Decreased miR-17-92 Cluster Correlates With the Senescent Features and Disrupted Oxidative Homeostasis, and the Impaired Therapeutic E cacy of Adipose Tissue-derived Mesenchymal Stem Cells

Yelei Cen Zhejiang University First A liated Hospital State Key Laboratory for Diagnosis and Treatment of Infectious Diseases Jinjin Qi Zhejiang University First A liated Hospital State Key Laboratory for Diagnosis and Treatment of Infectious Diseases Liang Chen Zhejiang University School of Medicine First A liated Hospital Yanning Liu (  liuyanning@zju.edu.cn ) Zhejiang University First A liated Hospital State Key Laboratory for Diagnosis and Treatment of Infectious Diseases https://orcid.org/0000-0002-0554-6622 Guohua Lou Zhejiang University First A liated Hospital State Key Laboratory for Diagnosis and Treatment of Infectious Diseases


Background
Acute liver failure (ALF) is a life-threatening clinical syndrome characterized by rapid hepatocellular necrosis due to various acute injuries induced by hepatotoxic drugs, immune-mediated attack, or viral infections. Liver transplantation and arti cial liver therapies are presented as the main clinical treatments for ALF [1]. However, the shortage of available donor livers limits the widespread clinical application of liver transplantation, and multiple postoperative complications also limit the e cacy of arti cial liver therapy against ALF [2]. Increasing number of studies showed that mesenchymal stem cells (MSCs) can effectively treat ALF [3]. We have also determined the therapeutic e cacy of adipose tissue-derived mesenchymal stem cells (AMSCs) or AMSC-derived exosomes on various liver diseases including liver brosis, hepatocellular carcinoma, and ALF [4,5]. However, there are still some issues should be addressed before clinical application of MSCs on ALF, especially in the context of elderly donors. Several studies have shown that a decrease in the in vitro proliferation capacity and multidirectional differentiation potential of human bone marrow-derived MSCs with the increase in the age of the donors [6]. In addition, AMSCs from aged donors [7], as well as from individuals suffering from metabolic disorders [8] showed a diminished regenerative potential. Further study showed that MSCs isolated from aged donors exhibit senescent features, resulting in more senescence-associated β-galactosidase (SA-βgal)-positive cells and increased secretion of senescence-associated secretory phenotype (SASP) cytokines [9]. Although many studies have been shown the aging-associated cell senescence of MSCs and reported the reduced therapeutic potential of AMSCs from aged donors, the exact mechanism underlying the agingassociated cell dysfunction and therapeutic role of aged AMSCs on ALF remains unclear. In the current study, we compared the therapeutic e cacy of AMSCs from aged mice to those from young mice on ALF, and then investigate the potential mechanisms contributed to the aging-associated cell senescence of AMSCs. Finally, we explored the modi cation strategy to reverse the aging of AMSCs. Our study may provide an indicator for evaluating the AMSC quality for clinical usage and also provide a promising approach for improve the therapeutic e cacy AMSCs on ALF.

Methods
Isolation and identi cation of AMSCs C57BL/6J mice were purchased from Nanjing BioMedical Research Institute of Nanjing University. The experimental procedures were reviewed and approved by the Institutional Animal Care and Use Committee in our institution. Inguinal adipose tissues were obtained from male mice (8 weeks old as young mice, 20 months old as aged mice). Adipose tissues were digested with 0.075% collagenase type I (Sigma-Aldrich, St Louis, MO, USA) for 30 min, then centrifuged and washed with PBS for 2 times. Finally, the isolated cells were cultured in murine MesenCult TM Expansion Kit (Stemcell) containing 2 mM Lglutamine (Thermo Fisher Scienti c) and 1% antibiotic-antimycotic. Cells were maintained and expanded by 2-10 passages before usage. ALF mouse model and AMSCs treatment C57BL/6J Mice (aged 5-6 weeks) were intraperitoneally injected with Lipopolysaccharide (LPS, 10 µg/kg, Sigma-Aldrich) and D-galactosamine (GalN, 400 mg/kg, Sigma-Aldrich) to establish a mouse model of ALF. AMSCs from young mice (yAMSC) or aged mice (oAMSC), P2, P6 or P10 AMSCs from a culture-induced senescent cell model, and miR-17 or miR-20a-modi ed oAMSCs (oAMSC-M17, oAMSC-M20a) or control miRNA-modi ed oAMSCs (oAMSC-M67) were administered via the tail vein (2×10 5 , n = 6) immediately after LPS/GalN injection. The control group was administered with vehicle alone (n = 6).
At 6 h after LPS/GalN injection, the mice were sacri ced, and serum and liver samples were collected.
Serum was evaluated for biochemical parameters. The liver samples were evaluated for histochemistry and Western blot analysis.

