Mesenchymal Stem Cells Protect Against Ferroptosis via Exosome-Mediated Stabilization of Xct in Acute Liver Injury

Background: Ferroptosis, a newly recognized form of regulated cell death, was recently identied as a novel therapeutic target in tissue injury. Various studies have shown that administration of mesenchymal stem cells (MSC) is a promising therapeutic approach to repair liver injury. However, the role of ferroptosis in acute liver injury (ALI) and MSC-based therapy is unknown. Results: Here we found that CCl 4 induced elevated lipid reactive oxygen species (lipid-ROS) and mRNA levels of putative molecular markers of ferroptosis such as Ptgs2 and LOX genes. CCl 4 also downregulated the xCT protein levels resulting in the accumulation of lipid peroxidation and ferroptosis. MSC transplantation largely abolished CCl 4 -induced ferroptosis. Furthermore, the protective effects of MSC against ferroptosis were closely correlated with exosome-mediated stabilization of xCT. Administration of MSC-Exo restored the xCT protein level, decreased the elevated lipid-ROS level and Ptgs2 and LOX mRNA levels, and promoted liver restoration by inhibiting ferroptosis. Interestingly, in ALI mouse livers after MSC-Exo treatment, exosome-induced recovery of xCT protein was accompanied by upregulation of CD44 and OTUB1. The level of ubiquitinated xCT upregulated by CCl 4 was signicantly downregulated by OTUB1-mediated deubiquitination, and strong interactions of xCT with OTUB1 and CD44 proteins were detected. Conclusions: Taken together, our data indicate that MSC-Exo has a protective role against ferroptosis by maintaining xCT function. This provides a novel therapeutic strategy for ferroptosis-induced ALI. This study aimed to investigate the role of ferroptosis in ALI and the anti-ferroptotic effects of MSC and MSC-Exo. Carbon tetrachloride (CCl4) was administered in mice to induce ALI. The occurrence of ferroptosis was investigated by measuring the levels of lipid peroxidation, ROS, MDA and molecular markers of ferroptosis. Key axis proteins related to ferroptosis in the liver were analyzed by Western blot analysis, immunohistochemical staining, immunouorescence staining and co-immunoprecipitation. in ALI murine livers. Images were taken at 4 h. Data are the mean ± SD uorescence; *p < 0.05, **p < 0.001; n = 5. Individual images are presented. Schematic diagram of the protective effects of MSC-derived exosomes on maintaining xCT function during ferroptosis involved in CCl4-induced ALI.


