to those in the control group. In addition, the average locomotor velocity of the mice did not differ among the groups (Supplementary Fig. 4B), confirming that the observed differences in exploratory behavior were anxiety specific and not due to motor impairments. Consistent with previous findings, compared with control mice, PBS-injected CTE28 model mice exhibited a significantly reduced alternation% in the Y-maze test (p = 0.014). In contrast, compared with untreated CTE28 mice, exosome-treated mice (CTE28-EXO) showed a significantly greater alternation% (p = 0.0006), reaching levels comparable to those of controls (Fig. 3 C). These findings indicate that exosome treatment alleviated the delayed cognitive dysfunction induced by chronic traumatic encephalopathy (CTE).​ As shown in Fig. 3 D, in the three-chamber test, mice from all groups spent significantly more time investigating the cage with the stranger mouse than the empty cage, demonstrating intact social abilities across experimental conditions.​ During the post-test phase, when social cognition and discriminative ability were assessed, the ability of the CTE28 model mice to discriminate between the novel mouse and the familiar mouse was impaired. In contrast, compared with the control mice, the exosome-treated CTE28 mice (CTE28 + EXO) spent significantly more time investigating the novel mouse, indicating a behavioral pattern consistent with that of the control group mice (Fig. 3 F, G). These findings suggest that BMSC-exos effectively attenuated social cognition deficits in the CTE mice. We also conducted Morris water maze (MWM)​ experiments in an independent cohort of mice (Supplementary Materials and Methods). Compared with control mice, CTE28 mice exhibited longer escape latencies across training days during the acquisition phase, indicating impaired spatial learning. BMSC-exos improved this learning curve (Supplementary Fig. 4E). In the probe trial, CTE28 mice showed a longer latency to reach the platform, spent less time in the target quadrant ​and made fewer target platform crossings, confirming a deficit in spatial memory retention. These parameters were significantly restored in the BMSC-exo-treated group (Supplementary Fig. 7F–H). Assessment of the therapeutic efficacy of BMSC-exos constituted a longitudinal process. We also evaluated behavioral phenotypes at intermediate time points (7th and 14th days), and the workflow is detailed in Supplementary Fig. 5A. On the 7th day, there were no significant differences between the BMSC-exo-treated CTE28 + EXO group and the PBS-treated CTE28 group. However, compared with nonmodel control mice, mice in both CTE-exposed groups demonstrated anxiety-like behaviors (Supplementary Fig. 5B) and impaired social cognitive function (Supplementary Fig. 5E, F). By day 14, these behavioral abnormalities persisted in the mice in the CTE28 group, whereas the function of the mice receiving exosome treatment (CTE28 + EXO) was restored (Supplementary Fig. 5G–K). Alterations in neurogenesis in the hippocampal subgranular zone (SGZ) following BMSC-exos treatment The experimental timeline is summarized in Fig. 4 A. During the last 3 days of CTE modeling, mice received intraperitoneal injections of EdU to label proliferating cells. ​​Following the completion of modeling, we performed BMSC-exo interventions (once daily for seven consecutive days, CTE28 + EXO), while CTE mice in the placebo group (CTE28) and control group mice received PBS injections. Brain tissue samples from the mice were subsequently collected on different days post-modeling and subjected to sectioning and staining to investigate the effects of BMSC-exos on neurogenesis in the SGZ. Fig. 4. Open in a new tab BMSC-exos enhance neurogenesis in the hippocampal SGZ of CTE mice on the 28th day after modeling. A Experimental timeline: CTE modeling was performed over 4 weeks, with intraperitoneal EdU injections (100 mg/kg) administered during the last 3 days to label proliferating cells. At the end of modeling, the mice in the CTE28 + EXO group received daily tail vein injections of BMSC-exos for 7 consecutive days, whereas those in the CTE28 and control groups received PBS. Brain tissues were collected at the indicated time points for sectioning and staining. The data shown in this figure represent​ the analysis conducted on the 28th day after modeling. B , C Representative immunofluorescence images of SGZ staining. B Neural stem/progenitor cell markers (Sox2, red; Nestin, green; DAPI, blue) and an immature neuron marker (DCX, cyan) in the control, CTE28, and CTE28 + EXO groups. The third column shows Sox2-Nestin–double-positive cells (white arrows indicate representative cells). Scale bars: 50 μm. C EdU labeling (proliferating cells, red) in the control, CTE28, and CTE28 + EXO groups. Scale bars: 100 μm. D – F Quantitative analyses of SGZ cell populations. D There were no significant differences in the number of Sox2-Nestin–double-positive neural stem cells across the groups. E The number of DCX-positive immature neurons was significantly lower in the CTE28 group than in the control group and was rescued by treatment with BMSC-exos (CTE28 + EXO). F Increased numbers of proliferating EdU⁺ cells in the CTE28 and CTE28 + EXO groups vs. the control group, with the greatest increase observed in the CTE28 + EXO group. Data are presented as the mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001 As shown in Fig. 4 D, on the 28th day post-modeling (14 days after the final BMSC-exo injection), no significant differences​​ were observed in the number of Sox2-Nestin–double-positive cells​ within the SGZ across the experimental groups. However, ​​DCX-positive cell counts​​ were ​​significantly lower​​ in the CTE28 group than in the control group. This reduction was reversed by treatment with BMSC-exos (CTE28-EXO group), as DCX-positive cells recovered to levels comparable to those in the control group (Fig. 4 E). In contrast, compared with the control group, both the CTE28 and the CTE28-EXO groups exhibited significantly increased numbers of EDU-positive cells​​, with a more pronounced increase observed in the CTE28-EXO group (p = 0.0005) (Fig. 4 F). Representative immunofluorescence images of SGZ staining for each group are shown in Fig. 4 B, C. To determine the fate of the proliferating cells and assess their differentiation into mature neurons, we performed costaining for EdU and the mature neuronal marker NeuN in hippocampal sections (Supplementary Fig. 6). The proportion of EdU-positive cells that coexpressed NeuN was significantly greater in the BMSC-exo-treated group (CTE28 + EXO) than in the untreated CTE28 group. These results indicate that a greater fraction of the cells that proliferated during the injury phase successfully differentiated into mature neurons in the presence of BMSC-exos. In addition to the aforementioned analyses at 28 days post-modeling, we further evaluated neurogenesis in the SGZ by harvesting brain tissues at ​​additional time points (7th and 14th day)​​ for sectioning and immunofluorescence staining across all experimental groups. On the 7th day post-modeling (the initial phase after exosome administration was completed), compared with the control treatment, CTE induction significantly increased the number of Sox2-Nestin–double-positive NSCs (p = 0.04). Exosome treatment (CTE28-EXO) ​​did not further increase this transient elevation​​​​ (Fig. 5 A). The DCX-positive immature neuron count ​​remained comparable between the CTE28 and control groups​​, while compared with the control group, the CTE28-EXO group exhibited ​​significantly greater numbers of immature neurons (​p = 0.027; Fig. 5 B). Both the CTE28 and CTE28-EXO groups showed ​​more EdU-positive proliferating cells​​ than did the control group (Fig. 5 C). ​​On the 14th day post-modeling, the marker expression trends largely mirrored those observed on the 7th day (Fig. 5 D–F). However, a ​​notable difference in the DCX-positive cell count emerged​​: compared with the CTE28-EXO group, the CTE28-EXO group exhibited ​​significantly fewer DCX-positive immature neurons​​ (p = 0.002; Fig. 5 E). We synthesized the dynamic changes in the aforementioned markers across all time points (7th, 14th, and 28th days) to construct Fig. 5 G, where the band width​​ visually quantifies the ​​relative abundance of marker-positive cells​​ for each group,​​ horizontal comparisons​​ illustrate intergroup differences, and vertical trajectories​​ track longitudinal changes within each group. Fig. 5. Open in a new tab Dynamic changes in the expression of SGZ neurogenesis markers in CTE mice with BMSC-exos. Upper part: Experimental timeline: CTE modeling (4 weeks; 100 mg/kg EdU intraperitoneal injections on final 3 days). On day 28, mice in CTE28 + EXO group received BMSC-exos (i.v.) daily for 7 days, whereas those in the control and CTE28 groups received PBS. Tissues were harvested at 7, 14, and 28 days. A – C Data from the 7th day post-modeling. A The number of Sox2-Nestin–double-positive neural stem cells was significantly greater in the CTE28 and CTE28 + EXO groups than in the control group. B The number of DCX-positive immature neurons was comparable in the CTE28 group vs. the control group but elevated in the CTE28 + EXO group. C The number of EdU-positive proliferating cells was greater in both the CTE28 and CTE28 + EXO groups than in the control group. D – F Data from the 14th day post-modeling. D Sox2-Nestin–double-positive cells showed trends similar to those on the 7th day. E DCX-positive cells were more abundant in the CTE28-EXO group than in the control and CTE28 groups (*p < 0.05, **p < 0.01). F The number of EdU-positive cells was increased in the CTE28/CTE28 + EXO group. G Integrated analysis: Schematic representation of marker dynamics across the 7th/14th/28th days. Band width: Relative abundance of marker-positive cells. Horizontal axis: Intergroup comparisons. Vertical axis: Longitudinal changes within groups. Data are presented at the mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001 We further investigated the ​​acute-phase effects of CTE modeling on SGZ neurogenesis on the 3rd day post-modeling. Compared with control mice, CTE28 mice presented significantly more Sox2-Nestin–double-positive NSCs (p = 0.017), a comparable number of DCX-positive immature neurons (p = 0.37 > 0.05), ​​markedly more EdU-positive proliferating cells​​ (p = 0.0001). Comprehensive data are presented in Supplementary Fig. 7. Alterations in neuronal structural plasticity in the hippocampal CA1 zone following BMSC-exo treatment Stereotaxic injection surgery was performed 2 weeks prior to model initiation to allow for animal recovery and sparse labeling via viral expression. The injection