Isolation and detection of miRNA
Total RNA enriched with miRNAs was isolated from yAMSCs, oAMSCs and different passaged AMSCs by using a miRVana miRNA isolation kit (Thermo Fisher) according to the manufacturer's instructions.
Complementary DNA was synthesized from the isolated miRNAs by using TaqMan™ mmu-miRNA-speci c primers including miR-17-19 cluster (Thermo Fisher) and TaqMan™ MicroRNA Reverse Transcription Kit (Thermo Fisher). Real-time PCR was then performed following the manufacturer's instructions (Thermo Fisher) to examine the expression of miR-17-19 cluster, especially miR-17 and miR-20a. Data were normalized over the average cycle threshold (CT) value of U6, and the 2 −ΔΔCT method was used to determine the relative miRNA expression.

RNA isolation and qPCR
Total RNA was isolated from AMSCs by using TRIzol. First-strand cDNA was subsequently synthesized followed by qPCR analysis by using ABI Prism 7900 (Applied Biosystems) to examine the expression levels of TNF-, IL-6, MCP1, PAI-1, Mmp3 and Mmp13, β-actin were used as internal controls. The 2 −ΔΔCT method was used to determine the relative mRNA expression levels of these genes.

Liver histological and serum biochemical analysis
Liver tissues were processed for para n embedding and were sectioned into 4 µm sections. The sections were then routinely stained with hematoxylin and eosin (H&E staining). The serum levels of alanine aminotransferase aspartate (ALT) and aspartate aminotransferase (AST) were analyzed by standard analyzer DRICHEM 4000ie (FUJIFILM). Total bilirubin (TBiL) was measured using standard clinical chemistry techniques (Integra II; Roche, UK).

Cytokine detection by ELISA
The murine serum levels of the cytokines including TNF-α, IL-6, and MCP-1 (MultiSciences, Shanghai, China) were assayed by commercial ELISA kits following the manufacturer's instructions.
SA-β-gal staining SA-β-gal staining was used to determine cell senescence of AMSCs. Brie y, yAMSCs, oAMSCs and different passaged AMSCs were seeded at a density of 5 × 10 4 /well in a 6-well plate with complete medium and incubated for 24 h at 37°C in 5% CO 2 . Afterwards, the cells were xed with 4% paraformaldehyde (PFA, Sigma-Aldrich, USA) at room temperature for 20 min and subsequently stained by SA-β-gal staining kit (Beyotime, Shanghai, China) following the manufacturer's instructions. Positive senescent cells stained in blue were observed using an inverted microscope (AxioCam ERc 5s, Carl Zeiss, Germany).

Determination of reactive oxygen species (ROS) generation
AMSCs were stained with 5 µM H2DCF-DA (Life Technologies) for 10 min and then washed twice with ice-cold PBS and ROS generation was determined by FACS analysis. ROS generation was quanti ed as the increase in mean uorescence intensity (MFI).

Transfection of small interfering RNA (siRNA) and miRNA
SiRNA against c-Myc, TXNIP and scrambled siRNA were purchased from Guangzhou RiboBio Co., Ltd. and miR-17 and miR-20a mimics, miRNA mimic negative control (M-Ctrl), miR-17 and miR-20a inhibitors, and miRNA inhibitor negative control (I-Ctrl) were also purchased from RiboBio. RNAiMAX (Thermo Fisher Scienti c) was used for siRNA and miRNA transfection. At 75% con uence, AMSCs were transfected with siRNA (100 nm) and/or miRNA (50 nM) at 37˚C. At 48 h after transfection, the cells were harvested for qPCR and Western blot analysis.
Thioredoxin (TRX) activity assay TRX activity was measured by the Thioredoxin Fluorometric Activity Assay Kit (Cayman, Item No.500228) according to the manufacturer's instruction. Brie y, 0.5-1 mg/ml of cell lysates were prepared in cold buffer (100 mM Tris-HCl pH 7.5, 1 mM EDTA), supplemented with protease inhibitor. Then 20µl cell lysates were mixed with 10 µl TRX reductase, 10 µl NADPH and 40 µl assay buffer. After the reaction was performed at 37°C for 30 min, 20µl reconstituted Eosin-Labeled Insulin was added. Measure the uorescence at excitation and emission wavelength of 520 nm and 560 nm, once every minute for 60 min by Synergy Neo2 (BioTek).

