Human umbilical cord–derived mesenchymal stem cell exosomes ameliorate hepatocyte ferroptosis and sepsis-associated liver injury by restoring iron homeostasis in hepatocytes

Abstract

Mesenchymal stem cells-derived exosomes (MSCs-Exo) constitute critical mediators of intercellular communication between stem cells and other cell types. Among these, exosomes derived from human umbilical cord mesenchymal stem cells (HUMSCs-Exo) have attracted considerable attention owing to their potent immunomodulatory capacity, low surface antigenicity, and favorable biosafety profile. Accumulating evidence indicates that HUMSCs-Exo can attenuate ferroptosis in target tissues across a spectrum of pathological conditions, including inflammatory bowel disease and viral cardiomyopathy, thereby mitigating tissue injury, improving organ function, and prolonging host survival. Nevertheless, the regulatory effects of HUMSCs-Exo on sepsis-associated liver injury (SALI) and hepatocellular ferroptosis, as well as the underlying molecular mechanisms, remain largely undefined. In this study, we identified pronounced differential gene expressions between control and septic mice hepatocytes, particularly within pathways related to iron metabolism, lipid metabolism, and ferroptosis. Building on these findings, we systematically characterized the dynamic progression of hepatocyte ferroptosis in septic mice through integrated in vivo and in vitro experiments, demonstrating a strong positive correlation between ferroptotic activity and the severity of liver injury. Using wild-type male C57BL/6J mice and the murine normal hepatocyte cell line NCTC1469, we further demonstrated that HUMSCs-Exo markedly upregulate the expression of FPN, xCT, GPX4, and FTH1 while concomitantly downregulating ACSL4 and TFRC. These molecular alterations reduced intracellular lipid peroxidation and labile iron accumulation, thereby robustly attenuating sepsis-induced hepatocyte ferroptosis and conferring significant hepatoprotective effects. Finally, these findings were validated in the immortalized human normal hepatocyte cell line THLE-2, in which HUMSCs-Exo similarly suppressed lipopolysaccharide-induced ferroptosis and cellular injury by modulating intracellular lipid peroxide and free iron levels. Collectively, our results demonstrate that HUMSCs-Exo exert potent inhibitory effects on sepsis-induced hepatocyte ferroptosis and confer protection against liver injury, primarily through the restoration of intracellular iron homeostasis and the suppression of lipid peroxidation. This study provides a novel therapeutic strategy for the treatment of SALI and ferroptosis.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12951-026-04270-z.

Background

Sepsis is a life-threatening condition characterized by dysregulated host responses to infection [1], leading to acute organ dysfunction. Globally, it remains associated with high incidence, substantial mortality, and a significant disease burden [2–4]. The liver, as a central hub of systemic circulation, plays a dual role in the pathophysiology of sepsis—serving both as a defender and a victim. It orchestrates critical physiological functions [5], including the synthesis of acute-phase proteins, production of inflammatory cytokines, clearance of pathogens, and regulation of coagulation. However, it is also highly susceptible to injury during the early stages of sepsis [5, 6]. Hepatic insults manifest as hypoxic hepatitis due to ischemia and shock, cholestasis caused by impaired bile metabolism, hepatocellular damage driven by the interplay of immune activation and immunosuppression, and secondary sclerosing cholangitis. In severe cases, this progression may culminate in acute liver failure (ALF). Extensive clinical evidence has established a close association between sepsis-associated liver injury (SALI) and both short- and long-term mortality and adverse outcomes [5, 7–9]. These findings underscore the importance of delineating liver- and hepatocyte-specific alterations during sepsis, which is critical for improving survival and prognosis in affected patients.

Despite the complex and multifactorial pathogenesis of SALI, preclinical evidence suggests that septic hosts exhibit profound dysregulation of hepatic iron and lipid metabolism, accompanied by elevated oxidative stress [10, 11]. These metabolic disturbances [12, 13], particularly those involving the TFRC/FTH1/NCOA4/FPN axis, the SLC7A11 (xCT)/GSH/GPX4 antioxidant system, and the PUFAs/ACSL4 pathway, are intimately linked to ferroptosis [14, 15], an iron-dependent form of regulated cell death marked by excessive lipid peroxidation and/or intracellular ferrous iron accumulation, alongside redox imbalance. Collectively, these findings implicate hepatocellular ferroptosis as a potentially critical contributor to SALI and raise the possibility that therapeutic inhibition of ferroptosis may offer a novel strategy for mitigating liver injury in septic conditions. Nevertheless, the temporal dynamics of hepatocyte ferroptosis during sepsis and its correlation with liver injury severity remain incompletely undefined. Here, using in vivo experiments, we delineate the dynamic trajectory of hepatocyte ferroptosis in septic mice and establish its association with the severity of liver injury.

Concurrently, mounting evidence has positioned mesenchymal stem cells-derived exosomes (MSCs-Exo) as a promising paradigm in cell-free therapeutic interventions [16, 17]. Acting as pivotal mediators of intercellular communication, these nanoscale vesicles possess the intrinsic capability to traverse biological barriers and be selectively internalized by target cells, delivering a diverse cargo of bioactive molecules, including nucleic acids, proteins, and lipids, to exert multifaceted effects such as immunomodulation and tissue repair. Exosomes derived from human umbilical cord mesenchymal stem cells (HUMSCs-Exo) have attracted particular interest owing to potent immunoregulatory activity and low surface antigenicity [18–20]. HUMSCs-Exo have been shown across multiple disease models, including inflammatory bowel disease [21], diabetes [22], primary ovarian insufficiency [23] and viral myocarditis [24], to ameliorate tissue injury and organ dysfunction and extend host survival, in part by suppressing ferroptosis through modulation of key effectors such as ACSL4, GPX4 and SLC7A11 at the protein and/or transcript level. Nevertheless, whether HUMSCs-Exo attenuate SALI and hepatocyte ferroptosis, and the underlying mechanisms, remain unknown.

In this study, we sought to characterize the dynamic trajectory of hepatocyte ferroptosis in a murine model of sepsis and to elucidate its association with the severity of liver injury. Furthermore, we aimed to investigate the regulatory effects and underlying molecular mechanisms of HUMSCs-Exo on sepsis-induced hepatocyte ferroptosis and hepatic damage, both in vitro and in vivo, thereby providing novel mechanistic insights and translational perspectives for the treatment of SALI.

Experimental Flowchart

Methods and materials

Culture, isolation, and characterization of HUMSCs and their exosomes

HUMSCs and their exosomes were kindly provided by the National Human Genome Center (Northern), in collaboration with Hesheng Nuoshai Biopharmaceutical Technology Co., Ltd. (Beijing, China). Briefly, HUMSCs were cultured in a chemically defined medium supplemented with exosomes-depleted fetal bovine serum for 24 h. The medium was then replaced with fresh complete medium, and conditioned medium was collected after an additional 48 h. The supernatant was sequentially centrifuged at 300 g for 10 min, 2,000 g for 20 min, and 10,000 g for 30 min at 4 °C to remove cells and debris. Exosomes were subsequently pelleted by ultracentrifugation at 100,000 g for 70–90 min at 4 °C (Ultra-high speed freezing centrifuge) and stored at −80 °C until further use (Fig. S1A). As previously described [25], transmission electron microscopy (TEM) was employed to visualize the morphology and size of the isolated vesicles. NanoSight nanoparticle tracking analysis (NTA) was used to determine their particle size distribution and concentration. Western blotting (WB) was carried out to confirm the presence of exosomes-specific markers, including TSG101, CD63, and CD9, and the absence of the negative marker Calnexin. HUMSCs were derived from umbilical cord tissue obtained from healthy donors with written informed consent for perinatal tissue donation (Supplementary Material 2).