volume was optimized through systematic testing (100/150/200 nL). The results showed that 150 nL provided an optimal neuronal labeling density without inducing tissue deformation, thereby enabling reliable single-neuron structural reconstruction (Supplementary Fig. 8, ​ Supplementary video). We similarly established distinct observation time points, and the comprehensive experimental protocol is clearly outlined in Fig. 6 A. Fig. 6. Open in a new tab BMSC-exos restore CTE-induced structural deficits in hippocampal CA1 neurons. A Experimental timeline: Stereotaxic injection was performed 2 weeks before modeling, followed by CTE induction (4 weeks). Tissues were harvested at the indicated time points. The data shown in this figure represent​ the analysis conducted on the 28th day after modeling. B – D Representative images of CA1 pyramidal neurons (sparsely GFP-positive neurons): Top left: 20× magnification image showing dendrites (scale bar: 50 μm). Top right: schematic diagram of Sholl analysis. Bottom: 100× magnification for basal dendritic spine quantification (scale bar: 10 μm). E , F Morphometric quantification: E The total dendritic length was decreased in the CTE28 group compared with that in the control group, which was reversed by exosomes (CTE28 + EXO). F The number of dendritic branches was decreased in the CTE28 subgroup but was restored in the CTE28 + EXO subgroup. G Sholl analysis of neuronal complexity: Compared with both the control and CTE28 + EXO groups, the CTE28 group showed a mild reduction in complexity. Severe deficits appeared at intermediate distances (70 μm and 100–120 μm). *p < 0.05 Control group vs. CTE28 group; #p < 0.05 CTE28 group vs. CTE28 + EXO group. H Basal dendritic spine density: The spine density was significantly lower in the CTE28 group than in the control group. Treatment with BMSC-exos (CTE28 + EXO) restored the spine density to a level comparable to that in the control group, and it was significantly greater than that in the CTE28 group. Data are presented as the mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 On the 28th day post-modeling, compared with control mice, CTE28 mice exhibited a significantly lower total dendritic length (p = 0.0011) and lower dendrite branch number (p = 0.0002) in CA1 pyramidal neurons. These morphological impairments were ​​rescued by exosome treatment (CTE28-EXO), which restored the values to control levels (Fig. 6 E, F). Sholl analysis of ​​neuronal complexity​​ revealed that neurons from mice in the CTE28 group had ​​mildly reduced complexity​​ compared with those from the control and CTE28-EXO groups. Notably, ​​severe complexity deficits​​ occurred at intermediate distances (70 μm, 100–120 μm) from the soma (Fig. 6 G). ​​Dendritic spine density​​ was markedly lower in the CTE28 group than in the control and CTE28-EXO groups (p = 0.0462; Fig. 6 H). Collectively, these data indicate that CTE modeling induces structural damage to CA1 neurons, whereas BMSC-exos effectively reverse these pathological alterations. Similarly, ​​to assess the ​​structural plasticity of CA1 neurons, we also conducted ​​intermediary time-point analyses​​ on the 7th and 14th days post-modeling. ​​Quantitative morphometry, including the total dendrite length, dendrite number, Sholl analysis, and spine density, revealed no significant intergroup differences​ (Supplementary Fig. 9). Evaluation of ​​acute-phase CTE effects​​ on the 3rd day post-modeling indicated no​​ obvious alterations in neuronal structural plasticity​​ (Supplementary Fig. 10), suggesting that ​​CTE-induced structural damage develops progressively rather than acutely. Alterations in CTE-associated biochemical pathologies following BMSC-exo treatment To elucidate the molecular mechanisms underlying the behavioral and structural improvements, we analyzed hippocampal tissues on the 28th day after modeling. Western blot analysis revealed that the level of phosphorylated Tau (p-Tau) and the level of the postsynaptic density protein PSD-95 were significantly greater and lower, respectively, in the CTE28 group than in the control group (Fig. 7 B, C). BMSC-exos effectively reversed these pathological changes. ​​Notably, the expression of the presynaptic protein synaptophysin remained comparable across all groups (Fig. 7 D), suggesting a specific vulnerability of the postsynaptic compartment in our CTE model.​​ Fig. 7. Open in a new tab BMSC-exos modulate CTE-related protein expression at 28th day post-modeling. A Representative Western blot images of proteins from hippocampal tissues. B Phosphorylated tau (p-Tau) levels were significantly greater in the CTE28 group than in the control group and were attenuated by BMSC-exo treatment. C The level of the postsynaptic density protein PSD-95 was significantly decreased in the CTE28 group, and BMSC-exo treatment restored its level to that in the control group. D The expression of the presynaptic protein synaptophysin remained comparable across all groups. E The expression of the astrocyte marker GFAP was significantly elevated in the CTE28 group, and BMSC-exo treatment did not alter this increase. F The expression of the microglial marker Iba1 was significantly increased in the CTE28 group, and BMSC-exo treatment significantly reduced its expression level. Data represent the mean ± SEM; *p < 0.05, **p < 0.01 We next assessed neuroinflammation. CTE28 injury induced significant activation of both microglia (elevated Iba1 expression) and astrocytes (elevated GFAP expression) (Fig. 7 E, F). BMSC-exo treatment significantly attenuated microglial activation, as evidenced by reduced Iba1 levels. ​​However, the elevated GFAP level was not significantly reduced by the treatment, indicating a differential effect on glial cell populations.​​ Furthermore, analysis of hippocampal homogenates using the Meso Scale Discovery V-PLEX Platform (MSD, USA, Kit #K15048D-2) revealed that CTE28 injury upregulated the expression of several proinflammatory cytokines, including IL-12p70, KC/GRO, and TNF-α, whereas BMSC-exo treatment suppressed this cytokine surge (Supplementary Fig. 11). ​​ Discussion Chronic traumatic encephalopathy (CTE) is a progressive neurodegenerative disorder driven by repetitive mild traumatic brain injuries and is prevalent among contact sports athletes and military personnel. Despite its significant public health burden, research on CTE pathogenesis and therapeutics has progressed slowly [ 2 , 26 ]. The development of valid animal models is crucial for advancing CTE research, which requires that these models faithfully recapitulate the core clinical features of human CTE (such as cognitive decline, anxiety, etc.), as well as the characteristic delayed symptom onset.​​ Patients with CTE typically exhibit dementia-like symptoms after a latent period of several years following the initial trauma, often without any prior symptoms before clinical onset [ 27 ]. Existing models often cannot mimic the chronicity and behavioral complexity of CTE, with many exhibiting acute neuronal injury but not the ​​delayed neurobehavioral sequelae​​ of clinical CTE [ 26 , 28 , 29 ]. In the current study, we addressed these gaps by establishing a novel murine CTE model optimized through ​​prolonged, low-intensity repetitive head impacts​​ (28 daily closed-head injuries; CTE28). This model successfully simulates core clinical features of CTE, as follows: ​​Delayed symptom onset: Anxiety-like behaviors and cognitive impairments emerged acutely (3rd day), resolved transiently (7th and 14th day), and reemerged chronically (day 28), mirroring the ​​latency period​​ observed in human CTE where symptoms manifest years after injury exposure. However, in our investigation of the therapeutic efficacy of exosomes, PBS-treated CTE28 mice exhibited an exacerbation of anxiety-like behavior. This likely reflects the ​​aversive nature of restraint stress during tail vein injections, which is known to acutely dysregulate limbic neurocircuitry and monoamine transmission [ 30 , 31 ]. Notably, restraint stress has been empirically demonstrated to induce significant affective disturbances in murine models, including increased anxiety-like phenotypes [ 32 , 33 ].​​ This underscores the need for stress-minimized administration routes in future CTE therapeutic studies. ​​ Notably, performance in the three-chamber test reflects a composite of several behavioral domains, including social motivation, social recognition memory, general novelty-seeking tendency, and anxiety-related avoidance. To disentangle these potential contributors, we first examined sociability during the pre-test phase (stranger mouse vs. empty cage). All groups showed intact social motivation, ruling out a general lack of interest in social stimuli as an explanation for subsequent deficits. The post-test phase then assessed social memory and novelty preference (familiar mouse vs. novel mouse). Furthermore, the possibility that reduced social investigation was merely a consequence of generalized anxiety was mitigated by the normal social approach behavior seen in the pre-test phase. Notably, a paradoxical discrepancy emerged between cognitive performance in the Y-maze and social cognition during the three-chamber test (post-test); whereas Y-maze spontaneous alternation deficits showed transient recovery at intermediate timepoints, social cognition impairment persisted throughout the observation period.