Construction of miRNA-modi ed AMSCs
The miR-17 (LV-miR-17) and miR-20a (LV-miR-20a) expression vectors and the control vector cel-miR-67 (LV-cel-miR-67) were constructed by cloning their precursor sequences, together with a short upstream and a short downstream anking region, into lentiviral expression vectors. Before infection, 1 × 10 6 AMSCs were seeded in 10 ml complete medium overnight and then transfected with 100 nM LV-miR-17, LV-miR-20a or LV-cel-miR-67 at a multiplicity of infection of 10 in the presence of polybrene (8 µg/ml; Sigma-Aldrich, USA) for 24 h. Stable transfectants were used in the subsequent experiments.

Statistics
Statistical evaluation was performed using independent samples t-tests between two groups and using one-way analysis of variance with Tukey's post hoc test between three or more groups. All data are presented as the mean ± SD. P < 0.05 was considered signi cant.

Results
AMSCs from aged mice show an abated protective effect on acute liver failure We rstly determined the therapeutic e cacy of AMSCs in LPS/GalN-induced ALF mouse model. As shown in Figure 1A and B, LPS/GalN induced signi cant liver injury, as indicated by dramatically elevated serum ALT, AST, and TBiL levels accompanied by massive hepatic necrosis, while administration of the AMSCs from young mice (yAMSC) resulted in a remarkable reduction in serum ALT, AST, and TBiL levels ( Fig. 1A) and less hepatic necrosis (Fig. 1B). The elevated serum levels of in ammatory cytokines, such as TNF-α, IL-6, and MCP-1 (Fig. 1C), in LPS/GalN-injected mice were also signi cantly reduced by yAMSCs administration compared with vehicle-treated mice at 6 h after LPS/GalN injection. However, administration of the AMSCs from old-aged mice (oAMSC) showed an abated protective effect on ALF as compared to yAMSCs administration ( Fig. 1A-C). These data indicate that aging-related change of cell characteristics may affect the therapeutic e ciency of AMSCs on ALF. AMSCs from aged mice show senescent features To determine whether the different e ciency of AMSCs on ALF related to stem cell aging, we analyzed the morphology and cell senescent features of yAMSCs and oAMSCs at passage 2. YAMSCs displayed a characteristic spindle-like morphology, whereas oAMSCs were larger and irregular ( Fig. 2A). There were more SA-β-gal-stained cells in cultured oAMSCs as compared to yAMSCs (Fig. 2B). Moreover, CCK-8 assay showed that the proliferation capacity of oAMSCs was signi cantly lower than that of yAMSCs ( Fig. 2C). Furthermore, the protein levels of cell senescent molecular markers, p16 and p21, in cultured oAMSCs were signi cantly increased than that in yAMSCs (Fig. 1D). The mRNA levels of SASP markers including Mmp3, Mmp13, PAI-1, MCP-1, TNF-, and IL-6 were also remarkably increased in oAMSCs than that in yAMSCs (Fig. 1E). These data demonstrated that the AMSCs from aged mice displayed a senescent phenotype, compared to the AMSCs from young mice.
MiR-17-92 cluster associates with the cellular senescence of AMSC by targeting p21 MiRNA has been found to play critical roles in many age-related diseases, including cardiovascular and neurological diseases [10]. Differentially expressed miRNAs in cellular senescence has already been reported in cultured broblasts [11]. However, their role in MSC senescence and aging has not been fully characterized. Since among the reported cell senescence-related miRNAs, miR-17, 18a, 19a, 19b-1, 20a, and 92a-1, are all derived from one miRNA cluster, the miR-17-92 cluster (Fig. 3A), we then examined whether this miRNA cluster is related to AMSC senescence. RT-qPCR analysis con rmed that the expression levels of the miR-17-92 cluster miRNAs, especially miR-17 and miR-20a, were obviously decreased in oAMSCs compared to that in yAMSCs (Fig. 3B). The downregulated levels of miR-17 and miR-20a were also observed in the culture-induced senescence of AMSCs by comparing their levels between the passage 2 (P2) and the P7 of AMSCs (Fig. 3C). We next investigated whether the miR-17 and miR-20a by themselves were su cient for overcoming senescence of AMSC. The proliferation capacities of yAMSCs transduced with miR-17 or miR-20a inhibitors were signi cantly decreased as compared to those transduced with control inhibitors (Fig. 3D). While, the accumulation of SA-β-gal in the miR-17 or miR-20a inhibitor-transfected yAMSCs were markedly increased (Fig. 3E). These data indicate that the decreased miR-17-92 cluster maybe related to the cellular senescence of AMSCs and miR-17 and miR-20a are both necessary and su cient for affecting AMSC senescence, and thus are the key components mediating the anti-senescence activity of miR-17-92. Since the miR-17-92 cluster is recognized as cell cycle regulators via direct targeting of p21 [12], encoded by the CDKN1A gene (Fig. 3F), we then investigated the regulation of miR-17 and miR-20a on p21 by the transfection with their mimics or inhibitors. As anticipated, overexpression of miR-17 and miR-20a by transfection with their mimics both reduced p21 protein levels in oAMSCs (Fig. 3G). On the contrary, by transfection with miR-17 and miR-20a inhibitors, the p21 protein levels were increased accordingly in yAMSC (Fig. 3H). These data strongly indicate that the upregulated expression of p21 in senescent AMSCs may be caused by the reduced expression of miR-17-92 cluster.
