A simple and efficient strategy for cell‐based and cell‐free‐based therapies in acute liver failure: hUCMSCs bioartificial liver

Abstract Acute liver failure (ALF) is a life‐threatening condition. Cell‐based and cell‐free‐based therapies have proven to be effective in treating ALF; however, their clinical application is limited by cell tumorigenicity and extracellular vesicle (EV) isolation in large doses. Here, we explored the effectiveness and mechanism of umbilical cord mesenchymal stem cells (hUCMSCs)‐based bioartificial liver (hUCMSC‐BAL), which is a simple and efficient strategy for ALF. D‐galactosamine‐based pig and mouse ALF models were used to explore the effectiveness of hUCMSC‐BAL and hUCMSC‐sEV therapies. Furthermore, high‐throughput sequencing, miRNA transcriptome analysis, and western blot were performed to clarify whether the miR‐139‐5p/PDE4D axis plays a critical role in the ALF model in vivo and in vitro. hUCMSC‐BAL significantly reduced inflammatory responses and cell apoptosis. hUCMSC‐sEV significantly improved liver function in ALF mice and enhanced the regeneration of liver cells. Furthermore, hUCMSC‐sEV miRNA transcriptome analysis showed that miR‐139‐5p had the highest expression and that PDE4D was one of its main target genes. The sEV miR‐139‐5p/PDE4D axis played a role in the treatment of ALF by inhibiting cell apoptosis. Our data indicate that hUCMSC‐BAL can inhibit cytokine storms and cell apoptosis through the sEV miR‐139‐5p/PDE4D axis. Therefore, we propose hUCMSC‐BAL as a therapeutic strategy for patients with early ALF.

3][4][5] The release of metabolites, inflammatory mediators, and toxic metabolites caused by ALF and the ability of residual liver cells to repair liver function determine the prognosis of ALF.That is, hepatocyte injury is attributed to inflammatory mediators and toxic metabolites released by early ALF, which results in an imbalance of homeostasis and the immune environment of surrounding normal hepatocytes. 1This causes the death of surrounding normal hepatocytes and produces a cascade reaction, resulting in a large number of hepatocyte deaths and irreversible liver injury.Therefore, controlling the inflammatory response and homeostasis may be a mechanism of treating early ALF.
At present, liver transplantation is the only effective treatment for ALF patients; however, its application is limited by a shortage of liver donors, complex surgery, high costs, and the need for lifelong immunosuppressant use. 6Artificial livers are a key area of research in ALF treatment that have emerged in recent years, which include nonbioartificial, bioartificial, and hybrid artificial livers.Although nonbioartificial livers are widely used in clinical applications, they cannot significantly improve the survival time of patients with ALF, except with high plasma exchange. 7Conversely, bioartificial and combined artificial livers are limited by their seed cells and bioreactors, and most of them are considered to still be in the preclinical stage, and thus have not been applied in clinical settings except the bioartificial liver of HepArt Medical Devices (AMC-BAL) and Vital Therapies (ELAD). 8wever, large clinical trials with AMC-BAL 9 and ELAD 10 did not show a survival benefit.Although scientists have extensively studied seed cells from various sources and their corresponding bioreactors, [11][12][13][14][15][16][17][18] most previous literature focused on liver cell function replacement therapy and are limited by the lack of human liver cell sources and the tumorigenicity of liver tumor cell lines, as well as safety and ethical concerns.Nevertheless, bioartificial livers that act solely on the cytokine storm, homeostasis regulation, and immune regulation have not previously been reported.
Mesenchymal stem cell-based and cell-free-based therapies have proven to be effective in treating ALF.0][21][22][23][24] However, MSC transplantation, typically local or systemic intravenous injection of MSCs, may cause the release of inflammatory factors, an enhanced risk of opportunistic infections and cancer, allogeneic production of antibodies, and other adverse consequences. 25In contrast, veinous infusion of MSCs may cause blood vessel clots.Interestingly, an increasing number of studies have shown that extracellular vesicles (EV), mainly small EV (sEV, derived from MSCs regulate immunity and inflammatory response and promote cell proliferation. 25[28] To address these questions and demonstrate the value of human umbilical cord mesenchymal stem cells (hUCMSCs) for regenerative medicine in ALF, we first constructed a bioartificial liver based on hUCMSCs (hUCMSC-BAL), which we then verified using a pig ALF model.Not only could hUCMSC-BAL avoid adverse reactions caused by cells entering the patient's body, but also sEV and cytokines secreted by hUCMSCs could enter the patient's body through the bioreactor semi-permeable membrane to play a therapeutic role.To further explore the specific mechanism of hUCMSC-BAL, we extracted sEV and factors in the circulating culture medium of the hUCMSCs bioreactor and constructed a mouse ALF model using sEV from the hUCMSCs bioreactor for treatment.sEV transcriptome analysis revealed that miR-139-5p was the most highly expressed miRNA and that PDE4D is one of its main target genes.These results were verified in L02 and HepaRG cell ALF model.

