Previously, we reported a system to enrich mouse fetal hepatic progenitor cells (HPCs) by forming cell aggregates. In this study, we sorted two cell populations, CD49f+Thy1−CD45− cells (CD49f-postive cells) and CD49f±Thy1+CD45− cells (Thy1-positive cells), from the cell aggregates using a flow cytometer. CD49f-positive cells stained positive for endodermal specific markers such as α-fetoprotein (AFP), albumin (ALB), and cytokeratin 19 (CK19), and are thus thought to be HPCs. However, Thy1-positive cells were a morphologically heterogeneous population; reverse-transcription polymerase chain reaction (RT-PCR) and immunocytochemical analyses revealed the expression of mesenchymal cell markers such as α-smooth muscle actin, desmin, and vimentin, but not of AFP, ALB, or CK19. Therefore, Thy1-positive cells were thought to be of a mesenchymal lineage. When these two cell populations were co-cultured, the CD49f-positive colonies matured morphologically and stored a significant amount of glycogen. Furthermore, real-time RT-PCR demonstrated an increased expression of tyrosine amino transferase and tryptophan oxygenase mRNA, and transmission electron microscopy confirmed that co-cultured cells produced mature hepatocytes. However, when CD49f-positive cells were cultured alone or when the two populations were cultured separately, the CD49f-positive cells did not mature. These results indicate that CD49f-positive cells are primitive hepatic endodermal cells with the capacity to differentiate into hepatocytes, and that Thy1-positive cells promote the maturation of CD49f-positive cells by direct cell-to-cell contact. In conclusion, we were able to isolate CD49f-positive primitive hepatic endodermal cells and Thy1-positive mesenchymal cells and to demonstrate the requirement of cell-to-cell contact between these cell types for the maturation of the hepatic precursors. (HEPATOLOGY 2004;39:1362–1370.)
Because of a shortage of donors for liver transplantation, cell transplantation has been explored as a useful bridge or alternative therapy. Hepatocyte transplantation can improve liver function sufficiently to extend the waiting time for liver transplantation.1–4 However, using this as an effective clinical therapy requires the development of a cell source other than donated organs. Therefore, research is currently being conducted on hepatic stem and progenitor cells. In general, progenitor cells are highly expandable in vitro, easily cryopreservable, and quite resistant to hypoxic conditions.5 Hepatic progenitor cells (HPCs) mature rapidly into adult hepatocytes in quiescent liver6 and have far greater regenerative capacity than do adult hepatocytes in retrorsine-treated liver.7 However, very little functional analysis of transplanted HPCs has been performed, and it is currently unknown whether transplanted HPCs can improve liver dysfunction. To transplant fully functional cells, it may be necessary for immature cells to be matured in vitro before transplantation. Therefore, the development of an in vitro maturation system is important.
Many studies of the maturation of primitive hepatic endodermal cells in vitro have demonstrated the requirement for maturation not only of soluble factors, such as fibroblast growth factors,8 oncostatin M,9 and hepatocyte growth factor,10 but also cell-to-cell contact between parenchymal cells and nonparenchymal cells.11–14 Recently, Nagai et al.11 reported that cell-to-cell contact between hepatic stellate cells (HSCs) and liver epithelial cells induced the differentiation of the liver epithelial cells into a hepatocytic lineage. Nitou et al.12 reported that the coexistence of fetal mouse hepatoblasts and nonparenchymal cells was essential for their mutual survival, proliferation, differentiation, and morphogenesis.
Previously, we designed a system to enrich mouse fetal HPCs.15 In this system, fetal HPCs are enriched by the formation of cell aggregates, which is dependent on homophilic binding of cell adhesion molecules such as E-cadherin. This system enables us to enrich the viable HPCs while limiting cell damage. Examining the antigenic profiles of HPCs is crucial to isolate only the HPCs. Therefore, the purpose of this study was to identify and characterize the cell populations contained in the cell aggregates and to examine the interactions among these populations.
C57BL/6J Jms Slc mice were obtained from SLC (Hamamatsu, Japan). Animals were maintained at a constant temperature of 18°C to 20°C and in a 12-hour light–dark cycle. They were housed at Kyoto University, and all animal experimental procedures were performed according to the Animal Protection Guidelines of Kyoto University.
Isolation and Culture of Fetal HPCs.