Animal housing and establishment of the murine model

Specific pathogen-free (SPF) male wild-type (WT) C57BL/6J mice (8 weeks old, 20–25 g; SiPeiFu, Beijing, China) were housed under standardized conditions (12/12-h light/dark cycle, temperature 24 ± 2 °C, humidity 60 ± 5%) with ad libitum access to pellet food and water. Animal grouping and experimental procedures are illustrated in the experimental workflow. Briefly, in the phase one, we initially conducted transcriptomic sequencing on hepatocytes from the control group and the CLP-24 h group. Subsequently, the mice were randomly assigned to the control group, CLP-6 h, CLP-12 h, CLP-24 h, and CLP-48 h groups, with samples collected at the corresponding time points for subsequent analysis and detection. In the phase two, animals were randomly allocated to four groups: Control-6 h, Control-24 h, CLP-6 h, and CLP-24 h, with all mice receiving a tail-vein injection of 1 × 10⁶ DiR-labelled HUMSCs. In the phase three, mice were randomly assigned to a control group, a CLP group, and an HUMSCs-Exo-treated group. Mice in the HUMSCs-Exo-treated group received a single tail-vein injection of HUMSCs-Exo at a dose of 3 × 10⁹ vesicles per mouse, suspended in 150 µl phosphate-buffered saline (PBS), whereas mice in the CLP group received an equal volume of PBS via tail-vein injection. All animal experiments were conducted in accordance with the principles of the Declaration of Helsinki, approved by the Animal Care and Use Committee of the General Hospital of the People’s Liberation Army (EAC-IACUC-2025F21-027), and performed in compliance with the ARRIVE guidelines. (Supplementary Material 2)

The animal model was established as follows: (1) Cecal Ligation and Puncture (CLP) Model [26]: Briefly, mice were fasted for 12 h prior to surgery with free access to water. Anesthesia was induced with 2.5% tribromoethanol. Following fixation of the limbs, a 1.0 cm midline laparotomy was performed inferior to the xiphoid process to expose the abdominal cavity and cecum. The cecum was ligated approximately 1.0 cm from its distal end using 3 − 0 silk suture, then punctured once through-and-through with a 21G needle, and a small amount of fecal content was gently extruded. The abdominal wall was subsequently closed in two layers. Postoperatively, mice received a subcutaneous injection of 1 mL pre-warmed (37 °C) sterile saline and were kept under thermal support with continuous monitoring. (2) DiR-labelled HUMSCs tail-vein injection experiment [27]: Briefly, mice in the CLP group received a single tail-vein injection of DiR-HUMSCs at 2 h after CLP surgery, at a dose of 1 × 10⁶ cells per mouse. Mice in the control group were administered an equivalent dose of DiR-HUMSCs via tail vein injection at the same time point. (3) HUMSCs-Exo tail-vein injection experiment [28]: Briefly, mice in the HUMSCs-Exo-treated group received a single tail-vein injection of HUMSCs-Exo at 2 h after CLP surgery, at a dose of 3 × 10⁹ vesicles per mouse.

Cell culture and establishment of in vitro models

The NCTC1469 murine normal hepatocyte cell line [29] and the THLE-2 immortalized normal human hepatocyte cell line [30] (EallBio Biomedical Technology Co., Ltd., Beijing, China) were used in this study. NCTC1469 cells were cultured in specialized NCTC1469 medium (Cat. No. 03.20272), and THLE-2 cells in dedicated THLE-2 medium (Cat. No. 03.202306), both obtained from EallBio. Cells were maintained in a humidified incubator at 37 °C with 5% CO₂. Cell grouping and treatment protocols are thoroughly presented in the experimental workflow. Lipopolysaccharide (LPS) (L2880, Merck, Darmstadt, Germany), Erastin (HY-15763, MCE, New Jersey, USA), Deferoxamine (HY-B1625, MCE, New Jersey, USA) and HUMSCs-Exo [31] were utilized to induce and modulate ferroptosis-related responses. All cell lines were authenticated via short tandem repeat (STR) profiling and confirmed to be free of mycoplasma contamination (Supplementary Material 2).

Histopathological examination

Histological analyses included hematoxylin and eosin (H&E) staining (GP1031, Servicebio, Wuhan, China) and enhanced Prussian blue staining with DBA intensification (GP1069, Servicebio, Wuhan, China). Briefly, liver tissues were harvested and fixed in 4% paraformaldehyde, followed by routine paraffin embedding, sectioning, staining, dehydration, and mounting. Tissue sections were examined using the CaseViewer platform (Servicebio, Wuhan, China) to assess hepatic injury, inflammatory infiltration, and iron accumulation. As previously described [32], liver injury was graded on a 0–4 scale based on the severity of parenchymal necrosis (Supplementary Material 1).

Biochemical analysis of liver function

Serum biomarkers for liver function included lactate dehydrogenase (LDH), aspartate aminotransferase (AST), alanine aminotransferase (ALT), total bilirubin (TBIL), direct bilirubin (DBIL), albumin (ALB), alkaline phosphatase (ALP), gamma-glutamyl transferase (γ-GT), and total bile acids (TBA). At designated time points, serum and/or cell culture supernatants were collected from each group and analyzed using an automated biochemical analyzer (Jiancheng Bioengineering, Nanjing, China).

Oxidative stress assessment

Oxidative stress markers included malondialdehyde (MDA) (G4300/G4302, Servicebio, Wuhan, China), ferrous iron (Fe²⁺) (G4323, Servicebio, Wuhan, China), total iron (G4301, Servicebio, Wuhan, China), reduced glutathione (GSH) (G4305, Servicebio, Wuhan, China), and oxidized glutathione (GSSG) (G4304, Servicebio, Wuhan, China). All assays were conducted strictly in accordance with the manufacturers’ instructions. Briefly, weighed liver tissues or cultured cells were homogenized in an appropriate volume of extraction buffer, followed by centrifugation to collect the supernatant, which was kept on ice prior to analysis. Reaction reagents were added to standards and samples, and absorbance was measured in 96-well plates at specific wavelengths using a microplate reader. Concentrations were calculated based on standard curves according to the provided formulas.

