Extracellular vesicles derived from human embryonic stem cell-MSCs ameliorate cirrhosis in thioacetamide-induced chronic liver injury

2.1 Generation and characterization of ES-MSCs

Human embryonic stem cell line, Royan H6 (Baharvand et al., 2006) (Royan Stem Cell Bank, Tehran, Iran) was cultured on matrigel-coated plates in Dulbecco's modified Eagle's medium/F12 supplemented with 20% knockout serum replacement, 2 mM GlutaMAX, 0.1 mM non-essential amino acids (NEAAs), 1% insulin-transferrin-selenium (ITS) (all from Life Technologies, Invitrogen, Carlsbad, CA), 0.1 mM β-mercaptoethanol (Sigma-Aldrich, St. Louis, MO) and 100 ng/ml basic fibroblast growth factor (bFGF, Royan Biotech, Tehran, Iran) as described before. Every other day, the medium was changed and every 5 days the colonies were splitted using collagenase IV (Sigma-Aldrich). In order to induce ESCs differentiation into MSCs, spontaneous differentiation was performed. Briefly, ESC colonies were dissociated using 0.05% trypsin/EDTA (Invitrogen) and plated on 6-cm² non-adherent culture petri dish at 2 × 10⁵ cells/cm² in ESC medium supplemented with 10 μM ROCK inhibitor Y-27632 (Calbiochem, Darmstadt, Germany) in the absence of bFGF. The medium was refreshed every other day to generate embryoid bodies (EBs) after 7 days. Then, generated EBs were plated on gelatin-coated plates in Alpha Modified Eagle's Medium (α-MEM, Life technologies) supplemented with 10% fetal bovine serum (FBS, Hyclone) and 2 mM L-glutamine for 7–10 days. The outgrowing differentiated cells were collected and subsequently re-plated and cultured for additional days. Then, resulted ES-MSCs of passage 0 were passaged at 80% confluency until 16 passages. Expanded ES-MSCs of passage 3 were used for in-vitro and in-vivo experiments. The conditioned medium of ES-MSCs of passage three to seven was collected for EV purification.

For growth rate, 1 × 10⁴ cells/cm² were cultured in T25 cm² tissue culture flasks (TPP, Germany). Doubling time was calculated according to the following formula:

The human BM- and AD-MSCs were provided by Royan Stem Cell Bank and cultured in Alpha Modified Eagle's Medium (α-MEM, Life technologies) supplemented with 10% fetal bovine serum (FBS, Hyclone) and 2 mM GlutaMAX at the same density.

2.2 Multilineage differentiation

Osteogenic and adipogenic differentiation capacity of ES-MSC were evaluated 14 and 21 days post-incubation in osteogenic and adipogenic induction medium, as described before (Lotfinia et al., 2016).

2.3 Mixed lymphocyte reaction (MLR) assay

To evaluate immunomodulatory effect of ES-MSCs, 1 × 10⁵ cells were exposed to mitomycin-C (10 μg/ml) for 2 hr. Then, mitomycin-C-treated ES-MSCs were seeded on a 96-well plate for 24 hr. In addition, we isolated PBMCs from two HLA-mismatched donors by Ficoll-Paque™ density gradient protocol. As we described before (Pourgholaminejad et al., 2016), the proliferation of allogenic stimulator PBMCs (S), was suppressed with mitomycin-C (25 μg/ml) following 45 min treatment. Responder PBMCs (R) were incubated with 25 μM Carboxyfluorescein succinimidyl ester (CFSE) (Molecular probe, Invitrogen) for fluorescence labeling. For MLR assay, three experimental groups were considered (i.e., R, R + S, R + S + ES-MSCs). For group R, only 1 × 10⁵ responder PBMCs were cultured as control negative. In group R + S, both stimulator and responder cells were co-cultured together and considered as positive control. In R + S + ES-MSCs group, we co-cultured R + S in the presence of pre-seeded ES-MSCs in which the proliferation of responder PBMCs was detected based on the dilution of CFSE, using flow-cytometry analysis (BD FACS caliber™ cytometer and FlowJo 7-6-1 software). In order to show the EVs-induced immunosuppression, we performed another experiment in which 10 μg of ES-MSC EVs were used instead of cells. Finally, three groups of R, R + S and R + S + ES-MSC EVs were compared to each other regarding the percentage of the proliferated PBMCs.

