Human Muse Cells, Nontumorigenic Phiripotent-Like Stem Cells, Have Liver Regeneration Capacity through Specific Homing and Cell Replacement in a Mouse Model of Liver Fibrosis

Abstract

Muse cells, a novel type of nontumorigenic pluripotent-like stem cells, reside in the bone marrow, skin, and adipose tissue and are collectable as cells positive for pluripotent surface marker SSEA-3. They are able to differentiate into cells representative of all three germ layers. The capacity of intravenously injected human bone marrow-derived Muse cells to repair an immunodeficient mouse model of liver fibrosis was evaluated in this study. The cells exhibited the ability to spontaneously differentiate into hepatoblast/hepatocyte lineage cells in vitro. They demonstrated a high migration capacity toward the serum and liver section of carbon tetrachloride-treated mice in vitro. In vivo, they specifically accumulated in the liver, but not in other organs except, to a lesser extent, in the lungs at 2 weeks after intravenous injection in the liver fibrosis model. After homing, Muse cells spontaneously differentiated in vivo into HepPar-1 (71.1±15.2%), human albumin (54.3±8.2%), and anti-trypsin (47.9±4.6%)-positive cells without fusing with host hepatocytes, and expressed mature functional markers such as human CYP1A2 and human Glc-6-Pase at 8 weeks after injection. Recovery in serum, total bilirubin, and albumin and significant attenuation of fibrosis were recognized with statistical differences between the Muse cell-transplanted group and the control groups, which received the vehicle or the same number of a non-Muse cell population of MSCs (MSCs in which Muse cells were eliminated). Thus, unlike ESCs and iPSCs, Muse cells are unique in their efficient migration and integration into the damaged liver after intravenous injection, nontumorigenicity, and spontaneous differentiation into hepatocytes, rendering induction into hepatocytes prior to transplantation unnecessary. They may repair liver fibrosis by two simple steps: expansion after collection from the bone marrow and intravenous injection. A therapeutic strategy such as this is feasible and may provide significant advancements toward liver regeneration in patients with liver disease.

Introduction

Liver diseases are a source of significant morbidity, mortality, and health care costs worldwide^1,2. Chronic liver diseases, such as liver cirrhosis and hepatitis types B and C, are characterized by a loss of functional liver cells and the formation of fibroses². While the most effective therapy for such severe liver failure is liver transplantation, this approach is hindered by organ donor shortage, excessive cost, and the use of immunosuppressive agents. Stem cell transplantation is an alternative approach to the treatment of such liver diseases, and a variety of candidates have been investigated to date^3-5. Embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs) wield the potential to generate functional hepatocytes. However, the use of these cells in patients with liver disease is limited by their inherent tumorigenicity, cost performance, and ethical issues, particularly in the case of ESCs.

Mesenchymal stem cells (MSCs) are reported to reduce liver fibrosis and inflammation in chronic liver diseases through their trophic and anti-inflammatory effects^6-9. They are considered candidate stem cells for the treatment of various liver diseases, but their capacities for homing into damaged liver tissue and for differentiation into liver cells in vivo are relatively low^10-13.

Recently, unique stem cells, termed multilineage-differentiating stress-enduring (Muse) cells, have been demonstrated to be pluripotent-like, stress-tolerant stem cells^14-16. They express pluripotency markers such as octamer-binding transcription factor 4 (Oct3/4), sex-determining region Y-box 2 (Sox2), and Nanog, and are positive for stage-specific embryonic antigen-3 (SSEA-3), a well-known surface marker for human ESCs^14-16. They are present in the bone marrow (BM), comprising ∼0.03% of the mononucleated fraction. They also comprise 1%-3% of cultured MSCs, such as BM-MSCs and commercially available fibroblasts. Approximately 30 ml of fresh human BM aspirate yields ∼1 million Muse cells by day 3 in culture, suggesting potential clinical application¹⁴. Muse cells have the ability to self-renew and differentiate into cells of all three germ layers from a single cell and have the ability to differentiate into a-fetoprotein⁺ and albumin⁺ cells in high proportions in vitro when a cytokine induction system is implemented¹⁵. Importantly, their telomerase activity is very low. Muse cells do not have tumorigenic propensities, nor do they undergo teratogenesis in vivo. In contrast to ESCs and iPSCs, injection of Muse cells into the testis of severe combined immunodeficiency (SCID) mice did not lead to formation of teratomas even after 6 months of treatment, consistent with the fact that they reside in normal adult tissues such as BM and adipose tissue^14-17.

As shown in muscle degeneration, skin injury, and partial hepatectomy models, Muse cells are unique in that they are able to home to the site of damage when supplied locally or by intravenous injection and then differentiate into cells compatible with the tissue into which they homed^14,18. Several other groups also confirmed that locally injected Muse cells repair skin ulcers of a diabetes mellitus model by generating new dermal and epidermal cells and were also shown to replenish new neuronal cells by spontaneous differentiation, which could participate in the reconstruction of the pyramidal tract and sensory circuit in a stroke model^19-21. In these studies, Muse cells did not require induction and/or gene manipulation for generating purposive cells since naive Muse cells, without manipulation, have the intrinsic capacity to migrate to the damaged tissue, integrate, and spontaneously differentiate into tissue-compatible cells. Therefore, Muse cells can be directly administrated to patients after collection from BM or other sources. Based on this evidence, Muse cells have been posited as an optimal source for stem cell therapy using a simple method such as collection from BM and supply by intravenous injection for chronic liver diseases, if regenerative effects and functional recovery are robustly confirmed.

This study aimed to validate critical components of these processes including migration ability toward the damaged liver, in vivo dynamics after intravenous injection, engraftment into the damaged liver, differentiation, contribution to the replenishment of new hepatocytes, liver tissue repair, and functional recovery in a SCID mouse model of liver fibrosis by using human BM Muse cells. Xenotransplantation of human Muse cells into a mouse model has advantages compared to allogeneic and autogeneic cell transplantation in some aspects since human-specific markers can be used to confirm the integration and distribution of the transplanted human cells in mouse tissue. For this, a mouse model of liver fibrosis was created using SCID mice that do not reject human cells; human BM Muse cells were infused into the mice.

Materials and Methods

Preparation of Human Muse and Non-Muse Cells

Human BM-MSCs were purchased from Lonza (Basel, Switzerland). Cells were cultured at 37°C, 5% CO₂ in minimal essential medium Eagle α modification (α-MEM; Sigma-Aldrich, St. Louis, MO, USA) containing 10% fetal bovine serum (FBS; HyClone, Logan, UT, USA), 0.1 mg/ml kanamycin (Invitrogen, Carlsbad, CA, USA), and 1% GlutaMAX (Life Technologies, Carlsbad, CA, USA). Cells were subcultured at a ratio of 1:2 after reaching 90%-100% confluence using 0.25% trypsin-ethylenediaminetetraacetic acid (Trypsin-EDTA; Life Technologies). Cells at the fourth to eight subculture were used in this study. The growth curve of BM-MSCs is provided in Figure 1A.

Figure 1.