Western blotting

The following primary antibodies were used: TFRC Mouse Monoclonal Antibody (1:1000; #46222, CST, MA, USA), ACSL4 Rabbit Polyclonal Antibody (1:1000; 22401-1-AP, Proteintech, Wuhan, China), xCT Rabbit Monoclonal Antibody (1:1000; #98051, CST, MA, USA), GPX4 Rabbit Monoclonal Antibody (1:1000; #52455, CST, MA, USA), FTH1 Rabbit Monoclonal Antibody (1:1000; #3998, CST, MA, USA), FPN1 Rabbit Polyclonal Antibody (1:500; A01953-2, Boster, Wuhan, China), and β-actin Mouse Monoclonal Antibody (1:10000; HX1827, Huaxingbio, Beijing, China). Secondary antibodies included AMCA-conjugated Goat Anti-Mouse IgG (H + L) (1:5000; HX2109, Huaxingbio, Beijing, China) and AMCA-conjugated Goat Anti-Rabbit IgG (H + L) (1:5000; HX2108, Huaxingbio, Beijing, China). Detailed protocols are provided in Supplementary Material 1.

ELISA analysis

Enzyme-linked immunosorbent assays (ELISAs) were employed to determine levels of inflammatory cytokines in murine liver tissues and/or serum, including interleukin-6 (IL-6; 88-7064-88, Thermo Scientific, MA, USA), interleukin-1β (IL-1β; 88-7013 A-88, Thermo Scientific, MA, USA), and tumor necrosis factor-α (TNF-α; 88-7324-88, Thermo Scientific, MA, USA). All procedures were conducted in strict accordance with the manufacturers’ protocols. Briefly, standards and appropriately diluted samples were added to the ELISA plates, followed by successive blocking, incubation, and washing steps. Absorbance was measured in 96-well plates using a microplate reader, and cytokine concentrations were calculated based on standard curves.

Confocal laser scanning microscopy

Intracellular lipid peroxidation, labile ferrous iron levels and reactive oxygen species (ROS) were evaluated using BODIPY^™ 581/591 C11 (1:1000; D3861, Invitrogen, MA, USA), FerroOrange (1:1000; F374, Dojindo, Japan) and Reactive Oxygen Species Assay Kit (1:1000; CA1410, Solarbio, Beijing, China) respectively. Adherent cells were incubated with either probe in complete medium at 37 °C for 20 min in the dark. Hoechst 33,342 nuclear stain (1:200; #639, ImmunoChemistry Technologies, LLC, CA, USA) was then added for an additional 10 min of incubation. After staining, cells were washed three times with 1× PBS. Images were captured using a Leica TCS SP8 confocal laser scanning microscope (Leica Microsystems, Germany) under appropriate excitation/emission settings.

Statistical analysis

All statistical analyses and graphical presentations were conducted using GraphPad Prism version 6.0 (GraphPad Software, La Jolla, CA, USA). Each experiment was independently repeated at least three times. Data were first tested for normality. For comparisons involving three or more groups, one-way or two-way analysis of variance (ANOVA) was employed as appropriate. Pearson correlation analysis was used to evaluate linear relationships between variables. Data are presented as mean ± standard error of the mean (SEM), and a P value < 0.05 was considered statistically significant.

(Additional experimental details are provided in Supplementary Material 1.)

Results

Significant alterations in ferroptosis-, iron metabolism-, and lipid metabolism–related gene expression in hepatocytes from septic mice

First, we initiated RNA sequencing of hepatocytes isolated from septic and control mice to identify differentially expressed genes (DEGs) associated with sepsis and to evaluate potential alterations in relevant signaling pathways. Sample correlation analysis, principal component analysis (PCA), and heatmap visualization (Fig. 1A and B) demonstrated strong positive correlations among samples within each group, indicating high biological reproducibility, while overall gene expression patterns were broadly comparable across experimental groups. Subsequent analysis using the DESeq2 pipeline with Benjamini-Hochberg multiple-testing correction identified a total of 2,355 significantly DEGs (Fig. S2A), including 1,288 upregulated and 1,067 downregulated genes. Volcano plots (Fig. 1C) and MA plots (Fig. 1D) illustrated the statistical significance and expression magnitude of these genes, respectively. Notably, compared with controls, hepatocytes from septic mice exhibited upregulation of the ferroptosis-related gene TFRC and PTGS2, and downregulation of SLC40A1 and GCLC. Reactome pathway annotation (Fig. S2B), GO level 2 annotation and enrichment analyses (Fig. 1E) revealed that these DEGs were predominantly involved in immune system process, metabolic process, and catalytic activity, with significant enrichment in pathways related to the negative regulation of lipid metabolism, iron ion transmembrane transport, iron ion transport, and reactive oxygen species metabolism. Furthermore, KEGG level 2 annotation and enrichment analyses (Fig. 1F) indicated that these genes play key roles in lipid metabolism, cell death, and infectious disease, with ferroptosis emerging as a significantly enriched pathway (adjusted P < 0.05)

Fig. 1.

Transcriptomic profiling of hepatocytes from septic and control mice (each n = 4). A. Correlation matrix and PCA demonstrated high intra-group consistency and clear separation between the CLP and control groups. B. Heatmap illustrated global gene expression patterns across samples in both groups. C and D. Volcano plot and MA plot compared CLP versus control groups, highlighting four ferroptosis-related genes with significant differential expression: upregulated genes (TFRC, PTGS2) and downregulated genes (SLC40A1, GCLC). E. GO annotations analysis level 2 and GO enrichment analysis showed significant enrichment in pathways between the CLP and control groups. F. KEGG annotations analysis level 2 and KEGG enrichment analysis indicated involvement in several processes between the CLP and control groups. All enrichment results showed statistical significance with Padjust < 0.05

Temporal dynamics of hepatocellular ferroptosis and characteristics of liver injury in septic mice

To delineate the dynamic trajectory of hepatocyte ferroptosis and characterize hepatic injury in septic mice, hepatocytes, liver tissues, and serum samples were collected at 6 h, 12 h, 24 h, and 48 h after CLP induction for comprehensive analyses (Fig. 2A). Serum biochemical assays (Fig. 2B) revealed that AST and ALT levels peaked between 12 h and 24 h post-surgery, followed by a partial recovery of liver function indices at 48 h. H&E staining, DBA-enhanced Prussian blue staining, and quantitative liver histological scoring (Fig. 2C; Fig. S2C) demonstrated markedly exacerbated inflammatory infiltration in septic livers compared with controls, accompanied by hepatocellular swelling and necrosis, disorganization of hepatic cords, bile duct dilatation, and focal haemorrhage. Notably, aberrant iron accumulation was evident within hepatic tissues, with the most severe tissue injury and iron deposition observed at 24 h post-CLP. Subsequent assessment of inflammatory mediators and oxidative stress markers in liver tissues (Fig. 2D) showed progressive increases in MDA, Fe²⁺, and the Fe²⁺/total iron ratio from 6 h to 24 h after CLP, concomitant with a time-dependent decline in GSH levels and the GSH/GSSG ratio, with the most pronounced alterations at 24 h (P < 0.0001). These parameters exhibited partial normalization by 48 h. In parallel, hepatic expression of the pro-inflammatory cytokines IL-1β, TNF-α, and IL-6 was significantly elevated in septic mice (P < 0.0001). Following liver perfusion, hepatocyte isolation, and protein extraction (Fig. S2D), WB analysis (Fig. 2F) revealed that, relative to controls, septic hepatocytes exhibited upregulation of the ferroptosis-promoting proteins TFRC and ACSL4, alongside downregulation of the ferroptosis-suppressive proteins FPN, xCT, FTH1, and GPX4. Across the septic time-course groups, TFRC and ACSL4 expression increased in a time-dependent manner from 6 h to 24 h post-CLP, whereas FPN, xCT, FTH1, and GPX4 displayed a corresponding time-dependent decrease. Finally, Pearson correlation analyses between serum AST and ALT levels and hepatic Fe²⁺ and MDA concentrations (Fig. 2E; Fig. S2E) demonstrated a strong positive association between liver injury and ferroptosis-related indices. Specifically, AST correlated with Fe²⁺ and MDA with r values of 0.7018 (P < 0.0001) and 0.7993 (P < 0.0001), respectively, while ALT correlated with Fe²⁺ and MDA with r values of 0.7515 (P < 0.0001) and 0.7345 (P < 0.0001), respectively. Concomitantly, at 2 h following CLP, septic mice received a single intraperitoneal administration (100 µl) of Z-DEVD-FMK (5 mg kg⁻¹), Necrosulfonamide (20 mg kg⁻¹), Disulfiram (50 mg kg⁻¹), or Ferrostatin-1 (10 mg kg⁻¹) (Fig. S1B). Serum samples and primary hepatocytes were harvested 24 h post-surgery to further delineate the contribution of hepatocyte ferroptosis to SALI. As illustrated in Fig. S1C and Fig. S1D, hepatocyte viability was increased across all inhibitor-treated groups compared with the septic cohort, accompanied by partial restoration of hepatic biological function. Notably, relative to the other inhibitor groups, the ferroptosis-specific inhibitor conferred a markedly greater improvement in hepatocyte survival (P < 0.01) and substantially attenuated sepsis-induced hepatic injury (P < 0.0001).