2.4 Purification and characterization of EVs

To isolate EVs, passage 3–7 of ES-MSCs were cultured in T150 culture flasks up to 70% cell confluency. Then, FBS-enriched medium was replaced with MSC basal medium supplemented with EV-free FBS plus 2 mM GlutaMAX. Forty-eight hours post incubation with EV-free medium, the conditioned medium (CM) was collected and stored at −80 °C. To purify EVs, CM were submitted to two light spin centrifugation (300g for 10 min and 2500g for 25 min at 4 °C) to remove debris and dead cell bodies. Then, the remaining supernatant was centrifuged at 20,000g for 25 min to isolate micro-sized vesicles. After that, the supernatant was subjected to ultracentrifugation at 10,0000g for 2 hr at 4 °C. At this point, the supernatant was discarded and the EV pellet was washed with PBS using ultracentrifugation for additional 2 hr similar to the previous step. Finally, the supernatant was removed and the pellet was re-suspended in PBS and stored at −80 °C. We used fixed angle Type 45Ti rotor and Optima™ L-100XP ultracentrifuge instrument. The protein concentration of EVs was measured using BSA protein assay kit (Pierce, ThermoFisher, Rockford, IL). To determine the size and relative intensity of ES-MSC EVs, we used dynamic light scattering (DLS) equipment (Malvern, Worcestershire, UK). The spheroid morphology of vesicles was determined using scanning electron microscopy. Moreover, the expression of general EV markers including CD81, CD63, and TSG101 were evaluated using Western blot standard protocol described by Thery et al. (2006). In order to do a qualitative assessment of targeted delivery and engraftment of EV to the liver, the EV pellet was suspended in 1 ml diluent C in a vial and 4 μl PkH-26 red florescence linker membrane lipophilic dye was added to 1 ml diluent C in another vial and mixed well to disperse. After 5 min of incubation, staining was stopped using 1%BSA and it was subjected to additional ultracentrifugation to pellet EVs. Then, PKH-labeled EVs were used for in-vivo engraftment and also for in-vitro primary hepatocyte uptake. They were observed using fluorescence microscope (IX71; Olympus) and KODAK in-vivo F series imaging system.

2.5 Histopathological analysis

Histopathology analyses performed by two blinded pathologists on four slides for each group. The liver tissue was fixed using 10% formalin, processed, and paraffin-embedded. Then, 6-μm thickness tissues were prepared using microtome and subsequently stained with hematoxylin and eosin (H&E) and masson-trichrome (MT).

Histopathology data according to histological grading and staging (Ishak's score) were classified to piecemeal necrosis (grade 0–4), confluent necrosis (grade 0–6), focal lytic necrosis (grade 0–4), portal inflammation (grade 0–4) and fibrosis (grade 0–6).

2.6 Animal models

Male wistar rats (150–200 g body weight) were kept in single cages under 12 hr light/12 hr dark cycles and controlled humidity and temperature condition in Royan animal facility. To induce cirrhosis, all animals received intraperitoneal injection of 200 mg/kg thioacetamide (TAA) twice weekly for 16 weeks. Ultrasound imaging was performed for confirmation of cirrhosis after 16 weeks. Then, cirrhotic animals were assigned to four groups including ES-MSC (n = 4), ES-MSC EVs (generated from the same lot of the cells; n = 4), fluorescence labeled ES-MSC EVs (n = 1), vehicle (received PBS; n = 4) against normal (n = 3). Next, 4 × 10⁶ cells and 350 μg EVs were dissolved in 400 μl injection water and infused intra-splenicly under the guidance of ultrasound. To evaluate targeted delivery of vesicles, fluorescence-labeled ES-MSC EVs were injected and animals were subjected to KODAK in-vivo F series for live imaging 24 hr-post administration. Other animals were followed up for 4 weeks. After sonography imaging at week 20, animals were euthanized, the excised liver tissues were deep freezed in liquid nitrogen and animal sera were collected for additional assays.

All animal studies and procedures were already approved by the Royan Institutional Review Board and Institutional Ethics Committee of Royan Institute (No: IR.ACECR.ROYAN.REC.1394.11).

2.7 Statistical analysis

All experiments were conducted in at least three independent repeats. All data were showed as mean ± SD or mean ± SEM and one way analysis of variance (ANOVA) was used to determine differences between groups with Tukey post-hoc test and the viability with student's t test. p-values <0.05 were considered significant.

3.1 Generation and characterization of ES-MSCs

In order to produce ES-MSCs, human ES cells were cultured in ES medium in the absence of bFGF to form EBs. Then, EBs were plated in gelatin-coated plates and cultured in MSC medium. Spontaneous differentiating of EBs resulted in outgrowth of ES-MSCs. They were further passaged in order to acquire a homogenous population that had spindle-shaped morphology (Figure 1A). The resulting cells could be passaged 16 times. The procedure of MSCs isolation from ESCs is reproducible. The morphology and yield in independent experiments were similar. These cell lines were then passaged 14 times without noticeable differences in doubling time (Figure 1B) and fold changes (Figure 1C). The growth rate of the cells started to decrease at passage 14 in both lines (Figures 1B and 1C). Then, we characterized ES-MSC line 1.