Background
The liver has antibacterial/antiviral and drug detoxi cation functions and plays a central role in regulating homeostasis [1]. Acute liver injury (ALI) is the most prominent cause of liver disease and is associated with high morbidity and mortality [2]. A variety of hepatotoxic factors, including viruses, lipid deposition, and drugs, can induce ALI, and 3.5% of deaths worldwide result from liver disease [3,4]. Therefore, it is crucial to elucidate the underlying mechanisms of ALI and to develop appropriate treatment strategies. Carbon tetrachloride (CCl 4 ) is a typical hepatotoxic compound, widely used to evaluate the pathological mechanism of ALI [5]. We aimed to explore the potential mechanism of ALI using a CCl 4 -induced mouse model and to explore possible effective treatment strategies.
Much attention has been focused on applying mesenchymal stem cells (MSC) in biomedicine due to their potential to promote liver regeneration and repair liver injury. MSC are a heterogeneous subset of stromal cells that can be easily extracted from adipose, bone marrow, and synovial tissues and that differentiate into numerous cell lineages according to speci c biomedical applications. The immunological properties of MSC, including their immunoregulatory, anti-in ammatory and immunosuppressive abilities, suggest a potential role in immune tolerance [6]. Due to these properties, MSC are considered an ideal cell source for tissue regeneration applications. Previously, we showed that transplantation of MSC attenuated CCl 4induced ALI by modulating the hepatic immune system [7]. Initial research suggested that MSC exert their therapeutic effects by engrafting to the site of injury. However, further studies showed that only a small percentage of injected MSC reach their targets and that extracellular vesicles such as exosomes released by MSC have vital and bene cial effects during treatment [8,9]. In addition, exosomes released from MSC contain special types of RNAs, lipids, and proteins, which are required to restore physiological conditions and assist cell and tissue regeneration and proliferation [10,11]. MSC-derived exosomes (MSC-Exo) exhibit similar functions to those of the cells from which they originate and are applied as MSC-based cell-free therapy.
In this study, CCl 4 not only induced acute injury but also promoted ferroptosis in the liver. The hepatotoxic effects of CCl 4 are signi cant and are related to its metabolic conversion in the liver to reactive intermediates, such as trichloromethyl free radicals. In addition, lipid peroxidation stimulation was observed in the liver shortly after administering CCl 4 to mice in vivo [12]. A high accumulation of products from lipid peroxidation initiates ferroptosis. In addition, various studies suggest that ferroptosis might be a new type of cell death associated with liver diseases such as viral hepatitis, drug-induced liver injury, alcoholic liver disease, non-alcoholic steatohepatitis, and hemochromatosis [13][14][15][16].
Ferroptosis is a newly identi ed form of regulated cell death driven by perturbation of the glutathione (GSH)-dependent lipid-hydroperoxide-scavenging network [17,18]. It is characterized by overproduction of lipid hydroperoxides and an increased cellular labile iron pool [19]. Lipid peroxidation is regulated by system XC − , which transports glutamate out of the cell in exchange for transport of cysteine, an important precursor for GSH synthesis, into the cell [20]. Glutathione is vital for glutathione peroxidase 4 (GPX4) activity, which protects cells against ferroptosis by converting toxic lipid hydroperoxides into nontoxic lipid alcohol [21]. Thus, inhibition of GPX4 can result in the accumulation of lipid hydroperoxides that cause cell ferroptosis. On the other hand, system XC − is a cysteine-glutamate antiporter, consisting of a light-chain subunit (SLC7A11/xCT) and a heavy-chain subunit (SLC3A2). xCT exhibits transporter activity highly speci c for glutamate and cysteine and thus plays an important role in providing cysteine for the biosynthesis of GSH, a vital antioxidant in cells. Emerging strategies targeting ferroptosis have been developed to treat organ and tissue injury [22][23][24]. As the main detoxi cation organ, the liver plays a central role in regulating homeostasis. Furthermore, it was shown that MSC-derived exosomes alleviate oxidative stress-induced dysfunction in mouse livers in vivo [25].
Hence, in this study, we investigated the role of ferroptosis in MSC-based treatment for ALI in vitro and in vivo. Our results showed that xCT was signi cantly downregulated in an ALI mouse model, and its downregulation promoted CCl 4 -induced hepatocyte ferroptosis. MSC transplantation and MSC-Exo treatment largely abolished ferroptosis and protected the liver from injury. Moreover, MSC-Exo-mediated recovery of xCT protein was accompanied by upregulation of CD44 and OTUB1 in ALI mouse livers. The stability of xCT was closely related to OTUB1-mediated deubiquitination. Finally, our studies indicate that exosomes from MSC suppressed hepatocytes ferroptosis to mediate liver repair in ALI mice by maintaining xCT function.

Results
Ferroptosis drove CCl 4 -induced mouse models of ALI As shown in Fig. 1A, typical histopathological changes of ALI were observed microscopically in the CCl 4administered mouse livers and compared with those in the PBS and oil groups (left and middle), including diffuse hepatic necrosis. To investigate the contribution of ferroptotic cell death in CCl 4 -induced ALI, we performed real-time PCR analysis of putative molecular markers of ferroptosis, which showed that CCl 4 treatment induced robust increases in the mRNA levels of liver prostaglandin-endoperoxide synthase 2 (Ptgs2), 15-LOX, 12-LOX, and 5-LOX (Fig. 1B). Next, we measured lipid peroxidation in living liver cells by ow cytometry using the C11-BODIPY 581/591 uorescent probe, a canonical index of ferroptosis. CCl 4 signi cantly increased the lipid-ROS levels in the mouse livers of the CCl 4 -treated group compared with the PBS and oil groups (Fig. 1C). The release of oxidized lipid mediators is a reported characteristic of ferroptosis [26]. Membranes with arachidonic acid enrichment may facilitate ferroptosis by releasing arachidonic acid metabolites (hydroxyeicosatetraenoic acid) during cell death. The levels of 20-hydroxyeicosatetraenoic acid (HETE), 15(R)-HETE, 15(S)-HETE, 12-HETE, 11-HETE, 8-HETE, and 18-HETE were increased in the livers of mice 48 h after CCl4 injection (Fig. 1D). These ndings suggest that ferroptosis is a crucial driver of CCl 4 -induced liver injury and mortality in mice.