Fig. 2.

Temporal dynamics of hepatocellular ferroptosis and characteristics of liver injury in septic mice. A. Animal experimental model diagram. WT male C57BL/6J mice were randomly divided into sham-operated controls and septic groups. Septic mice were euthanized at 6 h, 12 h, 24 h, and 48 h post-CLP for the collection of serum, liver tissue, and hepatocytes for downstream analyses. B. Serum AST and ALT levels in control and septic mice (each n = 12). C. Histological assessment of liver injury in control and septic mice via H&E staining (scale bar = 50 μm), DBA-enhanced Prussian blue staining (scale bar = 20 μm). D. Quantification of oxidative stress and inflammatory markers in liver tissues of control and septic mice (each n = 6). E. Pearson correlation analyses between serum AST and ALT levels and hepatic Fe²⁺ concentrations of septic mice (24 h post -CLP) (each n = 42). F. Expression of ferroptosis-related proteins in hepatocytes from control and septic groups (each n = 8). All data were tested for normal distribution and analyzed using one-way ANOVA. Results are presented as mean ± standard error of the mean (SEM); error bars = SEM. Significant differences between groups are indicated as **** P < 0.0001, *** P < 0.001, ** P < 0.01, * P < 0.05; ns = not significant

Dose- and time-dependent characteristics of LPS-induced ferroptosis in NCTC1469 cells

An in vitro sepsis model was established using the murine normal hepatocyte cell line NCTC1469 (Fig. 3A) to delineate the time- and dose-dependent characteristics of ferroptosis under varying LPS concentrations and exposure durations. First, a two-way analysis of variance was utilized to evaluate the interactive effects of LPS concentration and exposure time on hepatocellular injury. As shown in Fig. 3B, LPS-induced release of AST and ALT exhibited a pronounced time-dependent pattern, with progressive aggravation of cellular injury over time. Among the experimental groups, hepatocellular damage was most pronounced following 24 h of stimulation with 250 ng/mL LPS. To define the dose-response relationship of LPS-induced ferroptosis, NCTC1469 cells were exposed to increasing concentrations of LPS (0, 250 ng, 500 ng, 1 µg, and 2 µg/mL) for 24 h, after which samples were collected for analysis. As shown in Fig. 3C and E, compared with controls, all LPS-treated groups exhibited a marked reduction in cell viability, accompanied by significant increases in intracellular MDA, lipid peroxides, and Fe²⁺ levels, along with a concomitant depletion of GSH. In parallel, the expression of the ferroptosis-promoting proteins ACSL4 and TFRC was significantly upregulated. Moreover, with increasing LPS concentrations, ferroptosis-promoting factors progressively increased whereas ferroptosis-suppressive factors declined, indicating a clear dose-dependent induction of ferroptosis. To further delineate the time-response relationship, cells were exposed to 250 ng/mL LPS and harvested at 6 h, 12 h, 24 h, and 48 h. As illustrated in Fig. 3D and F, from 6 h to 24 h of stimulation, cell viability progressively decreased, while intracellular MDA, lipid peroxide, and Fe²⁺ levels steadily increased and GSH levels declined markedly. Consistently, ACSL4 and TFRC expression was significantly elevated over this interval. Notably, these ferroptosis-associated alterations were partially attenuated at 48 h of LPS exposure.

Fig. 3.

Time- and dose-dependent characteristics of LPS-induced ferroptosis in NCTC1469 mouse hepatocytes. A. Cell experimental model diagram. The NCTC1469 murine hepatocyte cell line was used to evaluate ferroptosis under LPS stimulation. Cells were divided into unstimulated controls and LPS-treated groups. LPS concentrations included 0 ng, 250 ng, 500 ng, 1 µg, and 2 µg (/mL), with stimulation durations of 6 h, 12 h, 24 h, and 48 h. Supernatants and cells from each group were collected for subsequent analysis. B. AST and ALT release in control and LPS-treated cells (each n = 6), and the interaction effects of LPS concentration and exposure time on AST and ALT secretion. C. Dose-dependent relationship of LPS-induced ferroptosis in NCTC1469 cells following 24 h LPS exposure. Assessed endpoints included cell viability, MDA, Fe²⁺, and GSH levels (each n = 6), as well as lipid peroxidation and ferrous iron accumulation visualized by laser confocal microscopy (scale bar = 50 μm). D. Time-dependent of LPS-induced ferroptosis (LPS concentration: 250 ng/mL), with measurements of cell viability, MDA, Fe²⁺, and GSH (each n = 6), as well as lipid peroxidation and ferrous iron accumulation visualized by laser confocal microscopy (scale bar = 50 μm). E. Dose-dependent expression of ferroptosis-related protein in NCTC1469 cells following 24 h LPS exposure (each n = 6). F. Time-dependent expression of ferroptosis-related proteins in response to 250 ng/mL LPS stimulation (each n = 6). All data were tested for normality and analyzed using one-way or two-way ANOVA. Results are presented as mean ± SEM; error bars = SEM. Significant differences between groups are indicated as **** P < 0.0001, *** P < 0.001, ** P < 0.01, * P < 0.05; ns = not significant

HUMSCs-Exo attenuates hepatocellular ferroptosis and confers hepatoprotection in sepsis

First, DiR-HUMSCs were administered to control and septic mice via tail-vein injection (Fig. 4A) to characterize their spatiotemporal biodistribution in vivo. As shown in Fig. 4B and C, in vivo fluorescence imaging revealed that at 6 h post-surgery, only a limited accumulation of HUMSCs was detected in the livers of septic mice compared with controls. In contrast, at 24 h post-surgery, hepatic localization of HUMSCs in septic mice was markedly increased compared with that observed at 6 h postoperatively (P < 0.0001), accompanied by a significantly higher mean fluorescence intensity per unit weight in excised liver tissues relative to controls (P < 0.0001). Consistent with these observations, fluorescence microscopy of frozen liver sections (Fig. 4D and E) demonstrated substantial accumulation of DiR-HUMSCs within the hepatic parenchyma of septic mice at 24 h, whereas less signal was observed in control animals. Subsequently, HUMSCs-Exo were isolated by differential ultracentrifugation to further investigate their hepatoprotective effects in sepsis and to explore the underlying mechanisms.