​​ This dissociation likely stems from the increased complexity of social cognitive processing—wherein CTE mice exhibited temporary behavioral compensation in simpler tasks (e.g., alternation in the Y-maze test and basic sociability in the pre-test) but failed to restore higher-order social novelty discrimination. However, this type of cognitive dysfunction is relatively subtle and typically does not affect general social functions; thus, it may not have received sufficient attention in CTE patients [ 34 ]. This highlights the potential importance of investigating early symptoms of CTE from the perspective of social-related cognitive impairment. Notably, models with shorter injury durations (CTE1 and CTE7) failed to sustain chronic symptoms, underscoring the necessity of prolonged injury exposure to recapitulate the insidious progression of CTE. These results could provide an internal basis for our model selection. In addition to the behavioral phenotypes, our model exhibited CTE-relevant pathological alterations, notably an increase in p-Tau levels at the 28-day post-modeling time point. The combination of biochemical (Western blot for Thr181) and histopathological (IHC for Ser202/Thr205) evidence suggested that the model captured key aspects of tau pathology relevant to CTE. This may provide a useful platform for evaluating potential therapeutic interventions aimed at mitigating tau-related pathophysiology in a CTE context. However, the current study did not investigate long-term (e.g., 3–12 months post-injury) neuropathological hallmarks, such as progressive perivascular tau pathology, astroglial scarring at sulcal depths, or cavum septum pellucidum. The investigation of such chronic progression represents an important direction for future research to fully validate the model's pathological specificity. The primary strength of our CTE28 paradigm lies in its ability to provide a therapeutically tractable platform that recapitulates the delayed-onset functional deficits within a feasible experimental timeframe, thereby offering a robust system for mechanistic and therapeutic exploration. Mesenchymal stem cells (MSCs) derived from various sources have been demonstrated to be effective in the treatment of nerve injuries [ 35 ]. BMSC-exos have emerged as a promising “cell-free therapeutic strategy” in neural regenerative medicine owing to their low immunogenicity, high biocompatibility, and ability to cross the blood–brain barrier [ 9 ]. These nanoscale vesicles carry bioactive molecules such as proteins, mRNAs, and miRNAs, which mediate anti-inflammatory responses [ 36 ], promote neuroplasticity [ 12 , 37 ], and enhance angiogenesis [ 38 – 40 ]. Previous evidence has suggested the therapeutic potential of mesenchymal stem cell-derived exosomes in neurodegenerative pathologies [ 41 – 43 ]. However, there is a scarcity of direct evidence supporting the therapeutic application of BMSC-exos for CTE, which possesses a neuropathological profile distinct from other neurodegenerative disorders. Our current study could provide important evidence​ supporting the potential of BMSC-exos as a therapeutic strategy for CTE. On the basis of our previously optimized animal models, we directly investigated the therapeutic effects of BMSC-exos on CTE model mice by performing longitudinal assessments across multiple time points to evaluate dynamic neurobehavioral and neuropathological outcomes.​ The results demonstrated that BMSC-exos effectively ameliorated key behavioral phenotypes in CTE model mice, including anxiety-like behaviors (reduced center time, decreased center entries, and increased thigmotaxis in the open-field test), cognitive dysfunction (decreased alternation% in the Y-maze test), and social novelty discrimination deficits (impaired ability to distinguish novel mice in the three-chamber test). ​​Compelling evidence has indicated that BMSC-exos can ameliorate cognitive deficits in neurodegenerative pathologies such as Alzheimer's disease (AD) [ 16 ] and Parkinson's disease (PD) [ 17 ]. ​​The restoration of social-related cognition (assessed in the post-test phase of the three-chamber test) by BMSC-exos was not immediate but manifested at 7 days postinjection (14th day post-modeling). This delayed efficacy may reflect the time required for exosomal trafficking across the blood–brain barrier (BBB) [ 44 ] and/or intracellular processing to activate neuroprotective pathways [ 45 , 46 ].​ The PKH-67 tracking data provided direct evidence that intravenously infused BMSC-exos could reach the hippocampal formation and be internalized by both neurons and microglia. This may support a multifaceted mechanism of action: a direct effect​ on neuronal structural plasticity and survival via the delivery of exosomal cargo and an indirect, modulatory effect​ on the neuroinflammatory milieu by influencing microglial activity. BMSC-exos have therapeutic effects on neurodegenerative pathologies through multiple mechanisms, including but not limited to modulating neuroinflammation [ 36 ], promoting neurogenesis [ 24 , 47 ], and enhancing angiogenesis [ 38 – 40 ]. Among these, neurogenesis represents a particularly promising direction. Although neurodegenerative diseases exhibit heterogeneous pathogenesis and symptoms (e.g., AD, PD, and CTE), they universally involve the degeneration or death of central neurons. BMSC-exos orchestrate neural repair through the synergistic delivery of neurotrophic factors (BDNF, GDNF, and VEGF) and proneurogenic miRNAs (e.g., miR-133b) to stimulate NSC proliferation, axonal regrowth, and synaptogenesis [ 12 , 37 ]. A previous study demonstrated that BDNF-mediated BMSC-exos could increase neurogenesis and suppress apoptosis in rats following brain injury [ 48 ]. Furthermore, BMSC-exos can deliver miR-133b to astrocytes and neurons, regulating gene expression profiles to promote neurite remodeling and functional recovery after stroke [ 49 ]. Therefore, we explored the possible mechanisms through which BMSC-exos exert therapeutic effects from the perspective of neurogenesis.​ Notably, at 28 days post-modeling—a critical time point coinciding with symptom re-emergence in CTE mice and the peak therapeutic efficacy of BMSC-exos—BMSC-exos did not alter the number of Sox2-Nestin–double-positive NSCs in the SGZ but significantly restored the population of DCX-positive immature neurons, which was markedly reduced in the CTE28 group.​ Dynamic monitoring of these indicators revealed that the number of NSCs initially increased following establishment of the CTE model. Similar findings have been reported in other studies, where murine models subjected to traumatic brain impact or stroke also exhibited an increase in NSC populations [ 50 – 52 ]. In addition, in the initial days following the administration of BMSC-exos (7 days post-modeling, which coincides with the initial phase after the completion of exosome administration), CTE injury alone increased the NSC pool, and BMSC-exo treatment did not further amplify this expansion. However, our results demonstrated a gradual decline in NSC numbers over time, after which they eventually returned to baseline levels. Throughout this process, BMSC-exos did not directly promote NSC expansion. Specifically, the increase in NSCs observed in the CTE28 + EXO group was attributable to CTE modeling rather than exosome administration.​​ ​​Neural progenitor cells (immature neurons expressing DCX), derived from the differentiation of NSCs, represent a committed lineage in the neuronal differentiation pathway and serve as proliferative cells during early neuronal development [ 7 , 8 ]. These immature neurons migrate to the hippocampus and integrate into existing neural circuits by forming novel synaptic connections, thereby participating in hippocampus-dependent processes, such as learning and memory functions [ 6 , 53 ].​ In CTE model mice, DCX expression in the SGZ remained comparable to control levels during the early post-modeling phase. However, a significant decline was observed by the 28th day post-modeling. Integrated with the dynamic changes in NSCs, this phenomenon likely reflects a ​​compensatory activation of neurogenesis, wherein the expanded NSC pool differentiates into immature neurons [ 54 , 55 ]. Nevertheless, the efficiency of NSC-to-immature neuron conversion appears to be impaired under CTE pathology, as evidenced by the absence of a concurrent increase in DCX-positive immature neurons despite transient NSC expansion. Consequently, DCX expression decreased significantly once NSC numbers returned to baseline levels. Conversely, BMSC-exo treatment significantly elevated DCX expression in CTE model mice. Notably, at 28 days post-modeling—when NSCs had returned to baseline levels—DCX expression remained comparable to control levels and was significantly higher than that in untreated CTE28 mice. These results suggest that BMSC-exos enhance neuronal differentiation efficiency and immature neuron survival rather than merely expanding the NSC pool. We concurrently observed that, regardless of BMSC-exo treatment, the number of EdU-positive cells in the SGZ of mice subjected to CTE modeling was significantly greater than that in the control group. EdU (5-ethynyl-2'-deoxyuridine) serves as an S-phase marker of DNA synthesis, functioning similarly to BrdU, and can label all proliferating cell types [ 56 ]. Notably, in the absence of colabeling with cell type–specific markers (such as GFAP, Iba1, or Olig2), the observed increase in EdU-positive cells should be interpreted as a general increase in cellular proliferation rather than as specific evidence for enhanced neurogenesis. Under physiological conditions, EdU-positive cells in the SGZ are predominantly related to neurogenesis (e.g., neural stem/progenitor cells and immature neurons). However, under pathological conditions such as injury, inflammation, or aging, the proportions of EdU-positive glial cells (e.g., activated microglia or oligodendrocyte precursor cells) and vascular endothelial cells increase markedly [ 56 – 58 ]. This shift in cellular composition may explain the observed discrepancy between EdU labeling trends and other neurogenesis-specific markers (e.g., DCX, Sox2, or Nestin) in the CTE model. The detection of EdU-NeuN–double-positive cells confirms that the proliferating cells differentiated into mature neurons. The significantly increased proportion of these cells in BMSC-exo-treated mice demonstrates its role in promoting not only neurogenesis (increased DCX) but also the subsequent maturation and survival of newborn neurons. The observed restoration of neuronal structural plasticity in the hippocampal CA1 region following BMSC-exo treatment represents another mechanism underlying the amelioration of cognitive and behavioral deficits in our CTE model. CA1 is a major output node critical for memory processing. Our data demonstrate that CTE modeling induced significant dendritic atrophy and spine loss in CA1 pyramidal neurons after a relatively long period of time (28th day), and these effects were effectively reversed by the systemic administration of BMSC-exos. Our spine analysis quantified the total density, not the subtype-specific morphology (e.g., mushroom, thin). While this would provide a robust measure of overall synaptic structural integrity, future studies employing higher-resolution imaging could elucidate whether BMSC-exos preferentially rescue specific spine subtypes. The concurrent increase in the postsynaptic density protein PSD-95 could provide complementary evidence for the restoration of mature, functional synapses following BMSC-exo treatment. The functional significance of this structural restoration was also supported by the results of the hippocampus-dependent Morris water maze test. Deficits in spatial learning and long-term memory were observed in CTE mice, and these deficits were significantly reversed by BMSC-exo treatment, providing evidence that the recovery of CA1 dendritic complexity and synaptic integrity could be linked to improvements in cognitive function. This structural recovery aligns with emerging evidence that exosomal cargo can modulate synaptic integrity and neuronal connectivity in neurodegenerative contexts [ 59 – 61 ]. ​​Proteomic sequencing also revealed that exosomes derived from stem cells contain multiple neuroprotective factors and promote neurogenesis as well as neuronal structural integrity [ 13 , 62 ].