Human umbilical MSC-derived exosomes improve intracerebral hemorrhage recovery via SIRT1-driven suppression of NF-κB/NOS2 signaling: coordinating microglial homeostasis and neuroprotection

Abstract

Intracerebral hemorrhage (ICH) remains a devastating neurological disorder with high mortality, driven primarily by uncontrolled neuroinflammation and secondary brain injury. Here, we show that human umbilical mesenchymal stem cell-derived exosomes (hUMSC-Exos) robustly promote functional recovery in a murine ICH model by reprogramming microglial biology and mitigating neuronal damage, via a mechanism dependent on the NAD⁺-dependent deacetylase SIRT1. Intranasal delivery of hUMSC-Exos enabled efficient uptake by perihematomal microglia, astrocytes, and neurons, reducing neuronal apoptosis and improving both sensorimotor and cognitive outcomes. Microglia-specific transcriptomic profiling revealed that hUMSC-Exos suppressed ICH-induced proinflammatory gene networks, particularly those governed by NF-κB/NOS2 signaling, while attenuating pathological microglial proliferation. Mechanistically, hUMSC-Exos upregulated SIRT1, which repressed NF-κB nuclear translocation and subsequent NOS2 expression. Pharmacological inhibition of SIRT1 with EX527 abrogated key beneficial effects of hUMSC-Exos: it reversed the suppression of microglial proliferation, restored neuronal apoptosis to ICH levels, and eliminated improvements in locomotor activity, anxiety-like behavior, and spatial learning/memory—assessed via open field and Morris water maze tests. Conversely, NOS2 blockade recapitulated the neuroprotective actions of hUMSC-Exos. Beyond anti-inflammatory effects, hUMSC-Exos promoted transcriptional programs linked to tissue remodeling and vascular regeneration, underscoring their dual role in mitigating injury and enhancing repair. Collectively, our study identifies a SIRT1-dependent axis through which stem cell-derived exosomes orchestrate microglial homeostasis and neuronal survival after ICH, establishing exosome-based therapy as a promising cell-free strategy for acute brain injury with translational potential.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-025-07430-1.

Introduction

Intracerebral hemorrhage (ICH) represents a devastating subtype of stroke, comprising 10–15% of all cerebrovascular events yet contributing disproportionately to global morbidity and mortality [1]. Unlike ischemic stroke, which is amenable to acute reperfusion therapies, ICH is characterized by cerebral vascular rupture and blood extravasation into brain parenchyma, triggering not only mass effect and cytotoxicity but also a delayed yet prolonged inflammatory cascade that exacerbates secondary brain injury over days to weeks [2]. Despite its clinical urgency, no pharmacotherapies targeting the post-ICH inflammatory response have advanced to clinical practice, highlighting a critical translational gap and unmet need for interventions that mitigate neuroinflammation while preserving neural repair mechanisms [3].

Microglia, the brain’s resident immune sentinels, orchestrate the post-ICH inflammatory niche. Following hemorrhage, microglia rapidly polarize into pro-inflammatory M1-like phenotypes, secreting cytokines (e.g., TNF-α, IL-1β), nitric oxide (NO), and reactive oxygen species that drive neuronal apoptosis and perihematomal edema [4, 5]. A central driver of this deleterious response is sustained activation of the NF-κB pathway in microglia, which transcriptionally upregulates inducible nitric oxide synthase (NOS2/iNOS) to propagate a feedforward neurotoxic loop [6, 7]. Notably, preclinical studies show that genetic or pharmacological inhibition of NOS2 reduces ICH-induced brain damage, validating this pathway as a therapeutic target.

Cell-derived exosomes have emerged as promising cell-free therapeutics in CNS disorders, leveraging their ability to cross the blood–brain barrier and deliver bioactive cargos (e.g., miRNAs, proteins, metabolites) to modulate local inflammation [8, 9]. While exosomes derived from human umbilical mesenchymal stem cells (hUMSC-Exos) exhibit neuroprotection in ischemic stroke and traumatic brain injury models [10, 11], their role in ICH remains underexplored—particularly regarding their impact on microglial transcriptional reprogramming.

The molecular complexity of neurological diseases is rooted in brain cellular heterogeneity, a challenge that traditional bulk RNA sequencing fails to resolve due to its inability to capture cell-specific transcriptional signatures [12]. Cell-type specific transcriptomics has emerged as a transformative approach, enabling precise dissection of spatiotemporal gene regulation in diseased brain tissues [13]. Fresh isolation of microglia, astrocytes, neurons, and endothelial cells via flow cytometry or magnetic sorting allows transcriptomic profiling at single-cell resolution [14, 15]. This reveals distinct molecular programs underlying cell-specific responses to pathological insults—for example, pro-inflammatory microglial subpopulations in neurodegeneration or astrocytic dysregulation of glutamate homeostasis and blood-brain barrier integrity [16]. By linking cell-specific transcriptional changes to functional outcomes, this approach illuminates causal relationships between genetic alterations and disease phenotypes [17, 18]. In preclinical research, cell-type specific transcriptomics enables granular evaluation of therapeutic interventions, facilitating target identification and efficacy assessment [19]. This precision is critical for developing cell-specific therapies that mitigate off-target effects and enhance treatment outcomes in neurological disorders.

In this study, we reported a comprehensive investigation integrating behavioral phenotyping, histological characterization, and single-cell transcriptomic profiling of microglia to elucidate the therapeutic efficacy of hUMSC-Exos in experimental intracerebral hemorrhage. Our findings demonstrated that intranasally administered hUMSC-Exos were efficiently internalized by microglia and diverse neural cell populations within the perihematomal niche, driving functional recovery while attenuating neuroinflammatory cascades. Transcriptomic profiling of isolated microglia revealed that hUMSC-Exos potently reversed ICH-induced proinflammatory transcriptional programs, with notable suppression of NF-κB/NOS2 signaling axes, and promoted microglial reprogramming toward an anti-inflammatory phenotype. Mechanistic studies identified SIRT1-dependent deacetylation as the key regulatory node, both in vitro and in vivo, through which hUMSC-Exos suppressed the NF-κB/NOS2 axis. Collectively, our integrated approach combining molecular profiling and functional assays reveals that hUMSC-Exos act as pleiotropic regulators of post-ICH neuroinflammation, positioning them as a promising cell-free therapeutic strategy for this critical unmet clinical need.

Methods

Animals

C57BL/6 mice (10–12 weeks old, 23–28 g body weight) were obtained from Slac Laboratory Animals Center (Shanghai, China). Animals were housed in a specific pathogen-free facility with controlled environmental conditions: temperature at 22 ± 1 °C, relative humidity of 65–70%, and a 12-h light/dark cycle. Standard rodent chow and water were provided ad libitum. All experimental protocols were approved by the Ethics Committee of Shanghai Public Health Clinical Center and conformed to the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Measures were taken to minimize animal suffering throughout the experiments, including optimized handling procedures and regular health monitoring.

ICH model

Intracerebral hemorrhagic (ICH) induction and subsequent interventions were executed following validated neurosurgical techniques. Briefly, C57BL/6 mice were anesthetized using isoflurane (2–3% induction, 1–2% maintenance) and securely mounted on a stereotaxic apparatus. A single burr hole was drilled into the skull at coordinates relative to bregma: anteroposterior − 0.2 mm, lateral 2.5 mm, and dorsoventral 3.5 mm, targeting the right basal ganglia. Using a Hamilton microsyringe, 30 µL of autologous blood was slowly infused at a precise rate of 2 µL/min, with the needle retained in place for an additional 10 min post-infusion to prevent reflux. After removal, the burr hole was sealed with bone wax, and the scalp incision was closed using 6 − 0 sutures. For sham-operated controls, the same surgical procedures were performed, excluding the blood injection step.