MSC transplantation alleviated CCl 4 -induced ferroptosis in ALI
Oxidative stress-induced excessive accumulation of lipid hydroperoxides ultimately results in ferroptosis. We showed that MSC treatment dramatically alleviated CCl 4 -induced ALI due to its immunomodulatory function [7]. Moreover, MSC play important roles in reducing oxidative stress [27]. Thus, we hypothesize that MSC may have a protective role against ferroptosis by promoting lipid-ROS scavenging in CCl 4induced ALI. Histopathological staining indicated that hepatic necrosis was dramatically improved following administration of MSC and ferrostatin-1 compared with the CCl 4 group, indicating the effect of MSC against CCl 4 -induced ferroptosis in ALI ( Fig. 2A). Compared with the CCl 4 , we observed signi cant decreases in the mRNA levels of Ptgs2, 12-LOX, 5-LOX, and 15-LOX, classic biomarkers of ferroptosis, in the MSC and Fer-1 groups (Fig. 2B). MSC treatment also signi cantly reduced the accumulation of CCl 4induced lipid hydroperoxides in the liver (Fig. 2C). Interestingly, according to the mRNA levels of liver Ptgs2 and LOXs, and lipid peroxidation in living cells, it is plausible that MSC and ferrostatin-1 have comparable effects on protection against ferroptosis. Compared with Fer-1, MSCs had a similar effect on protection against liver damage, as evidenced by C11-BODIPY 581/591 probe staining (Fig. 2C). The decreases in the Ptgs2, 12-LOX, 5-LOX, and 15-LOX mRNA levels were similar between the MSC and Fer-1 groups (Fig. 2B). It is worth noting that LOXs (and particularly 15-LOX) have been reported as essential regulators of ferroptotic cell death as they contribute to the cellular pool of lipid hydroperoxides [28,29]. Moreover, the increased levels of 20-hydroxyeicosatetraenoic acid (HETE), 15(R)-HETE, 15(S)-HETE, 12-HETE, 11-HETE, 8-HETE, and 18-HETE induced by CCl 4 was abrogated by MSC and Fer-1 treatment (Fig. 2D). These results suggest that the potential ability of MSC to downregulate peroxidation contributes to lipid-ROS reduction and inhibits the progression of ferroptosis.
The xCT protein level was downregulated in CCl 4 -induced ALI but upregulated following MSC treatment invivoandin vitro Next, we established an acute hepatocyte injury model to detect the mechanisms underlying xCT regulation in ALI treated with MSC in vitro. On days 1, 2, 3, and 7, we detected a dramatic decrease in cell viability using the Cell Counting Kit-8 when the concentration of CCl 4 increased to 10 mM ( Based on the above hepatocyte model, co-culturing experiments were performed to investigate the effects of MSC on damaged hepatocytes. As shown in Fig. 3D, AST and ALT levels were signi cantly increased in the CCl 4 group but dramatically reduced after MSC co-culture. We speculate that MSC have a protective effect in ALI via the paracrine pathway. Next, to ascertain whether co-culture with MSC also rescues the CCl 4 -induced decrease in the xCT protein level in primary hepatocytes, WB analysis was performed. Interestingly, similar to the in vivo ndings, the xCT protein level was reduced in the CCl 4 group compared with the PBS group but was restored by MSC co-culture (Fig. 3E). These results suggest that a paracrine factor mediates the regulation of xCT protein levels to facilitate the protective effects of MSC against ferroptosis.