| Large-scale expansion of hUCMSCs in the bioreactor
At present, an important challenge for bioartificial liver application is the large-scale production of seed cells (up to 10 9 -10 10 cells).Therefore, we constructed an hUCMSCs bioreactor culture system (fiber surface area 4000 cm 2 /1.2 m 2 , fiber outer cavity volume 20 mL/70 mL).hUCMSCs were implanted in the outer cavity of the hollow fiber of the bioreactor and attached to the surface of the hollow fiber (Figure 1h-k).The serum-free medium passed through the inner cavity of the hollow fiber using a peristaltic pump and exchanged substances through the semi-permeable membrane.sEV, secretory proteins, and other molecules could freely pass through the hollow fiber semi-permeable membrane, whereas cells and cell fragments could not.We determined the proliferation rate of cells based on the rate of glucose (GLU) decline and lactate increase in the culture medium (Figure 1 l,m).Because the rate of GLU decline is directly proportional to the number of cells in the bioreactor, we considered the expansion of hUCMSCs in the bioreactor to be complete when the rate of GLU decline was stable for 1 week.

| Establishment of the porcine ALF model
To verify the efficacy of hUCMSC-BAL, we induced an ALF model in miniature pigs by intravenous injection of D-gal at 0.40 g/kg body weight according to our previous study, 4 thereby establishing an animal model with similar characteristics to human ALF.The blood alanine transaminase (ALT), aspartate transaminase (AST), total bilirubin (TBIL), ammonia (Amm) levels, prothrombin time (PT), and the international normalized ratio (INR) in experimental animals increased significantly ( ps < 0.05) after D-gal injection, meeting the diagnostic criteria of ALF. 4 All ALF model pigs showed clinical ALF symptoms, such as anorexia, unstable gait, and yellow urine.These results confirm the successful construction of the ALF animal model.

| Prevention of ALF progression by 8-h hUCMSC-BAL treatment
Survival analysis showed that the survival rate of group C (5/5) was better than that of groups B (3/5) and A (0/5) (Figure 2c).All animals showed anorexia and activity reduction, their liver function-related indicators (Amm, ALT, AST, alkaline phosphatase (ALP), TBIL, and direct bilirubin (DBIL)) began to increase, blood GLU began to decline, and coagulation function (PT, activated partial thromboplastin time (APTT), and fibrinogen (Fib) began to deteriorate after 12 h of D-gal administration, but there was no significant difference between groups.Biochemical measurements showed that high-dose hUCMSC-BAL treatment reduced Amm concentrations at 24 h after D-gal administration compared with those in the ALF control and low-dose hUCMSC-BAL groups (Figure 2d).Pigs in group A showed obvious symptoms of HE 48 h after D-gal administration.However, none of the animals in group C presented with the characteristic symptoms of HE.Liver function (AST, ALT, TBIL, ALP, gamma-glutamyl transferase (GGT), and DBIL) in group C gradually recovered after hUCMSC-BAL treatment compared with that in groups A and B. While GLU continuously declined in groups A and B, ALB in groups B and C declined after hUCMSC-BAL treatment, then gradually recovered (Figure 2d).There were no significant differences in creatinine (CRE) levels among the three groups, except for a significant increase before death (Figure 2d).To evaluate the safety and efficacy of hUCMSC-BAL during treatment, we collected samples to detect biochemical indicators every 2 h during treatment and monitored the general condition, heart rate, respiratory rate, blood pressure, and oxygen saturation of the experimental animals.hUCMSC-BAL was stable, and no obvious abnormality was observed during treatment.There were no significant differences in GGT, GLU, ALB, BUN, CRE, or APTT between the two groups during hUCMSC-BAL treatment; however, significant differences were observed in Amm, ALT, AST, ALP, TBIL, and DBIL (Figure S2a-d).Vital signs remained stable during 8 h of treatment, except for the systolic blood pressure (SP) and diastolic blood pressure (DP), which decreased slightly after anesthesia, and no significant difference was observed between two groups (Figure S2e).The prefilter pressure, post-filter pressure, and transmembrane pressure of the plasma separator remained stable throughout hUCMSC-BAL treatment (Figure S2f).Cell death/live staining showed that cell viability was more than 90% before and after hUCMSC-BAL treatment (Figure S2g).

