Induction of umbilical cord blood–derived β2mc-Met+ cells into hepatocyte-like cells by coculture with CFSC/HGF cells



Several studies have indicated that adult stem cells derived from bone marrow (BM) and cord blood (CB) can differentiate into hepatocyte-like cells. This ability is important for the treatment of hepatic diseases with BM or CB as a potential approach. However, methods are still being developed for the efficient induction of stem cell differentiation and expansion to get enough cells to be useful. In the present study, we enriched a subset of umbilical cord blood β2mc-Met+ cells (UCBCCs) and investigated the combination effect of liver nonparenchymal cells (cirrhotic fat-storing cells [CFSCs]) and hepatocyte growth factor (HGF) on the induction of UCBCCs into hepatocyte-like cells. UCBCCs were cocultured with CFSC/HGF feeder layers either directly or separately using insert wells. Flow cytometric analysis showed that most UCBCCs were CD34+/−CD90+/−CD49f+CD29+Alb+AFP+. After cocultured with transgenic feeder layers for 7 days, UCBCCs displayed some morphologic characteristics of hepatocytes. Reverse-transcription polymerase chain reaction (RT-PCR) and immunofluorescence cell staining proved that the induced UCBCCs expressed several hepatocyte specific genes including AFP, Alb, CYP1B1 and cytokeratins CK18 and CK19. Furthermore, the induced cells displayed liver specific functions of indocyanine green (ICG) uptake, ammonium metabolism and albumin secretion. Hence, our data have demonstrated that UCBCCs might represent a novel subpopulation of CB-derived stem/progenitor cells capable of successful differentiation into hepatocyte-like cells when incubated with CFSC/HGF cells. In conclusion, not only HGF but also CFSCs and/or the secreted extracellular matrix (ECM) have been shown to be able to serve as essential microenvironment for hepatocyte differentiation. (Liver Transpl 2005;11:635–643.)

Cell therapy has emerged as a strategy for patients to repair damage and replace degenerative tissues. Over the past several decades, the transplantation of hematopoietic stem cells has been widely used clinically.1, 2 Recent studies have indicated that adult stem cells, especially the stem cells derived from bone marrow (BM) or cord blood (CB) could also differentiate into nonhematopoietic cells, especially hepatocytes both by transplantation under certain circumstances in vivo and culture in vitro.3–11 This plasticity provides vast potential for the treatment of disease such as liver failure with their own stem cells, although the mechanism underlying this plasticity remains to be determined.12

Hepatocyte growth factor (HGF), also known as scatter factor, as a broadly distributing and actual factor that promotes many tissue restorations, plays an essential role in liver development and regeneration.13, 14 Using gene engineering means with mammal cells as host expression system to produce recombined human hepatocyte growth factor (hHGF) is a potent approach for mass production of HGF with high bioactivity. To make more scientific and effective use of HGF, we constructed a transgenic cell line, cirrhotic fat-storing cell (CFSC)/HGF, to express hHGF stably and effectively in CFSCs, a hepatic nonparenchymal cell line that has been isolated and established from CCl4- injured rat liver by Greenwel et al.15

β2 microglobulin (β2m) is the most conserved protein expressed broadly on almost all nucleated cells in mammalian. However, many immortal cancer cells and the inner cell mass of the preimplantation blastula do not express β2m, so β2m cells might represent progenitors possessing proliferation potency.16 Soukiasian et al.17 have demonstrated β2m cells of BM to be capable of engraftment in areas of transmural myocardial scar, with de novo formation of cardiac myocytes, which verified the “stemness” character of the β2m cells further. Since HGF plays an essential role in liver development and regeneration, and all biological responses induced by HGF are elicited by binding to its receptor, c-Met, a transmembrane tyrosine kinase encoded by Met proto-oncogene,18 we assumed that β2mc-Met+ cells might have the potential to differentiate into hepatocytes.

Thus, in this study we purified the above subpopulation from CB, umbilical cord blood β2mc-Met+ cells (UCBCCs) and proved their potential of differentiation into hepatocyte-like cells at the sustainment of transgenic feeder layer CFSC/HGF. Here we provide the evidence that UCBCC might act as a promising cell source for cellular therapy for liver diseases in the future. The transgenic CFSC/HGF cell line might induce the differentiation by secretion of aim protein HGF and production of extracellular matrix (ECM) and other growth factors. Furthermore, the direct interaction of these two cell populations might play a significant role in the process.