Characterization of multilineage-differentiating stress-enduring (Muse) cells. (A) Growth curve of mesenchymal stem cells (MSCs) obtained by using the cell counting kit-8 (CCK8). Bone marrow (BM)-derived MSCs (1.3 × 10³ cells, p = 7) were seeded onto 96-well plates, and the plates were incubated [day (D)]. (B) Stage-specific embryonic antigen-3⁺ (SSEA-3⁺) Muse cells (gate P3) and SSEA-3⁻ non-Muse cells (gate P6) sorted from human BM-MSCs. (C) Quantitative polymerase chain reaction (qPCR) for octamer-binding transcription factor 4 (OCT4), sex-determining region Y-box 2 (SOX2), and Nanog. *p < 0.05, **p < 0.01, ***p < 0.001. (D) Cells expanded from single M-cluster on gelatin-coated dish. (E) Immunocytochemistry of cells expanded from a single M-cluster on gelatin-coated culture dish expressed hepatoblast/hepatocyte markers: Δ-like protein (DLK), α-fetoprotein, cytokeratin 19, and cytokeratin 18 (red). Blue indicates nuclei staining with 4′,6-diamidino-2-phenylindole (DAPI). Scale bars: 50 μm.

Cell sorting was done according to previous reports^14-17. Fluorescence-activated cell sorting (FACS) buffer was prepared as follows: 50 ml of total volume consisting of 44 ml of phosphate-buffered saline (PBS; without calcium chloride and magnesium chloride; Nacalai Tesque, Kyoto, Japan), 5 ml of 5% bovine serum albumin (BSA; Nacalai Tesque), and 1 ml of 100 mM EDTA (Nacalai Tesque). After detaching the cells with trypsin, they were suspended in FACS buffer at 5.0 × 10⁵ cells per 100 μl of buffer. Cells were divided into three groups: two as controls, namely, incubation without any antibodies (the experimental setup to monitor autofluorescence) and with secondary antibody only (to determine the level of background surface staining), were used for setting the gate; the third sample was incubated with primary and secondary antibodies for cell sorting (Fig. 1B). Anti-SSEA-3 antibody (1:100; Millipore, Bedford, MA, USA) was used as a primary antibody, and cells were incubated for 1 h at 4°C. After washing with FACS buffer three times, cells were incubated with secondary antibody, fluorescein isothiocyanate (FITC)-conjugated anti-rat immunoglobulin M (IgM) antibody (1:100; Jackson ImmunoResearch, West Grove, PA, USA), for 1 h at 4°C. After washing, cells were filtrated through a cell strainer (100 mm; BD Biosciences, San Jose, CA, USA) to eliminate clumps. The cells were analyzed and sorted by BD FACSAria^™ II Cell Sorter (BD Biosciences). SSEA-3⁺ and SSEA-3⁻fractions were determined using the control samples. SSEA-3⁺ Muse cells and SSEA-3⁻ non-Muse cells were sorted under low stream speed.

Formation of Muse Cell Cluster (M-Cluster) in a Single-Cell Suspension Culture

Muse cells were cultured in suspension using poly(2-hydroxyethyl methacrylate) (poly-HEMA; Sigma-Aldrich)-coated 96-well plates as previously described¹⁴. After the limiting dilution, each single cell was transferred into an individual well in 96-well plates with α-MEM containing 10% FBS and 1% GlutaMAX. On the next day of plating, wells without any cells or with multiple cells were eliminated from observation. At 7 days, formation of single Muse cell-derived clusters, termed M-clusters, was observed by phase-contrast microscopy. Each M-cluster was gently picked up and transferred onto 24-well gelatin (Sigma-Aldrich)-coated coverglass individually.

Quantitative Polymerase Chain Reaction (qPCR)

Human Muse and non-Muse cells (5 × 10⁴ cells) were cultured in α-MEM containing 10% FBS and 1% GlutaMAX for 1 day. In the case of M-clusters, they were collected after single-cell suspension culture for 7 days. Total RNA was extracted from these cells or M-clusters using a NucleoSpin RNA XS (Macherey-Nagel, Duren, Germany). First-strand cDNA was generated, using the SuperScript III synthesis kit. DNA was amplified using the Applied Biosystems (Carlsbad, CA, USA) 7500 Fast Real-Time PCR system according to the manufacturer's instructions. Primers for OCT4 (Hs00999632_g1), SOX2 (Hs01053049_s1), and Nanog (Hs04260355_g1) were obtained from Applied Biosystems. Data were processed using the ΔΔCT method.

Immunocytochemistry for M-Clusters

M-clusters were cultured on gelatin-coated wells for 10-14 days. Cells were fixed with 4% paraformaldehyde (PFA; Millipore) in PBS for 1 h at 4°C and were subjected to immunocytochemistry as described previously^16,17. In brief, samples were incubated with antibodies against α-fetoprotein (1:100; DAKO, Carpinteria, CA, USA), Δ-like protein (DLK; 1:500; Santa Cruz Biotechnology, Dallas, TX, USA), cytokeratin 18 (CK18; 1:100; Abcam, Cambridge, UK), and CK19 (1:100; Thermo Fisher Scientific, Waltham, MA, USA) at 4°C overnight. All antibodies were diluted in PBS/0.1% BSA solution. After washing with PBS three times, cells were incubated with secondary antibodies either of cyanine 3 (Cy3)-conjugated anti-mouse IgG or anti-rabbit IgG (1:500; Jackson ImmunoResearch) for 2 h at room temperature (RT). Nuclear staining was performed using 4′,6-diamidino-2-phenylindole (DAPI; 1:500; Sigma-Aldrich), and then cells were inspected using a confocal laser scanning microscope (C2+; Nikon, Tokyo, Japan).

Preparation of Acute Liver Damage Model for In Vitro and In Vivo Muse Cell Migration/Integration Capacity Assay

CB17/Icr-Prkdc<scid>/CrlCrlj (SCID) mice (8-10 weeks old; Charles River Laboratories, Yokohama, Japan) were used in this study. All animal experiments were treated according to the regulations of the Standards for Human Care and Use of Laboratory Animals of Tohoku University (Sendai, Japan).

Mice received a single intraperitoneal (IP) injection of 1.5 ml/kg of carbon tetrachloride (CCl₄; Wako, Osaka, Japan). CCl₄ was dissolved in olive oil (Wako) at 1:10.

For in vitro migration assay, the serum and liver were collected at 1, 24, and 48 h after IP injection of CCl₄ in SCID mice. The detailed method for the migration assay is described below.

For in vivo analysis of integration and differentiation capacities of Muse cells in the acute liver damage model, human Muse cells (2 × 10⁴ cells) and non-Muse cells (2 × 10⁴ cells) were injected into the tail vein of SCID mice 24 h after IP CCl₄ injection. After 2 weeks, every organ was collected for qPCR of human Alu sequence. At day 30, the liver was subjected to immunohistochemistry as described below. Five animals were prepared for each group at each time point.

Preparation of Liver Fibrosis Model and Cell Transplantation

SCID mice were used in this study. Animals received IP injections of 0.5 ml/kg of CCl₄ twice a week for 8 weeks. Muse cells (5 × 10⁴ cells), non-Muse cells (5 × 10⁴ cells), or the same volume of PBS as a control was injected into the tail vein of each mouse at 2, 4, and 6 weeks after the initial injection of CCl₄. There were eight animals in each group. The experimental procedures for generating the liver fibrosis model, the number of cells, and the timing of injection are described in Figure 4A.