MSC-derived exosomes inhibited ferroptosis in CCl 4 -induced ALIin vitro
To further examine the role of MSC in ALI, we ascertained the levels of ROS and MDA. As shown in  4B). These data indicate that MSC exert protective effects via paracrine mechanisms in ALI. Recent studies have shown that MSC produce exosomes, which can ameliorate tissue injury via the delivery of its DNA, miRNA, and protein contents [30][31][32]. Therefore, we extracted and identi ed exosomes from MSC conditioned medium. We used refrigerated transmission electron microscopy (Fig. 4C) and WB analysis ( Supplementary Fig. S1D) to characterize vesicles recovered from the MSC-conditioned medium by differential ultracentrifugation and identi ed 30-100 nm vesicles that were morphologically similar to exosomes, which expressed the CD63 and CD81 exosomal markers. To determine whether hepatocytes can internalize MSC-Exo, we labeled exosomes with green lipophilic uorescent dye (PKH67), and they were subsequently co-cultured with CCl 4 -induced injury hepatocytes. The hepatocytes exhibited high uptake e ciency, as indicated by the green uorescent signal (Fig. 4D). Furthermore, after 24 h incubation with PKH67-labeled exosomes, more MSC-Exo aggregated on the cell membrane in CCl 4induced injured hepatocytes, suggesting that the MSC-Exo exert anti-ferroptotic effects by fusing with the recipient cell membrane or interacting with proteins on the cell membrane. Next, we measured the lipid peroxidation of the primary hepatocytes with CCl 4 -induced injury after MSC-Exo treatment. Similar to erastin, CCl 4 induced signi cant upregulation of lipid-ROS level, while MSC-Exo treatment decreased the upregulated level of lipid-ROS, similar to Fer-1 treatment (Fig. 4E). Moreover, CCl 4 induced a dramatic increase in the MDA level, similar to erastin, while MSC-Exo reduced the increased MDA level, similar to Fer-1, in hepatocytes (Fig. 4F). This data suggested that MSC-derived exosomes inhibited ferroptosis in CCl 4 -induced ALI in vitro.

MSC-derived exosomes inhibited ferroptosis in CCl 4 -induced ALI invivo
Having shown that MSC-Exo inhibited ferroptosis in CCl 4 -induced liver injury in vitro, we investigated the effects of MSC-Exo in CCl 4 -induced ferroptosis in vivo. Interestingly, MSC-Exo also ameliorated hepatic necrosis and downregulated the increased ALT/AST levels induced by CCl 4 in ALI, which is consistent with the protective effects of MSC ( Fig. 5A and B). In addition, measurement of MDA levels and real-time PCR analysis of molecular markers of ferroptosis showed that MSC-Exo treatment signi cantly downregulated MDA and liver Ptgs2, 15-LOX, 12-LOX, and 5-LOX mRNA levels induced by CCl 4 (Fig. 5C and D). These ndings suggest that MSC-Exo also inhibit ferroptosis in CCl 4 -induced liver injury in vivo.