Fetal HPCs were obtained from E13.5 fetal livers and were enriched by formation of cell aggregates. The isolation and culture of the cell aggregates was performed as described previously.15 Dissociating the cell aggregates into single cells is technically difficult. Therefore, cell aggregates selected by gravity sedimentation were inoculated on type-I collagen-coated culture plates (Becton Dickinson Co., Ltd., Lincoln Park, NJ). After 24 hours of incubation, the aggregates adhered to the plates and extended as monolayer colonies. After removing hematopoietic cells by washing twice with phosphate buffered saline (PBS), adherent cells were incubated with trypsin-ethylenediaminetetraacetic acid solution (Sigma Chemical Co., Ltd., St. Louis, MO) for 12 minutes. The dissociated cells were washed three times with PBS containing 3% fetal calf serum (FCS; ICN, Aurora, OH) and were used for fluorescence-activated cell sorter (FACS) analysis or FACS sorting. A flow diagram describing the formation of cell aggregates and FACS analysis and cell sorting is shown in Fig. 1.
The dissociated cells were incubated with each antibody at 4°C for 30 minutes, were washed three times, and were resuspended in 3% FCS-PBS. The following antibodies were used, all diluted at 1:100: anti-Thy1-fluorescein isothiocyanate (FITC; Immunotech, Marseille, France), CD49f (integrin α6)-phycoerythrin (PE), CD45 (leukocyte common antigen)-allophycocyanin (APC), CD29 (integrin β1)-FITC, CD16-FITC, CD31 (PECAM-1)-FITC, CD34-FITC, Flk1 (VEGF-R2)-PE, CD44-APC, CD106-FITC, c-Kit-APC, CD71 (transferrin receptor)-FITC, and Sca-1-PE monoclonal antibody (Pharmingen, San Jose, CA). For anti-CD105 (endoglin; Pharmingen) and anti-c-Met (hepatocyte growth factor receptor) monoclonal antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) diluted at 1:100, the second antibody was biotin-conjugated anti-mouse immunoglobulin G diluted at 1:100, and visualization was performed using streptavidin-APC (Pharmingen) diluted at 1:100. Then, the cells were analyzed with an FACSCalibur (Beckton Dickinson, San Jose, CA). Gating was implemented based on isotypic control staining profiles.
Cell Sorting by FACS and Culture of the Separated Cells.
The dissociated cells were incubated with Thy1-FITC, CD49f-PE, and CD45-APC monoclonal antibody at 4°C for 30 minutes and prepared as described above. Then, the cells were separated using a FACSVantage (Becton Dickinson). After separation, the collected cells were washed twice with 3% FCS-PBS, resuspended in Dulbecco's modified essential medium (Gibco BRL, Grand Island, NY) with 10% FCS, 20 mmol/L HEPES, 25 mmol/L NaHCO3, 0.5 mg/L insulin, 10−7 mol/L dexamethasone (Wako Pure Chemical, Osaka, Japan), 10 mmol/L nicotinamide (Wako Pure Chemical), 2 mmol/L L-ascorbic acid phosphate (Wako Pure Chemical), 20 μg/L deleted form of hepatocyte growth factor (kindly provided by Snow Brand Product Co., Osaka, Japan), 100 units/mL penicillin G, and 0.2 mg/mL streptomycin. Then, the cells were inoculated on type-I collagen-coated 24-well plates (Becton Dickinson) at a density of 2 × 104/well. To evaluate the interaction between the two separated cell fractions, coculture was performed as follows. In method 1, a mixture of both cell fractions (1:1 in cell density) was inoculated on type-I collagen-coated 24-well plates. In method 2, both cell fractions were cultured separately on type-I collagen-coated 24-well plates using BIOCOAT Cell Culture Inserts (Becton Dickinson). After 16 hours, the culture media were changed, and thereafter, the media were changed every 2 to 3 days.
Immunocytochemistry for α-fetoprotein (AFP), albumin (ALB), and cytokeratin 19 (CK19) was performed as previously described.15 For immunocytochemistry of desmin and α-smooth muscle actin (α-SMA), the cultured cells were washed twice with PBS and fixed in 3.3% formalin for 12 minutes at room temperature. Nonspecific binding was blocked with 10% skim milk (Snow Brand Product Co.) and 0.4% bovine serum albumin (Sigma-Aldrich Chemie Co., Ltd., Steinheim, Germany) in 0.1% saponin (Sigma Chemical) in PBS. Then, endogenous avidin and biotin were blocked with an avidin and biotin blocking kit (Vector Laboratories, Inc., Burlingame, CA). Subsequently, the cells were incubated with the primary antibodies for 16 hours at 4°C followed by incubation with the biotin-conjugated secondary antibody for 30 minutes at 37°C. The primary antibodies were antihuman α-SMA (DAKO Japan, Kyoto, Japan) diluted at 1:200 and antihuman desmin (DAKO Japan) diluted at 1:100. The secondary antibody was biotin-conjugated antimouse immunoglobulin G (DAKO Japan) diluted at 1:500, and visualization was performed using streptavidin-conjugated Texas Red-X (Molecular Probes, Inc., Eugene, OR). Nuclear counterstaining was performed with 4′, 6-diamidino-2-phenylindole. The signal was detected using a fluorescence microscope (Axiovert 135; Carl Zeiss Vision Co., Ltd., Hallbergmoos, Germany).