Fig. 4.

Spatiotemporal distribution of DiR-HUMSCs in septic and control mice. A. Animal experimental model diagram. Septic and control mice received equivalent doses of DiR-HUMSCs (1 × 10⁶ cells per mouse) via tail vein injection at 2 h post-surgery. In vivo imaging was performed at 6 h and 24 h post-injection to visualize systemic and liver-localized fluorescence. B and C. At 6 h and 24 h post-injection, in vivo imaging and average weight fluorescence intensity of liver tissue in vitro of control group and CLP group (each n = 5). D and E. At 6 h and 24 h post-injection, fluorescence imaging results of frozen liver tissue slices from control group and model mice CLP group (scale bars = 500 μm, 100 μm, and 20 μm) (each n = 3). All data were tested for normality and analyzed using two-way ANOVA. Results are presented as mean ± SEM; error bars = SEM. Significant differences between groups are indicated as **** P < 0.0001, *** P < 0.001, ** P < 0.01, * P < 0.05; ns = not significant

Characterization of HUMSCs-Exo is presented in Fig. S2F. WB confirmed the presence of canonical exosomal markers CD9, CD63, and TSG101, and the absence of the negative marker Calnexin. TEM revealed classic saucer-shaped vesicles with intact membranes and no aggregation. NTA determined an average vesicle diameter of 72.1 nm with a particle concentration of 1.16 × 10¹¹ particles/mL. We subsequently conducted both in vivo and in vitro experiments to investigate the impact of HUMSCs-Exo on hepatocellular ferroptosis and SALI.

In the in vitro experiments (Fig. 5A), cells were assigned to a control group, an LPS-treated group, an LPS + Erastin (ferroptosis inducer) group, an LPS + DFO (iron chelator) group, an LPS + HUMSCs-Exo (Exo) group, an LPS + Erastin + Exo group, and an LPS + DFO + Exo group. As shown in Fig. 5B-D, LPS exposure markedly reduced cell viability and substantially increased the release of AST and ALT, while intracellular levels of MDA, Fe²⁺, and ROS were significantly elevated and GSH levels were markedly depleted. Treatment with HUMSCs-Exo or DFO effectively reversed these alterations, significantly limiting aberrant intracellular accumulation of lipid peroxides, labile iron and ROS level, alleviating cellular injury, and restoring cell viability. Moreover, WB analysis (Fig. 5E) demonstrated that, similar to Erastin stimulation, LPS exposure induced pronounced dysregulation of ferroptosis-related protein expression. In contrast, treatment with HUMSCs-Exo or DFO significantly downregulated the expression of ACSL4 and TFRC (P < 0.0001) while upregulating FPN, xCT, GPX4, and FTH1 (P < 0.0001), effectively attenuating hepatocyte ferroptosis.

Fig. 5.

Protective effects of HUMSCs-Exo on LPS-induced ferroptosis and cellular injury in NCTC1469 mouse hepatocytes. A. Cell experimental model diagram. Cells were divided into sevsen groups: unstimulated control group, LPS-stimulated (250 ng/mL) group, LPS + Erastin (5 µM)-stimulated group, LPS + DFO (5 µM) group, LPS + HUMSCs-Exo (Exo) group, LPS + Erastin + Exo group, LPS + DFO + Exo group. As for HUMSCs-Exo, 1 × 10⁸ vesicles/mL was administered once at 2 h post-LPS stimulation; in the DFO group, a single dose of DFO (5 µM) was administered at the same time point. Supernatants and cells were collected 24 h after stimulation for analysis. B. AST and ALT levels in culture supernatants from each group (each n = 6). C. Measurements of intracellular MDA, Fe²⁺, and GSH levels, as well as cell viability in each group (each n = 6). D. Lipid peroxidation, ferrous iron accumulation and ROS visualized by laser confocal microscopy (scale bar = 50 μm). E. Expression levels of ferroptosis-related proteins in each group (each n = 6). All data were tested for normality and analyzed using one-way ANOVA. Results are presented as mean ± SEM; error bars = SEM. Significant differences between groups are indicated as **** P < 0.0001, *** P < 0.001, ** P < 0.01, * P < 0.05; ns = not significant

In the in vivo experiments, mice in the HUMSCs-Exo-treated (Exo) group received a single tail-vein injection of HUMSCs-Exo (3 × 10⁹ vesicles per mouse) at 2 h after CLP surgery, and samples were collected at 24 h postoperatively for subsequent analyses (Fig. 6A). As shown in Fig. 6B, exosomes treatment markedly improved survival outcomes, with 7-day survival rates of 33.33% and 80% in the CLP and Exo-treated groups, respectively (P = 0.0175), and 12-day survival rates of 26.7% and 80%, respectively (P = 0.0039). Histopathological staining, liver injury scoring, and serum biochemical analyses (Fig. 6C and Fig. S2G) demonstrated that, compared with septic mice, Exo-treated mice exhibited substantially attenuated hepatocellular necrosis, reduced inflammatory infiltration, diminished haemorrhage, and decreased iron deposition within hepatic tissues, accompanied by a pronounced suppression of the exaggerated inflammatory response induced by sepsis. Consistently, serum biochemical profiles (Fig. 6D) revealed that exosomes administration significantly reduced the release of ALT, AST, TBIL, DBIL, ALP, TBA, and γ-GT, while ALB synthesis (P < 0.0001). These findings indicate a broad restoration of hepatic biosynthetic capacity, as well as hepatobiliary secretion and transport functions. Furthermore, exosomes treatment (Fig. 6E) significantly decreased hepatic levels of MDA, Fe²⁺, and the Fe²⁺/total iron ratio (P < 0.0001), while markedly increasing GSH content and the GSH/GSSG ratio (P < 0.0001). In parallel (Fig. 6E), hepatic concentrations of the pro-inflammatory cytokines TNF-α, IL-1β, and IL-6 were significantly reduced (P < 0.0001). WB analysis (Fig. 6F) further confirmed that, relative to septic controls, exosomes treatment downregulated the expression of the ferroptosis-promoting proteins ACSL4 and TFRC while upregulating FPN, xCT, GPX4, and FTH1, thereby substantially attenuating sepsis-induced hepatocyte ferroptosis.

Fig. 6.