Behavioral testing

Sensorimotor function was evaluated by blinded researchers using rotarod, foot fault, and wire hanging tests from days 1–28 post-ICH. Mice underwent 3 days of behavioral training prior to testing, with those displaying abnormal baseline behavior excluded from analysis.

Cell culture and in vitro model establishment

hUMSCs were isolated from human umbilical cords procured from Guangzhou Saliai Stem Cell Science and Technology Co., Ltd. Tissues were processed under sterile conditions, and cells were cultured in complete DMEM/F12 medium (Gibco) supplemented with 10% fetal bovine serum (FBS, Gibco). Cultures were maintained at 37 °C in a humidified 5% CO₂ incubator, and passages 3–5 were used for experiments to ensure phenotypic consistency. The BV2 murine microglial cell line (Shanghai Fuheng Biotechnology) was cultured in DMEM containing 10% heat-inactivated FBS, 100 U/mL penicillin, and 100 µg/mL streptomycin. Media were refreshed every 2–3 days, and cells were passaged at 70–80% confluence. For all in vitro microglial experiments (including cell seeding, hemin stimulation, immunofluorescence, and Transwell migration assays), BV2 cells or primary microglia were plated in 24-well plates. For in vitro ICH modeling, BV2 cells were stimulated with 20 µM hemin (Sigma-Aldrich) for 24 h under standard culture conditions.

Isolation and culture of primary microglia

Primary microglia were isolated from postnatal day 1–3 C57BL/6J mouse cortices. Following meninges removal, tissues were minced and digested with 0.25% trypsin-EDTA (Gibco) at 37 °C for 20 min, quenched with DMEM/F12 containing 10% FBS, and triturated. The suspension was filtered through a 70-µm strainer, and cells were plated in T75 flasks with glial medium (DMEM/F12, 10% FBS, 1% penicillin-streptomycin, 10 ng/mL M-CSF; PeproTech). After 10–12 days, microglia were detached by gentle shaking (200 rpm, 2 h at 37 °C) and confirmed by Iba1 immunostaining.

Exosome isolation, characterization, and administration

Exosomes (Exos) were isolated from hUMSC-conditioned medium using differential ultracentrifugation. Briefly, culture supernatants were centrifuged at 2,000 × g for 20 min and 10,000 × g for 30 min (4 °C) to remove debris, filtered through 0.22-µm membranes, and ultracentrifuged at 110,000 × g for 70 min (4 °C). Pellets were washed in PBS and ultracentrifuged again. Exos were resuspended in PBS, quantified via BCA assay, and stored at -80 °C. Characterization was performed using transmission electron microscopy (TEM; Hitachi HT7700) for morphology, nanoparticle tracking analysis (NTA; NanoSight NS300) for size distribution, and Western blotting for CD63, Alix, and TSG101. For in vitro labeling, Exos were incubated with PKH26 (Sigma-Aldrich) per manufacturer instructions. For intranasal delivery, mice were anesthetized with isoflurane post-ICH and administered 20 µg Exos in 10 µL PBS/nostril using a micropipette, alternating nostrils every 30 s for uniform uptake. Treatments were administered daily for 3 days starting immediately after surgery. In vitro, recipient cells (including BV2 murine microglial cells and primary mouse microglia) were seeded in 24-well plates with 800 µL of culture medium per well, and incubated with exosomes to achieve a final exosome concentration of 4 µg/mL for 24 h.

Drug treatment

To elucidate the functional roles of NOS2, SIRT1, and their downstream signaling pathways in ICH in vivo, mice subjected to ICH received specific pharmacological agents. The selective NOS2 inhibitor, 1400 W (Sigma-Aldrich), was administered via intraperitoneal (IP) injection at a dose of 20 mg/kg body weight once daily for three consecutive days. Meanwhile, the SIRT1 inhibitor, EX-527, was also given by IP injection at a dose of 10 mg/kg body weight once daily for three days. This treatment regimen was designed to target the respective molecular pathways during the critical post-ICH period, when inflammation and neurodegenerative processes are actively ongoing. For in vitro studies, BV2 murine microglial cells or primary microglial cells were utilized. To inhibit NOS2, the selective inhibitor 1400 W was added to the cell culture medium at a concentration of 5 µM. For SIRT1 inhibition, EX-527 was applied at a concentration of 0.5 µM. These inhibitors were added 1–2 h prior to the inflammatory stimulus, which was induced by treating the cells with Hemin. The inhibitors were allowed to act on the cells for 24 h, enabling the observation of their effects on the cellular responses to the inflammatory challenge.

Isolation of microglia from brain tissue

Three days after surgery, mice were anesthetized with 1% sodium pentobarbital (IP) and transcardially perfused with saline. Brain tissues were dissected and dissociated using the Adult Mouse Brain Dissociation Kit (Miltenyi Biotec, 130-092-628) per the manufacturer’s protocol. Microglia were isolated via magnetic-activated cell sorting (MACS) using CD11b MicroBeads (Miltenyi Biotec, 130-093-636), yielding highly purified populations for downstream analyses.

Immunofluorescence staining

Three days post-surgery, mice were anesthetized with 2% isoflurane and euthanized by decapitation. Brains were extracted and fixed in 4% paraformaldehyde/PBS at 4 °C overnight, then cryoprotected in 30% sucrose/PBS until sunk. Coronal Sect. (25 μm) were cut using a Leica CM 1900 freezing microtome. Free-floating sections were washed in PBS, blocked with 5% normal goat serum/1% Triton X-100/PBS for 1 h, and incubated overnight at 4 °C with primary antibodies: rabbit anti-Iba1 (Abcam, ab178846), chicken anti-GFAP (Abcam, ab4674), rabbit anti-NeuN (Abcam, ab177487), goat anti-Iba1 (Abcam, ab5076), rabbit anti-SIRT1 (Abcam, ab189494), rabbit anti-iNOS (CST, 13120), rabbit anti-p-NF-κB p65 (CST, 3033), rat anti-CD16 (BD Biosciences, 553142), and mouse anti-Arg1 (Santa Cruz, SC-271430). After washing, sections were incubated with Alexa Fluor-conjugated secondary antibodies (Jackson ImmunoResearch) for 1.5 h at RT, counterstained with DAPI, mounted with anti-fade medium, and imaged using an Olympus VS120 scanner for whole sections. High-resolution confocal imaging was performed on a Nikon AX Ti2E, with analysis via ImageJ.

Quantitative real-time PCR (qRT-PCR)

Total RNA was extracted from cells and brain tissues using TRIzol reagent (Invitrogen) per the manufacturer’s protocol. RNA quality was verified via NanoDrop spectrophotometry, ensuring A260/A280 ratios of 1.8–2.0. cDNA synthesis was performed using the HiScript III All-in-one RT SuperMix (Vazyme, R333), and qRT-PCR was conducted on a QuantStudio 5 system (Thermo Fisher Scientific) with ChamQ Universal SYBR qPCR Master Mix (Vazyme, Q711). Primer sequences are detailed in Table S1, and relative gene expression was calculated using the 2^^−ΔΔCt method, normalized to Gapdh.