Figure 4.

Functional and histological evaluations in the liver fibrosis model. (A) The liver fibrosis severe combined immunodeficiency (SCID) mouse model was carried out by intraperitoneal (IP) injection of carbon tetrachloride (CCl₄; 0.5 ml/kg) twice per week for up to 8 weeks. Multilineage-differentiating stress-enduring (Muse) and non-Muse cells (5 × 10⁴ cells) were infused via the tail vein at 2, 4, and 6 weeks, and data were collected at 8 weeks [intravenous injection (IV)]. (B, C) Serum total bilirubin (B) and serum albumin (C) levels in the Muse, vehicle, and non-Muse groups at 8 weeks of infusion. (D, E) Evaluation of the liver fibrotic area by staining with (D) Sirius red and (E) Masson's trichrome staining at 8 weeks. (F–H) ELISA for matrix metalloproteinase-1 (MMP-1), MMP-2, and MMP-9 in human Muse and non-Muse cells. **p < 0.01, ***p < 0.001. Scale bars: 50 μm.

Migration Assay

The migratory capacity of Muse and non-Muse cells was assessed using 24-well Matrigel invasion chambers (BD Biosciences) according to the manufacturer's protocol. Serum and liver tissue were collected from the SCID mouse models at 1, 24, and 48 h after 1.5 ml/kg of CCl₄ injection, as described above (see “Preparation of Acute Liver Damage Model for In Vitro and In Vivo Muse Cell Migration/Integration Capacity Assay”). Normal serum and liver tissues obtained from intact SCID mice were used as controls. The lower chamber was filled with 750 μl of α-MEM containing 10% mouse serum or 50 mg of liver tissue (Fig. 2A). A cell suspension (Muse or non-Muse 2.5 × 10⁴ cells/500 μl of α-MEM containing 10% FBS) was added to the upper insert. After 22 h of incubation (37°C, 5% CO₂), the nonmigrated cells that remained at the upper surface of the inserts were removed with a cotton swab. The cells that migrated through the membrane to the lower surface of the membrane were stained using a Diff-Quik kit (Sysmex, Hyogo, Japan). The nuclei of the migrated cells were counted under a light microscope, observing 10 random fields (200x) (Fig. 2A).

Figure 2.

In vitro migration of multilineage-differentiating stress-enduring (Muse) cells toward the serum and liver tissue of carbon tetrachloride (CCl₄)-treated animals. (A–C) In vitro migration assay. (A) Outline of the experiment. Human Muse or non-Muse cells were placed in the upper chamber of the semipermeable membrane, and either the serum or the liver tissue collected at 1, 24, and 28 h after intraperitoneal (IP) injection of CCl₄, or from intact animal, was placed in the lower chamber. After 22 h of incubation, the number of migrated cells through the semipermeable membrane was counted. Muse cells migrate to the (B) serum and (C) liver tissue with higher efficiency than that observed in non-Muse cells. (D) Western blot for C-X-C chemokine receptor type 4 (CXCR4) and hepatocyte growth factor receptor (c-Met) in human Muse and non-Muse cells. β-Actin was used as control. (E) Migration of Muse and non-Muse cells toward the serum 24 h after CCl₄ injection with or without the CXCR4 antagonist AMD3100. **p < 0.01, ***p < 0.001.

The effect of the C-X-C chemokine receptor type 4 (CXCR4) on the migration activity of Muse and non-Muse cells was assessed by placing a CXCR4 antagonist (AMD3100; 1 μg/ml; ab120718; Abcam) into the lower chamber. Muse and non-Muse cells (2.5 × 10⁴ cells) were placed in the upper chamber and were incubated at 37°C, 5% CO₂, for 24 h with or without AMD3100 in 10% mouse serum (24 h after CCl₄ injection) in α-MEM.

Western Blot

Lysates from isolated human Muse and non-Muse cells were obtained. Fifteen micrograms of total protein was loaded into each channel and separated using a 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) gel and then transferred to polyvinylidene difluoride (PVDF) membranes using standard protocols, after which they were incubated with antibodies against CXCR4 (x2000; NBP1-77067; Novus Biologicals, San Diego, CA, USA) and hepatocyte growth factor (HGF) receptor (c-Met; x500; ab51067; Abcam). Signals were visualized by chemiluminescence (ECL; Pierce, Waltham, MA, USA), and the signals were quantified by ImageQuant LAS 4000 mini (GE Healthcare, Buckinghamshire, UK). β-Actin (x10,000; analyzed with an antibody from Abcam) served as the loading control.

Assessment of Human Alu Sequence

Human Muse or non-Muse cells (5 × 10⁴ cells) were injected into the tail veins of liver damage models (24 h after CCl₄ injection) and intact mice. Genomic DNA from each organ (brain, heart, lung, liver, spleen, pancreas, kidney, muscle, and BM) was collected at 2 weeks after cell injection using the REDExtract-N-Amp Tissue PCR kit protocol (Sigma-Aldrich). To quantify human genomic DNA in each organ, qPCR assays were used to amplify human Alu repeats according to a previous protocol with minor modifications²². PCRs were performed in a volume of 20 μl containing 10 μl of TaqMan Universal Master Mix II (Applied Biosystems), 900 nM forward and reverse primers, 250 nM TaqMan probe, and 10 ng of target template. Reactions were incubated at 50°C for 2 min, 95°C for 10 min, and then amplified for 45 cycles, with each cycle composed of an incubation step at 95°C for 15 s followed by 60°C for 1 min. Standard curves were generated by serially diluting human genomic DNA extracted from human MSCs, which was mixed with mouse genomic DNA obtained from mouse liver. The sequences of the PCR primers and probe used for detection were as follows: primers, 5′-CATGGTGAAACCCCGTCTCTA-3′ (sense) and 5′-GGGTTCAAGCGATTCTCCTG-3′ (anti-sense); TaqMan probe, 5′-FAM-ATTAGCCGGGCGTG GTGGCG-TAMRA-3′. To compare the results of all experiments, 1 ng of human genomic DNA was set as a gene expression level of 1.

Hematological Examination

Blood samples were collected from the inferior vena cava of liver fibrosis model mice from Muse, non-Muse, and vehicle groups at 8 weeks. Serum was collected by centrifugation of blood samples (4°C, 1,500 × g, 15 min) and was subjected to measurement of total bilirubin and serum albumin using a DRY-CHEM 7000V (Fuji-Film, Tokyo, Japan).

Quantitative Analysis of Liver Fibrosis

At 8 weeks after treatment, all mice were fixed by cardiac perfusion with 4% PFA in PBS under deep anesthesia. The liver was removed and placed in 4% PFA for an additional 5 h. Then 3-μm-thick paraffin sections were prepared. Sirius red staining and Masson's trichrome staining were performed to analyze the liver fibrosis area. Paraffin sections were stained with either a Picrosirius Red Stain Kit (Polysciences, Inc., Warminster, PA, USA) or Masson's trichrome (Muto Pure Chemicals, Tokyo, Japan) according to the manufacturer's instructions. All slides were observed under a microscope. The red area in Sirius red staining and the blue area in Masson's trichrome staining, considering the fibrotic area, were measured using NIS elements (Nikon). The mean percentage of the fibrotic area in each mouse was calculated by examining 10 randomized high-magnification fields (200x).