c-Myc-regulated miR-17-92 is involved in the senescence of AMSC It is known that miR-17-92 can be directly trans-activated by c-Myc, which is related to the stemness of variety of cells [13]. We therefore hypothesized that c-Myc-regulated miR-17-92 maybe involved in the senescence of AMSC. As anticipated, the expression level of c-Myc was higher in yAMSC than that in oAMSCs and was also remarkably decreased from P2 to P7, a culture-induced senescent cell model of AMSCs (Fig. 4A). The decreased c-Myc level was consistent with the downregulated miR-17-92 level during cellular senescence. To further consolidate the role of c-Myc in controlling miR-17-92, we speci cally inhibited c-Myc expression by siRNA in AMSCs (Fig. 4B). We found that down-regulation of c-Myc reduced miR-17 and miR-20a levels whereas increased p21 level and SA-β-gal accumulation in AMSCs and subsequently reduced the proliferation capacity of AMSCs (Fig. 4B-D). In addition, by further transfection of miR-17 or miR-20a mimics into AMSCs, we found that c-Myc siRNA-induced upregulation of p21 and -reduced proliferation capacity could be markedly reversed by increasing in miR-17 or miR-20a levels ( Fig. 4E and F). These observations indicated that c-Myc-induced miR-17-92 cluster plays a dominant role in the senescence of AMSC by regulating p21.
Decreased miR-17-92 cluster associates with the dysregulated redox system in senescent AMSCs Disrupted oxidative homeostasis has been characterized in senescent MSCs. High ROS level is responsible for age-related loss of cellular functions and also represents an important cause of cellular senescence [14]. We found that the ROS level was higher in oAMSCs than that in yAMSCs and was also increased in the AMSCs from P2 to P7 (Fig. 5A). Redox regulation by TRX plays a crucial role in responses to oxidative stress [15]. As shown in Fig. 5B, the expression lever of TRX showed no obvious difference between oAMSCs and yAMSCs or between P2 and P7 AMSCs, whereas the activity of TRX was markedly decreased in the aging or replicative senescent AMSCs (Fig. 5C).
Our previous study showed that miR-17 could suppress the expression of thioredoxin-interacting protein (TXNIP) by directly targeting its mRNA 3' UTR (Fig. 5D). Since TXNIP is known as a negative regulator of TRX [15], we guessed the abnormal redox status and the decreased TRX activity in senescent AMSCs may due to the disrupted suppression of TXNIP by miR-17, which was declined in senescent AMSCs. As expected, TXNIP expression was upregulated in oAMSCs and P7 AMSCs compared to that in yAMSCs and P2 AMSCs, respectively (Fig. 5B). In addition, reduce miR-17 and miR-20a levels by transfection with their inhibitors could upregulate TXNIP expression consistent with the reduction of TRX activity in AMSCs, whereas miR-17 and miR-20a mimic transfection could downregulate TXNIP expression consistent with an increase in TRX activity in AMSCs ( Fig. 5E and F). To further determine the role of TXNIP in regulating the oxidative homeostasis of AMSCs, we transfected a siRNA against TXNIP into oAMSCs and then examined the ROS level and the TRX activity. TXNIP knock-down partly reversed the increased ROS level and restored the TRX activity in oAMSCs, further substantiating a role for TXNIP in redox regulation (Fig. 5G-I). Moreover, TXNIP knock-down also reversed the decreased TRX activity and the enhanced ROS level in AMSCs by miR-17 or miR-20a inhibitor transfection ( Fig. 5J and K). This clearly indicates that the dysregulated redox system in senescent AMSCs may due to the abnormal expression of TRX inhibitor TXNIP, which is negatively regulated by miR-17-92 cluster.
MiR-17 and miR-20a levels correlated with the therapeutic e cacy of AMSCs on acute liver failure The above studies have shown that miR-17-92 cluster, especially miR-17 and miR-20a, is associated with AMSC senescence. We then determine whether the reduced therapeutic e cacy of senescent AMSCs on ALF was correlated with the decreased miR-17-92 level. The miR-17-92 cluster level of AMSCs was detected at passage 2, 6, and 10 by qPCR. As anticipated, the miR-17 and miR-20a levels were gradually declined from P2 to P10 (Fig. 6A). Administration of P2 AMSCs had better protective effects on LPS/GalN-induced ALF as compared to P6 AMSCs-treated mice but administration of P10 AMSCs, no protection against liver injury were evident as determined by serum ALT, AST and TBiL levels (Fig. 6B) and H&E staining (Fig. 6C).
To further determine whether miR-17-92 cluster, especially miR-17 and miR-20a, modi cation can improve senescent AMSCs therapeutic e cacy on ALF, we constructed miR-17-or miR-20a-modi ed AMSCs by infection of LV-miR-17 (oAMSC-M17) and LV-miR-20a into oAMSCs (oAMSC-M20a). oAMSCs infection with LV-cel-miR-67, which has no known mRNA binding targets in mouse, was served as control (oAMSC-M67). As shown in Fig. 6E and F, the pathologic alteration and the upregulated serum ALT, AST, and TBIL induced by LPS/GalN could more effectively attenuated by oAMSC-M17 or oAMSC-M20a administration compare to oAMSC-M67 administration. These data suggest that the declined miR-17-92 cluster level may be correlated with the diminished therapeutic e cacy of senescent AMSCs on ALF and modi cation with miRNAs belong to this cluster may act as a strategy to enhance the therapeutic potential of AMSCs.