RNA sequencing and data analysis

Microglia isolated from three experimental groups, each with three biological replicates, underwent RNA sequencing. Total RNA was extracted and processed to construct cDNA libraries using the NEBNext Ultra RNA Library Prep Kit for Illumina (NEB, USA) following the manufacturer’s instructions. Sequencing was carried out on an Illumina HiSeq platform with a PE150 configuration. The raw FASTQ files were initially evaluated for quality by FastQC. Subsequently, Trim Galore was employed to remove adapter sequences, reads of low quality, and short fragments. The clean reads were then aligned to the mouse reference genome (UCSC mm10) using HISAT2, and SAMtools were used for further processing of the alignments. Gene - level read counts were obtained via FeatureCounts, and TPM (Transcripts Per Million) values were calculated. Differential gene expression analysis was performed using DESeq2. Genes with an adjusted P-value < 0.05 and |Log2FC| >1 were considered differentially expressed. To understand the functional implications of these differentially expressed genes (DEGs), Gene Ontology (GO) enrichment analysis was conducted with ShinyGO 0.76.3, which helped in categorizing the DEGs based on their biological processes, molecular functions, and cellular components.

Cytokine quantification by ELISA

Inflammatory cytokines in BV2 microglial culture supernatants were quantified using ELISA kits (R&D Systems) per manufacturer protocols. Briefly, 96-well plates pre-coated with biotinylated antibodies against iNOS, TGF-β, IL-10, TNF-α, and IL-1β were loaded with standards and samples. After incubation, unbound material was washed away, and enzyme-conjugated polyclonal antibodies were added. Following a 2-h RT incubation and washing, substrate solution was added for 30 min, then reactions were stopped, and absorbance read at 450 nm on a microplate reader. Cytokine concentrations were calculated from standard curves, with all samples assayed in triplicate.

Western blotting analysis

Tissue, cell, and exosome proteins were extracted using RIPA lysis buffer and quantified with a Bradford assay. Lysates were denatured, separated by SDS-PAGE, and transferred to PVDF membranes (Millipore). Membranes were blocked, then incubated overnight at 4 °C with primary antibodies, including anti-TSG101 (Abcam, ab125011), anti-CD63 (Abcam, ab271286), anti-ALIX (Abcam, ab117600), anti-SIRT1 (Abcam, ab189494), anti-p-NF-κB p65 (CST, 3033), and anti-NF-κB p65 (CST, 8242). After TBST washes, membranes were incubated with HRP-conjugated secondary antibodies for 1.5 h at RT. Protein bands were detected using an enhanced chemiluminescence kit (Thermo Scientific) and imaged with a chemiluminescence system. Band intensities were analyzed by ImageJ, and all experiments were repeated in triplicate.

Transwell migration assay

Transwell migration was assessed using 24-well plates with 8.0-µm pore membranes (Corning). Cells resuspended in serum-free medium were seeded in the upper chamber, and 500 µL complete medium was added to the lower compartment. After 24 h, LPS + IFN-γ, Exos, LP17, or SB203580 were introduced to the lower chamber. Following an additional 24 h incubation, cells were fixed with 4% paraformaldehyde, stained with 0.25% crystal violet (Beyotime), and non-migrated cells were removed. Migrated cells on the lower membrane were imaged and quantified.

Statistical analysis

All experiments and data analyses were conducted with investigator blinding to reduce bias. Statistical analyses were performed using Prism 10 (GraphPad). One-way or two-way ANOVA was used for multi-group comparisons, as per the experimental design. Quantitative data are presented as mean ± standard error of the mean (SEM) or mean ± standard deviation (SD). Statistical significance was set at p < 0.05 and denoted as *p < 0.05, **p < 0.01, and ***p < 0.001.

Results

Intranasal delivery of hUMSC-derived exosomes enables widespread cellular uptake in the perihematomal brain following intracerebral hemorrhage

Intracerebral hemorrhage presents a critical challenge for therapeutic delivery due to compromised blood-brain barrier integrity. Here, we demonstrate that intranasally administered hUMSC-Exos exhibit efficient brain targeting and broad cellular tropism in an ICH mouse model, providing a non-invasive strategy for central nervous system delivery. Transmission electron microscopy confirmed the classical cup-shaped morphology of isolated hUMSC-Exos (Fig. 1A; Fig. S1), while nanoparticle tracking analysis revealed a homogeneous size distribution with a mean diameter of 145 nm (Fig. 1B). Western blot analysis validated expression of exosomal markers Alix, CD63, and TSG101, confirming their extracellular vesicle identity (Fig. 1C). These biophysical properties suggest optimal suitability for nasal-to-brain transport.

Fig. 1.

Characterization and brain uptake of hUMSC-derived exosomes following intranasal administration after intracerebral hemorrhage. A Transmission electron microscopy (TEM) shows the characteristic cup-shaped morphology of isolated hUMSC-derived exosomes. B Nanoparticle tracking analysis (NTA) shows a consistent size distribution of exosomes with a peak diameter around 140–150 nm. C Western blot analysis confirms the presence of classical exosome markers: Alix, CD63, and TSG101. D Representative immunofluorescence images show the colocalization of PKH26-labeled exosomes with DAPI in perihematomal regions. E Quantification of PKH26-labeled hUMSC-Exos in brain Sections 12 h after intranasal delivery demonstrates dose-dependent accumulation in the ischemic penumbra, peaking at 20 µg. n = 5 mice per group. NS = not significant, *p < 0.05, ***p < 0.001. F-H Immunofluorescence staining reveals co-localization of PKH26⁺ hUMSC-Exos with Iba1⁺ microglia (F), GFAP⁺ astrocytes (G), and NeuN⁺ neurons (H) in the perihematomal regions. I In vitro uptake of PKH26-labeled hUMSC-Exos by BV2 microglia under control or Hemin-stimulatedconditions. J Primary microglia (pMG) also exhibit enhanced uptake of hUMSC-Exos following Heminstimulation, as indicated by colocalization of PKH26 signal with Iba1⁺ cells

Following intranasal administration of PKH26-labeled hUMSC-Exos, fluorescence imaging revealed robust accumulation in the perihematomal striatum at 24 h post-ICH. Dose-response analysis showed maximal uptake at 20 µg (Fig. 1D–E). This distribution pattern bypassed traditional BBB restrictions, highlighting the nasal route as an effective delivery pathway for neurotherapeutics. High-resolution immunofluorescence demonstrated efficient uptake of hUMSC-Exos by multiple neural cell types. PKH26-labeled exosomes co-localized with Iba1⁺ microglia, GFAP⁺ astrocytes, and NeuN⁺ neurons in the perihematomal region (Fig. 1F–H). This broad tropism suggests multifaceted mechanisms of action across immune and parenchymal cells. In vitro studies in BV2 microglia and primary mouse microglia (pMG) revealed enhanced exosome internalization under Hemin-stimulated conditions, mimicking the ICH microenvironment. Confocal microscopy showed robust co-localization of PKH26 signal with Iba1⁺ cells (Fig. 1I–J). This suggests that injury-induced cellular states may promote exosome internalization, optimizing therapeutic efficacy.