BM, bone marrow; CB, cord blood; HGF, hepatocyte growth factor; CFSC, cirrhotic fat-storing cells; β2m, β2 microglobulin; UCBCCs, umbilical cord blood β2mc-met+ cells; ECM, extracellular matrix; PBS, phosphate-buffered saline; MACS, magnetic bead cell sorting; IgG, immunoglobulin G; ICG, indocyanine green; AFP, α-fetoprotein; Alb, albumin.

Materials and Methods

Isolation of Umbilical Cord Blood Cells

Human cord blood samples from healthy, full-term deliveries were drained from the end of the cord into glass bottles with 20 U/mL preservative-free heparin. The blood was first diluted 1:1 with phosphate-buffered saline (PBS, pH 7.4) and then subsided with a quarter volume of 0.5% methafibrin (Sigma, St. Louis, MO) for 30 minutes to precipitate erythrocytes. Cells were pelleted by centrifugation and resuspended in PBS. Mononuclear cells were isolated by fractionation of the cells through density gradient centrifugation on a Ficoll-Paque (P =1.077, Amersham Biosciences, Piscataway, NJ) for 20 minutes at 400g. The interface containing the mononuclear cells was collected and stored at 4°C until further use.

Two-Step Magnetic Bead Cell Sorting (MACS)

The UCBCCs were isolated by a two-step indirect MACS technique. In brief, the mononuclear cells previously labeled with rabbit anti-β2m antibody were incubated with super- magnetic microbeads conjugated with anti-rabbit immunoglobulin (IgG), and then passed through a depletion column placed in a magnetic field (Miltenyi Biotec, Bergisch Gladbach, Germany). The unlabeled cells (β2m cells) passing through the column freely were incubated with rabbit anti–c-Met antibody, and then were conjugated with anti-rabbit IgG magnetic beads. Lastly, they were passed through a magnetic selection column. The UCBCCs were retained in the column under the high gradient magnetic field. After the column was removed from the magnetic field, UCBCCs were washed off the column with PBS. Viability of the cells was consistently over 95%, as determined by trypan blue exclusion.

Flow Cytometric Analysis

The UCBCCs surface phenotype was characterized using standard flow cytometric analysis. Briefly, 0.3 to 0.5 × 106 cells were stained for 20 minutes at room temperature with the appropriate fluorescein isothiocyanate conjugated monoclonal antibodies of CD34, CD90, CD49f, CD29, albumin (Alb), α-fetoprotein (AFP) or the control monoclonal antibodies of anti-human IgG (Serotec, Oxford, UK) For intracytoplasmic AFP staining, cells were fixed in 80% ethanol after permeabilization with 0.1% Triton X-100/PBS. Following immunostaining, cells were washed and resuspended in fluorescence-activated cell-sorting (FACS) fix (1% formaldehyde) and then analyzed using a Becton Dickinson FACSCalibur machine (BD Biosciences, Bedford, MA) and CellQuest software.

Establishment of CFSC/HGF Feeder Layers

A CFSC/HGF cell strain was constructed by transduction of the hepatic stellate cell line, CFSC (granted by Greenwel et al.15), with recombined retroviral vector pMSCV-HGF reported previously.19 In brief, the recombined pMSCV-HGF plasmid was constructed by inserting the amplified human HGF fragments into Hpa I/Bgl II sites of linearized retroviral vector pMSCVneo (R&D Systems, Minneapolis, MN). Then, after the CFSCs were infected with pMSCV-HGF–containing medium for four times in every 24-hour interval and screened with G418 for 2 weeks, the transgenic cell strain CFSC/HGF was constructed finally following HGF-expressing G418R clone selection, subculture, and identification with enzyme-linked immunosorbent assay. The control cell strain CFSC/neo was prepared similarly, using CFSCs infected with empty vector pMSCV-neo.

The feeder layers were established by irradiation of 80% confluent normally cultured CFSC/HGFs or CFSC/neo with 18 Gy of 60Co-γ radiation. Cultures were maintained in a DMEM/F12 complete medium supplemented with 10% fatal bovine serum at 37°C in a humidified air containing 5% CO2.