| hUCMSC-BAL therapy alleviated peripheral inflammation
ALF animals exhibit features of systemic inflammation, progression to multiple organ dysfunction, and immune imbalance.In this study, all inflammatory cytokines were significantly altered in the early stage of F I G U R E 1 hUCMSCs acquisition and identification, and large-scale expansion in hollow fiber bioreactor.(a) Schematic diagram of acquisition and identification, and large-scale expansion of hUCMSCs, which were first expanded to 10 7 -10 8 in the culture flask, then inoculated into the hollow fiber bioreactor until amplified to 10 9 -10 10 ; (b) hUCMSCs were isolated by the tissue block method; (c) P5-generation hUCMSCs morphology under an optical inverted microscope.(d) hUCMSCs differentiated into fat and stained with Oil red O (Â200).(e) Osteogenic differentiation of hUCMSCs stained with alizarin red (Â200).(f) Cartilage differentiation of hUCMSCs stained with safranin O (Â200).(g) Flow cytometry showed that hUCMSCs highly expressed CD73, CD90, and CD105 (both ≥ 95%) but did not express CD34, CD45, CD19, CD11b, or HLA-DR (both ≤ 5%); (h) The results of live and dead staining of hUCMSCs in the bioreactor; (i) hUCMSCs in the bioreactor under a light microscope; (j) Hollow-fiber superfine structure in the bioreactor.(k) Adhesion of hUCMSCs to hollow fiber in the bioreactor.(l-m) Changes in GLU, Lac, and PO 2 in the bioreactor during hUCMSCs culture in (l) Group B and (m) Group C. GLU, glucose; Lac, lactate; PO 2 , oxygen partial pressure.Scale bar 200 μm.
ALF, suggesting that drug-induced ALF is closely related to the inflammatory cascade.The levels of pro-inflammatory cytokines (IFN-γ, TNF-α, IL-1α, IL-1β, IL-6, IL-8, IL-12, and IL-18) in group A significantly increased and peaked before animal death.Most pro-inflammatory cytokines in group C decreased slightly after hUCMSC-BAL treatment, which may be related to hemodilution and filtration.Subsequently, the above factors increased slightly again, and then reached a peak value at 48 h after modeling before decreasing, with the values lower than those for groups A and B at the same time.With the recovery of liver function, the above inflammatory cytokines gradually returned to baseline levels in groups B and C (Figure .3a).The levels of anti-inflammatory cytokines (IL-10) and growth factors (hepatocyte growth factor (HGF) and vascular endothelial growth factor VEGF) increased after ALF modeling and peaked at 48 h, with no significant difference among the three groups (Figure 3b).Based on the baseline levels of inflammatory cytokines, heat maps at 72 h after modeling showed that pro-inflammatory factors (IFN-γ, TNF-α, IL-1α, IL-1β, IL-6, IL-8, IL-12, and IL-18) increased the most in group A, followed by group B, whereas no significant increase was observed in group C. The increase in IL-18 was the most obvious, and the difference between the groups was the largest.Anti-inflammatory cytokines (IL-10) and related growth factors (HGF and VEGF) did not increase significantly, and there was no significant difference between groups (Figure 3c).

| hUCMSC-BAL therapy alleviates liver injury, enhances liver regeneration, and alleviates HE
To evaluate the pathological changes in ALF animals, hematoxylineosin (H&E), Ki67, and TUNEL staining and electron microscopy were performed after the animals died or after the end point of observation.TUNEL staining and TUNEL-positive cells showed that hepatocyte apoptosis in group C was lower than that in groups A and B (Figure 4a,f).H&E staining showed extensive hemorrhagic necrosis, disorder of the hepatic lobule structure, and vacuoles in group A, which was more severe than that in groups B and C (Figure 4b).The HE scores in group C were lower than those in groups A and B (Figure 4d).Ki67 staining and Ki67-positive cells showed that hepatocyte regeneration in group B was higher than that in groups A and C (Figure 4c,e).
Electron microscopy showed that endoplasmic reticulum and mitochondrial damage in group C was less than that in groups A and B (Figure 4g).However, Masson, Sirius red, PAS, and Oil red staining in the three groups showed no significant differences (Figure S3).Gross necroscopy specimens and H&E staining of extrahepatic organs in the three groups showed no significant changes (Figure S4).The above results showed that hUCMSC-BAL can inhibit cell apoptosis, promote liver cell regeneration, and prolong the survival of ALF animals.