The non-invasive intranasal route combined with broad cellular uptake positions hUMSC-Exos as a promising platform for ICH therapy. By bypassing BBB limitations and targeting multiple cell types in the injury microenvironment, these findings lay the foundation for clinical translation of exosomal therapeutics via the nasal route in acute neurological injuries.

hUMSC-derived exosomes promote neurological recovery and preserve white matter integrity via suppression of neuronal apoptosis and enhancement of microglial Myelin debris clearance after ICH

To evaluate the effects of hUMSC-Exos on neurological deficits and white matter injury following ICH in mice, we conducted a series of experiments. Administration of hUMSC-Exos significantly improved sensorimotor function across a 35-day post-ICH period. Compared to ICH controls, hUMSC-Exos-treated mice exhibited marked enhancements in wire hanging, foot fault, and rotarod test performance (Fig. 2A–C), reflecting restored motor coordination and balance. Histological analysis at day 35 revealed preservation of neural architecture in the ipsilateral striatum. Immunofluorescent staining showed robust upregulation of myelin basic protein (MBP) and neurofilament-H (NF-H), indicating maintenance of myelination and axonal integrity, respectively (Fig. 2D–F). Mechanistically, hUMSC-Exos exerted early neuroprotective effects as early as day 3 post-ICH. Co-immunostaining for cleaved caspase-3 and NeuN revealed a significant reduction in apoptotic neurons within the perihematomal region of hUMSC-Exos-treated mice (Fig. 2G, I). Concomitantly, NeuN staining alone confirmed preservation of neuronal density in these areas (Fig. 2H, J), reinforcing the anti-apoptotic role of hUMSC-Exos.

Fig. 2.

hUMSC-Exos promote neurofunctional recovery and reduce neuronal apoptosis following ICH. A-C Sensorimotor deficits were evaluated before (pre) and up to 35 days after ICH or Sham surgery by wire hanging, fault foot and rotarod test. n = 12 mice per group. D Representative images of MBP/NF-H immunostaining 35 days after ICH. E-F Quantification of the fluorescence intensity and area of MBP and NF-H in the STR. n = 6 mice per group. G, I Representative images of cleaved-caspase 3/NeuN immunostaining 3 days after ICH, as reflected by the quantification. n = 5 mice per group. H, J Representative images of NeuN immunostaining 3 days after ICH, as reflected by the quantification. n = 5 mice per group. All values are expressed as mean ± SD. The data were analyzed using two-way analysis of variance or one-way analysis of variance test. ***p < 0.001

Beyond acute neuroprotection, microglial phagocytic activity is crucial for clearing degenerated myelin debris during the chronic phase of ICH to facilitate functional recovery. To investigate the impact of hUMSC-Exos on this process, we assessed the co-localization of the microglial marker Iba1 and degenerated MBP (dMBP) at 35 days post-ICH. In Sham mice, minimal Iba1⁺dMBP⁺ co-localization was observed, whereas ICH mice exhibited increased microglial phagocytosis of myelin debris. Notably, hUMSC-Exos administration further enhanced the percentage of Iba1⁺dMBP⁺ area relative to the total dMBP⁺ area compared to ICH controls (Fig. S2), demonstrating that hUMSC-Exos potentiate microglial clearance of myelin debris in the chronic phase.

Collectively, these findings establish hUMSC-Exos as a multi-faceted therapeutic agent: they mediate acute neuroprotection via inhibition of neuronal apoptosis, promote clearance of myelin debris by microglia during the chronic phase, and confer sustained functional recovery through preservation of white matter integrity. These results highlight hUMSC-Exos as a promising translational strategy for mitigating neurological sequelae after ICH.

hUMSC-derived exosomes reprogram microglial polarization and suppress neuroinflammation after intracerebral hemorrhage

ICH triggers a dysregulated neuroinflammatory response dominated by microglial activation, which exacerbates secondary brain injury. Here, we show that hUMSC-Exos potently modulate microglial phenotype and dampen inflammatory cascades after ICH. Intracerebral delivery of hUMSC-Exos significantly remodeled microglial polarization in the perihematomal striatum. At days 3 and 7 post-ICH, immunofluorescence analysis revealed a marked decrease in pro-inflammatory microglia (Iba1⁺/CD16⁺) and a corresponding increase in anti-inflammatory microglia (Iba1⁺/Arg1⁺) in hUMSC-Exos-treated mice compared to ICH controls (Fig. 3A–D). This phenotypic shift was associated with a 40–60% reduction in M1 marker co-expression and a 2.5-fold increase in M2 marker coverage, as quantified by confocal microscopy. Quantitative RT-PCR at day 3 post-ICH validated the functional shift in microglial activation. hUMSC-Exos treatment led to significant downregulation of canonical pro-inflammatory genes (CD16, IL6, iNOS, IL1β; Fig. 3E) and upregulation of anti-inflammatory associated transcripts (TGF-β, YM1/2, CD206, Arg1; Fig. 3F). Notably, the mRNA levels of IL1β and iNOS were suppressed by 70–80%, while Arg1 and CD206 showed 3–4-fold increases, indicating a robust transition toward a neuroprotective microglial state. Concomitant with phenotypic reprogramming, hUMSC-Exos attenuated microglial phagocytic activity. Immunofluorescence staining at day 3 revealed reduced co-localization of Iba1 with the phagocytic markers CD68 and Mac2 in the perihematomal region (Fig. 3G–J), with quantitative analysis showing a 50% decrease in phagolysosomal puncta.

Fig. 3.

hUMSC-derived exosomes modulate microglial polarization and suppress neuroinflammation after ICH. A, C Representative images of Iba1/CD16 and Iba1/Arg1 immunofluorescence in the peri-hematoma STR 3 days after ICH, as reflected by the quantification. n = 5 per group. B, D Representative images of Iba1/CD16 and Iba1/Arg1 immunofluorescence in the peri- hematoma STR 7 days after ICH, as reflected by the quantification. n = 5 per group. E The mRNA level of CD16, IL6, INOS, IL1β of peri-hematoma tissue following 3d after ICH. n = 5 per group. F The mRNA level of TGF-β, YM1/2, CD206, Arg1 of peri-hematoma tissue following 3d after ICH. n = 5 per group. G, I Representative images of CD68/Iba1 immunofluorescence in the peri-hematoma STR 3 days after ICH, as reflected by the quantification. n = 5 per group. H, J Representative images of Iba1/Mac2 immunofluorescence in the peri-hematoma STR 3 days after ICH, as reflected by the quantification. n = 5 per group. All values are expressed as mean ± SD. The data were analyzed using one-way analysis of variance test. *p < 0.05, **p < 0.01, ***p < 0.001, ns = not significant

The dual effects of hUMSC-Exos—promoting anti-inflammatory polarization and dampening phagocytic hyperactivity—collectively reshape the post-ICH immune microenvironment. By reducing pro-inflammatory cytokine release and preserving neural tissue from excessive phagocytic clearance, hUMSC-Exos likely mitigate secondary neuronal injury. These findings establish hUMSC-Exos as a novel therapeutic modality to orchestrate microglial functional plasticity and combat neuroinflammation after ICH.

hUMSC-derived exosomes suppress hemin-induced microglial activation and shift phenotype toward anti-inflammatory states in vitro

ICH releases hemoglobin-derived hemin, triggering robust pro-inflammatory microglial activation that exacerbates secondary brain injury. While in vivo studies link hUMSC-exosomes to microglial polarization, their direct effects on hemin-challenged microglia remain undefined. Here, we dissect the anti-inflammatory mechanisms of hUMSC-Exos in primary and BV2 microglial models of hemin-induced injury. Immunofluorescence revealed that Hemin significantly increased CD16⁺/Iba1⁺ pro-inflammatory microglia (Fig. 4A, H left), while hUMSC-Exos markedly reduced CD16 expression. Conversely, Arg1⁺/Iba1⁺ anti-inflammatory microglia were elevated following exosome treatment (Fig. 4B, H right), suggesting a phenotypic shift toward a reparative state. Consistently, ELISA measurements demonstrated that Hemin-induced secretion of pro-inflammatory cytokines—iNOS, IL-1β, and TNF-α—was significantly suppressed by hUMSC-Exo co-treatment (Fig. 4C–E). In parallel, anti-inflammatory cytokines IL-10 and TGF-β were upregulated following exosome administration (Fig. 4F–G). These effects were reproducible in BV2 microglial cells, where exosome treatment downregulated Il-6 and Il-1β mRNA expression, and enhanced Il-10 transcription (Fig. 4I), reinforcing the broad anti-inflammatory efficacy of hUMSC-Exos.