Coculture of UCBCCs With Transgenic Feeder Layers

Cocultures were performed with 2 different methods as follows: (1) The UCBCCs were cultured directly onto feeder layers noted above at seeding density of 5 × 104 cells/cm2; (2) The Transwell dual chambers with the top one coated with feeder layers and the lower one seeded with UCBCCs on Matrigel (R&D Systems) were used in separate coculture (Fig. 1). The complete medium was changed into a conditional one supplemented with nicotinamide 0.61 g/L, insulin-transferrin-selena sodium, dexamethasone 0.1 μmol/L, bovine serum albumin 2 g/L, glucose 1 g/L, galactose 2 g/L, ornithine 0.1 g/L, proline 0.03 g/L, glutamine 0.73 g/L (Invitrogen Life Technologies, Carlsbad, CA), and some microelements. The medium was replaced semiquantitively every 3 days.

Figure 1.

Protocol for coculture of UCBCCs and CFSC/HGFs. CFSC/HGF and UCBCC cocultures were performed as follows. Method A: UCBCCs were cultured directly on CFSC/HGF feeder layer. Method B: Separated by a semipermeable polyterafluoroethylene (PTFE) membrane (0.4 μm), CFSC/HGFs (top chamber) and UCBCCs (lower chamber) were cocultured in Transwell dual chambers. The two fashions were cultured in the conditional medium with medium changed every 3 days. Phenotype, morphology, and function were examined after 7 days of coculture.

Electron Microscopy

Cells were washed with PBS, fixed with 3% glutaraldehyde in 100 mmol/L sodium cacodylate buffer for 48 hours, post-fixed in 1% osmium tetroxide, dehydrated in graded alcohols, and embedded in Epon812 in sequence. Ultrathin sections were stained with uranyl acetate and lead citrate and viewed with a CM120 Electron Microscope (Philips, Amsterdam, the Netherlands).

Immunofluorescence Staining

After washing 3 times with PBS, cells were fixed with cold acetone for 30 minutes. The fixed cells were permeabilized with 0.1% Triton X-100/PBS for 7 minutes on ice and then incubated in PBA (3% bovine serum albumin/PBS) for 30 minutes for blocking. Indirect immunofluorescence staining for albumin with secondary fluorescein isothiocyanate (FITC) anti-rabbit IgG and for cytokeratins 8 and 18 (CK8&18) with secondary TRITC anti-mouse IgG (Santa Cruz Biotechnology, Santa Cruz, CA) were performed. To visualize the coexistence of albumin and CK8&18, 3-dimensional digital reconstruction images were obtained using Radiance 2100 confocal system (Bio-Rad Laboratories, Hercules, CA) in conjunction with a TE300 microscope (Nikon, Kingston, UK) with a factor N computer zoomed image in a single optical plane.

Reverse-Transcription Polymerase Chain Reaction (RT-PCR)

Total RNA was extracted from 3 × 105 cells using TRIzol (Sigma) according to manufacturer's instructions. For complementary DNA synthesis, the mixtures were incubated at 37°C for 60 minutes. Subsequently, PCR was performed with 30 cycles at 94°C for 30 seconds, 55°C for 30 seconds, and 72°C for 30 seconds. The primers for infant or adult hepatocyte specific AFP, Alb, CK18, CK19, cytochrome P450 (CYP) 1B1 and the positive control α-tubulin are listed in Table 1. The PCR products were subjected to 1% agarose gel electrophoresis and visualized by staining with ethidium bromide.

Table 1. Primers Used for Reverse-Transcription Polymerase Chain Reaction (RT-PCR)
Primer namePrimersLength (bp)
  1. Abbreviation: bp, base pair.


Indocyanine Green (ICG) Uptake Study

After the cells were washed with PBS, ICG solution was added to the induced UCBCCs at a final concentration of 1 mg/mL. Preliminary experiments indicated that there was no adverse effect on cell viability at this concentration. The cells were incubated with ICG at 37°C for 15 minutes, rinsed 2 times with PBS, and then cultured with complete medium again. The hepatocyte specific uptake and secretion of ICG in induced UCBCCs were examined microscopically.