MSC-Exo protect against ferroptosis via stabilization of xCT in ALI
We investigated the underlying mechanisms of the MSC-Exo-mediated protective effects against CCl 4induced ferroptosis. Accumulating evidence indicates that the stability of xCT, which is the functional subunit of a glutamate-cysteine transporter that resides on the cell membrane, is modulated by CD44v8-10 [20,33]. CD44 is a cell surface marker of MSCs [7] and a critical molecule inducing rm adhesion between MSC and injured liver tissues, which mediate the recruitment of MSC to CCl 4 -induced liver injury [34]. Moreover, we detected CD44 protein both in MSC and MSC-derived exosomes ( Supplementary Fig. S1D). Therefore, we hypothesized that the CD44-xCT axis is functionally related to the anti-ferroptotic effects of MSC-Exo treatment. As shown in Fig. 6A, xCT protein levels in acute injured hepatocytes treated with 0, 20, 40, 80, and 160 µg MSC-Exo were measured by WB analysis. Interestingly, the exosome-induced recovery of xCT protein was accompanied by upregulation of CD44 and OTUB1, key regulators of xCT function and CD44-mediated effects in ferroptosis, in a dose-dependent manner. OTUB1, an ovarian tumor deubiquitinase family member, was previously reported to directly interact with and regulate xCT stability [35]. Furthermore, CCl 4 administration increased the ubiquitination of xCT, while MSC-Exo treatment downregulated the ubiquitination of xCT (Fig. 6B). These data suggest that CCl 4induced ubiquitination abolished the stabilization of xCT, leading to a downregulated level of xCT in ALI. However, MSC-Exo treatment promotes the stabilization of xCT by reducing its ubiquitination, resulting in upregulation of xCT in ALI.
To determine the mechanisms by which MSC-Exo exerts protective effects against ferroptosis, we performed immuno uorescence and co-immunoprecipitation analyses. Immuno uorescence analysis suggested that CD44 and xCT were co-expressed and increased in ALI after MSC-Exo treatment compared with the CCl 4 and PBS groups (Fig. 6C). Next, to measure the interaction under physiological conditions, we performed co-immunoprecipitation analysis targeting endogenous proteins expressed in liver tissues of ALI after MSC-Exo treatment. As shown in Fig. 6D, the endogenous CD44 and OTUB1 proteins were co-precipitated by a xCT-speci c antibody, while endogenous xCT was co-precipitated by a CD44-speci c and OTUB1-speci c antibody, respectively ( Fig. 6E and F). Thus, we further speculated that MSC-Exo treatment promotes increases in CD44 and OTUB1 proteins and stabilizes xCT by directly interacting with and reducing its ubiquitination in CCl 4 -induced ALI.
MSC-Exo in the circulation localized more readily in the liver in CCl 4 -induced ALI Next, we hypothesized that exosomes participate in tissue crosstalk during MSC-Exo treatment, resulting in exosome uptake in the recipient tissues. To determine the biodistribution of MSC-Exo after entering the circulation during treatment, we isolated MSC-Exo from MSC-conditioned medium and labeled them with lipophilic carbocyanine DiR. Next, the labeled MSC-Exo were intravenously administered via the tail to ALI mice, which were subjected to whole-body intravital imaging. In vivo uorescent imaging showed that DiR-labeled MSC-Exo targeted the injured and normal livers at 6 h post-injection. There was also more uorescence in ALI mice compared with the PBS and oil groups (Fig. 7A), indicating that MSC-Exo in circulation were more likely to aggregate in damaged livers. We performed uorescent imaging of the lung, heart, kidney, liver, and spleen of ALI mice treated with DiR-labeled MSC-Exo to con rm this. As shown in Fig. 7B, the experiments in mice receiving labeled MSC-Exo showed signi cantly higher uorescent signals in the liver compared with other organs, suggesting that MSC-Exo in circulation localized more readily in the liver in ALI mice. Neutralizing antibodies against CD44 reduced MSC-Exo targeting to the diseased liver by 30.85% (P < 0.05) (Fig. 7C). These results indicated that the effect of MSC-Exo on protection against ferroptosis was associated with the function of CD44.