Fig. 4.

hUMSC-derived exosomes attenuate pro-inflammatory activation and promote anti-inflammatory polarization in microglia in vitro. A Immunofluorescence staining of primary microglial cells (pMGs) treated with control, Hemin, or Hemin + hUMSC-Exos. Iba1 labels total microglia (red); CD16 (green, A) and Arg1 (yellow, B) mark pro- and anti-inflammatory phenotypes, respectively. DAPI stains nuclei (blue). C–E ELISA quantification of pro-inflammatory cytokines (iNOS, IL-1β, and TNF-α) in supernatants of pMG cultures. n = 3 per group. F–G ELISA of anti-inflammatory cytokines IL-10 and TGF-β in the same conditions. N = 3 per group. H Quantification of CD16⁺/Iba1⁺ and Arg1⁺/Iba1⁺ microglia from immunofluorescence images. n = 3 per group. I qRT-PCR of inflammatory gene expression (Il-6, Il-1β, Il-10) in BV2 cells under the same treatments. n = 5 per group. Data are shown as mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001; one-way ANOVA test

These in vitro data elucidate a cell-autonomous mechanism by which hUMSC-Exos suppress hemin-induced microglial pro-inflammatory activation and promote an anti-inflammatory phenotype. By targeting both phenotypic markers and cytokine signaling, exosomes rewire microglial function toward neuroprotection—an effect that aligns with in vivo observations of reduced neuroinflammation after ICH. This dual regulation of microglial polarization and cytokine secretion establishes hUMSC-Exos as a potent modulator of innate immune responses in the injured brain, supporting their therapeutic potential for ICH and other neuroinflammatory disorders.

Global RNA-seq analysis uncovers hUMSC-derived exosomes reprogram inflammatory transcriptional signatures in microglia after intracerebral hemorrhage

While phenotypic and cytokine-level changes induced by hUMSC-Exos in microglia after ICH have been explored, the global transcriptional reprogramming underlying these effects remains undefined. To address this, we performed RNA-sequencing on microglia isolated from sham-operated, ICH, and ICH + hUMSC-Exo–treated mice (Fig. 5A). We successfully isolated and purified microglia, as verified by RT-qPCR analysis of cell-type-specific markers (oligodendrocytes, neurons, astrocytes, and microglia) that demonstrated enrichment of microglial markers and reduced expression of markers for other neural cell populations (Fig. S3). RNA-seq data quality control was validated by saturation analysis demonstrating sufficient sequencing depth to capture transcriptomic complexity (Fig. S4) and by violin plots and Venn diagrams showing consistent gene expression distributions and biologically relevant overlaps of expressed genes across Sham, ICH, and ICH + Exo groups (Fig. S5). Principal component analysis demonstrated distinct transcriptional clustering across the three groups: the ICH cohort segregated completely from sham controls, whereas hUMSC-Exo–treated microglia partially overlapped with the sham transcriptome (Fig. 5B), indicating broad reversal of ICH-induced transcriptional alterations. Differential expression analysis showed a marked increase in both upregulated and downregulated genes in ICH vs. sham (Fig. 5C, left), while hUMSC-Exo treatment reversed a large subset of these changes (Fig. 5C, right). Volcano plots and hierarchical clustering further confirmed that hUMSC-Exos broadly attenuated the ICH-induced transcriptional activation of proinflammatory programs (Fig. 5D–G). GSEA revealed significant enrichment of inflammatory pathways—including responses to LPS, biotic stimuli, and tumor necrosis factor production—in ICH microglia compared to sham (Fig. 5H). Notably, these inflammatory programs were downregulated following hUMSC-Exo treatment, which instead promoted gene sets associated with TGF-β response, cell adhesion, and tissue remodeling (Fig. 5I), indicating a transcriptional reprogramming toward an anti-inflammatory and reparative phenotype.

Fig. 5.

hUMSC-Exos reshape global microglial transcriptional landscape and attenuate inflammatory signatures following ICH. A Schematic workflow illustrating microglial isolation, total RNA extraction, and RNA-seq library construction from sham, ICH, and ICH + hUMSC-Exo groups. B Principal component analysis (PCA) of transcriptomic profiles distinguishes Sham, ICH, and ICH + Exo groups. C Bar plot showing the number of significantly upregulated (red) and downregulated (blue) genes in pairwise comparisons. D-E Volcano plot (D) and hierarchical clustering heatmap (E) of differentially expressed genes (DEGs) in ICH vs. Sham microglia. F-G Volcano plot (F) and heatmap (G) showing reversal of ICH-induced transcriptional changes after hUMSC-Exo treatment. H Gene set enrichment analysis (GSEA) of ICH vs. Sham highlights enrichment of proinflammatory programs such as LPS response, biotic stimulus, and TNF production. I GSEA of ICH + Exo vs. ICH shows enrichment of anti-inflammatory and reparative pathways, including TGF-β signaling, cell adhesion, and tissue remodeling. NES = normalized enrichment score; FDR = false discovery rate

These global transcriptomic data elucidate the molecular underpinnings of hUMSC-Exo–mediated microglial reprogramming after ICH: exosomes not only dampen pro-inflammatory transcriptional cascades but also activate pathways linked to tissue repair. By rewiring the microglial transcriptome at a genome-wide scale, hUMSC-Exos establish a pro-reparative immune microenvironment, providing a mechanistic framework for their in vivo neuroprotective effects. This transcriptional reprogramming highlights hUMSC-Exos as a potent modulator of microglial function in acute brain injury.

Trajectory analysis identifies exosome-mediated suppression of inflammatory modules and iNOS⁺ microglia in the ICH brain

While transcriptomic profiling revealed broad anti-inflammatory effects of hUMSC-Exos on microglia after ICH, the molecular trajectories of gene expression changes and key pathological cell states remain undefined. To dissect these mechanisms, we integrated trajectory analysis, gene ontology (GO) enrichment, and protein-level validation. Comparative GO analysis of microglial transcriptomes (ICH vs. sham) identified strong enrichment of proinflammatory and proliferative pathways, including mitotic cell cycle, cytokine response, and innate immune activation (Fig. 6A). In contrast, genes downregulated in ICH + Exo vs. ICH were predominantly associated with pathological remodeling—specifically cell migration, adhesion, angiogenesis, and vascular development (Fig. 6B). This suggests hUMSC-Exos not only dampen inflammation but also repress injury-driven maladaptive processes. Heatmaps of selected genes confirmed robust upregulation of proinflammatory mediators (e.g., Tnf, Trem1, Nos2, Ccr7) in ICH microglia, with marked downregulation following hUMSC-Exo treatment (Fig. 6C–D). Notably, Nos2 (encoding iNOS, a key mediator of oxidative stress in microglia) exhibited significant increase in ICH, which was reduced after exosome administration, foreshadowing targeted suppression of iNOS⁺ microglia.