Urea and Albumin Production in Culture Supernatant

After the UCBCCs were induced for 7 days in fresh serum-free medium, they were exposed to 0.15 mol/L NH4Cl for 8 hours. The urea and albumin levels of culture supernatant were detected with OLYMPUS AU-600 automatic biochemical analysis instrument (Olympus, Tokyo, Japan).


Data are shown as means ± SD, and the results were analyzed by ANOVA and Fisher's exact tests with SPSS software.


Phenotype of Freshly Isolated UCBCCs

The UCBCCs accounted for 2.5% ± 1.2% of total mononuclear cells of cord blood samples (n = 6). As shown in Figure 2, 17.45% and 11.9% of the freshly isolated UCBCCs expressed the hematopoietic stem cell markers CD34 and CD90; 89.49% and 97.39% of them contained the α6 β1 integral subunits CD49f and CD29; and 98.19% and 71.93% of them expressed hepatocyte-specific AFP and albumin, respectively.

Figure 2.

Flow cytometric analysis of UCBCCs. UCBCCs isolated with two-step MACS were stained with FITC-conjugated monoclonal antibodies of CD34, CD90, CD49f, CD29, AFP, Alb, or human IgG as control after fixed in 80% ethanol (for AFP, cells were fixed after permeabilization with 0.1% Triton X-100/PBS), then analyzed with flow cytometry. Percentages of FITC-labeled cells are shown in each panel. Establishment of the gate was based on the profile of the negative control.

Morphologic Changes During Coculture

The G418R transgenic monoclones were screened out and the cell strain with satisfactory HGF expression was determined with enzyme linked immunosorbent assay (9.76 ng/mL), and then was subcultured extendedly (Fig. 3A-B). After being cocultured for 4 days with transgenic feeder layer (CFSC/HGFs), parts of UCBCCs became enlarged. The enlargement became obvious to the seventh day, with some cells aggregated with each other and mitotic figures presented, which morphologically resembled hepatocytes, with big, round and bi-/multi-nucleus and abundant cytoplasm (Fig. 3C-I). We took count of differentiated cells for 10 fields under a microscope with the relative magnification of ×200. Numbers of differentiated cells in the direct cocultures and in the Transwell cultures were 37.9% ± 5.7% and 28.8% ± 4.2%, respectively (P < 0.05). The presence of hepatocyte-like morphology, including plentiful round or elliptical mitochondria, endoplasmic reticulum, glycogenic granules and microvilli on the surface of the cells, was confirmed with transmission electron microscope (Fig. 3J).

Figure 3.

Morphologic views. Appearance of transgenic feeder layers and UCBCCs induced on CFSC/HGFs for 7 days. (A) The G418R transgenic monoclonals (original magnification, ×40) were screened out. (B) The satisfactory HGF expression cell strain was subcultured extendedly (original magnification, ×200). Following coculture of the UCBCCs with CFSC/HGFs for 7 days, some of the UCBCCs became enlarged. Cell aggregates and mitotic cells with big, round and bi-/multi-nucleus and abundant cytoplasm were present. (C-E) UCBCCs cultured directly on feeder layers (original magnifications, ×100 [C]; ×200 [D]; and×400 [E]). (F-H) UCBCCs and CFSC/HGFs cultured separately in a dual chamber intervening with semipermeable PTFE membrane (original magnifications, ×100 [F]; ×200 [G]; and ×400 [H]); thick arrows point out the bi-nucleus hepatocyte-like cells, which differentiated from UCBCCs. (I) UCBCCs and CFSC/neos cultured separately for 7 days (original magnification, ×200). (J) The hepatocyte-specific ultrastructures of plentiful round or elliptical mitochondria, endoplasmic reticulum, glycogenic granules, and microvillus (indicated with thin arrows) on the surface of the cells were confirmed in induced UCBCCs with a transmission electron microscope (original magnification, ×12,800).

Induced UCBCCs Expressing Hepatocyte Specific Genes

Immunofluorescence detection indicated that differentiated UCBCCs distributed along with CFSC/HGFs, and expressed hepatocyte-specific Alb and CK8&18 (Fig. 4). Freshly isolated UCBCCs did not express CK18 and CYP1B1 messenger RNA, and expressed low levels of AFP, Alb, and CK19. Following coculture with CFSC/HGF feeder layers for 7 days, the expression of above genes in differentiated UCBCCs was appeared and increased, which were not expressed in the CFSC/HGF cells (Fig. 5).