Elisa

Human Muse and non-Muse cells were cultured at a cell density ofl × 10⁵ cells in 3.5-cm dishes overnight. Cells were washed with PBS and cultured further for 48 h in α-MEM without serum. The secretion of matrix metalloproteinase-1 (MMP-1), MMP-2, and MMP-9 was measured after collecting the culture media and quantified using an enzyme-linked immusorbent assay (ELISA) according to the manufacturer's instructions (MMP-1: ab100603, MMP-2: ab100606, MMP-9: ab100610; Abcam).

Immunohistochemistry

Cryosections (8-10 μm thick) were prepared for immunohistochemistry. After antigen retrieval [by incubation with 1% SDS (Nacalai Tesque) in PBS for 5 min], samples were incubated with either anti-human mitochondria antibody (1:100; Abcam) or anti-human Golgi complex antibody (1:100; Abcam), diluted with PBS/0.1% BSA solution at 4°C overnight, and then with Alexa 488-conjugated anti-mouse IgG or anti-rabbit IgG (1:500; Jackson ImmunoResearch) at RT for 2 h.

For double staining, samples were first incubated with either anti-human mitochondria or anti-human Golgi antibodies followed by secondary antibodies, and then further incubated either with anti-α1-anti-trypsin antibody (α1-AT; 1:200; Thermo Fisher Scientific), anti-human hepatocyte antibody (HepPar-1; 1:200; DAKO), or anti-human albumin antibody (1:100; Bethyl Laboratories, Montgomery, TX, USA) at 4°C overnight. Samples were then incubated with Cy3-conjugated anti-rabbit IgG, anti-mouse IgG, or anti-goat IgG antibodies (1:500; Jackson ImmunoResearch) at RT for 2 h, and observed under a confocal laser microscope after nuclei staining with DAPI. Calculation of a positive cell ratio was implemented by examining 10 randomized high-magnification fields (200x). Cell counting was performed using National Institutes of Health (NIH) ImageJ software (Bethesda, MD, USA).

Fluorescence In Situ Hybridization (FISH)

Cryosections (8 μm thick) were prepared. Species-specific green signal probes for mice and red signal probes for humans were purchased from Chromosome Science (Sapporo, Japan). The experimental procedure was conducted according to the manufacturer's instructions. Slides were stained with DAPI and observed under a confocal laser microscope.

Reverse Transcription Polymerase Chain Reaction (RT-PCR)

Total RNA was extracted from the livers of each mouse and purified using the RNeasy Mini Kit (Qiagen, Valencia, CA, USA). First-strand cDNA was generated by reverse transcription of total RNA using the SuperScript III synthesis kit (Invitrogen) according to the manufacturer's protocol.

The PCRs were performed with Ex Taq DNA polymerase (Takara Bio, Otsu, Japan) using standard temperature cycling conditions. The used primers and their cycling conditions were as follows: human β-actin [annealing temperature (T_m) 55°C, 40 cycles (C)], 5′-AGGCGGACTATGACTTAGTTGCGTTACACC-3′ (sense) and 5′-AAGTCCTCGGCCACATTGTGAACT TTG-3′ (antisense); mouse β-actin (T_m 55°C, 32 C), 5′-ACCCTAAGGCCAACCGTGAAAAGATGAC-3′ (sense) and 5′-CCGCTCGTTGCCAATAGTGATGACCT-3′ (antisense); human albumin (T_m 54°C, 36 C), 5′-CACAAGCCCAAGGCAACAAAA-3′ (sense) and 5′-AGTGCTGTACCACTCTATTAG-3′ (antisense); human cytochrome P450, family 1, subfamily A, polypeptide2 (CYP1A2) (T_m 58°C, 36 C), 5′-GGCACTTCGACCCTTACAAT-3′ (sense) and 5′-TTCAACCAGAGGTTCCT GTG-3′ (antisense); human glucose-6-phosphatase (Glc-6-Pase) (T_m 58°C, 36 C), 5′-GAAAGATAAAGCCGACCTACA-3′ (sense) and 5′-GCAGCAGATAAAATCCGATG-3′ (antisense). Human adult liver (Clontech Laboratories, Inc., Mountain View, CA, USA) was used as a positive control, and the liver of the vehicle group was used as a negative control.

Mechanism of Muse Cell Differentiation Into Hepatoblast/Hepatocyte-Lineage Cells

Human Muse cells were cultured with conditioned medium either of intact or etoposide-treated apoptotic mouse Hepa-1-6 cells [CRL-1830; American Type Culture Collection (ATCC), Manassas, VA, USA] or were cocultured with either intact or etoposide-treated apoptotic Hepa-1-6 cells. Culture medium contained 10% FBS in Dulbecco's modified Eagle's medium (DMEM; Sigma-Aldrich). Etoposide was used at 50 μM. The proportion of cells for coculture of human Muse cells and mouse Hepa-1-6 was 1:3. At day 3, 1 week, 2 weeks, and 3 weeks, total RNA was purified and subjected to qPCR for detection of human-specific hepatoblast/hepatocyte markers. The primers used included human Sox17 (Hs00751752_s1), human CK18 (Hs02827483_g1), human prospero homeobox protein 1 (Prox1; Hs00896394_m1), human α-fetoprotein (Hs01040598_m1), and human β-actin (Hs060665_g1). Total RNA from mouse fetal liver was used as the negative control and human fetal liver (Takara Bio) as positive control. The expression level of each sample was normalized by human β-actin, and then the ratio for the normalized value of the expression level in human fetal liver was calculated in each marker. Mouse fetal liver samples that reacted with the human-specific primers were always under the detection level.

Statistical Analysis

The Student's t-test was used to assess the significance between two groups. The statistical significance of differences among three or more groups was assessed using one-way analysis of variance (ANOVA) with Bonferroni's multiple comparison tests. Statistical analysis was performed with JMP statistical software (SAS Institute, Cary, NC, USA). A value of p < 0.05 was considered a significant difference between groups.

Results

Characterization of Muse Cells

Human Muse cells positive for SSEA-3 (∼2% of human BM-MSCs) and non-Muse cells negative for SSEA-3 were isolated by cell sorting as described (Fig. 1B)¹⁴. When subjected to single-cell suspension culture, only Muse cells generated single M-clusters, which are similar to ESC-derived embryoid bodies formed in suspension, while none of the non-Muse cells formed such clusters, as described previously (not shown)¹⁴. We investigated gene expression of pluripotency markers in naive adherent Muse cells, naive adherent non-Muse cells, and M-clusters formed in suspension. We demonstrated that naive Muse cells showed higher levels of gene expression for the pluripotency markers OCT4, SOX2, and Nanog compared with naive non-Muse cells, and notably, SOX2 and Nanog in naive non-Muse cells fell below the detection threshold. Remarkably, the expression of OCT4, SOX2, and Nanog in M-clusters became approximately 9-, 54-, and 35-fold higher, respectively, than those in naive Muse cells, with statistically significant differences (Fig. 1C). The ability of Muse cells to differentiate into triploblastic lineage cells was confirmed as previously described (not shown)¹⁴. Muse cells were already reported to differentiate into α-fetoprotein⁺ and albumin⁺ cells when they were induced with HGF and fibroblast growth factor 4 (FGF-4)¹⁵. We examined in this study whether cells expanded from M-clusters on gelatin-coated culture (Fig. 1D) spontaneously differentiate into cells expressing hepatoblast/hepatocyte markers without cytokine induction. They included cells positive for DLK (1.5±0.6%), α-fetoprotein (3.0±0.8%), CK19 (1.7±0.4%), and CK18 (2.0±0.9%) (Fig. 1E). Since non-Muse cells do not form clusters in single-cell suspension, they were directly plated onto a gelatin culture soon after isolation by cell sorting and cultured for the same time period as the M-clusters. However, non-Muse cells did not show any expression of hepatoblast/hepatocyte lineage markers (not shown). Therefore, Muse cells are posited to have a higher potential for differentiation into hepatoblast/hepatocyte lineage cells.