​ The absence of severe acute structural damage in CA1 neurons reflects ​​multilayered neurobiological resilience, where endogenous stabilizers (Reelin, CSPGs), metabolic adaptations, and transient plasticity suppression may collectively delay morphological decline [ 63 , 64 ]. This silent interval underscores the time-dependent pathogenesis of CTE, wherein patients typically remain asymptomatic for years or decades after initial trauma exposure before clinical onset, [ 27 ] ​​and whereby initial compensatory mechanisms eventually succumb to cumulative damage factors​ [ 65 , 66 ] (e.g., impaired maturation of immature neurons, neuroinflammation cascades), ​​culminating in the chronic-phase neuronal structural deficits and behavioral abnormalities previously discussed​​. This window of intact structural plasticity represents a critical opportunity for early intervention with neuroprotective agents (e.g., BMSC-exos) to halt progression. We acknowledge the limitation of not directly tracking newborn granule cells, which could directly link neurogenesis to learning and memory. Future studies may use more specialized lineage-tracing approaches (e.g., pulse labeling methods based on genetically engineered mice [ 67 ]) to investigate the maturation and integration of newborn neurons and their relationships with cognitive function. While our data highlight the significant restoration of hippocampal neurogenesis and neuronal structural plasticity as fundamental cellular correlates of functional recovery, these processes are unlikely to occur in isolation. To elucidate the potential mechanisms through which BMSC-exos confer these restorative effects, we performed biochemical analyses to investigate tau pathology, synaptic integrity, and neuroinflammation.​​ The analysis revealed that our CTE model is associated with significant tau hyperphosphorylation and a ​​specific loss of the postsynaptic scaffolding protein PSD-95​​, while the presynaptic protein synaptophysin remained intact, suggesting a specific postsynaptic vulnerability in our model. The specific restoration of PSD-95, along with the reduction in p-tau, suggests that BMSC-exos may also promote synaptic and neuronal homeostasis. The therapeutic efficacy of mesenchymal stem cell-derived exosomes (MSC-exos) is recognized to be multifaceted and to involve mechanisms such as immunomodulation, angiogenesis, and metabolic regulation [ 36 , 38 , 68 , 69 ]. The miRNAs they carry also regulate various processes, including neuronal apoptosis, M2 microglial polarization, and autophagy, through pathways such as the Wnt/β-catenin and TLR4/NF-κB pathways [ 36 , 70 ]. The potential of BMSC-exos to reduce tau hyperphosphorylation in CTE may involve several synergistic mechanisms. Exosomes may modulate the phosphorylation balance by targeting pivotal tau kinases, such as GSK-3β and CDK5, and phosphatases, such as PP2A [ 71 – 73 ]. Additionally, they may exert indirect effects by ameliorating the chronic neuroinflammatory milieu and promoting autophagy. Notably, in Alzheimer's disease models, MSC-derived exosomes have shown promise in reducing Aβ and p-tau accumulation and alleviating astrogliosis, effects associated with the modulation of pathways such as the PI3K/Akt/mTOR pathway [ 74 ]. However, whether an identical mechanistic framework operates in CTE remains to be determined. BMSC-exo treatment significantly attenuated the chronic neuroinflammatory state in CTE mice, as evidenced by reduced microglial activation (marked by Iba1) and decreased levels of proinflammatory cytokines. Future studies using more specific phenotypic markers will be valuable for further delineating the exact nature of microglial polarization induced by BMSC-exos in a mouse model of CTE. ​​However, the sustained high level of GFAP following BMSC-exo treatment requires careful interpretation. This persistent elevation should not be automatically equated with maladaptive gliosis. Growing evidence suggests that astrocytes can adopt a spectrum of phenotypes, ranging from harmful (A1, neurotoxic) to beneficial (A2, neuroprotective), both of which are associated with upregulated GFAP expression [ 75 ]. In the context of our findings—where BMSC-exos successfully attenuated microglial activation and proinflammatory cytokine secretion and promoted functional and structural recovery—the persistent GFAP signal may paradoxically indicate a modulated, potentially protective astrocytic response. It is plausible that BMSC-exos aid in shifting the astrocyte population toward an A2-like, neuroprotective phenotype that contributes to trophic support, extracellular matrix remodeling, and the containment of neuroinflammation, thereby facilitating the observed neural repair. Definitive confirmation of this hypothesis awaits future investigation using specific A1/A2 marker analysis. By dampening the chronic neuroinflammatory state characteristic of CTE, BMSC-exos may create a conducive milieu that supports the survival and integration of newborn neurons and preserves the existing synaptic architecture. However, the precise cellular targets remain undefined. Future work utilizing techniques such as cell-specific knockout models, advanced in vitro co-cultures, or single-cell RNA sequencing will be critical to determine the relative contribution of BMSC-exos actions on neurons, astrocytes, microglia, and other cell types within the complex neural environment. In summary, our study optimized a CTE model that recapitulates the delayed symptom onset characteristic of human CTE, providing a validated platform for therapeutic screening. We then evaluated the therapeutic potential of BMSC-exos, demonstrating their efficacy in alleviating CTE-associated behavioral deficits through the dynamic restoration of hippocampal neurogenesis and/or neuronal structural plasticity. However, limitations include that the precise molecular drivers within BMSC-exos (e.g., specific neuroprotective proteins or miRNAs) responsible for the observed therapeutic actions, as well as the definitive causal links between exosome-induced plasticity and functional recovery, have not been fully elucidated. The synergistic action of these cargoes may contribute to the differentiation efficiency of neural stem/progenitor cells into immature neurons and support their subsequent maturation. While the ultracentrifugation method used in this study is well-established, it may potentially co-isolate non-vesicular contaminants, such as protein aggregates [ 76 ]. Our rigorous characterization via NTA, TEM, and Western blot could support that the observed therapeutic effects are primarily attributable to BMSC-exos. Future investigations employing advanced techniques such as size-exclusion chromatography will be valuable to further minimize contaminants and precisely delineate the active cargo. Furthermore, the precise mechanisms by which intravenously injected exosomes cross the blood–brain barrier to reach the hippocampus (e.g., via transcytosis or injury-induced permeability) and the exact timing of their arrival in the CNS remain to be fully characterized.​ Future work may address these gaps to elucidate therapeutic strategies. Our study suggests that BMSC-exos could enhance neurogenesis; however, the methodology employed did not allow us to trace the fate of individual cells over time or elucidate the precise mechanistic contributions of BMSC-exos to specific phases of neurogenesis. Future studies utilizing longitudinal lineage-tracing techniques could help clarify the dynamic cellular behaviors and underlying mechanisms involved. In addition, the absence of an uninjured control group treated with BMSC-exos alone in the current study was also a limitation. This design choice was made to prioritize resources for evaluating therapeutic efficacy within the CTE model and to maintain the feasibility of our longitudinal behavioral assessment protocol. Future research could refine the experimental design to further validate the specificity of our findings. Collectively, the findings of this study provide foundational insights for developing exosome-based interventions against CTE while also offering novel therapeutic perspectives for other neurodegenerative diseases.​ Supplementary Information 13287_2026_4943_MOESM1_ESM.tif (7.5MB, tif) Supplementary material 1. Supplementary Fig. 1. The schematic of the modeling device and supplementary behavioral assessments in an independent cohort of mice at 28 days post-modeling. (A) The schematic of the modeling device. Key components are labeled: 1, baseplate; 2, mounting rod; 3, rotatable tube clamp; 4, plastic trough; 5, foam board lining; 6, protective net pouch; 7, PVC tube with side openings. The red dashed circle​ on the mouse's head indicates the permissible target area​ for the impact. (B) Percentage of exploration time during the training phase of the novel object recognition test (NORT). Both control and CTE28 mice showed comparable exploration times for the two identical objects, indicating that there was no inherent object preference bias. (C) Recognition index in the test phase of the NORT. Compared with control mice, CTE28 mice exhibited a significantly lower recognition index (p < 0.05), indicating impaired nonspatial recognition memory. (D) Immobility time in the forced swim test (FST). Compared with control mice, CTE28 mice exhibited a significantly increased immobility time (p < 0.05), suggesting increased depressive-like behavior. (E) Representative Western blot images (top) and quantitative analysis (bottom) of p-tau (Thr181) (50 kDa) protein levels in hippocampal tissues normalized to that of GAPDH (36 kDa). Compared with control mice, CTE28 mice had significantly greater p-tau levels (p < 0.05), confirming the presence of a key CTE-related neuropathology in the model. (E–I) Representative immunohistochemical images (left, scale bar: 50 μm) and corresponding quantification (right) of p-tau (Ser202/Thr205) positive area in the cerebral cortex (CTX) and hippocampal dentate gyrus (DG). CTE28 mice showed a significant increase in p-tau immunoreactivity in both regions (p < 0.05). The data are presented as the mean ± SEM. *p < 0.05. 