ES-MSCs at different passages were evaluated to characterize their surface antigen markers (Figure 1D and Supplementary Figure S1A). ES-MSCs expressed CD90, CD105 (endoglin, SH2), CD106, CD73 (SH3), and CD44. These cells were negative for CD34, CD45, and CD11b. This consistent antigen expression profile maintained intact in all consecutive passages. ESC markers, SSEA-4, OCT4, and Nanog did not express on ES-MSCs (Figure 1D and Supplementary Figure S1A, B).

To confirm the multipotency of ES-MSCs, we used osteogenic and adipogenic media to differentiate them. After 3 weeks, the majority of the cells were differentiated into osteoblasts (Figure 1E).

Adipogenic differentiation demonstrated the accumulation of small cytoplasmic vesicles that are lightly stained by oil red O (Figure 1F). For further characterization, the expression profile of differentiated cells were assessed using qRT-PCR. The results showed upregulation of transcription factors involved in osteogenesis and adipogenesis [bone-specific alkaline phosphatase (ALP), Collagen 1α (COL1α), Osteocalcin (OCN) and RUNX2) depicted in Figure 1G and lipoprotein lipase (LPL), PPARγ and Adiponectin depicted in Figure 1H].

ES-MSCs were also not able to stimulate PBMC response in co-culture condition (Supplementary Figure S3A). The immunophenotyping data of ES-MSCs and their similar differentiation capacity revealed that these cells presented the characteristics of MSCs.

3.2 ES-MSCs showed better immunomodulatory activity compared to BM- and AD-MSCs

MSCs should be able to modulate the activity of immune cells directly by regulating inflammatory cytokine secretion (Uccelli et al., 2008). To assess this, we analyzed the proliferation of stimulated PBMCs co-cultured with ES-MSCs as well as BM or AD-derived MSCs and detected CFSE dilution using flow cytometry following MLR assay (Figure 2A and Supplementary Figure S2A). CFSE labeled-PBMCs of the first donor were stimulated to proliferate using a second donor PBMC population according to allogenic mixed lymphocyte reaction (MLR) assay. After IL2 stimulation, in the absence of MSCs, PBMCs as responder cells were highly proliferated. However, co-culturing with MSCs from different sources resulted in inhibition of proliferation of stimulated PBMCs (R + S) (a: p < 0.05 vs. R + S; A: p < 0.0001 vs. R + S). Interestingly, ES-MSCs were able to suppress PBMC proliferation more effectively compared to BM-MSCs and AD-MSCs (p < 0.0005), suggesting that ES-MSCs have better activity in regulating immune cells rather than BM-MSCs and AD-MSCs.

In addition, we observed that not only ES-MSCs from early passages (passage 3) exhibit immunomodulatory effect, but also cells from later passages (passages 7 and 12) showed similar suppression effect (Figure 2B, and Supplementary Figure S2B).

The crucial mechanism of immunomodulation of MSCs lies in MSC-conditioned medium containing soluble factors and EVs rather than cell-cell contacts (Caplan & Correa, 2011; Lou et al., 2017). Therefore, we evaluated the therapeutic effect of MSC-conditioned medium regarding both soluble factors and EVs.

Initially, to understand the mechanism underlying immunosuppressive properties of ES-MSCs, we analyzed one pro-inflammatory cytokine (IFN-γ) and two anti-inflammatory cytokines (TGF-β and IL-10) in the conditioned medium of each group following MLR test by ELISA. The secretion of IFN-γ from stimulated PBMCs decreased after incubation with MSCs from different sources. The secretion of TGF-β was higher in the presence of ES-MSCs (Figure 2C). In case of IL-10, a slight increase in cytokine secretion was only found as a consequence of incubation with ES-MSCs (Figure 2D). Figure 2E shows inhibitory effect of ES-MSCs on IFN-γ secretion compared to BM- and AD-MSCs. (a: p < 0.01 vs. R + S; A: p < 0.0001 vs. R + S; b: p < 0.01 vs. ES-MSC; and B: p < 0.0001 vs. ES-MSC).

The data showed that ES-MSCs were able to have impact on immune cells, suggesting that ES-MSCs have greater capability of regulating immune cell activities than BM-MSCs and AD-MSCs. Therefore, we continued our experiments using MSCs and ES-MSC-derived EVs (ES-MSC EVs).