HE features cerebral edema and neuronal dysfunction, often
induced by hyperammonemia.In addition to directly causing cerebral edema, hyperammonemia also induces neuroinflammation, mediating cognitive impairment.The brain water content was used to evaluate cerebral edema and was the lowest in group C compared with that in the other groups ( p < 0.05) (Figure 4h,i), and the Amm was lower in group C than in the other groups (p < 0.05) (Figure 4j).These findings suggest that hUCMSC-BAL therapy has a significant protective effect on the brain by alleviating cerebral edema and Amm.

| hUCMSC-sEV express the morphological characteristics and specific markers of sEV
To further explore the mechanism of hUCMSC-BAL in the treatment of ALF, we successfully extracted sEV from the circulating culture medium of the hUCMSCs bioreactor (Figure 5a), and found no significant difference in the protein content of the inner and outer hollow fibers in the bioreactor (Figure 5g).The NanoView results showed that the average particle size of hUCMSC-sEV was approximately 56 nm (Figure 5b-d).Transmission electron microscopy showed that the sEV were disc-shaped, with a diameter of approximately 100 nm (Figure 5e).
Western blotting results confirmed that hUCMSC-sEV highly expressed the sEV protein markers CD9, TSG101, CD81, syntenin, and CD63, but did not express the negative protein marker calnexin (Figure 5f).The above experimental results prove that the sEV isolated from the culture supernatant of the hUCMSCs bioreactor have the typical morphology, good uniformity, and characteristics of sEV.
To confirm that hUCMSC-sEV could be taken up by the liver, we labeled hUCMSC-sEV with DIR, which is a fluorescent dye that binds to the phospholipid bilayer membrane, and observed them in vivo and in vitro under a fluorescence microscope within 48 h.The results showed that the liver could take up hUCMSC-sEV after injection through the tail vein (Figure 5h) and that hUCMSC-sEV could be taken up by L02 and AML-12 cells (Figure 5i).

| hUCMSC-sEV relieves ALF in mice
To further explore the relevant mechanisms, we successfully constructed a mouse ALF model using D-gal and LPS and treated mice   with hUCMSC-sEV (Figure 6a).Compared with control, hUCMSC-sEV injection significantly alleviated liver injury, reduced hepatocyte apoptosis, and promoted hepatocyte regeneration in ALF mice (Figure 6b-e).

| hUCMSC-sEV miRNA sequencing and bioinformatic analysis
To further explore the specific mechanisms of hUCMSC-BAL, we performed high-throughput sequencing of hUCMSC-sEV (Figure S5a).
The results showed that hUCMSC-sEV were rich in a variety of miR-NAs, among which miR-139-5p was the most abundant (Figure S5b).
Functional enrichment can better determine the dominant pathway controlled by hUCMSC-sEV.GO analysis of the targets genes of the top 10 miRNAs showed that "regulation of transcription from RNA polymerase II promoter" and "protein phosphorylation" in the biological process component (Figure S5c), "nucleoplasm" and "cytosol" in the cellular component (Figure S5d), and "protein binding" and "transcription factor activity, sequence-specific DNA binding" in the molecular function component had the highest significance (Figure S5e).KEGG pathway enrichment was dominated by the Rap1, MAPK, and cAMP signaling pathways (Figure S5f).In summary, these results indicate a possible contribution of hUCMSC-sEV miRNAs to the inhibition of hepatocyte apoptosis.

| Conservatism of miR-139-5p expression
To clarify the conserved expression of miR-139-5p and provide a basis for subsequent clinical application, we detected the expression of miR-139-5p in hUCMSCs from different passages, culture methods, and sources.The results showed that miR-139-5p was stably expressed by hUCMSCs from different passages, culture methods, and sources (Figure 7a-c).Interestingly, miR-139-5p expression in the bioreactor was significantly higher than that in the plate culture (Figure 7c).

| PDE4D was a direct target of miR-139-5p
The results of the target genes of miR-139-5p predicted by online miRNA target gene prediction websites showed that PDE4D, LCOR, ZBTB34, NR5A2, and DCBLD2 were the main target genes (Figure S6).Among them, PDE4D can regulate the phosphorylation of Akt, which is known to inhibit several pro-apoptotic Bcl-2 family members, such as Bax. 29Moreover, by binding site prediction, we found that miR-139-5p bound to the PDE4D 3 0 -UTR.Furthermore, the results of dual-luciferase reporter assays also confirmed this observation (Figure 7d,e).