Muse Cells Efficiently Migrate to and Integrate Into Damaged Liver

We examined the capacity of human Muse cells to migrate toward the serum and liver tissue of liver damage model in vitro. For this, single IP injection of CCl₄ was performed in SCID mice in order to make the acute liver damage model. The serum and damaged liver tissues were collected at 1, 24, and 48 h after CCl₄ injection. With the serum of an intact mouse, a very small number of Muse and non-Muse cells migrated, and there was no statistical difference observed between them (Fig. 2B). The serum from 1 h after CCl₄ injection (1 h-CCl₄) slightly increased migration of both Muse and non-Muse cells, while there were no statistical differences observed between the intact and 1 h-CCl₄ in both Muse and non-Muse cells (Fig. 2B). With the serum of the 24-h CCl₄, however, the number of Muse cells that migrated substantially increased with statistically significant differences between the intact and 1-h CCl₄ (both p < 0.001). The increase in migration of cells at 24 h was ∼12-fold higher than that in the 1-h CCl₄ in Muse cells, while a comparably drastic change was not recognized in non-Muse cells. A statistically significant difference was also recognized between Muse and non-Muse cells at 24 h (p < 0.001), with a migration rate of Muse cells three- to fourfold higher than that in non-Muse cells. With the serum from 48-h CCl₄, the magnitude was not as high as that seen in the 24-h CCl₄; however, a significant difference was still recognized between Muse and non-Muse cells (p < 0.01) (Fig. 2B).

Similar to the serum, the liver tissue from 24-h CCl₄ demonstrated the highest rate of Muse cell migration with statistically significant differences in comparison to non-Muse cells (p < 0.001) and to Muse cells in the intact, 1-h CCl₄, and 48-h CCl₄ liver tissues (all at p < 0.001) (Fig. 2C). The number of Muse cells that migrated toward the liver at 24-h CCl₄ was approximately sevenfold higher than at 1-h CCl₄ and was approximately fourfold higher than the migration rate of non-Muse cells at 24-h CCl₄. Strong migration was not detected in non-Muse cells (Fig. 2C). These results demonstrate that, in sharp contrast to non-Muse cells, Muse cells have a high migration activity toward the serum and liver in the CCl₄ liver damage model.

Several factors have been reported to control migration of stem cells such as MSCs, particularly CXCR4 and c-Met, ligands for stromal cell-derived factor-1 (SDF-1) and HGF, respectively. These factors are reported to play important roles in the homing of stem cells into the damaged liver^23,24. In Western blot, the expression of CXCR4 and c-Met signals was nearly the same between human Muse and non-Muse cells (Fig. 2D). Since Muse cells are a subpopulation of MSCs and the CXCR4–SDF-1 axis is a key player in MSC migration, we then evaluated whether migration of Muse cells is affected by AMD3100, a CXCR4 antagonist. As shown in Figure 2E, while not completely abrogated, migration of Muse cells toward the serum of 24-h CCl₄ was substantially suppressed by AMD3100 (p < 0.001). Suppression of migration was also recognized in non-Muse cells (p < 0.01) (Fig. 2E).

The in vivo dynamics of intravenously injected (tail vein) human Muse and non-Muse cells were analyzed in both intact and CCl₄ liver damage mice at 2 weeks. The qPCR of the human-specific Alu sequence in each organ was performed in order to survey the distribution of human Muse and non-Muse cells. In intact mice, a low signal for Alu sequence was detected in the lungs in both Muse cell- and non-Muse cell-injected mice, while the signal was under the detection level in other organs (Fig. 3A). The CCl₄ liver damage model after Muse cell injection showed the highest level of Alu sequence expression in the liver and, to a lesser extent, in the lungs, while the signal was undetectable in other organs at 2 weeks. In contrast, in non-Muse cell-injected mice, the signal for Alu sequence fell under the detection threshold in the liver as well as in other organs except the lungs (Fig. 3B).

Figure 3.

In vivo dynamics of multilineage-differentiating stress-enduring (Muse) and non-Muse cells in 24-h carbon tetrachloride (CCl₄) liver damage model. (A, B) Quantitative polymerase chain reaction (qPCR) of a human-specific Alu sequence both in intact (A) and CCl₄-treated mice (B) injected with human Muse and non-Muse cells at 2 weeks (BM, bone marrow). (C) H-Golgi immunostaining in the liver at 30 days of infusion. Since integration of Muse cells that expressed the H-Golgi signal (green) was heterogenous, representative pictures of high, moderate, and low integration areas were demonstrated. A minimal number of non-Muse cells was detected. Blue indicates nuclei staining with DAPI. (D) The proportion of H-Golgi⁺ cells relative to the total number of cells/mm² in the liver section. ***p < 0.001. (E, F) Because the injured liver is quite heterogenous, (B), (D), and (E), showing the integration of Muse cells into the damage tissue, were selected for demonstrative purposes. Integrated human cells positive for human mitochondria (E; green) and H-Golgi (F; green) expressed human albumin (E; red) and anti-human hepatocyte antibody (HepPar-1) (F; red), respectively, in the damaged liver of the Muse cell-injected animals (30 days). Blue indicates nuclei staining with DAPI. Scale bars: 50 μm (main figures), 25 μm (magnified boxes).

We further analyzed integration and differentiation of human Muse cells in the same liver damage model at day 30 using anti-human golgi complex (H-Golgi) or antihuman mitochondria. Since the histology of the injured liver is heterogeneous, some areas with serious damage had higher integration of human Muse cells, while other areas with moderate and slight damage had less integration (Fig. 3C). We therefore randomly selected 10 different areas from samples with different degrees of damage and measured the number of integrated cells. A robust number of H-Golgi⁺ cells (1.89±0.65% of total cells in 1-mm² liver section) was detected in the Muse cell-injected group; this number was about 48-fold higher than that in the non-Muse cell-injected group (0.04±0.08%) (p < 0.001) (Fig. 3C and D). In histological analysis, H-Golgi⁺ cells were mainly distributed around vessels in the liver in the Muse cell group, suggesting that intravenously injected Muse cells integrated into the liver from the vessels themselves (Fig. 3C). Integration of human Muse cells into the damaged liver was also confirmed by detection of human-specific mitochondria (Fig. 3E). Double staining with H-Golgi/human mitochondria and hepatocyte markers demonstrated that 49.8±1.9% of human mitochondria⁺ cells were positive for human-specific albumin, and 80.4±3.2% of H-Golgi⁺ were positive for the progenitor/mature liver cell marker HepPar-1 (Fig. 3E and F).