3.3 The characterization of ES-MSC EVs

Since the therapeutic effect of MSC-conditioned medium might be also induced by EVs, in the next step, we tried to isolate EV population from 48-hr conditioned medium of ES-MSC supernatant by applying conventional ultracentrifugation procedures.

ES-MSC EVs were purified as described before (Thery et al., 2006). Scanning electron microscopy (SEM) showed uniform spheroid morphology of EVs (Figure 3A). Dynamic light scattering analysis indicated an average size of 186.43 ± 18.48 nm with a polydispersity index (PDI) of 0.39 ± 0.05, demonstrating a homogenous population of EVs (Figure 3B).

Western blot and flow cytometry results confirmed expression of EV markers, TSG101, CD81 and CD63 in ES-MSC EVs (Figures 3C and 3D). CD81 was not detected in other MSC EVs (Figure 3C). EVs from other sources did not stimulate immune cell proliferation as effective as the ES-MSCs (Figure S3B).

After allogenic stimulation, in the absence of MSCs, PBMCs highly proliferated. However, administration of EVs from different sources of MSCs inhibited the proliferation of PBMCs (Figure 3E, Supplementary Figure S2C; p < 0.0001 all vs. R + S). Additionally, ES-MSC EVs were able to inhibit PBMC proliferation more effectively compared to BM-MSCs (p < 0.005 BM- vs. ES- and AD-MSC), suggesting that ESC-MSC EVs have higher capability of regulating immune cell activities compared to those derived from BM.

3.4 ES-MSCs and ES-MSC EVs can reduce fibrosis

To test the therapeutic activity of ES-MSC EVs, they were administrated to cirrhotic rats. The model was established after intraperitoneal injection of Thioacethamide (200 mg/kg) twice/week for 16 weeks (Figure 4A). All animals survived after injection with the exception of two. The injured rats treated with ES-MSC or ES-MSC EVs and evaluated after 4 weeks. All injections were carried out under the ultrasound guide through intrasplenic route as a safe and non-invasive procedure instead of surgery.

The livers treated with vehicle were pale with micro nodular features compared to the normal liver tissue which had a soft smooth surface (Figure 4B). ES-MSC or ES-MSC EV–treated livers had gross changes with less nodular morphology (Figure 4B). Histologically, the hematoxylin and eosin (H&E) staining of liver tissues for the vehicle group indicated severe hepatic necrosis and fibrosis with most areas showing periportal or periceptal interface hepatitis (piecemeal necrosis), centro-lobular necrosis with marked portal-portal and portal-central bridging (confluent necrosis) and ductal hyperplasia based on Ishak's scoring (Ishak et al., 1995). However, ES-MSC or EV-treated groups showed reduction of hepatocyte piecemeal, confluent necrosis and ductal proliferation in histological examinations.

Moreover, evaluation of H&E staining showed infiltrating of immune cells into fibrotic tissues secondary to necrosis in control group. On the other hand, infiltration of immune cells was markedly reduced in ES-MSC and EV groups compared to vehicle-treated rats.

MT staining also showed a robust fibrosis with thick bundles of collagen deposition surrounding the hepatocytes leading to a pseudo-lobules pattern. This phenomenon was observed 4 weeks after vehicle infusion (Figure 4C). On the other hand, moderate necrosis and fibrosis detected in ES-MSC or ES-MSC EV treated groups (Figure 4C). Quantitative analyses of pictures of both pathologic and Image Pro Plus software evaluations of the livers of both ES-MSC or ES-MSC EV groups showed significant improvement of necrosis and fibrosis in comparison to the vehicle-treated animals (a: p < 0.001 vs. ES MSC, and A:p < 0.0001 vs. ES MSC EV; Figures 4D and 4E). However, improvement of necrosis and fibrosis was significantly less marked in ES MSC EV compared to ES-MSC (b: p < 0.01).

Furthermore, immunostaining of collagen showed significant improvement of livers in ES-MSC and ES-MSC EV groups (p < 0.0001 for both cases; Figures 4F and 4G). There was no significant difference between EV and cell treated groups concerning the aforementioned assays.

Moreover, we tested the echogenicity of liver parenchyma and portal vein diameter (PVD) by ultrasound. Heterogeneous and extensive coarse parenchymal echogenicity with irregular surface, enlarged portal vein and splenomegaly were observed 16 weeks post-induction. However, 4 weeks after treatment with ES-MSCs and ES-MSC EVs, ultrasound examination showed homogenous echogenicity of liver parenchyma and reduction of the liver nodularity while the echogenicity and surface nodularity did not change in vehicle group (Supplementary Figure S4A–C).