| sEV miR-139-5p/PDE4D axis plays an important role in ALF treatment
To further explore the effect of the miR-139-5p/PDE4D axis on ALF treatment, we used L02 cells and HepaRG cells to construct ALF models and then transfected them with miR-139-5p mimics, miR-139-5p inhibitors, and miR-negative control (NC).The results showed that miR-139-5p could significantly reduce the expression levels of PDE4D, SIRT1, p-AKT, Bax and Caspase 3 as well as significantly increase the expression level of Bcl-2 (Figures 7f,g and S7a,b).Furthermore, we constructed UCMSCs with stable overexpression or knockdown of miR-139-5p and L02 and HepaRG cell lines with stable overexpression or knockdown of PDE4D.The results shown that the sEV miR-139-5p/PDE4D axis plays an important role in ALF treatment (Figures 7h-k and S7c,d).Therefore, we believe that hUCMSC-BAL may improve ALF liver injury by inhibiting hepatocyte apoptosis through the sEV miR-139-5p/PDE4D/SIRT1/Bcl-2 pathway (Figure 8).

| DISCUSSION
In this study, we successfully constructed an hUCMSCs-based bioreactor and verified it using a porcine ALF model.The results showed that the hUCMSCs bioreactor significantly improved liver function in ALF, reduced inflammation and liver cell death, and prolonged ALF animal survival time.The pathogenesis of ALF is different at each stage; therefore, it is necessary to select appropriate treatment methods according to the relevant pathological changes.Our study demonstrated that the hUCMSCs bioreactor can significantly reduce liver cell death in the early stage of ALF by regulating the homeostasis and immune environment around normal liver cells through sEV, and that the sEV miR-139-5p/PDE4D axis may play an important role in this.In the middle and late stages of ALF, hepatocyte functional replacement therapy and homeostasis regulation can be used simultaneously for treatment.
Selection of the optimal time to begin treatment and duration of treatment is an important step in bioartificial liver therapy that has been debated in previous studies.For example, previous studies have reported optimal times to begin treatment of 2 h, 30 12 h, 13 24 h, 14,17 36 h, 31 and 48 h 12,15 after modeling, and optimal treatment durations of 3 h, 14,17 6 h, 12,13,31 8 h, 15,32 and 12 h. 33Theoretically, the earlier the treatment begins, the longer the treatment duration and the better the treatment effect.In this study, D-gal was used to induce the ALF model; therefore, the timing of bioartificial liver treatment should be selected after the drug is completely metabolized.A previous study showed that D-gal was completely absorbed 12 h after injection 14 ; thus, in our study, hUCMSC-BAL therapy was started 12 h after D-gal injection and continued for 8 h.
The seed cells and bioreactors are the two most important elements in bioartificial livers.At present, many seed cells are related to liver cell functional replacement therapy, including human hepatocytes, 34 liver tumor cell lines, 10 human-induced hepatocytes, 14,16 and porcine hepatocytes.environment for the growth and metabolism, substance exchange, and immune isolation of seed cells.Previous studies have shown that hollow fiber bioreactors have the disadvantage of insufficient oxygenation, which is not conducive to the growth of commonly used seed cells. 35However, because the tissue oxygen concentration of hUCMSCs is less than 5%, hypoxic culture in vitro can effectively simulate the in vivo microenvironment and help maintain the proliferation, metabolism, and other physiological processes of hUCMSCs.
Therefore, the low-oxygen environment of a hollow-fiber bioreactor is beneficial for the growth and proliferation of hUCMSCs.In this study, we constructed an hUCMSC bioreactor that can amplify hUCMSCs on a large scale (by up to 10 11 ) for clinical treatment, while maintaining the activity and function of hUCMSCs.Moreover, we detected the content of sEV proteins in the culture medium of the hollow fibers inside and outside the cavity, and the results showed no significant difference.hUCMSC-BAL was constructed to avoid adverse reactions caused by hUCMSCs infusion in patients and to allow hUCMSCs derivatives (sEV and free factors) to enter patients through the bioreactor semi-permeable membrane to play a therapeutic role.Therefore, hUCMSC-BAL therapy compensates for the difficulties and deficiencies in hUCMSCs transplantation and the isolation and purification of hUCMSCs-derived sEV.
HE is a syndrome characterized by impaired brain function caused by acute and chronic liver disease. 36Although the pathophysiology of HE is not fully understood, it is linked to the presence of inflammatory mediators and Amm. 37Impaired detoxification due to ALF leads to the accumulation of multiple toxins, which then induce the release of inflammatory cytokines, eventually leading to multiorgan failure. 38psis with multiorgan failure and HE are the two major causes of death in patients with ALF. 38In our study, HE syndrome was observed in group A at 48 h and in group B at 60 h, but not in group C. Furthermore, the levels of Amm, TBIL, and inflammatory mediators in group C after hUCMSC-BAL treatment were significantly lower than those in groups A and B. These results showed that hUCMSC-BAL can significantly reduce the levels of inflammatory mediators, improve the microenvironment of normal hepatocytes, avoid cytokine storms, protect the detoxification of hepatocytes, and prevent multiorgan failure and HE.
sEV are a type of EV formed through an endocytic process and released through a multivesicular body into the extracellular space; 3][44] In our study, we found that hUCMSC-BAL could release sEVs into the body through a bioreactor semipermeable membrane to rescue ALF.To explore the underlying mechanisms, we extracted sEV from the hUCMSCs-based bioreactor to treat the mouse ALF model and found that these sEV had an obvious protective effect on liver injury.Furthermore, high-throughput sequencing of hUCMSC-sEV showed that they are rich in a variety of miRNAs, among which miR-139-5p was the most abundant.Finally, we predicted the target genes of miR-139-5p and found that PDE4D was one of the main target genes that could inhibit cell apoptosis.
To further explore the effect of the miR-139-5p/PDE4D axis in ALF treatment, we used L02 and HepaRG cells to construct ALF models and then transfected them with miR-139-5p mimics, miR-139-5p inhibitors, and miR-NC.The results showed that miR-139-5p could significantly reduce the expression levels of PDE4D, SIRT1, p-AKT, Bax, and Caspase 3 as well as significantly increase the expression level of Bcl-2.Therefore, we believe that hUCMSC-BAL may improve ALF liver injury by inhibiting hepatocyte apoptosis through the sEV miR-139-5p/PDE4D/SIRT1/Bcl-2 pathway.Furthermore, we constructed UCMSCs with stable overexpression or knockdown of miR-139-5p and L02 and HepaRG cell lines with stable overexpression or knockdown of PDE4D.The results showed that the sEV miR-139-5p/PDE4D axis plays an important role in ALF treatment.
However, our study has some limitations.First, the small number of experimental animals may lead to certain errors owing to the small sample size; thus, the experimental results require further verification by increasing the sample size.Second, clinical ALF is an extremely complex disease, and the ALF model induced by D-gal may not completely simulate clinical ALF; thus, further clinical verification is required.Third, the time required for the large-scale expansion of hUCMSCs to the number of clinical treatments is relatively long; thus, further study should attempt to shorten the amplification time.
Fourth, subject to experimental conditions we only use male pigs in our manuscript which may produce some deviation due to gender.
Lastly, although we report the therapeutic effects of hUCMSC-BAL treatment, probably via sEV release, as well as the important role of the sEV miR-139-5p/PDE4D axis, the related mechanisms require further research.
In conclusion, we successfully developed an in vitro bioartificial liver based on hUCMSCs, which can inhibit inflammatory responses, regulate homeostasis to reduce liver cell death, promote liver cell regeneration, and prolong the survival time of animals with ALF.We think that the sEV miR-139-5p/PDE4D axis plays an important role in this process.Further randomized clinical trials are required to verify whether hUCMSC-BAL treatment has therapeutic benefits in patients with ALF.