All these results indicate that Muse cells have a much higher capacity for migration to and accumulation in the damaged liver both in vitro and in vivo, and for differentiation into human-specific albumin⁺ and HepPar-1⁺ cells in vivo, while non-Muse cells did not exhibit such an ability.

Muse Cells Improve Function and Attenuate Fibrosis in Liver Fibrosis Model

The procedure for creating the liver fibrosis model and infusion of cells is shown in Figure 4A. No tumor formation was observed in Muse, non-Muse, or vehicle groups for up to 8 weeks (data not shown). At 8 weeks, the serum total bilirubin was highly significantly lower in the Muse group (0.26±0.05 mg/dl) compared to that in the vehicle (0.74±0.05 mg/dl; p < 0.001) and non-Muse (0.48±0.12 mg/dl; p < 0.001) groups, which represents about 2.8-fold lower in the Muse group compared to the vehicle group. While to a lesser magnitude than the Muse group, the total bilirubin in the non-Muse group was ∼1.5-fold lower than that in the vehicle group (p < 0.001), suggesting moderate recovery (Fig. 4B). The serum albumin level in the Muse group was the highest among the three groups (2.99±0.11 g/dl), with a highly significant statistical difference compared to the vehicle group (2.65±0.08 g/dl; p < 0.001) and the non-Muse group (2.81±0.06 g/dl;p < 0.01). To a lesser extent, the non-Muse group showed moderate recovery in the serum albumin compared to the vehicle group (p < 0.01) (Fig. 4C).

As fibrotic tissue was mainly composed of collagen type I/III, the extent of fibrosis was evaluated by Sirius red and Masson's trichrome staining. A widespread fibrotic area with a typical internodular septum was seen in the vehicle group at 8 weeks. In contrast, the fibrotic area was the smallest in the Muse group. Based on Sirius red staining, the Muse group (0.75±0.15% to total area per section) showed the smallest fibrotic areas in comparison to the vehicle (2.91±0.35%) and non-Muse (1.86±0.13%) groups, both with statistically significant differences (p < 0.001), improving fibrosis by 75% compared to vehicle group (Fig. 4D). A significant difference was seen between the non-Muse and vehicle groups (p < 0.001), and the fibrosis in the non-Muse group corresponded to 36% improvement compared to the vehicle group, suggesting a moderate effect of non-Muse cells (Fig. 4D). Similar results were observed in Masson's trichrome staining: the Muse group exhibited the lowest fibrotic area among the three groups (0.73±0.15%) with highly statistically significant differences compared to the vehicle (1.90±0.12%; p < 0.001) and non-Muse (1.11±0.15%; p < 0.01) groups, improving fibrosis by 62% compared to the vehicle group (Fig. 4E). The non-Muse group demonstrated a significant difference compared to the vehicle group (p < 0.001), but to a lesser extent than the Muse group, which corresponded to 42% improvement compared to the vehicle group (Fig. 4E).

Matrix metalloproteases (MMPs) are known to degrade and remove extracellular molecules from the tissue, and MMP-1, MMP-2, and MMP-9 are involved in fibrolysis and/or suppression of fibrosis²⁵. Both Muse and non-Muse cells showed the ability to produce MMP-1 and MMP-2 to a similar extent, while MMP-9 production was only recognized in Muse cells (Fig. 4F–H).

These results indicate that the human Muse cell group was more efficient than the vehicle and non-Muse cell groups in improving liver function determined by serum total bilirubin and albumin and attenuation of fibrosis in a mouse liver fibrosis model for up to 8 weeks.

Muse Cells Provide New Hepatocytes by Spontaneous In Vivo Differentiation in Liver Fibrosis Model

Human Muse cells were detected in higher numbers in areas around the vessels, while very small numbers of non-Muse cells were detected at 8 weeks (Fig. 5A). Similar to the acute model shown in Figure 3B, intravenously injected Muse cells integrated into the liver fibrosis model from the vessels themselves. The extent of Muse cell integration was not homogenous, and there was a tendency for higher integration in severely damaged areas (Fig. 5A). Therefore, 10 different randomly selected areas were subjected to measurement. The Muse group demonstrated a higher percentage of H-Golgi⁺ cells per total cell number in a 1-mm² section (5.78±2.39%), while that in the non-Muse group was very low (0.27±0.12%), with a statistically significant difference (p < 0.001), representing about 21 times higher numbers of H-Golgi cells in the Muse group (Fig. 5A and B).

Figure 5.

Differentiation of human multilineage-differentiating stress-enduring (Muse) cells in the liver fibrosis model. (A) Immunohistochemistry of H-Golgi⁺ cells located in the damaged liver at 8 weeks. A substantial number of H-Golgi⁺ cells (green) were detected around the vessels in the Muse group, while they were almost undetectable in the non-Muse group. Since integration of Muse cells was heterogenous, representative pictures of high, moderate, and low integration areas were demonstrated. Blue indicates nuclei staining with DAPI. (B) Proportion of H-Golgi⁺ cells relative to the total number of cells/mm² in the liver section. ***p < 0.001. (C) Expression of anti-human hepatocyte antibody (HepPar-1; red) in H-Golgi⁺ cells (green), (D) human albumin (red) in H-Golgi⁺ cells (green), and (E) human anti-trypsin (red) in human mitochondria⁺ cells (green). Scale bars: 50 μm (main figures), 25 μm (magnified boxes).

Immunohistochemistry was further performed in the Muse group. Muse cells that were positive for H-Golgi and human mitochondria were detected in the liver, expressing HepPar-1 (71.1±15.2% of H-Golgi⁺ cells) (Fig. 5C), human albumin (54.3±8.2% of H-Golgi⁺ cells) (Fig. 5D), and human anti-trypsin (47.9±4.6% of human mitochondria⁺ cells) (Fig. 5E). Therefore, integrated human Muse cells are suggested to differentiate spontaneously into hepatocyte marker-positive cells after integration.

Previous studies have posited that BM-derived hepatocytes in the damaged liver are occasionally formed by cell fusion^26,27. In order to examine whether the aforementioned differentiation in this study was a result of cell fusion or not, we conducted FISH analysis to investigate the existence of cell fusion between host hepatocytes (mouse origin, green color coded) and infused Muse cells (human origin, red color coded) (Fig. 6A). Neighboring sections of each FISH sample were subjected to double staining of H-Golgi and HepPar-1 in order to determine whether FISH signals were derived from differentiated human Muse cells. As a result, only 2.6±0.2% of H-Golgi⁺/HepPar-1⁺ human Muse cells that would approximately match with cells in FISH were suggested to be generated by cell fusion. Conversely, this suggests that approximately 97% of human Muse cells incorporated into the mouse liver tissue and differentiated into hepatocyte marker-positive cells without cell fusion.

Figure 6.