Discussion
Ferroptosis is a newly recognized non-apoptotic form of regulated cell death characterized by overwhelming accumulation of lipid hydroperoxides, which contribute to a wide range of pathologies, including cancer, neurodegeneration, and tissue injury [17,36,37]. Accumulating evidence suggests that patients with liver disease are likely to suffer from ferroptosis-regulated cell death [36,38]. Recently, it was reported that ferroptosis occurs in CCl 4 -induced liver brosis [39]. In this study, besides increased liver necrotic areas, lipid hydroperoxides were also dramatically upregulated, suggesting accumulation of lipid-ROS in ALI. Furthermore, we observed a signi cant increase in ferroptosis-relevant genes, such as Ptgs2 and LOXs. Ptgs2, also known as cyclooxygenase-2, is induced in cells undergoing ferroptosis [21]. Thus, these results suggest that ferroptosis-mediated cell death occurs in CCl 4 -induced acute liver injury. We also showed that MSCs protect against ferroptosis-induced cell death, since we saw similar protective effects of ferrostatin-1 and MSC treatment on reducing lipid-ROS accumulation and ferroptosis-relevant gene expression in the ALI mouse model. The protective effects of MSCs were signi cantly attenuated when MSCs were simultaneously injected with a ferroptosis inducer such as erastin. Thus, our results show evidence of the protective role of MSCs against ferroptosis, based on the lipid hydroperoxides and ferroptosis gene expression levels. This work should be followed up with further investigation of the antiferroptotic mechanisms of MSCs.
Recently, increasing evidence has shown that most of the MSC-mediated bene cial effects in the treatment of disease are attributed to the functions of exosomes released by MSCs. MSC-Exo are extracellular vesicles 30-100 nm in diameter containing MSC-derived bioactive molecules (including mRNAs, miRNAs, chemokines, cytokines, enzymes, and other functional proteins) that regulate the survival and proliferation of recipient cells. Interestingly, exosomes from human umbilical cord bloodderived MSCs protect against ALI via intercellular communication [25]. Therefore, we investigated the route and mechanisms underlying the hepatoprotective role of MSC-Exo.
As a cysteine/glutamate antiporter, xCT (commonly known as SLC7A11) is a key modulator of glutathione biosynthesis and antioxidant defense in homeostasis. Emerging evidence suggests that upregulation of xCT promotes cell growth by suppressing ferroptosis. In this study, MSC-Exo signi cantly reduced the elevated levels of ALT/AST and restored the reduced xCT protein level in ALI. In vitro, MSC-Exo downregulated the increased MDA level and lipid ROS (speci c biomarkers of ferroptosis) accumulation in hepatocytes induced by CCl 4 . Consistent with these ndings, we con rmed that MSC-Exo treatment reduces MDA levels and ferroptosis-relevant gene expression to protect the liver from injury in vivo. Thus, we hypothesize that MSC-Exo inhibits ferroptosis in CCl 4 -induced ALI in vitro and in vivo.
Additionally, the exosome-induced recovery of xCT protein was accompanied by upregulation of CD44 and OTUB1 and strong interactions of xCT with CD44 and OTUB1 in ALI following MSC-Exo treatment.
These data indicate that MSC-Exo might protect against ferroptosis via the CD44-xCT axis. More interestingly, MSC-Exo in the circulation localized more readily in the damaged liver. These results have vital implications concerning how MSC-Exo-mediated anti-ferroptotic effects are regulated in ALI.

Conclusions
In summary, we showed that ferroptosis occurs in liver tissues in an ALI mouse model. In addition, xCT was signi cantly reduced in ALI mouse liver tissues and CCl 4 -induced acute injured hepatocytes. MSC transplantation and MSC-Exo administration largely abolished ferroptosis and protected the liver from injury by maintaining xCT function. Furthermore, the protective effects of MSC-Exo against ferroptosis were correlated with the CD44-xCT axis. Together, these data indicate that MSC-Exo inhibit ferroptosis primarily by regulating the stabilization of xCT protein in ALI and strongly suggest that the ferroptosisrelevant pathway is a promising target, leading to valuable therapeutic strategies to treat ALI (Fig. 7D).

Methods
Aim, design and setting of the study This study aimed to investigate the role of ferroptosis in ALI and the anti-ferroptotic effects of MSC and MSC-Exo. Carbon tetrachloride (CCl4) was administered in mice to induce ALI. The occurrence of ferroptosis was investigated by measuring the levels of lipid peroxidation, ROS, MDA and molecular markers of ferroptosis. Key axis proteins related to ferroptosis in the liver were analyzed by Western blot analysis, immunohistochemical staining, immuno uorescence staining and co-immunoprecipitation.

Preparation of MSCs
MSCs were obtained from compact bones of 1-week-old C57BL/6 male mice. Cells from passages 3-5 were used in the experiments. Detailed information is shown in the Supplementary Materials.
Mouse models of ALI ALI induction was performed in 6-8-week old age-matched C57BL/6J male mice (n = 10-12 per group) by intraperitoneal injection of 3 mL/kg CCl 4 (Sigma-Aldrich Trading Co., Shanghai, China) in coconut oil (v/v, 50%). Control and negative control mice were injected with PBS and coconut oil, respectively. At 6 h after injection of CCl 4 , the mice were divided into four groups: CCl 4 group, injected with 100 µL PBS (supplemented with 2% mouse serum) through the tail vein; MSC group, injected with 5 × 10 5 MSCs suspended in 100 µL PBS (supplemented with 2% mouse serum) through the tail vein; Fer-1 group, intraperitoneally injected with ferrostatin-1 (a ferroptosis inhibitor, 2.5 µmol/kg body weight); MSC + erastin group, simultaneous intraperitoneal injection of erastin (a ferroptosis inducer, 30 mg/kg body weight) twice every other day. The mice were sacri ced 48 h after MSC injection, and serum and liver tissues were collected for subsequent analysis.