In vivo protective effects of HUMSCs-Exo against hepatocellular ferroptosis and liver injury in septic mice. A. Animal experimental model diagram. WT male C57BL/6J mice were randomly divided into three groups: sham-operated (control), sepsis (CLP), and HUMSCs-Exo-treated (Exo). In the Exo-treated group, mice received a single intravenous injection of HUMSCs-Exo (3 × 10⁹ vesicles/mouse, dissolved in 150 µL PBS) via the tail vein 2 h post-surgery. The sepsis group received an equal volume of PBS (150 µL) at the same time point. Serum, liver tissues, and isolated hepatocytes were collected 24 h post-procedure for subsequent analyses. B. Kaplan–Meier survival analysis of mice in sham-operated control (n = 10), CLP-induced sepsis (n = 15), and HUMSCs-Exo treatment (n = 15) groups. C. Representative H&E staining (scale bar = 50 μm) and enhanced Prussian blue staining with DBA enhancement (scale bar = 20 μm) of liver tissues from each group. D. Comprehensive hepatic function panel (eight biochemical indicators) in serum samples from each group (each n = 6). E. Quantification of hepatic oxidative stress and inflammatory markers (each n = 6), including Fe²⁺, Fe²⁺ of total Fe, GSH, GSH/GSSG, MDA, IL-1β, IL-6, and TNF-α. F. Immunoblotting for ferroptosis-related protein expression in hepatocytes isolated from each group (each n = 6). All data were assessed for normality and analyzed using one-way ANOVA. Results are expressed as mean ± SEM; error bars = SEM. Significant differences between groups are denoted as follows: **** P < 0.0001, *** P < 0.001, ** P < 0.01, * P < 0.05; ns = not significant.

Regulatory effects of HUMSCs-Exo on ferroptosis and injury in THLE-2 human hepatocytes

Subsequently, an in vitro sepsis model was established using the immortalized human normal hepatocyte cell line THLE-2 to further validate the protective effects of HUMSCs-Exo against LPS-induced ferroptosis and cellular injury in human hepatocytes. First, we characterized the time- and dose-dependent features of ferroptosis in THLE-2 cells under varying LPS concentrations and exposure durations (Fig. 7A), together with associated patterns of cellular injury. As shown in Fig. 7B, AST and ALT release increased progressively over time across all LPS-treated groups, with markedly elevated levels observed in the 500 ng/mL LPS group at 24 h compared with controls (P < 0.0001). Dose-response analyses (Fig. 7C) demonstrated that after 24 h of exposure to increasing LPS concentrations, cell viability was significantly reduced, accompanied by pronounced depletion of intracellular GSH and aberrant, dose-dependent increases in MDA, lipid peroxides, and Fe²⁺ levels. To delineate the time–response relationship, THLE-2 cells were exposed to 500 ng/mL LPS and harvested at sequential time points. As illustrated in Fig. 7D, cell viability declined progressively with prolonged exposure, whereas ferroptotic activity increased in a clear time-dependent manner. Pearson correlation analyses (Fig. 8A) revealed a strong positive association between cellular injury and ferroptosis, with correlation coefficients consistently exceeding 0.700 (P < 0.0001). On this basis, we further evaluated the protective effects of HUMSCs-Exo on ferroptosis and cell injury in THLE-2 cells (Fig. 8B). The experimental results, as shown in Fig. 8C and F, demonstrate that, compared to the LPS or Erastin stimulation groups, treatment with HUMSCs-Exo or DFO significantly enhanced hepatocyte viability, reduced AST and ALT release, and markedly decreased intracellular levels of MDA, lipid peroxides, Fe²⁺, and ROS. Concurrently, GSH levels were significantly increased. Furthermore, in the Exo-treated group, the expression of ACSL4 and TFRC in hepatocytes was significantly downregulated (P < 0.0001), while the expression levels of FPN, xCT, GPX4, and FTH1 were significantly upregulated (P < 0.0001), leading to a marked reduction in hepatocyte ferroptosis.

Fig. 7.

Time- and dose-dependent characteristics of LPS-induced ferroptosis in THLE-2 human hepatocytes. A. Cell experimental model diagram. The THLE-2 human hepatocyte cell line was used to evaluate ferroptosis under LPS stimulation. Cells were divided into unstimulated controls and LPS-treated groups. LPS concentrations included 0 ng, 125 ng, 250 ng, 500 ng, and 1 µg (/mL), with stimulation durations of 6 h, 12 h, 24 h, and 48 h. Supernatants and cells from each group were collected for subsequent analysis. B. AST and ALT release in control and LPS-treated cells (each n = 6), and the interaction effects of LPS concentration and exposure time on AST and ALT secretion. C. Dose-dependent relationship of LPS-induced ferroptosis in THLE-2 cells following 24 h LPS exposure. Assessed endpoints included AST, ALT, cell viability, MDA, Fe²⁺, and GSH levels (each n = 6), as well as lipid peroxidation and ferrous iron accumulation visualized by laser confocal microscopy (scale bar = 50 μm). D. Time-dependent of LPS-induced ferroptosis (LPS concentration: 500 ng/mL), with measurements of AST, ALT, cell viability, MDA, Fe²⁺, and GSH (each n = 6), as well as lipid peroxidation and ferrous iron accumulation visualized by laser confocal microscopy (scale bar = 50 μm). All data were tested for normal distribution and analyzed using one-way ANOVA. Results are presented as mean ± SEM; error bars = SEM. Statistical significance between groups is indicated as follows: **** P < 0.0001, *** P < 0.001, ** P < 0.01, * P < 0.05; ns = not significant

Fig. 8.

Protective effects of HUMSCs-Exo on LPS-induced ferroptosis and cellular injury in THLE-2 human hepatocytes. A. Pearson correlation analyses between AST and ALT levels and Fe²⁺ and MDA concentrations of THLE-2 human hepatocyte cell line (500 ng/mL LPS; Cultivate for 24 hours) (each n = 60). B. Cell experimental model diagram. Cells were divided into sevsen groups: unstimulated control group, LPS-stimulated (500 ng/mL) group, LPS + Erastin (5 μM)-stimulated group, LPS + DFO (5 μM) group, LPS + HUMSCs-Exo (Exo) group, LPS + Erastin + Exo group, LPS + DFO + Exo group. As for HUMSCs-Exo, 1 × 10⁸ vesicles/mL was administered once at 2 hours post-LPS stimulation; in the DFO group, a single dose of DFO (5 μM) was administered at the same time point. Supernatants and cells were collected 24 hours after stimulation for analysis. C. AST and ALT levels in culture supernatants from each group (each n = 6). D. Lipid peroxidation, ferrous iron accumulation and ROS visualized by laser confocal microscopy (scale bar = 50 µm). E. Measurements of cell viability, intracellular MDA, Fe²⁺, and GSH levels in each group (each n = 6). F. Expression levels of ferroptosis-related proteins in each group (each n = 6). All data were tested for normal distribution and analyzed using one-way ANOVA. Results are presented as mean ± SEM; error bars = SEM. Statistical significance between groups is indicated as follows: **** P < 0.0001, *** P < 0.001, ** P < 0.01, * P < 0.05; ns = not significant