Fluorescence in situ hybridization (FISH) and functional analyses of the multilineage-differentiating stress-enduring (Muse) group (8 weeks). (A) Upper row shows FISH, and lower row shows immunohistochemistry for H-Golgi (green)/anti-human hepatocyte antibody (HepPar-1; red) in the neighboring section. In FISH, green indicates the mouse chromosome, red indicates the human chromosome. As a consequence, *1 circle is considered the mouse cell, *2 and *3 circles are the human cells without fusion with mouse cells, and *4 circle is the human–mouse fused cells. Since sections were made at 8- to 10-μm thickness, the neighboring section do not exhibit exactly the same location of nucleoli or shape of cytoplasm. However, *1, *2, and *3 cells in FISH can be projected to corresponding cells in immunohistochemistry. The *4 cell was not projected to immunohistochemistry. The *1 cell is negative for H-Golgi. The *2 and *3 cells are double positive for H-Golgi⁺ and HepPar-1, thus H-Golgi⁺ human Muse cells are suggested to differentiate into hepatocyte marker HepPar-1⁺ cells without fusing with mouse hepatocytes. Blue indicates nuclei staining with DAPI. Scale bars: 25 μm. (B) Reverse transcription polymerase chain reaction (RT-PCR) of human-specific albumin, human cytochrome P450, family 1, subfamily A, polypeptide2 (CYP1A2), human glucose-6-phosphatase (Glc-6-Pase), and human and mouse β-actin. The human liver was used as a positive control, and the liver of the vehicle group was used as a negative control.

Expression of human-specific mature functional hepatocyte markers, such as human-specific albumin, human cytochrome P450 1A2 (CYP1A2; an enzyme involved in drug metabolism)²⁸, and human Glc-6-Pase (an enzyme related to free glucose formation)²⁹ were determined by RT-PCR. These makers showed a high level of expression in the livers of the Muse group, while those of the non-Muse and vehicle groups did not express these markers. Notably, the human β-actin signal was under the detection level in the non-Muse-transplanted liver (Fig. 6B), consistent with histological data of the non-Muse group in Figure 5A and B, demonstrating that the majority of non-Muse cells did not remain in the liver after infusion.

In order to verify the key element involved in the spontaneous differentiation of Muse cells into hepatoblast/hepatocyte lineage cells, we analyzed the expression of human-specific hepatoblast/hepatocyte markers in a human Muse cell–mouse Hepa-1-6 cell system. Naive human Muse cells did not express human Sox17, CK18, Prox1, or AFP (Fig. 7). Neither of the conditioned media of intact or etoposide-treated apoptotic mouse Hepa-1-6 cells induced expression of human hepatoblast/hepatocyte markers in human Muse cells (Fig. 7A and C). Combination of human Muse cells and intact mouse Hepa-1-6 in coculture also did not induce human marker expression in human Muse cells (Fig. 7B). However, when human Muse cells were cocultured with etoposide-treated apoptotic mouse Hepa-1-6, human CK18 appeared at day 3, human Sox17 and AFP at 1 week, and human Prox1 at 2 weeks (Fig. 7D), suggesting the importance of direct interaction between Muse cells and damaged hepatocytes rather than humoral factors produced by host hepatocytes for spontaneous differentiation of Muse cells into hepatocytes.

Figure 7.

Differentiation of human multilineage-differentiating stress-enduring (Muse) cells into hepatocyte lineage cells by coculture with apoptotic hepatocytes. Expression of human-specific sex-determining region Y-box 17 (Sox17), human cytokeratin 18 (CK18), human prospero homeobox protein 1 (Prox1), and human a-fetoprotein (AFP) in quantitative polymerase chain reaction (qPCR). Since primers were specific for human, mouse fetus liver was consistently negative for all markers while human fetus liver was positive. Human Muse cells, originally negative for all the markers, became positive for CK18 at day 3, Sox17 and AFP at 1 week, and Prox1 at 2 weeks only when cocultured with etoposide-treated apoptotic mouse Hepa-1-6. On the other hand, supply of conditioned medium from intact or apoptotic mouse Hepa-1-6 and coculture with intact mouse Hepa-1-6 did not induce human-specific marker expressions.

Discussion

Human Muse cells are unique in their efficient homing rate into damaged tissue simply by intravenous injection, capacity for survival in the hostile environment of damaged tissue due to their stress tolerance, in vivo differentiation into cells compatible with the tissue into which they homed, and nontumorigenicity³⁰. The present study described the specific characteristics of intravenously injected human Muse cells, such as ability to migrate to and integrate into the damaged liver, and secondly, the high efficiency with regard to tissue regeneration, functional recovery, and attenuation of fibrosis in a liver fibrosis SCID mouse model.

Our in vitro study showed the ability of Muse cells to spontaneously differentiate into cells positive for hepatoblast/hepatocyte markers, namely, DLK, α-fetoprotein, CK18, and CK19, on gelatin culture. Human Muse cells were further shown to be successfully integrated into the damaged liver and differentiated in vivo into cells expressing human albumin, HepPar-1, and human anti-trypsin, as well as human functional markers, the detoxification enzyme human CYP1A2, and human Glc-6-Pase. These results are consistent with previous studies reporting that Muse cells are capable of crossing oligolineage boundaries from the mesoderm to an ectodermal lineage such as neuronal cells, melanocytes, and keratinocytes, or to an endodermal lineage such as cells positive for GATA-6, α-fetoprotein, CK7, and albumin, while non-Muse cells are not^14-16,31.

The in vitro migration assay indicated that Muse cells are capable of perceiving signals that are produced by the damaged liver and released into the blood stream, unlike non-Muse cells. Such specific Muse cell migration capability could be partially mediated by specific receptor(s) present in Muse cells, and our data suggested that CXCR4 and possibly c-Met are candidate receptors that may mediate migration of Muse cells. Notably, the CXCR4 antagonist AMD3100 substantially suppressed in vitro migration of Muse cells toward the 24-h CCl₄ serum. Thus, CXCR4 is suggested to be one of the players that mediate Muse cell migration. However, the extent of the suppression was not 100% and Muse cell migration still occurred; non-Muse cells also expressed CXCR4 and their migration, while not as outstanding as that seen in Muse cells, was also suppressed substantially by AMD3100. These data suggest that other critical factor(s) are also responsible for the specific migration and integration of Muse cells. Since in vitro migration of Muse cells peaked at 24 h but sharply declined at 48 h (Fig. 2B and C), signal(s) playing a central role in mediating Muse cell migration may appear in the peripheral blood promptly after damage and quickly decrease after several days. These results strongly suggest the Muse cells respond promptly to the signal(s) produced by the damaged tissue, which explains the efficiency of Muse cell treatment for acute liver damage.

SDF-1, a ligand of CXCR4, is reported to maintain high levels in the liver at least until 3 weeks after damage, and thus the dynamics of SDF-1 seems to be different³². Critical factor(s) are yet to be clarified and should be explored in future studies. Along with in vitro study, efficient accumulation of Muse cells into the damaged liver in vivo was confirmed by the detection of a human-specific Alu sequence. The specific migration and integration capacities of human Muse cells into the damaged liver were further supported by immunohistochemistry and RT-PCR; H-Golgi⁺ and anti-human mitochondria⁺ cells were detected in the damaged livers of the Muse group using immunohistochemistry, along with human β-actin gene expression in RT-PCR.