13287_2026_4943_MOESM2_ESM.tif (1.8MB, tif) Supplementary material 2. Supplementary Fig. 2. Behavioral consequences of CTE1 and CTE7. (A) The latency to fall in the rotarod test across post-modeling intervals indicates comparable motor function among all groups. (B) The total locomotion distance in the open-field test confirms intact basal motor activity. (C) CTE7 mice exhibited anxiety-like behavior, which manifested as a significantly reduced central zone time (p < 0.05 vs. control & CTE1) on post-modeling days 3 and 14, whereas CTE1 mice remained unaffected. (D) Spontaneous alternation percentage in the Y-maze test reveals no persistent cognitive impairment in the CTE groups across the timepoints. (E-F) Sociability phase: All groups showed intact social preference (p < 0.05 stranger vs. empty cage). (G-H) Social novelty phase: CTE7 mice displayed persistent social recognition deficits, as demonstrated by a lack of novelty preference (novel vs. familiar mouse exploration time: p > 0.05) at all time points. The data are presented as the mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001. 13287_2026_4943_MOESM3_ESM.tif (18.2MB, tif) Supplementary material 3. Supplementary Fig. 3. Characterization of BMSCs and derived exosomes (BMSC-exos). (A) Phenotypic characterization by flow cytometry. The BMSCs were positive for CD29, CD44, and Sca-1 (Ly-6A) and negative for CD31 and CD117. (B) Functional characterization by trilineage differentiation. Adipogenic (oil red O staining, lipid droplets, left), osteogenic (alizarin red S staining, calcium deposition, middle), and chondrogenic (alcian blue staining, proteoglycans, right) differentiation confirmed cell multipotency. (C) Nanoparticle tracking analysis showing the size distribution of BMSC-exos (inset: nanoparticle image). (D) Western blot analysis of exosomal markers in the supernatant (lane 1), BMSCs (lane 2), and BMSC-exos (lane 3). (E) TEM image of BMSC-exos showing a typical cup-shaped morphology. Scale bars: B, 200 μm (left & middle), 50 μm (right); E, 100 nm. These data confirm the successful isolation and identity of BMSCs and exosomes. 13287_2026_4943_MOESM4_ESM.tif (671.6KB, tif) Supplementary material 4. Supplementary Fig. 4. BMSC-exos ameliorated anxiety-like and cognitive deficits in CTE mice at 28th day post-modeling. (A) Number of entries into the center zone during the first 5 min of the open-field test. Compared with the control mice, the CTE28 mice showed a significant decrease in the number of center zone entries, which was reversed by treatment with BMSC-exos. (B) Average locomotor velocity during the test. No significant differences were observed among the groups, indicating that motor function was not impaired. (C, D) Time spent in the side (C) and corner (D) areas, reflecting thigmotaxis. Compared with control mice, CTE28 mice spent significantly more time in these peripheral zones, which was attenuated by BMSC-exo administration. (E)​ Morris water maze performance during the training phase. Escape latency across training days are shown. (F-H) Spatial memory assessed in the probe trial. The latency to first reach the platform (F), the number of target platform crossings (G) and the time spent in the target quadrant (H) are shown. CTE28 mice exhibited prolonged latencies, made fewer crossings over the former platform location and spent less time in the target quadrant (p < 0.05), indicating impaired memory retention. These deficits were rescued by BMSC-exo treatment. The data are presented as the mean ± SEM. *p < 0.05. 13287_2026_4943_MOESM5_ESM.tif (1.5MB, tif) Supplementary material 5. Supplementary Fig. 5. Effects of BMSC-exos on the behavior of CTE model mice on the 7th and 14th days. (A) Schematic workflow of the longitudinal behavioral assessment. (B-F)​ Behavioral assessments on the 7th day. (B) Time spent in the center area during the open-field test, indicating anxiety-like behavior. Compared with the control group, both the CTE28 and the CTE28 + EXO groups showed significant decreases. (C-D) Pre-test of the three-chamber test. All groups spent significantly more time investigating the cage containing the stranger mouse than the empty cage (p < 0.05 vs. empty cage), indicating intact social motivation. (E-F) Exploration time and social novelty index in the post-test of the three-chamber test, indicating social cognitive function. Mice in both CTE-exposed groups exhibited impaired social novelty preference compared with controls. No significant difference was observed between the CTE28 and CTE28 + EXO groups at this time point. (G-K)​ Behavioral assessments on the 14th day. (G)​ Time spent in the center area. Compared with the control group, the CTE28 group remained significantly impaired, while the CTE28 + EXO group showed restoration to a level comparable to that of the controls. (H-I) Pre-test of the three-chamber test. (J-K) Exploration time and social novelty index in the post-test of the three-chamber test. The CTE28 group continued to show significant impairment, whereas the CTE28 + EXO group exhibited restored social novelty preference. Data represent the mean ± SEM. *p < 0.05, **p < 0.01. 13287_2026_4943_MOESM6_ESM.tif (16.4MB, tif) Supplementary material 6. Supplementary Fig. 6. BMSC-exos promote the differentiation of newborn cells in the hippocampus of CTE mice at 28 days post-modeling. (A) Schematic timeline of the experimental procedure, including CTE modeling, BMSC-exo administration, and tissue collection. (B) Representative immunofluorescence images of hippocampal sections from the indicated groups showing EdU (red, proliferating cells), NeuN (green, mature neurons), and DAPI (blue, nuclei). Scale bar, 50 μm. (C) Quantification of EdU-positive cells in the hippocampal subgranular zone. The number of proliferating cells was greater in the CTE28 + EXO group than in the CTE28 group. (D) Quantification of the percentage of EdU-positive cells that coexpressed NeuN. The proportion of newborn cells differentiating into mature neurons was significantly greater in the CTE28 + EXO group than in the CTE28 group. Data are presented as the mean ± SEM. *p < 0.05. 13287_2026_4943_MOESM7_ESM.tif (31MB, tif) Supplementary material 7. Supplementary Fig. 7. Changes in neurogenesis during the acute phase on the 3rd day after CTE modeling. (A) Representative immunofluorescence images of the SGZ: DAPI (blue), Sox2 (red), Nestin (green), and DCX (cyan). Scale bars: 50 μm. (B) EdU labeling (proliferating cells, red) in the same groups. Scale bars: 100 μm. (C-E) Quantitative analyses (3rd day): (C) The number of Sox2-Nestin–double-positive neural stem cells was greater in the CTE28 group than in the control group. (D) The number of DCX-positive immature neurons showed no significant difference. (E) The number of EdU-positive proliferating cells was markedly increased in the CTE28 group compared with the control group. Data are shown as the mean ± SEM. *p < 0.05, ***p < 0.001. 13287_2026_4943_MOESM8_ESM.tif (3.5MB, tif) Supplementary material 8. Supplementary Fig. 8. Optimization of viral injection volume for sparse neuronal labeling in the hippocampal CA1 region.​ Representative maximum-intensity projection confocal micrographs of mouse hippocampal CA1 following the stereotaxic injection of graded viral volumes: 100 nL, ​​150 nL,​ and ​​200 nL​​. 13287_2026_4943_MOESM9_ESM.tif (745.5KB, tif) Supplementary material 9. Supplementary Fig. 9. Structural plasticity alterations in hippocampal CA1 neurons during intermediate phases after CTE modeling. (A-D) Data from the 7th day post-modeling: Quantitative morphometry of CA1 pyramidal neurons in the control (green), CTE28 (blue), and CTE28 + EXO (red) groups: (A) Dendrite number. (B) Total dendrite length. (C) Sholl analysis showing comparable dendritic complexity across groups (10–200 μm from the soma). (D) Basal spine density quantification. (E-H) Data from the 14th day post-modeling: Consistently null results for (E) the dendrite number, (F) the dendrite length, (G) Sholl analysis, and (H) basal spine density. Data are shown as the mean ± SEM. 13287_2026_4943_MOESM10_ESM.tif (1.8MB, tif) Supplementary material 10. Supplementary Fig. 10. Changes in acute-phase structural plasticity at 3 days after CTE modeling. (A) Experimental timeline: Stereotaxic surgery at -2 weeks, CTE modeling lasting 4 weeks, tissue collection on the 3rd day. (B-D) Morphometric analyses across groups: (B) Dendrite length quantification. (C) Dendrite number. (D) Sholl analysis showing comparable dendritic complexity across groups (10–200 μm from the soma). (E-F) Representative images. (G) Comparable spine density among groups. These results conclusively demonstrate the absence of acute structural pathology in the CTE group at this stage. Data are shown as the mean ± SEM. *p < 0.05. 13287_2026_4943_MOESM11_ESM.tif (593.2KB, tif) Supplementary material 11. Supplementary Fig. 11. Proinflammatory cytokines in the hippocampus at 28 days post-modeling. Hippocampal homogenates were analyzed using the Meso Scale Discovery (MSD) V-PLEX platform (Kit #K15048D-2). The tissue homogenates were adjusted to a uniform protein concentration of 3.08 mg/mL, and 50 µL of each sample was analyzed. Data are presented as the mean ± SEM. 13287_2026_4943_MOESM12_ESM.tif (35.8MB, tif) Supplementary material 12. Supplementary Fig. 12. In vivo biodistribution and cellular uptake of PKH67-labeled BMSC-exos in the brain of CTE28 mice. Representative immunofluorescence images showing the localization of intravenously administered PKH67-labeled BMSC-exos in the hippocampal DG region In the merged images: Yellow arrowheads​ indicate co-localization of PKH67 signal with Iba1, demonstrating exosome uptake by microglia; Purple arrowheads​ indicate co-localization of PKH67 signal with NeuN, demonstrating exosome uptake by neurons. Scale bars: 50 µm. Supplementary material 13. (19.8KB, docx) Supplementary material 14. (25MB, wmv) Acknowledgements We thank Hui Liu for ordering the experimental animals and managing the breeding environments and Jing Yang for ordering the reagents. Author contributions Peng Liu (First Author):​​ Conceptualization, methodology, investigation, formal analysis, visualization, writing—original draft. ​​Zhuangzhuang Bai (Co-First Author):​​ Investigation, data curation, validation, writing—original draft. ​​Yongfei Yang:​​ Methodology, investigation, software. ​​Xiao Li:​​ Resources, investigation, validation. Jinghua Xia: Investigation, validation. ​​Qian Yang (Corresponding Author):​​ Conceptualization, funding acquisition, supervision, project administration, writing—review and editing. All authors reviewed the manuscript. Funding This research was supported by grants from the Air Force Medical University Talent Support “Lingyun Project” (2020lyjhyq), the National Science Foundation of China (Nos. 82221001 and 82301678), the Key Research and Development Projects of Shaanxi Province (2023-ZDLSF-52), the Xi'an Medical University Doctoral Research Startup Fund Project (2020DOC15) and the Shaanxi Provincial Department of Education Scientific Research Program Project (24JK0648). Data availability No datasets were generated or analysed during the current study. Declarations Ethics approval and consent to participate All the animal procedures were approved by the Institutional Ethics Committee of the Fourth Military Medical University (Title: Research on the Pathological Mechanisms and Intervention Strategies for Chronic Traumatic Encephalopathy, Approval No. 20210150, date of approval: January 1, 2021); the experiments were carried out in accordance with the National Research Council’s Guide for the Care and Use of Laboratory Animals (8th ed., National Research Council, 2011). Competing interests The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Peng Liu, Zhuangzhuang Bai and Yongfei Yang have contributed equally to this work. References 1. McKee AC, Stein TD, Huber BR, Crary JF, Bieniek K, Dickson D, et al. Chronic traumatic encephalopathy (CTE): criteria for neuropathological diagnosis and relationship to repetitive head impacts. Acta Neuropathol. 