| Isolation and culture of hUCMSCs
After obtaining approval from the Ethics Committee of Zhujiang Hospital and patient consent, fresh umbilical cord was obtained from puerpera.The isolation and culture of hUCMSCs were performed according to a previously described method. 45Briefly, umbilical cord tissue was subjected to type I collagenase digestion, centrifugation, filtration, and other steps until hUCMSCs were obtained.The culture medium was changed every 3 days.The cells were passaged upon reaching 80% confluency(Figure 1b).To ensure cell viability, this experiment only used cells from passages 3-5.

| Large-scale amplification of hUCMSCs in the hollow fiber bioreactor
The hUCMSC bioreactor comprises a hollow fiber bioreactor, peristaltic pump, liquid storage bottle, and connecting pipeline (Figure 1a).
The identified hUCMSCs were first inoculated into plate culture bottles, expanded to 10 8 cells using customized serum-free medium, and then inoculated into the bioreactor culture system.As the rate of blood GLU decline is directly proportional to the number of cells in the bioreactor, the culture medium should be changed regularly according to the sugar level of the circulating culture medium (when the sugar level is less than 1.5 mmoL/L or the culture time is > 5 days) until the hUCMSCs are amplified to 10 9 -10 10 .The culture medium in the inner and outer chambers of the bioreactor was collected and stored in the refrigerator at À80 C for standby application.into the internal jugular vein and femoral vein, respectively.These catheters were then used for blood collection, maintenance of anesthesia, fluid infusion, and BAL treatment pathway construction.A large dose of D-gal did not provide a sufficient treatment window, and a small dose did not induce ALF, but allowed the animal to recover by itself.Therefore, according to previous studies 14,46 and our previous research results, 15 we selected a dose of 0.40 g/kg to construct the ALF model.
After injection of D-gal (t = 0 h), the catheter was closed with heparin saline (2500 U/ml) to keep it unobstructed and allow the animals to wake up naturally from anesthesia.Clinical observations were conducted every 6 h during the experiment.Blood samples were collected every 12 h for blood detection (albumin [ALB], AST, ALT, blood urea nitrogen [BUN], TBIL, CRE, GLU, Amm, etc.) and coagulation function assessment (Drim-7000i, FUJI, and URIT).All samples were stored at À80 C for further testing.Animals in all groups received 500 mL of normal saline after D-gal injection.When the animals stopped feeding and drinking because of ALF, they received 5% dextrose normal saline (DNS5) at 50 mL/h to satisfy their basic water and energy needs.
Twenty milliliters of 50% dextrose was injected when the blood GLU level fell below 5 mmol/L.The experimental timeline is shown in Figure 2a.The study endpoint was animal death or survival for 168 h.