Discussion
Recently, MSCs are considered to be a highly attractive therapeutic strategy on variety of liver diseases including ALF, whereas, MSCs isolated from aged individuals exhibit a reduced proliferative activity and differentiation potential, as well as senescent features, limits their use in autologous transplantation [16].
In addition, in vitro expansion of primary MSCs for enough transplantation cells may also leads to cellular replicative senescence and loss of multipotency [17]. The altered protein expression of senescent MSCs is well studied and characterized. While, the role of miRNAs in MSC senescence and aging have not been fully characterized. In the present study, we showed that the therapeutic e cacy of AMSCs on ALF was markedly reduced in the AMSCs from old-aged mice or high passage, which exhibiting senescent features. Then we found that the decrease in miR-17-92 cluster levels may be associated with AMSC senescence by affecting the expression of classic senescence protein and oxidative homeostasis in AMSCs. Furthermore, modi cation of AMSCs with miR-17 and miR-20a, the two key miRNAs in miR-17-92 cluster, could reverse the senescent features of oAMSCs and also remodeling the therapeutic potential of senescent AMSCs on ALF.
Increasing studies showed that miRNAs are key regulators and reliable markers of cellular senescence [10,18]. Numerous dysregulated miRNA have been identi ed in senescent MSCs and known to correlate with many age-related diseases. For example, miR-486-5p was increased in aged MSCs and could inhibit osteogenic and adipogenic differentiation and induce senescence of MSCs by targeting SIRT1 [19]. In addition, the increased miR-335 in aged MSCs as well as in γ-irradiation-induced senescent MSCs was known to effect MSC senescent features through inhibition of AP-1 activity [20]. In this study, we found that the miRNA expression patterns, including miR-17-92 cluster, between the AMSCs from old-aged mice or young mice were markedly different (data not shown). qPCR analysis further showed that the levels of miR-17 and miR-20a, both derived from the miR-17-92 cluster and belonging to the miR-17 family, were decreased more obviously in aged AMSCs and replicative senescent AMSCs. Thus, we choose miR-17-92 cluster, especially miR-17 and miR-20a, for further study. The miR-17-92 cluster has been rstly identi ed as an oncogenic miRNA cluster [21]. Overexpression of miR-17-92 has been detected in various types of human cancer and known to promote cell cycle progression and proliferation, inhibit apoptosis, and induce tumor angiogenesis [21]. Recently, miR-17-92 cluster is increasingly concerned in aging research and the decrease in miR-17 family has been observed in human aging and senescent MSCs. We further showed that declined miR-17-92 cluster level was contributed to the concomitant up-regulation of p21 during AMSC senescence. Thus, via direct targeting the key cell cycle protein p21, the miR-17-92 cluster seems to maintain cell proliferation and survival, which is correlated with the progenitor/regenerative state of AMSCs.