2023;145(4):371–94. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 2. Smith DH, Johnson VE, Trojanowski JQ, Stewart W. Chronic traumatic encephalopathy - confusion and controversies. Nat Rev Neurol. 2019;15(3):179–83. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 3. McAteer KM, Turner RJ, Corrigan F. Animal models of chronic traumatic encephalopathy. Concussion. 2017;2(2):Cnc32. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 4. Sugahara C, Kin K, Sasaki T, Sasada S, Kawauchi S, Yabuno S, et al. Repeated non-hemorrhagic and non-contusional mild traumatic brain injury in rats elicits behavioral impairment with microglial activation, astrogliosis, and tauopathy: reproducible and quantitative model of chronic traumatic encephalopathy. Brain Res. 2025;1850:149412. [ DOI ] [ PubMed ] [ Google Scholar ] 5. Ackermans NL, Varghese M, Wicinski B, Torres J, De Gasperi R, Pryor D, et al. Unconventional animal models for traumatic brain injury and chronic traumatic encephalopathy. J Neurosci Res. 2021;99(10):2463–77. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 6. Culig L, Chu X, Bohr VA. Neurogenesis in aging and age-related neurodegenerative diseases. Ageing Res Rev. 2022;78:101636. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 7. Moreno-Jiménez EP, Terreros-Roncal J, Flor-García M, Rábano A, Llorens-Martín M. Evidences for adult hippocampal neurogenesis in humans. J Neurosci. 2021;41(12):2541–53. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. Anacker C, Hen R. Adult hippocampal neurogenesis and cognitive flexibility - linking memory and mood. Nat Rev Neurosci. 2017;18(6):335–46. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 9. Andrzejewska A, Dabrowska S, Lukomska B, Janowski M. Mesenchymal stem cells for neurological disorders. Adv Sci. 2021;8(7):2002944. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 10. Ramos-Zaldívar HM, Polakovicova I, Salas-Huenuleo E, Corvalán AH, Kogan MJ, Yefi CP, et al. Extracellular vesicles through the blood-brain barrier: a review. Fluids Barriers CNS. 2022;19(1):60. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 11. Saikia B, Dhanushkodi A. Engineered exosome therapeutics for neurodegenerative diseases. Life Sci. 2024;356:123019. [ DOI ] [ PubMed ] [ Google Scholar ] 12. Xin H, Katakowski M, Wang F, Qian JY, Liu XS, Ali MM, et al. MicroRNA cluster miR-17-92 cluster in exosomes enhance neuroplasticity and functional recovery after stroke in rats. Stroke. 2017;48(3):747–53. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 13. Reza-Zaldivar EE, Hernández-Sapiéns MA, Gutiérrez-Mercado YK, Sandoval-Ávila S, Gomez-Pinedo U, Márquez-Aguirre AL, et al. Mesenchymal stem cell-derived exosomes promote neurogenesis and cognitive function recovery in a mouse model of Alzheimer’s disease. Neural Regen Res. 2019;14(9):1626–34. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 14. Rahmani A, Saleki K, Javanmehr N, Khodaparast J, Saadat P, Nouri HR. Mesenchymal stem cell-derived extracellular vesicle-based therapies protect against coupled degeneration of the central nervous and vascular systems in stroke. Ageing Res Rev. 2020;62:101106. [ DOI ] [ PubMed ] [ Google Scholar ] 15. Shu J, Jiang L, Wang M, Wang R, Wang X, Gao C, et al. Human bone marrow mesenchymal stem cells-derived exosomes protect against nerve injury via regulating immune microenvironment in neonatal hypoxic-ischemic brain damage model. Immunobiology. 2022;227(3):152178. [ DOI ] [ PubMed ] [ Google Scholar ] 16. Liu S, Fan M, Xu JX, Yang LJ, Qi CC, Xia QR, et al. Exosomes derived from bone-marrow mesenchymal stem cells alleviate cognitive decline in AD-like mice by improving BDNF-related neuropathology. J Neuroinflammation. 2022;19(1):35. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 17. Xu X, Li Z, Zuo H, Chen H, Gui Y. Mesenchymal stem cell-derived exosomes altered neuron cholesterol metabolism via Wnt5a-LRP1 axis and alleviated cognitive impairment in a progressive Parkinson’s disease model. Neurosci Lett. 2022;787:136810. [ DOI ] [ PubMed ] [ Google Scholar ] 18. Kane MJ, Angoa-Pérez M, Briggs DI, Viano DC, Kreipke CW, Kuhn DM. A mouse model of human repetitive mild traumatic brain injury. J Neurosci Methods. 2012;203(1):41–9. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 19. Holschneider DP, Guo Y, Roch M, Norman KM, Scremin OU. Acetylcholinesterase inhibition and locomotor function after motor-sensory cortex impact injury. J Neurotrauma. 2011;28(9):1909–19. [ DOI ] [ PubMed ] [ Google Scholar ] 20. Rein B, Ma K, Yan Z. A standardized social preference protocol for measuring social deficits in mouse models of autism. Nat Protoc. 2020;15(10):3464–77. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 21. Li Z, Liu C, Li S, Li T, Li Y, Wang N, et al. BMSC-derived exosomes inhibit dexamethasone-induced muscle atrophy via the miR-486-5p/FoxO1 axis. Front Endocrinol (Lausanne). 2021;12:681267. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 22. Zhang Y, Bai J, Xiao B, Li C. BMSC-derived exosomes promote osteoporosis alleviation via M2 macrophage polarization. Mol Med. 2024;30(1):220. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 23. Liu P, Chu Z, Lei G, Deng LS, Yang L, Dang YH. The role of HINT1 protein in morphine addiction: an animal model-based study. Addict Biol. 2021;26(2):e12897. [ DOI ] [ PubMed ] [ Google Scholar ] 24. Fallahi S, Zangbar HS, Farajdokht F, Rahbarghazi R, Ghiasi F, Mohaddes G. Mesenchymal stem cell-derived exosomes improve neurogenesis and cognitive function of mice with methamphetamine addiction: a novel treatment approach. CNS Neurosci Ther. 2024;30(5):e14719. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 25. Paxinos GF, Franklin K. The mouse brain in stereotaxic coordinates. Academic Press, 2003. 26. Xiong Y, Mahmood A, Chopp M. Animal models of traumatic brain injury. Nat Rev Neurosci. 2013;14(2):128–42. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 27. Guskiewicz KM, Marshall SW, Bailes J, McCrea M, Cantu RC, Randolph C, et al. Association between recurrent concussion and late-life cognitive impairment in retired professional football players. Neurosurgery. 2005;57(4):719–26. [ DOI ] [ PubMed ] [ Google Scholar ] 28. Gangolli M, Benetatos J, Esparza TJ, Fountain EM, Seneviratne S, Brody DL. Repetitive concussive and subconcussive injury in a human Tau mouse model results in chronic cognitive dysfunction and disruption of white matter tracts, but not Tau pathology. J Neurotrauma. 2019;36(5):735–55. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 29. Mannix R, Berglass J, Berkner J, Moleus P, Qiu J, Andrews N, et al. Chronic gliosis and behavioral deficits in mice following repetitive mild traumatic brain injury. J Neurosurg. 2014;121(6):1342–50. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 30. Echeverry MB, Guimarães FS, Del Bel EA. Acute and delayed restraint stress-induced changes in nitric oxide producing neurons in limbic regions. Neuroscience. 2004;125(4):981–93. [ DOI ] [ PubMed ] [ Google Scholar ] 31. Aceto G, Nardella L, Lazzarino G, Tavazzi B, Bertozzi A, Nanni S, et al. Acute restraint stress impairs histamine type 2 receptor ability to increase the excitability of medium spiny neurons in the nucleus accumbens. Neurobiol Dis. 2022;175:105932. [ DOI ] [ PubMed ] [ Google Scholar ] 32. Sánchez-Marín L, Flores-López M, Gavito AL, Suárez J, Pavón-Morón FJ, de Fonseca FR, et al. Repeated restraint stress and binge alcohol during adolescence induce long-term effects on anxiety-like behavior and the expression of the endocannabinoid system in male rats. Biomedicines. 2022;10(3):593. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 33. Xing B, Liu P, Jiang WH, Liu F, Zhang H, Cao GF, et al. Effects of immobilization stress on emotional behaviors in dopamine D3 receptor knockout mice. Behav Brain Res. 2013;243:261–6. [ DOI ] [ PubMed ] [ Google Scholar ] 34. Alosco ML, Mariani ML, Adler CH, Balcer LJ, Bernick C, Au R, et al. Developing methods to detect and diagnose chronic traumatic encephalopathy during life: rationale, design, and methodology for the DIAGNOSE CTE research project. Alzheimers Res Ther. 2021;13(1):136. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 35. Han Y, Li X, Zhang Y, Han Y, Chang F, Ding J. Mesenchymal stem cells for regenerative medicine. Cells. 2019;8(8):886. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 36. Quan J, Liu Q, Li P, Yang Z, Zhang Y, Zhao F, et al. Mesenchymal stem cell exosome therapy: current research status in the treatment of neurodegenerative diseases and the possibility of reversing normal brain aging. Stem Cell Res Ther. 2025;16(1):76. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 37. Reza-Zaldivar EE, Hernández-Sapiéns MA, Minjarez B, Gutiérrez-Mercado YK, Márquez-Aguirre AL, Canales-Aguirre AA. Potential effects of MSC-derived exosomes in neuroplasticity in Alzheimer’s disease. Front Cell Neurosci. 2018;12:317. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 38. Liu C, Yang TH, Li HD, Li GZ, Liang J, Wang P. Exosomes from bone marrow mesenchymal stem cells are a potential treatment for ischemic stroke. Neural Regen Res. 2023;18(10):2246–51. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 39. Zou Y, Liao L, Dai J, Mazhar M, Yang G, Wang H, et al. Mesenchymal stem cell-derived extracellular vesicles/exosome: a promising therapeutic strategy for intracerebral hemorrhage. Regen Ther. 2023;22:181–90. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 40. Chen J, Chopp M. Exosome therapy for stroke. Stroke. 2018;49(5):1083–90. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 41. Garbuzova-Davis S, Willing AE, Ehrhart J, Wang L, Sanberg PR, Borlongan CV. Cell-free extracellular vesicles derived from human bone marrow endothelial progenitor cells as potential therapeutics for microvascular endothelium restoration in ALS. Neuromolecular Med. 2020;22(4):503–16. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 42. He S, Wang Q, Chen L, He YJ, Wang X, Qu S. miR-100a-5p-enriched exosomes derived from mesenchymal stem cells enhance the anti-oxidant effect in a Parkinson’s disease model via regulation of Nox4/ROS/Nrf2 signaling. J Transl Med. 2023;21(1):747. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 43. Xie X, Song Q, Dai C, Cui S, Tang R, Li S, et al. Clinical safety and efficacy of allogenic human adipose mesenchymal stromal cells-derived exosomes in patients with mild to moderate Alzheimer’s disease: a phase I/II clinical trial. Gen Psychiatr. 2023;36(5):e101143. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 44. Chen CC, Liu L, Ma F, Wong CW, Guo XE, Chacko JV, et al. Elucidation of exosome migration across the blood-brain barrier model in vitro. Cell Mol Bioeng. 2016;9(4):509–29. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 45. Zhou Y, Wen LL, Li YF, Wu KM, Duan RR, Yao YB, et al. Exosomes derived from bone marrow mesenchymal stem cells protect the injured spinal cord by inhibiting pericyte pyroptosis. Neural Regen Res. 2022;17(1):194–202. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 46. Yan Q, Yin Y, Li X, Li M. Exosome-shuttled MYCBPAP from bone marrow mesenchymal stem cells regulates synaptic remodeling and ameliorates ischemic stroke in rats. J Chem Neuroanat. 2023;132:102309. [ DOI ] [ PubMed ] [ Google Scholar ] 47. Wei R, Zhang L, Hu W, Shang X, He Y, Zhang W. Zeb2/Axin2-enriched BMSC-derived exosomes promote post-stroke functional recovery by enhancing neurogenesis and neural plasticity. J Mol Neurosci. 2022;72(1):69–81. [ DOI ] [ PubMed ] [ Google Scholar ] 48. Xu H, Jia Z, Ma K, Zhang J, Dai C, Yao Z, et al. Protective effect of BMSCs-derived exosomes mediated by BDNF on TBI via miR-216a-5p. Med Sci Monit. 2020;26:e920855. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 49. Xin H, Li Y, Liu Z, Wang X, Shang X, Cui Y, et al. MiR-133b promotes neural plasticity and functional recovery after treatment of stroke with multipotent mesenchymal stromal cells in rats via transfer of exosome-enriched extracellular particles. Stem Cells. 2013;31(12):2737–46. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 50. Uyeda A, Muramatsu R. Molecular mechanisms of central nervous system axonal regeneration and remyelination: a review. Int J Mol Sci. 2020;21(21):8116. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 51. Li H, Ghorbani S, Oladosu O, Zhang P, Visser F, Dunn J, et al. Therapeutic reduction of neurocan in murine intracerebral hemorrhage lesions promotes oligodendrogenesis and functional recovery. J Neuroinflammation. 2025;22(1):2. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 52. Mao W, Yi X, Qin J, Tian M, Jin G. CXCL12 promotes proliferation of radial glia like cells after traumatic brain injury in rats. Cytokine. 2020;125:154771. [ DOI ] [ PubMed ] [ Google Scholar ] 53. Cope EC, Gould E. Adult neurogenesis, glia, and the extracellular matrix. Cell Stem Cell. 2019;24(5):690–705. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 54. Boldrini M, Fulmore CA, Tartt AN, Simeon LR, Pavlova I, Poposka V, et al. Human hippocampal neurogenesis persists throughout aging. Cell Stem Cell. 2018;22(4):589-599.e585. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 55. Heinrich C, Bergami M, Gascón S, Lepier A, Viganò F, Dimou L, et al. Sox2-mediated conversion of NG2 glia into induced neurons in the injured adult cerebral cortex. Stem Cell Rep. 2014;3(6):1000–14. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 56. Wang W, Wang M, Yang M, Zeng B, Qiu W, Ma Q, et al. Transcriptome dynamics of hippocampal neurogenesis in macaques across the lifespan and aged humans. Cell Res. 2022;32(8):729–43. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 57. Yamamoto M, Itokazu T, Uno H, Maki T, Shibuya N, Yamashita T. Anti-RGMa neutralizing antibody ameliorates vascular cognitive impairment in mice. Neurotherapeutics. 2025;22(2):e00500. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 58. Astakhova O, Ivanova A, Komoltsev I, Gulyaeva N, Enikolopov G, Lazutkin A. Traumatic brain injury promotes neurogenesis and oligodendrogenesis in subcortical brain regions of mice. Cells. 2025;14(2):92. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 59. Nakano M, Nagaishi K, Konari N, Saito Y, Chikenji T, Mizue Y, et al. Bone marrow-derived mesenchymal stem cells improve diabetes-induced cognitive impairment by exosome transfer into damaged neurons and astrocytes. Sci Rep. 2016;6:24805. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 60. Cao N, Liao T, Liu J, Fan Z, Zeng Q, Zhou J, et al. Clinical-grade human umbilical cord-derived mesenchymal stem cells reverse cognitive aging via improving synaptic plasticity and endogenous neurogenesis. Cell Death Dis. 2017;8(8):e2996. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 61. Xu S, Zhong A, Zhang Y, Zhao L, Guo Y, Bai X, et al. Bone marrow mesenchymal stem cells therapy regulates sphingolipid and glycerophospholipid metabolism to promote neurological recovery in stroke rats: a metabolomics analysis. Exp Neurol. 2024;372:114619. [ DOI ] [ PubMed ] [ Google Scholar ] 62. Ma X, Huang M, Zheng M, Dai C, Song Q, Zhang Q, et al. ADSCs-derived extracellular vesicles alleviate neuronal damage, promote neurogenesis and rescue memory loss in mice with Alzheimer’s disease. J Control Release. 2020;327:688–702. [ DOI ] [ PubMed ] [ Google Scholar ] 63. Kiroski I, Jiang Y, Gavrilovici C, Gao F, Lee S, Scantlebury MH, et al. Reelin improves cognition and extends the lifespan of mutant Ndel1 mice with postnatal CA1 hippocampus deterioration. Cereb Cortex. 2020;30(9):4964–78. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 64. Mironova YA, Giger RJ. Where no synapses go: gatekeepers of circuit remodeling and synaptic strength. Trends Neurosci. 2013;36(6):363–73. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 65. Ruchika F, Shah S, Neupane D, Vijay R, Mehkri Y, Lucke-Wold B. Understanding the molecular progression of chronic traumatic encephalopathy in traumatic brain injury, aging and neurodegenerative disease. Int J Mol Sci. 2023;24(3):1847. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 66. Faden AI, Loane DJ. Chronic neurodegeneration after traumatic brain injury: Alzheimer disease, chronic traumatic encephalopathy, or persistent neuroinflammation? Neurotherapeutics. 2015;12(1):143–50. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 67. Cheng X, Li Y, Huang Y, Feng X, Feng G, Xiong ZQ. Pulse labeling and long-term tracing of newborn neurons in the adult subgranular zone. Cell Res. 2011;21(2):338–49. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 68. Balic N, Nikolac Perkovic M, Milos T, Vuic B, Kurtovic Kodzoman M, Svob Strac D, et al. Extracellular vesicles as a promising tool in neuropsychiatric pharmacotherapy application and monitoring. Prog Neuropsychopharmacol Biol Psychiatry. 2025;139:111393. [ DOI ] [ PubMed ] [ Google Scholar ] 69. Wei B, Wei M, Huang H, Fan T, Zhang Z, Song X. Mesenchymal stem cell-derived exosomes: a promising therapeutic strategy for age-related diseases. Cell Prolif. 2025;58(5):e13795. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 70. Peng X, Cui H, Tan S, Wen B, Luo X, Chen S, et al. The engineered bone marrow mesenchymal stem cell-derived exosome: a new strategy for the treatment of inflammatory diseases. Int Immunopharmacol. 2025;162:115136. [ DOI ] [ PubMed ] [ Google Scholar ] 71. Khan MI, Jeong ES, Khan MZ, Shin JH, Kim JD. Stem cells-derived exosomes alleviate neurodegeneration and Alzheimer’s pathogenesis by ameliorating neuroinflamation, and regulating the associated molecular pathways. Sci Rep. 2023;13(1):15731. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 72. Zhu D, Liu S, Huang K, Wang Z, Hu S, Li J, et al. Intrapericardial exosome therapy dampens cardiac injury via activating Foxo3. Circ Res. 2022;131(10):e135–50. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 73. Podvin S, Jones A, Liu Q, Aulston B, Ransom L, Ames J, et al. Dysregulation of exosome cargo by mutant Tau expressed in human-induced pluripotent stem cell (iPSC) neurons revealed by proteomics analyses. Mol Cell Proteomics. 2020;19(6):1017–34. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 74. Ebrahim N, Al Saihati HA, Alali Z, Aleniz FQ, Mahmoud SYM, Badr OA, et al. Exploring the molecular mechanisms of MSC-derived exosomes in Alzheimer’s disease: autophagy, insulin and the PI3K/Akt/mTOR signaling pathway. Biomed Pharmacother. 2024;176:116836. [ DOI ] [ PubMed ] [ Google Scholar ] 75. Liddelow SA, Guttenplan KA, Clarke LE, Bennett FC, Bohlen CJ, Schirmer L, et al. Neurotoxic reactive astrocytes are induced by activated microglia. Nature. 2017;541(7638):481–7. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 76. Welsh JA, Goberdhan DCI, O’Driscoll L, Buzas EI, Blenkiron C, Bussolati B, et al. Minimal information for studies of extracellular vesicles (MISEV2023): from basic to advanced approaches. J Extracell Vesicles. 2024;13(2):e12404. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials 13287_2026_4943_MOESM1_ESM.tif (7.5MB, tif) Supplementary material 1. Supplementary Fig. 1. The schematic of the modeling device and supplementary behavioral assessments in an independent cohort of mice at 28 days post-modeling. (A) The schematic of the modeling device. Key components are labeled: 1, baseplate; 2, mounting rod; 3, rotatable tube clamp; 4, plastic trough; 5, foam board lining; 6, protective net pouch; 7, PVC tube with side openings. The red dashed circle​ on the mouse's head indicates the permissible target area​ for the impact. (B) Percentage of exploration time during the training phase of the novel object recognition test (NORT). Both control and CTE28 mice showed comparable exploration times for the two identical objects, indicating that there was no inherent object preference bias. (C) Recognition index in the test phase of the NORT. Compared with control mice, CTE28 mice exhibited a significantly lower recognition index (p < 0.05), indicating impaired nonspatial recognition memory. (D) Immobility time in the forced swim test (FST). Compared with control mice, CTE28 mice exhibited a significantly increased immobility time (p < 0.05), suggesting increased depressive-like behavior. (E) Representative Western blot images (top) and quantitative analysis (bottom) of p-tau (Thr181) (50 kDa) protein levels in hippocampal tissues normalized to that of GAPDH (36 kDa). Compared with control mice, CTE28 mice had significantly greater p-tau levels (p < 0.05), confirming the presence of a key CTE-related neuropathology in the model. (E–I) Representative immunohistochemical images (left, scale bar: 50 μm) and corresponding quantification (right) of p-tau (Ser202/Thr205) positive area in the cerebral cortex (CTX) and hippocampal dentate gyrus (DG). CTE28 mice showed a significant increase in p-tau immunoreactivity in both regions (p < 0.05). The data are presented as the mean ± SEM. *p < 0.05. 