Moreover, the average value of PVD in ES-MSC and EV-treated groups were significantly decreased compared to vehicle group (Figure 4H, Supplementary Figure S4, and video; p < 0.01 for both cases). Sonography findings confirmed histopathological results indicating significant regression of cirrhosis in affected animals following treatment.

Quantitative RT-PCR analysis of the livers showed up-regulation of Col1α, the major contributor to fibrosis, αSMA and TIMP1 in vehicle group. However, they were reduced in ES-MSC or ES-MSC EV-treated livers significantly (p < 0.01 and p < 0.0001, respectively; Figure 4I). We found that gene expression of collagenase enzymes (i.e., MMP9 and MMP13) which degrade fibrillar collagens, increased in ES-MSC or ES-MSC EV −treated animals compared to vehicle group (a: p < 0.01 and A: p < 0.0001 vs. vehicle, and b: p < 0.01 vs. ES-MSC EV; Figure 4I). These results indicated that both ES-MSCs and ES-MSC EVs could markedly reverse histological changes associated with cirrhosis through down-regulation of the major contributors to fibrosis namely, Col1α, αSMA, and TIMP1 and up-regulation of the key factors in the degradation of extracellular matrix namely, MMP9 and MMP13.

In addition, we measured liver enzymes in treated animals (Figure 4J). Significant reductions were observed in plasma levels of AST, ALT and GGT after 4 weeks in animals treated with ES-MSCs and ES-MSC EVs compared to the group treated with vehicle (p < 0.01).

These results indicated that transplantation of MSCs or their EVs had a therapeutic effect in liver injury.

3.5 ES-MSCs and ES-MSC EVs can improve viability and reduce hepatocyte apoptosis both in vitro and in vivo

To determine bio-distribution of EVs in cirrhotic rats, animals were treated with PKH-26-labeled EVs using ultrasound guide via intra-splenic route and subsequently subjected to fluorescence imaging. The fluorescence imaging analysis showed accumulation of EVs in the liver 24 hr post-transplantation (Figure 5A). For further confirmation, the animals were sacrificed and the spleens and livers were isolated for additional imaging. Interestingly, strong signal intensity were found in the liver; however, few signals were observed in the spleen (Figure 5B). Fluorescence microscopy also showed incorporation of PKH-26-labeled EVs into primary hepatocytes 2 hr after incubation (Figure 5C).

MTS analysis was used to explore the effect of EVs on the viability of primary hepatocytes. The results showed high absorbance indicating high number of viable hepatocytes even after 96 hr incubation with EVs compared to vehicle-treated ones (p < 0.01; Figure 5C).

To evaluate the anti-apoptotic effect of ES-MSCs or ES-MSC EVs in vivo, immunohistochemical evaluation showed a markedly reduction of Caspase-3 expression in liver sections post treatment (a: p < 0.0005, A: p < 0.0001, Figures 5E and 5F).

Moreover, assessment of BCL2 expression as an anti-apoptotic gene and BAX as a pro-apoptotic gene showed a significant increase in BCL2 and a significant reduction of BAX expression following incorporation of EVs and MSCs into liver cells (p < 0.01 and p < 0.001, respectively; Figure 5G). Therefore, it was confirmed that EVs could target the hepatocytes and exert anti-apoptosis effect on these cells.

3.6 ES-MSCs and ES-MSC EVs modulate pro- and anti-inflammatory cytokines in a cirrhotic model

Considering the immunomodulatory effects of ES-MSCs and their secreted EVs in MLR experiment, we investigated their anti-inflammatory capability in inhibiting immune responses, in-vivo.

Tumor necrosis factor-α (TNFα) and interleukin-2 (IL-2) pro-inflammatory cytokines are highly associated with liver fibrosis. Four weeks after treatment, qRT-PCR showed that the expression of TNFα and IL-2 were upregulated in the vehicle group. However, TNFα and IL-2 significantly declined in the ES-MSC and ES-MSC EV groups (p < 0.0001 for both cases; Figure 6A). Consistently, determination of TNFα by ELISA supported the qRT-PCR data (p < 0.01; Figure 6B).

On the other hand, the expression of anti-inflammatory cytokines, TGF-β1 and IL-10 were significantly increased in ES-MSC and ES-MSC EV groups compared to vehicle group (a: p < 0.01, A: p < 0.0001 vs. vehicle, and b: p < 0.05 vs. ES-MSC EV; Figure 6A). Determination of IL-10 by ELISA confirmed the qRT-PCR data (p < 0.05; Figure 6B).

Thus, ES-MSCs and ES-MSC EVs could modulate pro- and anti-inflammatory cytokines in an AAT-injured mouse model and promote hepatic recovery.