We next investigated how the miR-17-92 cluster expression was regulated. The oncogene c-Myc, also known as a stemness gene, has previously been reported to regulate cell proliferation, survival, apoptosis, and senescence, several of the key phenotypes associated with the addiction of miR-17-92 cluster [13,22]. We found that the decreased c-Myc level was consistent with the downregulated miR-17-92 level during AMSC senescence. Moreover, c-Myc knock down-induced upregulation of p21 and -reduced cellular proliferation could be markedly reversed by increasing in miR-17 or miR-20a levels in AMSCs. We therefore deduced that c-Myc-regulated miR-17-92 is involved in the regulation of p21 expression during AMSC senescence and miR-17 and miR-20a are the key miR-17-92 components for su cient abrogation of c-Myc reduction-induced p21 expression.
In addition to the regulation of p21 expression in cellular senescence, miR-17-92 cluster has been known to affect senescence by regulating several other genes, such as Smurf1 and SOSC1 [23,24]. Recently, disrupted oxidative homeostasis and high ROS level have been characterized in MSC senescence. TXNIP is the α-arrestin family protein that is induced by oxidative stress and is known to inhibit the key antioxidant protein TRX via a direct interaction [15]. Previous studies showed that age-dependent upregulation of TXNIP resulted in decreased stress resistance to oxidative challenge in primary human cells and in Drosophila [25]. However, the mechanism in regulating TXNIP expression during cell senescence is unclearly known yet. Our previous study has already con rmed that miR-17 could directly target and suppress TXNIP expression. Thus, the abnormal redox status and the decreased TRX activity in senescent AMSCs may also due to the abrogated suppression of TXNIP by decreased miR-17. As expected, TXNIP expression was upregulated concomitant with the decreased miR-17 and miR-20a levels, the decreased TRX activity, and the enhanced ROS levels in senescent AMSCs. Moreover, TXNIP interference reversed miR-17 or miR-20a inhibitor transfection-induced TRX activity reduction and ROS elevation in AMSCs. These data further indicate that miR-17-92-mediated TXNIP and redox system regulation also play important role in AMSC senescence.
Finally, we con rmed that administration of miR-17-or miR-20a-modi ed AMSCs could more effectively attenuate LPS/GalN-induce ALF as compared to the administration of control miRNA-modi ed AMSCs. Besides its role in mediating oxidative stress, TXNIP is also known to mediate NLRP3 in ammasome activation. Our previous study showed that exosome-mediated miR-17 shuttling is involved in the therapeutic effects of AMSC-derived exosomes against ALF by targeting TXNIP and consequent NLRP3 in ammasome blockage [26]. Thus, the improved therapeutic e cacy of AMSCs by miR-17 or miR-20a modi cation might also due to the more effectively inhibition on NLRP3 in ammasome activation. Additionally, excessive ROS generation contribute to the occurrence and development of liver failure. Thus, the miR-17-or miR-20a-modi ed AMSCs may also more effectively suppressed ROS level by exosomal miRNA-mediated TXNIP-TRX intervention.

Conclusions
This study showed that the decreased level of miR-17-92 cluster, due to the reduced expression of c-Myc, was associated with the senescent features of AMSCs and the reduced therapeutic e cacy of AMSC on ALF. Thus, the cellular miR-17-92 cluster level, especially the miR-17 and miR-20a levels, can be used as an index to evaluate the therapeutic potential of AMSCs. Moreover, miR-17-92 cluster miRNAsmodi cation may provide a new strategy for reversing the cellular senescence and improving the therapeutic e cacy of AMSCs.

Consent for publication
Not applicable.

Availability of data and materials
All data and materials are available upon request.

Competing Interests
The authors have declared that no competing interest exists.