13287_2026_4943_MOESM2_ESM.tif (1.8MB, tif) Supplementary material 2. Supplementary Fig. 2. Behavioral consequences of CTE1 and CTE7. (A) The latency to fall in the rotarod test across post-modeling intervals indicates comparable motor function among all groups. (B) The total locomotion distance in the open-field test confirms intact basal motor activity. (C) CTE7 mice exhibited anxiety-like behavior, which manifested as a significantly reduced central zone time (p < 0.05 vs. control & CTE1) on post-modeling days 3 and 14, whereas CTE1 mice remained unaffected. (D) Spontaneous alternation percentage in the Y-maze test reveals no persistent cognitive impairment in the CTE groups across the timepoints. (E-F) Sociability phase: All groups showed intact social preference (p < 0.05 stranger vs. empty cage). (G-H) Social novelty phase: CTE7 mice displayed persistent social recognition deficits, as demonstrated by a lack of novelty preference (novel vs. familiar mouse exploration time: p > 0.05) at all time points. The data are presented as the mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001. 13287_2026_4943_MOESM3_ESM.tif (18.2MB, tif) Supplementary material 3. Supplementary Fig. 3. Characterization of BMSCs and derived exosomes (BMSC-exos). (A) Phenotypic characterization by flow cytometry. The BMSCs were positive for CD29, CD44, and Sca-1 (Ly-6A) and negative for CD31 and CD117. (B) Functional characterization by trilineage differentiation. Adipogenic (oil red O staining, lipid droplets, left), osteogenic (alizarin red S staining, calcium deposition, middle), and chondrogenic (alcian blue staining, proteoglycans, right) differentiation confirmed cell multipotency. (C) Nanoparticle tracking analysis showing the size distribution of BMSC-exos (inset: nanoparticle image). (D) Western blot analysis of exosomal markers in the supernatant (lane 1), BMSCs (lane 2), and BMSC-exos (lane 3). (E) TEM image of BMSC-exos showing a typical cup-shaped morphology. Scale bars: B, 200 μm (left & middle), 50 μm (right); E, 100 nm. These data confirm the successful isolation and identity of BMSCs and exosomes. 13287_2026_4943_MOESM4_ESM.tif (671.6KB, tif) Supplementary material 4. Supplementary Fig. 4. BMSC-exos ameliorated anxiety-like and cognitive deficits in CTE mice at 28th day post-modeling. (A) Number of entries into the center zone during the first 5 min of the open-field test. Compared with the control mice, the CTE28 mice showed a significant decrease in the number of center zone entries, which was reversed by treatment with BMSC-exos. (B) Average locomotor velocity during the test. No significant differences were observed among the groups, indicating that motor function was not impaired. (C, D) Time spent in the side (C) and corner (D) areas, reflecting thigmotaxis. Compared with control mice, CTE28 mice spent significantly more time in these peripheral zones, which was attenuated by BMSC-exo administration. (E)​ Morris water maze performance during the training phase. Escape latency across training days are shown. (F-H) Spatial memory assessed in the probe trial. The latency to first reach the platform (F), the number of target platform crossings (G) and the time spent in the target quadrant (H) are shown. CTE28 mice exhibited prolonged latencies, made fewer crossings over the former platform location and spent less time in the target quadrant (p < 0.05), indicating impaired memory retention. These deficits were rescued by BMSC-exo treatment. The data are presented as the mean ± SEM. *p < 0.05. 13287_2026_4943_MOESM5_ESM.tif (1.5MB, tif) Supplementary material 5. Supplementary Fig. 5. Effects of BMSC-exos on the behavior of CTE model mice on the 7th and 14th days. (A) Schematic workflow of the longitudinal behavioral assessment. (B-F)​ Behavioral assessments on the 7th day. (B) Time spent in the center area during the open-field test, indicating anxiety-like behavior. Compared with the control group, both the CTE28 and the CTE28 + EXO groups showed significant decreases. (C-D) Pre-test of the three-chamber test. All groups spent significantly more time investigating the cage containing the stranger mouse than the empty cage (p < 0.05 vs. empty cage), indicating intact social motivation. (E-F) Exploration time and social novelty index in the post-test of the three-chamber test, indicating social cognitive function. Mice in both CTE-exposed groups exhibited impaired social novelty preference compared with controls. No significant difference was observed between the CTE28 and CTE28 + EXO groups at this time point. (G-K)​ Behavioral assessments on the 14th day. (G)​ Time spent in the center area. Compared with the control group, the CTE28 group remained significantly impaired, while the CTE28 + EXO group showed restoration to a level comparable to that of the controls. (H-I) Pre-test of the three-chamber test. (J-K) Exploration time and social novelty index in the post-test of the three-chamber test. The CTE28 group continued to show significant impairment, whereas the CTE28 + EXO group exhibited restored social novelty preference. Data represent the mean ± SEM. *p < 0.05, **p < 0.01. 13287_2026_4943_MOESM6_ESM.tif (16.4MB, tif) Supplementary material 6. Supplementary Fig. 6. BMSC-exos promote the differentiation of newborn cells in the hippocampus of CTE mice at 28 days post-modeling. (A) Schematic timeline of the experimental procedure, including CTE modeling, BMSC-exo administration, and tissue collection. (B) Representative immunofluorescence images of hippocampal sections from the indicated groups showing EdU (red, proliferating cells), NeuN (green, mature neurons), and DAPI (blue, nuclei). Scale bar, 50 μm. (C) Quantification of EdU-positive cells in the hippocampal subgranular zone. The number of proliferating cells was greater in the CTE28 + EXO group than in the CTE28 group. (D) Quantification of the percentage of EdU-positive cells that coexpressed NeuN. The proportion of newborn cells differentiating into mature neurons was significantly greater in the CTE28 + EXO group than in the CTE28 group. Data are presented as the mean ± SEM. *p < 0.05. 13287_2026_4943_MOESM7_ESM.tif (31MB, tif) Supplementary material 7. Supplementary Fig. 7. Changes in neurogenesis during the acute phase on the 3rd day after CTE modeling. (A) Representative immunofluorescence images of the SGZ: DAPI (blue), Sox2 (red), Nestin (green), and DCX (cyan). Scale bars: 50 μm. (B) EdU labeling (proliferating cells, red) in the same groups. Scale bars: 100 μm. (C-E) Quantitative analyses (3rd day): (C) The number of Sox2-Nestin–double-positive neural stem cells was greater in the CTE28 group than in the control group. (D) The number of DCX-positive immature neurons showed no significant difference. (E) The number of EdU-positive proliferating cells was markedly increased in the CTE28 group compared with the control group. Data are shown as the mean ± SEM. *p < 0.05, ***p < 0.001. 13287_2026_4943_MOESM8_ESM.tif (3.5MB, tif) Supplementary material 8. Supplementary Fig. 8. Optimization of viral injection volume for sparse neuronal labeling in the hippocampal CA1 region.​ Representative maximum-intensity projection confocal micrographs of mouse hippocampal CA1 following the stereotaxic injection of graded viral volumes: 100 nL, ​​150 nL,​ and ​​200 nL​​. 13287_2026_4943_MOESM9_ESM.tif (745.5KB, tif) Supplementary material 9. Supplementary Fig. 9. Structural plasticity alterations in hippocampal CA1 neurons during intermediate phases after CTE modeling. (A-D) Data from the 7th day post-modeling: Quantitative morphometry of CA1 pyramidal neurons in the control (green), CTE28 (blue), and CTE28 + EXO (red) groups: (A) Dendrite number. (B) Total dendrite length. (C) Sholl analysis showing comparable dendritic complexity across groups (10–200 μm from the soma). (D) Basal spine density quantification. (E-H) Data from the 14th day post-modeling: Consistently null results for (E) the dendrite number, (F) the dendrite length, (G) Sholl analysis, and (H) basal spine density. Data are shown as the mean ± SEM. 13287_2026_4943_MOESM10_ESM.tif (1.8MB, tif) Supplementary material 10. Supplementary Fig. 10. Changes in acute-phase structural plasticity at 3 days after CTE modeling. (A) Experimental timeline: Stereotaxic surgery at -2 weeks, CTE modeling lasting 4 weeks, tissue collection on the 3rd day. (B-D) Morphometric analyses across groups: (B) Dendrite length quantification. (C) Dendrite number. (D) Sholl analysis showing comparable dendritic complexity across groups (10–200 μm from the soma). (E-F) Representative images. (G) Comparable spine density among groups. These results conclusively demonstrate the absence of acute structural pathology in the CTE group at this stage. Data are shown as the mean ± SEM. *p < 0.05. 13287_2026_4943_MOESM11_ESM.tif (593.2KB, tif) Supplementary material 11. Supplementary Fig. 11. Proinflammatory cytokines in the hippocampus at 28 days post-modeling. Hippocampal homogenates were analyzed using the Meso Scale Discovery (MSD) V-PLEX platform (Kit #K15048D-2). The tissue homogenates were adjusted to a uniform protein concentration of 3.08 mg/mL, and 50 µL of each sample was analyzed. Data are presented as the mean ± SEM. 13287_2026_4943_MOESM12_ESM.tif (35.8MB, tif) Supplementary material 12. Supplementary Fig. 12. In vivo biodistribution and cellular uptake of PKH67-labeled BMSC-exos in the brain of CTE28 mice. Representative immunofluorescence images showing the localization of intravenously administered PKH67-labeled BMSC-exos in the hippocampal DG region In the merged images: Yellow arrowheads​ indicate co-localization of PKH67 signal with Iba1, demonstrating exosome uptake by microglia; Purple arrowheads​ indicate co-localization of PKH67 signal with NeuN, demonstrating exosome uptake by neurons. Scale bars: 50 µm. Supplementary material 13. (19.8KB, docx) Supplementary material 14. (25MB, wmv) Data Availability Statement No datasets were generated or analysed during the current study. Articles from Stem Cell Research & Therapy are provided here courtesy of BMC ACTIONS View on publisher site PDF (9.7 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top