Figure 4.

Double immunofluorescence labeling of UCBCCs after 7 days of culture on CFSC/HGF feeder layers. Three-dimensional digital reconstruction images of cells after 7 days in coculture were obtained using a confocal system. (A-C) The red, green, and blue signals (immunofluorescence) represent staining for CK8&18, Alb, and nucleus, respectively. (D) The yellow signals represent the induced UCBCCs for their staining of both CK8&18 and Alb, while the blue ones represent CFSC/HGF feeder layers for the cells stained with neither CK8/18 nor Alb, with the exception of DAPI-nuclear. (E-H) Bi-nucleus hepatocyte-like cells. TRITC, rhodamine conjugate; DAPI, 4′,6-diamidino- 2-phenylindole.

Figure 5.

Detection of hepatocyte-specific messenger RNA by reverse-transcription polymerase chain reaction (RT-PCR). After UCBCCs were cocultured with transgenic feeder layers for 7 days, total messenger RNA was extracted for reverse-transcription polymerase chain reaction. The α-tubulin messenger RNA was a housekeeping gene for internal normalization. Lanes 1-6 show AFP, Alb, CK18, CYP1B1, α-tubulin, and CK19, respectively. Panels A and B show after (A) and before (B) UCBCCs' coculture with CFSC/HGF feeder layers. Expression of hepatocyte-specific (AFP, Alb, CK18, CYP1B1, and CK19) messenger RNAs appeared and/or were increased after induction for 7 days. The experiments were repeated thrice, similar findings were observed, and the representative histograms are shown in the graph. (*P< 0.01 compared to before differentiation.) (C) CFSC/HGF cultured alone. Lanes 7 and 8 were hHGF and β-actin.

Induced UCBCCs Displaying Hepatocyte-Specific Functions

The induced enlarged UCBCCs turned to deep green following incubation with ICG for 15 minutes, but UCBCCs with no morphologic change and CFSC/HGFs did not take up ICG at all (Fig. 6). The color of stained cells faded out completely within 4 hours of culture with refreshed complete medium again. The extent of albumin and urea production was greater following direct coculture on CFSC/HGF feeder layers than when Transwell dual chambers were used. Fresh UCBCCs, CFSC/HGFs, and CFSC/neos produced little or only small amounts urea or albumin when cultured alone. UCBCCs cocultured with CFSC/neo produced some albumin and urea too, but levels were significantly less than when UCBCCs were cultured with the CFSC/HGF feeder layers (Fig. 7).

Figure 6.

ICG uptake test of UCBCCs cocultured on CFSC/HGF feeder layers was performed at the seventh day. The enlarged UCBCCs turned to deep green, while no change was presented in UCBCCs with no morphologic change and the transgenic feeder layers. (original magnification, ×400).

Figure 7.

Urea and albumin production in different culture system were detected. 1: Positive control (hepatocytes of human fetal liver cultured in conditional medium). 2 and 3: UCBCCs cocultured with CFSC/HGF and CFSC/neo feeder layers directly. 4 and 5: UCBCCs cocultured with CFSC/HGF and CFSC/neo feeder layers separately in Transwell dual chambers. 6 and 7: UCBCCs and CFSC/HGF cultured alone, respectively. Values are presented as means ± SD, n = 6. *P< 0.05.


At present, there are approximately 10 UCB banks and nearly 20,000 UCB samples in China. The ready availability of hematopoietic stem cells and their weak immunogenicity make UCB an ideal source of hematopoietic stem cells to be used clinically.20 The banked CB would be more valuable for it has been reported to be able to differentiate into functional hepatocytes as BM cells did. So far, several groups have proved some of CB cells might differentiate into hepatocytes in vivo or in vitro.21–23 However, which subset or subsets of CB may have this plasticity still remains to be determined.