Integrated human Muse cells recognized as H-Golgi⁺ and human mitochondria⁺ cells expressed hepatocyte markers, human albumin, HepPar-1, and human anti-trypsin in the liver fibrosis model. The expression of human albumin, human CYP1A2, and human Glc-6-Pase in the Muse cell-transplanted liver and functional recovery in the total bilirubin and serum albumin strongly suggest that human Muse cells can spontaneously differentiate into cells corresponding to functional human hepatocytes in the SCID mouse liver after homing and integration. These properties of Muse cells render pretransplantation induction unnecessary for treating patients with liver disease. The conditioned medium and coculture experiment suggested that the direct interaction between Muse cells and damaged host hepatocytes is the critical factor for spontaneous differentiation of Muse cells, rather than humoral factors produced by host hepatocytes. Indeed, the conditioned medium of both intact and apoptotic mouse Hepa-1-6 failed to induce expression of human-specific hepatoblast/hepatocyte markers in human Muse cells. Similarly, the direct interaction of human Muse cells and intact mouse Hepa-1-6 also failed to trigger Muse cell differentiation. The precise mechanisms of how Muse cells interact with damaged host hepatocytes, how they receive the instruction from damaged host hepatocytes, and how Muse cells can deliver mechanistic advantages are important points to be clarified in the future.

Previous reports indicated that BM-derived hepatocytes in the damaged liver model may be formed by cell fusion^26,27. However, the frequency of cell fusion reported in those studies seems to be very low²⁷. Cell fusion may occur in certain species by cell type-specific proteins (e.g., syncytin), particularly in the case of primates^33,34. In our system, most of the human signals (Muse derived) did not merge with the host murine signal in FISH. Furthermore, most of the liver cells generated by the human Muse cells express functional human markers including human albumin, anti-trypsin, human CYP1A2, and human Glc-6-Pase, suggesting that fusion of human–murine cells is unlikely to be a major event during differentiation of Muse cells into hepatocytes. However, further studies are needed to evaluate the involvement of cell fusion when allogeneic Muse cells are supplied to liver damage models.

Muse cells dramatically reduced fibrotic areas observed in both Sirius red and Masson's trichrome staining. These results demonstrate that Muse cells efficiently integrated into the damaged area, favoring either fibrolysis or attenuation of fibrosis. Of note, the number of cells we transplanted was not substantially large. Three infusions of 5 × 10⁴ Muse cells, however, were sufficient to increase levels of serum albumin, to ameliorate levels of serum bilirubin, and to attenuate liver fibrosis. The modest effect of non-Muse cells to improve function recovery could be attributed in part to trophic factors and/or cytokines secreted by these cells during the short time that they remained before disappearing from the host liver. This effect could be attributed to the similarity of non-Muse cells and MSCs. Nearly 99% of MSCs are in fact composed of non-Muse cells, so the outcome of non-Muse cell transplantation is considered very close to that of MSCs in terms of trophic and anti-inflammatory effects^6,35.

Muse cells could substantially attenuate liver fibrosis compared to non-Muse cells. Muse cells showed a very high level of MMP-9 protein expression in comparison to non-Muse cells in ELISA (Fig. 4H). On the other hand, no significant differences were detected in MMP-1 and MMP-2 between both cell types (Fig. 4F and G). These results suggested that MMP-9 participated in the attenuation of liver fibrosis in the Muse group liver. As seen in Figure 5A and B, Muse cells could remain in the host liver for up to 8 weeks, while non-Muse cells could not. Even though the MMP production ability of each cell is equivalent, Muse cells can survive and remain in the host liver and may effectively and consistently provide MMPs, including MMP-1 and MMP-2, whereas non-Muse cells that do not remain in the tissue may have limitations in this regard.

Potential Relevance of Muse Cells for Clinical Use

BM-derived cells, such as BM mononuclear cells (BM-MNCs) and BM-MSCs, are currently suitable for clinical application and industrial use and thus have been a subject of rigorous study^35-39. It has been reported that allogeneic transplantation of BM-MNCs into a rodent liver fibrosis model resulted in differentiation of transplanted BM-MNCs into hepatocytes, improvement of the serum albumin level, and amelioration of liver fibrosis, exerted by production of MMP-9 by transplanted cells^35,36. Similarly, previous clinical studies of autologous BM-MNC infusion (ABMi) showed promotion of hepatocyte proliferation and improvement of serum total protein and albumin levels while causing no adverse events³⁷. Moderate improvement in bilirubin and albumin levels was recognized in ABMi, while infused cells seemed not to generate a large number of new hepatocytes in the damaged liver^37-39.

BM-MSCs represent another BM-derived cell source that has been studied for cell-based therapy in liver disease models³⁹. BM-MSC transplantation for hepatic diseases has been performed in various animal models including a mouse model of steatohepatitis and a rat model of liver fibrosis^11,40 and has demonstrated nontumorigenicity. BM-MSCs are suggested to provide pleiotropic effects including cytokine production, immunosuppression, and regeneration^10,12,13. For example, in a clinical study involving patients with hepatitis B virus (HBV)-induced hepatic failure, the BM-MSC transplantation group showed improvements in hepatic function shortly after treatment relative to a control group without adverse effects⁴¹. Nevertheless, the BM-MSC-transplanted group and control group showed no difference in long-term prognosis and long-term liver function⁴¹.

Implementation of stem cell therapy is not only a matter of cell number. Although in-depth analysis is still needed, transplantation of whole BM cells or BM-MSCs, even those containing equivalent numbers of Muse cells as done in this study, may not yield the same results. With the presence of a large number of non-Muse cells, Muse cells in MSCs may not be able to fully perceive signals from the damage site and their efficiency may be weakened. Therefore, the effect of an equal number of Muse cells either as a pure population or highly diluted (several percentage of the total population) with non-Muse cells may be different. If so, it might be necessary to omit unnecessary cell populations and purify Muse cells in order to fully evoke their function.

Our results indicate that Muse cells (i) have a very high homing rate into the damaged liver and remain in the host tissue for a period of 8 weeks, (ii) integrate into the damaged liver simply by intravenous injection, (iii) spontaneously differentiate into hepatocytes in vivo, (iv) attenuate fibrosis, and (v) significantly improve liver function in liver fibrosis mouse model. These findings suggest that they are the feasible stem cell type for the treatment of liver disease. Since the Muse cells we infused into the blood stream were naive cells and were not induced into hepatocytes prior to transplantation, pretransplant induction at a cell processing center, unlike ESCs and iPSCs, might not be necessary for treating patients with liver disease. Muse cells are reported to show low telomerase activity comparable to that of somatic cells such as fibroblasts^15,16. Consistently, they did not generate teratomas when injected into immunodeficient mouse (Nog mouse) testes for up to 6 months, while ESCs and iPSCs formed teratomas at about 8 to 12 weeks^14-16,42, suggesting nontumorigenicity of Muse cells. This study also showed no tumor formation for up to 8 weeks. A longer validation period in the liver model will robustly confirm the safety of Muse cells in future studies. Muse cells, represented by their relevant regenerative performance, are expected to provide a clinically feasible and simple step approach consisting of collection from the BM by SSEA-3, expansion, and then injection into the blood stream. However, determining the most appropriate number of Muse cells for efficient treatment, an optimized preparation method for collecting Muse cells, the timing of transplantation, and methods for efficient Muse cell injection, ensuring that there is no pluripotent “contamination” in integrated cells, are all necessary for future investigation.