Discussion

SALI represents one of the earliest and most prevalent forms of organ dysfunction during sepsis and is closely linked to host survival and prognosis [5]. Nevertheless, the characteristic patterns of hepatocellular and hepatic injury during sepsis remain incompletely defined, and no specific therapeutic strategies targeting SALI are currently available. In the present study, we demonstrated that DEGs between control and septic mouse hepatocytes are predominantly involved in iron metabolism, lipid metabolism, and reactive oxygen species metabolism, with significant enrichment in ferroptosis-related signaling pathways. These findings are consistent with previous reports by Liu et al. [10] and Muñiz-Santos et al. [11], and further suggest a potential mechanistic link between hepatocyte ferroptosis and SALI. We subsequently characterized the dynamic progression of hepatocyte ferroptosis and the associated hepatic injury in septic mice. Our results showed that within 24 h after sepsis induction, liver injury and local inflammatory responses progressively worsened. During this period, hepatic levels of MDA, Fe²⁺, the Fe²⁺/total iron ratio, GSH, and the GSH/GSSG ratio exhibited pronounced alterations. Concurrently, the expression of ferroptosis-promoting proteins ACSL4 and TFRC was markedly upregulated, whereas ferroptosis-suppressive proteins, including FPN, xCT, GPX4, and FTH1, displayed a progressive decline. Collectively, these findings indicate a time-dependent intensification of hepatocyte ferroptosis in septic mice within the first 24 h, with a peak at approximately 24 h post-insult. We next interrogated the relationship between hepatocellular ferroptosis and the severity of hepatic injury in septic mice. On the one hand, Pearson correlation analysis demonstrated that hepatocyte-derived AST and ALT release was significantly and positively correlated with both MDA accumulation and intracellular Fe²⁺ levels. On the other hand, comparative pharmacological inhibition using multiple cell death-specific inhibitors revealed that, relative to other targeted inhibitors, the ferroptosis-specific inhibitor Ferrostatin-1 exerted a more pronounced protective effect against hepatic injury and hepatocellular death in septic mice. Collectively, these findings implicate that targeting hepatocyte ferroptosis may represent a novel and promising therapeutic strategy for SALI. In vitro, sepsis models established using the murine normal hepatocyte cell line NCTC1469 and the immortalized human normal hepatocyte cell line THLE-2 showed that LPS stimulation induced ferroptosis in both cell types in a time- and dose-dependent manner. Taken together, both in vivo and in vitro evidence supports the notion that sepsis induces hepatocyte ferroptosis and that targeting this form of regulated cell death may offer a promising avenue for the treatment of SALI. However, several limitations warrant consideration. Although we performed correlation analyses linking hepatocyte ferroptosis with hepatic injury in septic mice and with cellular injury in LPS-treated THLE-2 cells, our study did not employ graded concentrations of canonical ferroptosis inducers (such as Erastin or RSL3) to establish dose–response relationships. Moreover, we lacked correlative analyses between ferroptosis-related markers and liver injury in patients with clinical sepsis, which would more robustly substantiate the association—and potentially the causality—between sepsis-induced hepatocyte ferroptosis and hepatic injury. Addressing these gaps constitutes a key objective of our future work. In addition, although we observed partial attenuation of liver injury and reduced ferroptosis levels at 48 h post-surgery in septic mice, the underlying mechanisms remain unclear and may be related to the onset of immunosuppression or the engagement of alternative programmed cell death pathways.

Preclinical studies have shown that HUMSCs-Exo can suppress ferroptosis in target cells across diverse disease models, including inflammatory bowel disease [21], diabetes [22], primary ovarian insufficiency [23] and viral myocarditis [24], thereby mitigating tissue injury and organ dysfunction and extending host survival. However, whether HUMSCs-Exo modulate SALI and hepatocyte ferroptosis, and the mechanisms underlying such regulation, remain unknown. In our previous work, following tail-vein administration of DiR-HUMSCs to septic mice, we delineated their spatiotemporal biodistribution and observed pronounced hepatic accumulation between 6 h and 24 h post-surgery, with a progressive increase over time. Notably, at 24 h post-insult, hepatic HUMSCs abundance in septic mice was significantly greater than that in control animals. These observations are consistent with findings by Pan et al. [33], who reported that superparamagnetic iron oxide-labelled MSCs predominantly localize to the liver in septic mice, identifying it as a principal target organ of injury. Nevertheless, a substantial body of evidence suggests that MSCs exert their biological effects primarily through paracrine mechanisms, releasing exosomes that mediate intercellular communication and confer immunomodulatory and tissue-reparative functions [16, 34]. On this basis, we administered HUMSCs-Exo via tail-vein injection to further investigate their protective effects against SALI and to elucidate the underlying mechanisms. Our findings demonstrate that exosomes treatment, both in vivo and in vitro, markedly enhances hepatocyte viability, significantly reduces intracellular ROS, lipid peroxides, and labile iron accumulation, and concomitantly restores GSH levels. At the molecular level, exosomes therapy downregulated the expression of the ferroptosis-promoting proteins ACSL4 and TFRC, while upregulating the ferroptosis-suppressive proteins FPN, xCT, GPX4, and FTH1. Furthermore, exosomes administration substantially attenuated hepatic tissue injury and local or systemic inflammatory responses in septic mice, partially restored hepatic biological functions, and significantly improved survival outcomes. Nevertheless, although HUMSCs-Exo have been reported to attenuate ferroptosis in target cells across multiple conditions, including viral myocarditis, inflammatory bowel disease, pulmonary fibrosis, liver fibrosis and diabetes, by delivering bioactive cargos such as let-7a-5p [24], miR-129-5p [21], miR-486-5p [35], miR-499a-5p [36], and miR-548ai and miR-660 [37], the specific exosomal determinants responsible for modulating hepatocyte ferroptosis in our system remain undefined. This unresolved “black box” of exosomal constituents, including nucleic acids, lipids and proteins, warrants systematic interrogation and further suggests that engineering HUMSCs and their derived exosomes may represent a viable strategy to potentiate anti-ferroptotic activity. Thus, MSCs-Exo possess unique advantages, including higher cellular uptake efficiency, low immunogenicity, and low tumorigenic potential, which make them a promising therapeutic modality. Whether used alone or as carriers for active biological components or chemotherapeutic agents, exosomes demonstrate significant clinical translational value. Clinical data also support the overall safety of exosomes-based therapies, with a low incidence of severe adverse events [38, 39]. However, the precise and accurate protein characteristics or compositional profiles of exosome surface ligands and internal cargo, which may vary across different sources (including species, donors, and tissue origins), remain unclear. These variations could influence the clinical indications for exosomes therapies. Additionally, challenges such as low exosome yield, unstandardized quality control (including extraction, purification, and storage), subpopulation analysis, and the lack of a universally accepted “administration template” (e.g., dosing, timing, route, and frequency) currently limit the clinical translation and application of exosomes-based therapies [38, 40]. Furthermore, the spatiotemporal distribution characteristics of exogenous exosomes (e.g., those derived from non-parental stem cells) within the human body remain poorly understood, including their “drug” pharmacokinetics. These “allogeneic exosome therapies” require further clinical research and suggest that autologous-derived exosomes (e.g., individual plasma-derived exosomes) may offer an alternative direction for cell-free therapies. Therefore, despite the promising clinical translational value of MSCs-Exo therapy, extensive clinical validation is needed to realize its full potential. This includes, but is not limited to, the standardization of methods, the reproducibility of therapeutic processes, and the long-term stability and safety of efficacy evaluation.