Fig. 6.

Temporal transcriptome clustering identifies inflammation-associated gene modules suppressed by hUMSC-Exos after ICH. A–B Gene ontology (GO) enrichment analysis of significantly upregulated genes in ICH vs. Sham (A) and downregulated genes in ICH + Exo vs. ICH (B), highlighting inflammatory, mitotic, and vascular-associated pathways. C–D Heatmaps of selected representative genes showing upregulation in ICH vs. Sham (C) and downregulation in ICH + Exo vs. ICH (D), including inflammation- and proliferation-related targets. E Time-series clustering of gene expression profiles across Sham, ICH, and ICH + Exo groups, revealing 8 distinct dynamic expression profiles. F Gene count distribution across profiles, with Profile 5 containing 17 dynamically regulated genes and its expression trajectory highlighted below. G GO enrichment analysis of Profile 5 genes, indicating enrichment in inflammatory and innate immune response pathways, including lipopolysaccharide response, neutrophil chemotaxis, and superoxide metabolism. H Immunofluorescence staining of Iba1 (red) and iNOS (green) in perihematomal brain regions across Sham, ICH, and ICH + Exo groups. ICH induced marked upregulation of iNOS in Iba1⁺ microglia, which was attenuated after hUMSC-Exo treatment

Temporal clustering across Sham, ICH, and ICH + Exo conditions identified 8 distinct gene expression profiles (Fig. 6E). Profile 5 exhibited a “rise-then-fall” trajectory: robust upregulation after ICH, followed by significant suppression with hUMSC-Exos (Fig. 6F). GO analysis of Profile 5 genes revealed dominant enrichment in innate immune and oxidative stress pathways, including cell migration, superoxide metabolism, LPS response, and neutrophil chemotaxis (Fig. 6G)—pathways central to ICH-induced secondary injury. To confirm these transcriptional findings, we performed Iba1/iNOS co-immunofluorescence. ICH induced increase in iNOS⁺/Iba1⁺ microglia in the perihematomal region, whereas hUMSC-Exo treatment reduced this population (Fig. 6H). This confirms that exosomes directly target iNOS⁺ microglia—an effector cell state driving oxidative stress and neuroinflammation after ICH.

Together, trajectory analysis reveals hUMSC-Exos act by: (1) suppressing a core inflammatory gene module linked to innate immunity and oxidative stress; (2) repressing pathological remodeling pathways; and (3) directly inhibiting iNOS⁺ microglia. By intercepting these convergent pathogenic processes, exosomes rewire microglial function toward a reparative state, providing a mechanistic blueprint for their therapeutic efficacy.

SIRT1-mediated suppression of the NF-κB/NOS2 axis underlies the anti-inflammatory efficacy of hUMSC-derived exosomes in microglia after intracerebral hemorrhage

While hUMSC-Exos reprogram microglial polarization after ICH, the molecular circuitry linking exosome signaling to inflammatory suppression remains unclear. Here, we dissect a SIRT1-dependent pathway wherein exosomes repress NF-κB/NOS2 signaling to mitigate microglial neuroinflammation. In BV2 microglial migration assays, Hemin-induced chemotaxis was significantly reduced by hUMSC-Exos, as well as the effect of the NOS2 inhibitor 1400 W (Fig. 7A). Corresponding qPCR analysis revealed that hUMSC-Exos and 1400 W co-treatment synergistically downregulated pro-inflammatory transcripts (Nfkb1, Tnf, Cxcl1, Cxcl13; Fig. 7B), indicating NOS2 suppression as a core mechanism of exosome-mediated inflammation attenuation. In primary microglial (pMG) cultures, Hemin robustly induced NOS2 expression, as evidenced by iNOS immunofluorescence (Fig. 7C) and qPCR (Fig. 7D). Both hUMSC-Exos and 1400 W significantly reduced NOS2 levels, with the combination exerting additive effects, confirming NOS2 as a key effector of exosome action.

Fig. 7.

SIRT1-mediated repression of NF-κB/NOS2 signaling potentiates the anti-inflammatory effects of hUMSC-derived exosomes in microglia. A Migration assay of BV2 microglia showing the effect of hUMSC-Exo and NOS2 inhibition on cell migration following Hemin stimulation. n = 5 per group. B mRNA expression levels of pro-inflammatory genes Nkfb1, Tnf, Cxcl1, and Cxcl13 in BV2 microglial cells treated with Hemin, NOS2 inhibitor, and hUMSC-Exos. n = 5 per group. C Immunofluorescence staining for iNOS (yellow) and Iba1 (red) in primary microglial cells treated with Hemin, hUMSC-Exos, and NOS2 inhibitor. D Quantification of iNOS⁺ microglial cells, showing that NOS2 inhibition combined with exosome treatment significantly reduces the number of activated microglia. n = 5 per group. E NOS2 mRNA expression in primary microglial cells treated with hUMSC-Exos and SIRT1 inhibitor. n = 5 per group. F In vivo detection of NOS2 expression in mouse brain tissue following hUMSC-Exo and SIRT1 inhibition treatments. n = 5 per group. G Western blot analysis of SIRT1, p-p65, and p65 protein levels in brain tissue from ICH mice treated with hUMSC-Exos and SIRT1 inhibitors. n = 3 per group. H Immunofluorescence staining for SIRT1 and Iba1 in the perihematomal region of Sham, ICH, ICH + Exo and ICH + Exo + EX527 groups. I Immunofluorescence staining of NF-κB p65 and Iba1 in Sham, ICH, ICH + Exo and ICH + Exo + EX527 groups. J-K Quantification of SIRT1⁺/Iba1⁺ and NF-κB p65⁺/Iba1⁺ microglial cells in Sham, ICH, ICH + Exo and ICH + Exo + EX527 groups. n = 5 mice per group. Data are presented as means ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001, one-way ANOVA test. “ns” indicates no significant difference

To interrogate upstream regulators, we used the SIRT1 inhibitor EX527. In pMG, EX527 reversed the NOS2-suppressive effect of hUMSC-Exos, as shown by increased Nos2 mRNA levels (Fig. 7E), demonstrating SIRT1 dependence. In vivo, hUMSC-Exos reduced Nos2 expression in the perihematomal region of ICH mice, while EX527 co-treatment abrogated this effect (Fig. 7F), validating the SIRT1-NOS2 axis in the ICH brain. Western blotting revealed that hUMSC-Exos increased SIRT1 protein levels and decreased phosphorylated p65 (p-p65) in ICH mouse brains (Fig. 7G), indicating SIRT1 activation and NF-κB inhibition. Immunofluorescence confirmed SIRT1 upregulation in Iba1⁺ microglia of ICH + Exo mice (Fig. 7H), concomitant with reduced nuclear localization of NF-κB p65 (Fig. 7I–K). These data indicate that exosomes activate SIRT1 to suppress NF-κB nuclear translocation, thereby inhibiting NOS2 transcription.