| Inflammatory factor detection
All blood samples were collected, centrifuged at 4000 rpm for 10 min to collect plasma, and stored at À80 C in a refrigerator.Cytokines were detected by ELISA kit (Table S3).

| Histopathology and immunohistochemistry
After the experimental animals died or reached the study endpoint, animal tissue samples (heart, liver, spleen, lung, kidney, large intestine, small intestine, and brain) were collected, fixed, embedded in paraffin, stained with H&E, and the liver tissue was stained by immunohistochemistry. Ki-67 and TUNEL staining was conducted according to the manufacturer's instructions.Professional pathologists used the Suzuki classification system to observe and score H&E sections of the liver tissue.The score range was 0-4 points for liver parenchymal necrosis, hepatic sinus congestion, and cytoplasmic vacuolization.Each pathologist examined three liver sections from each animal and randomly selected three visual fields from each section for scoring.The average score of each animal was determined by the sum of all scores.The positive results of Ki-67 and TUNEL staining were counted and defined as the regeneration and apoptosis indices, respectively.Masson, Sirius red, PAS, and Oil red staining was also performed.

| Transmission electron microscopy
The obtained liver tissue was divided into 1 Â 1 cm tissue blocks and fixed in glutaraldehyde + paraformaldehyde electron microscopy solution for 10 min.The tissue blocks were then divided into 2 Â 2 mm blocks and fixed in electron microscopy solution.After overnight dehydration at 4 C, the slices were scanned using a transmission electron microscope.

| hUCMSC-sEV extraction and identification
Differential ultracentrifugation was used to extract the sEV.Briefly, centrifugal forces of 300, 2000, 10,000, and 100,000Âg were used to centrifuge the cell culture medium.The final pellet was resuspended in 200 μL of PBS.Transmission electron microscopy was used to observe the shape of the sEV.NanoView was used to measure the concentration and particle-size distribution of sEV.Western blotting was used to measure the positive protein markers CD9, TSG101, CD81, syntenin, and CD63, and the sEV negative protein marker calnexin after employing a BCA protein quantification kit to determine the protein concentration of sEV according to the manufacturer's instructions (Table S1).

| Fluorescence imaging tracing of sEV in vitro and in vivo
To identify the transport of sEV in vitro and in vivo, sEV were labeled with DIR or PHK26 fluorescent dye.DIR-labeled sEV were injected through the tail vein of mice, and their transport in vivo was observed within 48 h.PBS was injected as a control.PHK26-labeled sEV were added to the medium of L02 and AML-12 cells and observed under a fluorescence microscope within 48 h.

| hUCMSC-sEV treatment of the mouse ALF model
Mice in the experimental group (sEV group) and control group (PBS group) were intraperitoneally injected with D-gal (800 mg/kg, Sigma-Aldrich) and LPS (50 μg/kg, Sigma-Aldrich), respectively, to construct the ALF model.The sEV group was injected with hUCMSC-sEV via the tail vein 24 h before modeling, whereas the PBS group was injected with the same amount of PBS solution.The mice were sacrificed 1 and 3 h after modeling, and samples were collected for the relevant tests.
After anesthesia, cardiac blood was collected from the mice, allowed to stand for 2 h, and then centrifuged at 3500 rpm for 10 min.The supernatant was retained and stored at À80 C for later use.Serum ALP, ALT, AST, and BUN levels were determined using a dry biochemical analyzer (Catalyst One; IDEXX Laboratory).
After the experimental animals died or reached the study endpoint, animal liver tissue samples were collected, fixed, embedded in paraffin, and stained with H&E, and the liver tissue was stained with the relevant immunohistochemistry stains (Ki-67 and TUNEL) according to the manufacturer's instructions.
For coagulation function, PT, INR, and APTT increased at 12 h after D-gal administration, whereas Fib began to decrease.Because of heparinization during hUCMSC-BAL treatment, APTT and PT increased significantly, and Fib decreased significantly at 24 h, but recovered at 36 h in groups B and C. The coagulation function (PT, INR, APTT, and Fib) of the surviving animals in groups B and C reached a peak, and then returned to baseline, whereas the coagulation function of dead animals in groups A and B continued to worsen (Figure 2e).Overall, hUCMSC-BAL treatment reduced blood Amm, improved liver function and coagulation function, and stabilized the homeostasis environment in the D-gal-induced ALF porcine model.