In addition, current research increasingly focuses on preconditioning strategies for MSCs, such as inflammatory licensing [41], hypoxia induction [42, 43], and LPS stimulation [44, 45]. These preconditioning methods can, to some extent, enhance the biological functions of MSCs-Exo, including, but not limited to, promoting angiogenesis, anti-apoptotic activity, immune modulation, and tissue regeneration [46]. This is likely due to preconditioning MSCs, which significantly alters their secretome and influences the nucleic acids, proteins, and lipids contained within their exosomes. Furthermore, preconditioning MSCs may increase the loading of protective or inhibitory molecules related to anti-inflammatory, antioxidant, pro-angiogenic, and immune-regulatory signaling pathways in the exosomes. Alternatively, preconditioning may upregulate the expression of surface adhesion molecules, integrins, and chemotactic-related molecules on exosomes, enhancing their uptake by specific organs and target cells. However, we must further investigate the close relationship between preconditioning strategies and the phenotypic and functional alterations of MSCs-Exo, considering that different preconditioning conditions (including stimulus type, intensity, and duration) may introduce significant heterogeneity in exosome components and therapeutic effects, potentially influencing the safety and dose–response evaluation in in vivo and in vitro models. Therefore, exploring preconditioned MSCs and their derived exosomes in the context of SALI and other organ damages, as well as their role in ferroptosis and other forms of programmed cell death, is a direction that warrants further in-depth research in the future.

Ferroptosis is an iron-dependent form of regulated cell death in which cellular iron metabolism and iron homeostasis play pivotal roles [14, 15]. Previous studies have shown that transferrin receptor 1 (TFRC), a membrane protein broadly expressed across diverse cell and tissue types, promotes intracellular iron accumulation by binding circulating transferrin (Tf)–iron complexes and mediating their endocytosis, thereby closely linking TFRC activity to the initiation of ferroptosis [47]. Ferroportin (SLC40A1) is the only known iron exporter located on the mammalian cell membrane and reduces intracellular iron levels by facilitating iron efflux [48]. Ferritin heavy chain 1 (FTH1) functions as the iron-binding subunit of ferritin [49], sequestering labile Fe²⁺ within cells to regulate intracellular iron availability and limit the pool of free iron. He et al. reported that MSCs therapy modulates the hepcidin—ferroportin axis in CCl₄-induced ALF, thereby alleviating ALF by suppressing hepatocyte ferroptosis [50]. Similarly, Cheng and colleagues demonstrated that MSCs mitigate iron metabolism–driven ferroptosis in CCl₄-induced ALF mice by upregulating FPN1 expression [51]. In line with these findings, our in vivo experiments revealed that the expression of TFRC, FPN, and FTH1 in hepatocytes from septic mice exhibited marked time dependence, accompanied by a pronounced increase in intracellular labile iron. Within 24 h after CLP induction, progressive disease severity was associated with a stepwise downregulation of FPN and FTH1 and a concomitant upregulation of TFRC, indicating a profound disruption of iron homeostasis in hepatocytes during sepsis. Importantly, treatment with HUMSCs-Exo effectively reversed these aberrant protein expression patterns, significantly reduced intracellular labile iron accumulation in septic hepatocytes, inhibited ferroptosis, and preserved hepatic function. Consistent findings were obtained in vitro using the iron chelator DFO [52], which primarily reduces pathological iron deposition by chelating intracellular free iron. In sepsis-mimicking conditions, HUMSCs-Exo treatment, similar to DFO, markedly attenuated the increase in labile iron levels in both murine and human normal hepatocyte cell lines, thereby facilitating the restoration of intracellular iron homeostasis. In summary, our in vivo and in vitro data collectively demonstrate that hepatocyte ferroptosis in sepsis is tightly coupled to dysregulated intracellular iron homeostasis. HUMSCs-Exo restore iron balance by enhancing ferritin-mediated iron sequestration and promoting iron efflux, thereby reducing the intracellular pool of free iron. In parallel, HUMSCs-Exo treatment suppresses aberrant lipid peroxidation and re-equilibrates redox homeostasis, ultimately attenuating hepatocyte ferroptosis, ameliorating SALI, and improving survival outcomes in septic mice.

Conclusion

In summary, our findings demonstrate a strong positive association between hepatocyte ferroptosis and SALI in septic hosts. HUMSCs-Exo therapy mitigates sepsis-induced hepatocyte ferroptosis by restoring intracellular iron homeostasis and modulating iron and lipid metabolic pathways, thereby alleviating hepatic injury, preserving liver function, and ultimately improving host survival.

Abiationsbrev

MSCs-Exo

Mesenchymal stem cells-derived exosomes

HUMSCs

Human umbilical cord mesenchymal stem cells

HUMSCs-Exo

Human umbilical cord mesenchymal stem cells-derived exosomes

SALI

Sepsis-associated liver injury

ALF

Acute liver failure

CLP

Cecal Ligation and Puncture

LPS

Lipopolysaccharide

DFO

Deferoxamine

Fer-1

Ferrostatin-1

DEGs

Differentially expressed genes

LDH

Lactate dehydrogenase

AST

Aspartate aminotransferase

ALT

Alanine aminotransferase

TBIL

Total bilirubin

DBIL

Direct bilirubin

ALB

Albumin

ALP

Alkaline phosphatase

γ-GT

Gamma-glutamyl transferase

TBA

Total bile acids

IL-6

Interleukin-6

IL-1β

Interleukin-1β

TNF-α

Tumor necrosis factor-α

PTGS2

Prostaglandin-endoperoxide synthase 2

MDA

Malondialdehyde

Fe²⁺

Ferrous iron

GSH

Reduced glutathione

GSSG

Oxidized glutathione

TFRC/CD71

Transferrin Receptor 1

ACSL4

Acyl-CoA synthetase long-chain family member 4

xCT/SLC7A11

Solute Carrier Family 7 Member 11

FTH1

Ferritin Heavy Chain 1

FPN

Ferroportin

PLOOHs

Oxic phospholipid hydroperoxides

GO

Gene Ontology

KEGG

Kyoto Encyclopedia of Genes and Genomes

PCA

Principal component analysis

SEM

Standard error of the mean

WB

Western blot

TEM

Transmission electron microscopy

NTA

NanoSight nanoparticle tracking analysis

SPF

Specific pathogen-free

H&E

Hematoxylin and Eosin

Author contributions

HK, YY, MF, YW, LW contributed to the conception and design of the study. YW, ZD, YC and JX performed the experiments. YW and LW conducted the data analysis. YW and ZD drafted the manuscript. All authors reviewed and approved the final version of the manuscript.

Funding

This work was supported by grants for the Beijing Natural Science Foundation (No. 7254419), National Natural Science Foundation of China (No. 82472238), and National Key R&D Program of China (2024YFA1212000).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval

All animal experimental procedures were conducted according to a protocol approved by the Animal Care and Use Committee of the General Hospital of PLA.

Competing interests

The authors declare no competing interests.