Collectively, these findings delineate a hierarchical pathway: hUMSC-Exos activate SIRT1, which represses NF-κB signaling to inhibit NOS2-dependent microglial migration and pro-inflammatory cytokine production. This SIRT1-NF-κB-NOS2 axis represents a core molecular mechanism by which exosomes exert anti-inflammatory effects, providing a druggable target for modulating microglial function after ICH.

SIRT1 is required for hUMSC-Exos to regulate microglial proliferation, neuronal apoptosis, and behavioral outcomes after ICH

To determine if SIRT1 mediates the effects of hUMSC-Exos in ICH, we employed the specific SIRT1 inhibitor EX527. Immunofluorescence staining for the proliferation marker Ki67 showed that ICH increased the abundance of Ki67⁺Iba1⁺ cells in the perihematomal region compared to Sham controls, and hUMSC-Exos decreased Ki67⁺Iba1⁺ cell numbers. Notably, co-treatment with EX527 abolished the Exos-induced inhibition of microglial proliferation (Fig. 8A, D). For assessing neuronal apoptosis, TUNEL staining combined with the neuronal marker NeuN (Fig. 8B–C) revealed a substantial increase in TUNEL⁺NeuN⁺ apoptotic neurons in the ICH group at both 3 days and 7 days post-injury. hUMSC-Exos treatment significantly reduced the count of apoptotic neurons, while co-treatment with EX527 reversed this anti-apoptotic effect, restoring apoptosis to levels similar to those in the ICH group (Fig. 8D).

Fig. 8.

SIRT1 mediates hUMSC-Exos-induced regulation of microglial proliferation, neuronal apoptosis, and behavioral recovery after ICH. A Immunofluorescence staining for the proliferation marker Ki67 and microglial marker Iba1 in perihematomal brain sections to assess microglial proliferation. B-C TUNEL staining combined with neuronal marker NeuN to detect apoptotic neurons at 3 days (B) and 7 days (C) post-ICH. D Quantitative analysis of Ki67⁺Iba1⁺ microglia and TUNEL⁺NeuN⁺ apoptotic neurons. n = 6 mice per group. E Representative movement trajectories of experimental groups. F Quantification of total distance traveled (left) and time spent in the center zone (right) to evaluate locomotor activity and anxiety-like behavior. n = 10 mice per group. G Escape latency during the 5-day acquisition phase. H Escape latency on Day 34 of acquisition. n = 10 mice per group. I Representative swim paths during the probe trial. J Quantification of swimming speed (left), time spent in the target quadrant (middle), and frequency of crossing the original platform location (right) during the probe trial. n = 10 mice per group. Data are mean ± SEM. ns = not significant. *p < 0.05, **p < 0.01, ***p < 0.001, one-way ANOVA test

Behavioral assessments confirmed the dependency on SIRT1. In the Open Field Test (Fig. 8E–F), ICH mice exhibited reduced total distance traveled and decreased time spent in the center zone (a measure of anxiety-like behavior) relative to Sham group. hUMSC-Exos alleviated these deficits, increasing locomotor activity and central exploration—effects that were blocked by co-treatment with EX527. In the Morris Water Maze, ICH mice displayed prolonged escape latency during the acquisition phase (Fig. 8G–H) and impaired spatial memory (evidenced by reduced time in the target quadrant and fewer crossings of the original platform during the probe trial; Fig. 8I–J). hUMSC-Exos improved learning and memory, but co-treatment with EX527 abrogated these benefits, such that co-treatment group showed outcomes comparable to those in the ICH group.

Collectively, these data demonstrate that SIRT1 is essential for hUMSC-Exos to modulate microglial proliferation, suppress neuronal apoptosis, and facilitate sensorimotor and cognitive recovery after ICH, highlighting SIRT1 as a pivotal mediator of Exos-induced neuroprotective and functional effects.

Discussion

Our study establishes hUMSC-derived exosomes as a promising cell-free therapeutic for ICH, revealing a novel mechanism by which exosomes reprogram microglial inflammation through a SIRT1-NF-κB/NOS2 signaling axis. This work integrates in vivo, in vitro, and transcriptomic approaches, uniquely combining microglia-specific RNA-sequencing with functional validation in both cell lines and primary cells to dissect the neuroprotective effects of hUMSC-Exos. And exosome-mediated benefit is SIRT1-dependent, as pharmacological inhibition of SIRT1 abrogates exosomes’ ability to inhibit microglial proliferation, suppress neuronal apoptosis, and rescue sensorimotor and cognitive deficits after ICH. By demonstrating that intranasal delivery of exosomes mitigates secondary brain injury via anti-inflammatory reprogramming and SIRT1-dependent modulation of neurogenesis and apoptosis, our findings expand the therapeutic repertoire for ICH and provide a mechanistic framework that underscores SIRT1 as a tractable target for enhancing exosome-based therapies.

By focusing on exosomes and their target-specific actions in microglia, we provide a more precise molecular framework, which is essential for developing targeted therapies. Moreover, the use of intranasal delivery—noninvasive and translatable to humans—addresses a critical challenge in neurotherapeutics, where systemic administration often yields poor brain bioavailability. Recent advances in drug delivery have positioned intranasal administration as a noninvasive, efficient route for CNS-targeted therapies, bypassing the blood-brain barrier (BBB) and minimizing systemic exposure [20, 21]. Studies leveraging this approach demonstrate its versatility: RGD-modified exosomes enhance vascular repair in spinal cord injury [22], macrophage-derived exosomes deliver resveratrol to suppress neuroinflammation in multiple sclerosis [23], and BDNF-loaded exosomes promote remyelination [24]. Similarly, intranasal exosomes derived from 3D-cultured stem cells improve synaptic plasticity in Alzheimer’s models [25], while MALAT1-enriched exosomes extend the therapeutic window for traumatic brain injury [26]. Our work aligns with and expands these findings by demonstrating that intranasally delivered hUMSC-Exos exhibit broad tropism for microglia, astrocytes, and neurons in ICH, accumulating dose-dependently in perihematomal regions to reduce apoptosis and enhance functional recovery. This underscores the translational potential of intranasal exosome delivery for acute CNS pathologies.

Exosomes have emerged as promising therapeutics for stroke due to their innate ability to traverse the BBB and modulate neuroinflammation, angiogenesis, and cell survival [27–30]. In ischemic stroke, exosomes from M2 microglia deliver miR-124-3p to inhibit neuronal apoptosis via ROCK/PTEN/AKT signaling [31], while biomaterial-assisted exosome delivery (e.g., hyaluronic acid hydrogels) sustains neurovascular repair [32]. Engineered exosomes carrying BDNF further enhance neuroprotection by mitigating oxidative stress and microglial activation [33]. Conversely, macrophage-derived exosomes exacerbate injury via THBS1-induced endothelial ferroptosis [34], highlighting the need for precise exosome engineering. For hemorrhagic stroke, exosomes from mesenchymal stem cells (MSCs) or astrocytes attenuate ICH injury by modulating miR-150-3p/TRAF6 [35], miR-27a-3p/ARHGAP25/Wnt [36], or PINK1-dependent microglial polarization [37]. Our transcriptomic data extend these insights by revealing that hUMSC-Exos suppress NF-κB/NOS2 signaling in microglia, disrupting the cycle of inflammation and neuronal damage—a mechanism critical for ICH recovery.