F
I G U R E 4 hUCMSC-BAL treatment could reduce cell apoptosis and HE and promote liver regeneration.(a) TUNEL staining of liver tissue of ALF animals in three groups, scale bar 200 μm.(b) H&E staining of liver tissue of ALF animals in the three groups, scale bar 200 μm, plus illustration at high magnification.(c) Ki67 staining of liver tissue of ALF animals in the three groups, scale bar 200 μm, plus illustration at high magnification.(d) Liver injury score of the three groups; (e) Comparison of Ki67-positive cells in the three groups; (f) Comparison of TUNELpositive rates among the three groups; (g) Electron microscopic of liver tissue of ALF animals in the three groups; scale bar 200 μm.(h) H&E staining of brain tissue of ALF animals in the three groups, scale bar 200 μm, plus illustration at high magnification.(i) Brain water content in the three groups; (j) Dynamic changes in blood ammonia in the three groups (n = 5).*p < 0.05 using unpaired Student's t-test or one-way ANOVA; Group A: ALF + ST; Group B: ALF + ST + low dose hUCMSC-BAL; Group C: ALF + ST + high dose hUCMSC-BAL; HE: Hepatic encephalopathy; H&E: Hematoxylin-eosin.F I G U R E 5 Legend on next page.

FENG
5 hUCMSC-sEV identification and tracing in vivo and in vitro.(a) hUCMSC-sEV acquisition and identification diagram.(b) sEV size according to NanoView.(c) Representative fluorescence images of different capture spots demonstrating colocalization of CD63, CD81, CD9, and mIgG on a single sEV.(d) sEV concentration detection at different capture spots.(e) Results of electron microscopic examination of sEV.(f) Western blotting results of sEV markers.(g) Comparison of particle size, concentration, and protein content of secretions in the internal and external chambers of the bioreactor.(h) In vivo fluorescence imaging tracing of hUCMSC-sEV.(i) In vitro fluorescence imaging tracing of hUCMSC-sEV.F I G U R E 6 Evaluation of hUCMSC-sEV treatment in the mouse ALF model.(a) Experimental design timeline.(b) Changes in liver and renal function indexes at 1 h after modeling between the two groups for ALT, AST, ALP, and BUN (n = 3 in each group).(c) Changes in liver and renal function indexes at 3 h after modeling between the two groups for ALT, AST, ALP, and BUN (n = 3 in each group).(d) H&E and Ki67 staining of liver tissue at 1 and 3 h after modeling between two groups.(e) TUNEL staining of liver tissue at 1 and 3 h after modeling between two groups.H&E: Hematoxylin-eosin; *p < 0.05 by unpaired-tailed Student's t-test.F I G U R E 7 Legend on next page.
2-1.0 years old and weighing 20-30 kg, were purchased from Dongguan Songshan Lake Pearl Laboratory Animal Science and Technology Co., Ltd., license No. SCXK (Guangdong) 2017-0030.Twelve male C57/BL6 mice, 10-15 weeks old and weighing 25-35 g, were purchased from the Experimental Animal Center of Southern Medical University.Mice were kept in cages at 22 ± 1 C and exposed to light for 12 h (from 06:00 to 18:00) and dark for 12 h.All experimental animals were isolated and caged, fed with special food, and allowed to drink freely.All animals were fed adaptively for 1 week and then fasted for 12 h before experimentation.Animal experiments were approved by the Animal Ethics Committee of Zhujiang Hospital, Southern Medical University (LACE-2020-155).
Propofol was injected via a central venous catheter (0.15-0.20 mg/kg/min) to maintain slight anesthesia, and the blood pressure, heart rate, and respiratory rate were monitored during hUCMSC-BAL treatments.The hUCMSC-BAL device was connected to the ALF model pigs via a dual-chamber hemodialysis catheter inserted into the jugular and femoral veins (Figures2b and S1).Saline (500 mL) was injected before the treatment to prevent hypotension.
4.3.3|hUCMSC-BAL treatmenthUCMSC-BAL treatment was initiated 12 h after D-gal administration for 8 h.Pentobarbital sodium and sumianxin II were used to induce basal anesthesia.