The cells possessing potential for hepatic differentiation may express some early endodermal markers before full differentiation into hepatocytes. With this in mind and in order to forecast the hypothesis which we have mentioned at the beginning, we thought the population of β2mc-Met+ cells from CB might be the possible candidate. We checked their characters with flow cytometric analysis and found that the freshly isolated UCBCCs accounted for about 2.5% of the total nucleated cells, with consistent expression of the embryonic epithelium specific α6 (CD49f) integrin and β1 (CD29) integrin, fetal liver–specific AFP and some of them for liver-specific Alb, indicating that the cells were undifferentiated and might bear potential of differentiation into hepatocytes. Those above characteristics coincide with the view of Suzuki's report that coexpressing α6 and β1 integrin subunits indicated hepatic stem/progenitor cells in the fetal mouse liver24 and of Kubota's report that expression of variant forms of AFP transcripts may reflect the ability of BM progenitors to differentiate into endodermal cells.25 Furthermore, the fact that some of them express CD34 and CD90 is accordant with previous theories that some BM or CB stem cells carrying hematopoietic properties have shown to differentiate into hepatocytes,26, 27 or to exist a human stem cell population with both hematopoietic and hepatic potential.28 Recently published papers by the group of Ratajczak also provided the same perspective on BM not only as a home for hematopoietic stem cells but also a “hideout” for CD34+/Ac133+/CXCR4+ stem/progenitor cells, which could differentiate into hepatocytes.29

Although the mechanism of the adult stem cell differentiation into a hepatocyte lineage is not understood, most scientists believed that the microenvironment of the stem cells might regulate their differentiation.30 This phenomenon has been termed “milieu-influenced” differentiation. For example, BM cells' engraftment as hepatocytes using male-to-female BM transplantation in humans was demonstrated in response to liver damage, which might promote BM cells' differentiation into hepatocytes.31 Agents including sera from liver failure patients or animals,32 demethylating agents, HGF, oncostatin M (OSM), fibroblast growth factor (FGF), and stem cell factor (SCF) have been used successfully to mimic the ECM of the liver regenerative environment and induce BM cells differentiation into hepatocytes and/or cholangiocytes in vitro.16, 34–36

Besides cytokines and/or growth factors, the reciprocal interaction between hepatic parenchymal cells and nonparenchymal cells might determine the network balance of cytokines and liver microenvironment regulation.37 Hepatic stellate cells residing in the space of Disse, as the important nonparenchymal cells, could supply a crucial microenvironment for many physiologic and pathologic events, such as liver development, maturation and regeneration, by producing ECM, including fibronectin and laminin, and expressing various soluble growth factors that might regulate the proliferation and differentiation of hepatic stem cells.38, 39 To realize a higher differentiation degree of UCBCCs, we established a more favorable environment to combine hepatic stellate cells and the most important growth factor, HGF, actively by the establishment of a transgenic feeder layer through transduction of CFSC with recombined retroviral vector pMSCV-HGF. Then we cocultured UCBCCs with CFSC/HGF to support their differentiation through cell-matrix, cell-cytokine and/or cell-cell interactions.

By seeding UCBCCs on the CFSC/HGF feeder layers without any cytokine added, distinct morphological changes in some UCBCCs emerged after coculture for 4 days, and the changes increased even more as time went on. Morphological characteristics of hepatocyte occurred after 7 days of coculture in about 38% to 46% of UCBCCs, which were identified to possess hepatocyte-like functions. UCBCCs could also differentiate into hepatocyte-like cells by coculturing with CFSC/HGFs separately using Transwells, although the differentiation efficiency was lower than the former. Moreover, the tests showed limited contribution of control CFSC/neo to the induction, which proved the important sustainment of nonparenchymal cells and extracellular matrices on the stem cell differentiation and prompted us to consider the insufficient factor secretion in nontransfected hepatic stellate cells only.

In this study, we showed that UCBCCs might represent a novel subpopulation of adult stem cells capable of successful and substantial differentiation into hepatocyte-like cells, and the inducing conditions created by HGF/CFSCs could simulate liver regeneration environment in vivo and support that differentiation efficiently. We have not only validated the influence of appropriate environments on the fate of CB stem cells, providing a reasonable model of mechanism exploration of adult stem cell differentiation, but also possibly contributed to future large-scale production of functional hepatocytes for medical purposes.


The authors thank Dr. Buo Dong (Beijing Institute of Radiology) for the flow cytometric analysis, and Dr. Tao Zhou for the 3D digital confocal system analysis, and they appreciate the donation of umbilical cord blood samples from Beijing Great Wall Hospital.