The therapeutic effects of exosomes in ICH are closely tied to their modulation of microglial function. Polarized microglia dictate outcomes: M1-derived exosomes propagate inflammation, while M2-derived exosomes promote repair via autophagy and mitochondrial pathways [38]. Engineered exosomes (e.g., SIRPα variants) further enhance hematoma clearance and M2 polarization [39], and miR-124-loaded exosomes reduce leukocyte infiltration [21]. Our study corroborates these findings, demonstrating that hUMSC-Exos transcriptionally reprogram microglia to downregulate proinflammatory NF-κB/NOS2 networks, which drive cytotoxic mediator release (e.g., iNOS, TNF-α). This aligns with evidence that exosomal miR-133b suppresses RhoA/ERK1/2-CREB signaling to mitigate apoptosis [40], and miR-1228-5p improves mitochondrial function post-stroke [10]. Importantly, our data suggest a dual role for hUMSC-Exos: not only attenuating injury but also activating tissue remodeling and vascular regeneration pathways, offering a comprehensive strategy for ICH recovery. Our findings position hUMSC-Exos as a multifaceted therapy for ICH, capable of modulating microglial inflammation while promoting repair. The intranasal route enhances clinical feasibility, avoiding invasive delivery challenges. However, key questions remain, including the optimal exosome dosage and timing, the role of specific cargoes (e.g., miRNAs, proteins), and long-term safety [41]. Future work should explore combinatorial strategies, such as biomaterial scaffolds for sustained release [42] or exosomes engineered to target additional pathways (e.g., ferroptosis [43–45]). By addressing these gaps, exosome-based therapies may transition toward precision medicine for stroke and other CNS disorders.

Emerging evidence highlights SIRT1, a NAD⁺-dependent deacetylase, as a pivotal mediator of neuroprotection in stroke, modulating oxidative stress, inflammation, and microglial activation. In ischemic stroke models, SIRT1 activation—via pharmacological agents (e.g., resveratrol, SRT1720) or natural compounds (e.g., bergenin)—attenuates neuroinflammation by suppressing NF-κB signaling and enhancing antioxidant defenses (SOD, Nrf2) [46, 47]. Similarly, transcutaneous electrical acupoint stimulation (TEAS) reduces infarct volume and neuronal apoptosis via SIRT1-dependent inhibition of the BRCC3/NLRP3 inflammasome and FOXO3a activation [48]. Beyond ischemia, SIRT1’s role extends to hemorrhagic stroke, where Arc, a synaptic plasticity regulator, exerts neuroprotection in subarachnoid hemorrhage (SAH) by preserving SIRT1/Nrf2 signaling [49]. Our findings align with and expand this paradigm, demonstrating that hUMSC-Exos activate SIRT1 to suppress NF-κB nuclear translocation, thereby reducing NOS2 expression and pro-inflammatory cytokine release—a mechanism critical for mitigating microglial-driven neurotoxicity in ICH.

Recent studies reveal that long noncoding RNAs (lncRNAs) fine-tune SIRT1 activity, offering novel therapeutic avenues for stroke. SNHG8, a neuroprotective lncRNA, mitigates ischemic injury by sponging miR-449c-5p and miR-425-5p, thereby preserving SIRT1/FoxO1 signaling and suppressing NF-κB-mediated neuroinflammation [50, 51]. This regulatory axis not only dampens microglial activation but also reinforces blood-brain barrier (BBB) integrity, highlighting lncRNAs as master regulators of post-stroke recovery. Our work complements these insights, showing that hUMSC-Exos epigenetically modulate SIRT1—an observation supported by the reversal of exosomal benefits upon SIRT1 inhibition. These data suggest that exosomal cargo (e.g., miRNAs or lncRNAs) may indirectly regulate SIRT1, akin to endogenous lncRNA-mediated mechanisms, though further profiling is needed to identify specific mediators.

Microglial behavior dictates stroke outcomes, with M1 pro-inflammatory phenotypes exacerbating injury and M2 anti-inflammatory phenotypes promoting repair. SIRT1 activation—whether by melatonin (via HMGB1/TLR4/NF-κB suppression), Apelin-13 (via SIRT1/NF-κB axis), or engineered exosomes (e.g., SIRPα variants)—drives microglial polarization toward the M2 state, reducing cytotoxic cytokine release (e.g., TNF-α, IL-1β) [52, 53]. Our study reinforces this concept, demonstrating that hUMSC-Exos suppress NF-κB/NOS2 signaling in microglia via SIRT1, mirroring effects seen with pharmacological SIRT1 agonists. Notably, the dependence of exosomal effects on SIRT1 (evidenced by inhibition experiments) positions SIRT1 as a non-redundant node in exosome-mediated immunomodulation, bridging extracellular vesicle therapy with intrinsic epigenetic regulation. The convergence of SIRT1-centric pathways—spanning lncRNA regulation, microglial reprogramming, and epigenetic modulation—underscores its potential as a master therapeutic target for stroke. Our findings with hUMSC-Exos further validate this approach, demonstrating that cell-free therapies can harness endogenous SIRT1 mechanisms to combat neuroinflammation.

Despite these advances, our study has limitations that warrant consideration. First, while we demonstrate a functional role for SIRT1 in microglia, the specific cargo within hUMSC-Exos that triggers SIRT1 upregulation remains undefined. Identifying key microRNAs, proteins, or metabolites responsible for this effect will be crucial for optimizing exosome-based therapies. Second, the murine ICH model used here reflects early-stage injury, and the long-term effects of hUMSC-Exos on scar formation, angiogenesis, and cognitive recovery remain to be explored. Additionally, while we show uptake by multiple cell types, the relative contribution of astrocyte or neuronal exosome signaling to the observed neuroprotection requires further dissection. Finally, species-specific differences in exosome uptake and signaling pathways necessitate validation in larger animal models before clinical translation.

Conclusions

In conclusion, we identify hUMSC-Exos as a potent cell-free therapeutic for ICH, driven by a SIRT1-dependent mechanism that reprograms microglial inflammation to restrain neurotoxicity and foster brain repair. Mechanistically, hUMSC-Exos converge on SIRT1 to modulate NF-κB/NOS2 signaling in microglia, unraveling a molecular link between stem cell-derived nanovesicle signaling and microglial functional polarization. These results not only position hUMSC-Exos as a promising precision therapy for ICH but also exemplify how exosome-based strategies can target immune cell dysregulation in acute neurological injuries, paving the way for rational design of nanomedicine in neuroinflammation.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

The study was conceived by DWR and JZ. DWR and JZ were responsible for designing and executing the experiments, analyzing the data, and drafting the manuscript. ZYZ, LCW, HL and YQW provided assistance with the experimental procedures and contributed to the data analysis. The animal experiments were conducted with the help of HL and CL. YNM, JH and ESW provided overall study conceptualization, experimental supervision, and manuscript revision. All authors reviewed and approved the final version of the manuscript.

Funding

This work was supported by the Shanghai Jinshan District Health System Medical Key Specialty Construction Project (No. JSZK2023B01), the Jinshan District Health Specialized Research Project (No. JSKJ-KTMS-2024-03), and the Youth Research Start-up Fund Project of Jinshan Hospital Affiliated to Fudan University (No. JYQN-JC-202208, JYQN-LC-202309 and JYQN-JC-202508).

Data availability

Not applicable.

Declarations

Ethical approval

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of Shanghai Public Health Clinical Center.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.