Isolation and characterization of exosomes
MSCs were cultured in exosome-free medium, and the exosomes were isolated and puri ed by ultracentrifugation. Brie y, MSC conditioned medium was collected after 48 h of culture and centrifuged at 2,000 × g for 20 min to remove debris and cells. The supernatant was collected and transferred to a new sterile tube and centrifuged at 10,000 × g for 15 min. The supernatant was ltered using a 0.22-µmpore sterile lter, followed by ultracentrifugation at 110,000 × g for 70 min at 4°C to obtain exosome pellets, which were resuspended in PBS and stored at − 80°C. We used the BCA protein assay kit (Thermo Fisher Scienti c, Waltham, MA, USA; UD283191) to determine the protein content of the concentrated exosomes. Exosomes were identi ed by Western blot (WB) analysis of the marker proteins CD63 and CD81. The morphology of the exosomes was observed using a 120 kV refrigerated transmission electron microscope (Tecnai Spirit Bio-TWIN, FEI Company, Hillsboro, OR, USA). The MSC-Exo concentration used to treat liver injury was 8 mg/kg body weight (n = 10) in the animal experiments.

Isolation and incubation of primary hepatocytes
Primary hepatocytes were harvested from 6-8-week-old C57BL/6 mice according to the rapid two-step method using collagenase perfusion [40]. We cultured the primary hepatocytes in William's E Medium containing 10% FBS (Gibco Biosciences, Waltham, MA, USA), 10 mg/mL streptomycin, 100 IU/mL penicillin, 1% ITS liquid medium supplement (Sigma-Aldrich, I3146), and 40 ng/mL dexamethasone, and observed the state of the primary hepatocytes daily.  Table S1). All experiments were performed in triplicate, and melting curve analysis was performed to monitor the speci city.

WB analysis
The Immuno uorescence staining and co-immunoprecipitation We used mouse liver slides or fragments containing primary cells. The slides were incubated with primary antibodies at 4°C overnight. The following day, after incubation with the corresponding secondary antibody, the nuclei were stained with DAPI before observing by confocal laser microscopy. Coimmunoprecipitation analysis was carried out using the Pierce Co-Immunoprecipitation Kit (26149; Thermo Fisher) according to the manufacturer's instructions.

Exosome labeling and tracking in mice
The details of the exosome labeling and tracking procedures are provided in the Supplementary Materials.

Statistical analysis
The results are presented as means ± standard deviation (SD) and were analyzed by GraphPad Prism software (GraphPad Software Inc., San Diego, CA, USA). Student's t-test was performed to compare values between two groups. We performed one-way ANOVA followed by Tukey's post-hoc test to compare data obtained from multiple groups. *p < 0.05 and **p < 0.001 were considered to indicate statistical signi cance.
The   were abrogated by MSC and Fer-1 treatment. Signi cance was calculated by one-way ANOVA with Tukey's post-hoc test. *p < 0.05 or **p < 0.001 indicates a signi cant difference between groups.

Figure 3
The protein level of xCT is downregulated in CCl4-induced ALI and upregulated following MSC treatment in vivo and in vitro. (A) Cell viability was measured on days 1, 2, 3, and 7 after hepatocytes were treated     Exosomes in the circulation were more readily localized in the liver in ALI mice. (A) Whole-body uorescence imaging of C57BL/6 male mice treated with 8 mg/kg labeled exosomes from MSC conditioned medium. Images were taken 6 h after tail vein injection. (B) Quantitative uorescent imaging of lung, heart, kidney, liver, and spleen from recipient mice receiving labeled exosomes from MSC conditioned medium. (C) Neutralizing antibodies against CD44 reduced hepatic engraftment of MSC-Exo