Total RNA was extracted from the separated cells just after sorting, E13.5 fetal liver, and adult liver using an RNeasy kit (QIAGEN, Chatsworth, CA), according to the manufacturer's instructions. Complementary DNA was synthesized from total RNA using the Omniscript RT kit (QIAGEN) and amplified using specific primer pairs and AmpliTaq Gold DNA polymerase (Perkin Elmer, Foster City, CA). The PCR conditions were as follows: 95°C for 15 minutes, followed by 30 cycles of 94°C for 15 seconds, 60°C for 30 seconds, and 72°C for 1 minute, and a final extension at 72°C for 10 minutes. Desmin, α-SMA, and vimentin were used as mesenchymal cell markers. Vascular endothelium (VA)-cadherin, platelet endothelial cell adhesion molecule (PECAM), and fetal liver kinase 1 (Flk1) were used as endothelial cell markers. CD16 was used as a Kupffer cell marker. Primers used for amplification are listed in Table 1.
Table 1. Primer Sequences Used
mRNA expression was quantified by real-time RT-PCR using ABI Prism 7700 (Applied Biosystems, FosterCity, CA). One-step RT-PCR reactions were performed in 96-well plates containing, in each well, 100 ng total RNA together with 0.1 μmol/L of each of the sense and antisense primers, and 0.2 μmol/L probe in a total volume of 25 μl. All reactions were run in triplicate. After the RT reaction at 48°C for 30 minutes, the reaction mixtures were heated at 95°C for 10 minutes, followed by 40 cycles at 95°C for 15 seconds and 60°C for 1 minute. The comparative threshold cycle (CT) method against the expression level of glyceraldehydes-3-phosphate dehydrogenase was used to determine relative quantities. Tyrosine amino transferase (TAT) was used as a perinatal hepatocyte marker gene, and tryptophan oxygenase (TO) was used as a mature hepatocyte marker gene. mRNA from fetal and adult liver was used as a positive control for AFP, TAT, and TO expression. Primers and probes used are listed Table 2.
Table 2. TaqMan Probe and Primer Sequences
Periodic Acid–Schiff (PAS) Staining Analysis of Cultured Cells.
To examine whether the cultured cells produced and stored glycogen, as an indicator of functional maturation, PAS staining was performed. The cultured cells were fixed in 3.3% formalin for 12 minutes, and intracellular glycogen was stained using a PAS staining solution (Muto Pure Chemicals Co., Ltd., Tokyo, Japan), according to the standard protocol.
Transmission Electron Microscopy.
CD49f-positive cells and Thy1-positive cells were taken either immediately after their separation or after they had been co-cultured for 30 days, and they were fixed in 2% glutaraldehyde-PBS for 1 hour, postfixed in 2% osmium tetroxide, and embedded in Epon-812 resin (TAAB, UK). Ultra-thin sections were cut using an ultramicrotome and were stained with uranyl acetate. The sections were examined by transmission electron microscopy (H-7000; Hitachi, Tokyo, Japan).
Flow Cytometric Fractionation of Cell Aggregates.
FACS analysis using antibodies against CD49f, Thy1, and CD45 determined that three major populations existed in the cell aggregates: (1) CD49f+Thy1-CD45- cells (CD49f-positive cells; 48.51 ± 7.34%); (2) CD49f±Thy1+CD45− cells (Thy1-positive cells; 23.79 ± 7.92%); and (3) CD49f+(low and high)Thy1−CD45+ cells (CD45-positive cells; 24.78 ± 8.18%; Fig. 2A–D). Both CD49f- and Thy1-positive cells expressed c-Met, the latter more strongly (Fig. 2E, F). However, none of the cells in the aggregates expressed c-Kit (Fig. 2G). Because CD45 is a common leukocyte antigen, and the CD45-positive cells isolated by FACS sorting did not attach to the culture dish, but remained floating, we believe that the CD45-positive cells were hematopoietic cells contaminating the cell aggregates. Therefore, these cells were excluded from further study, and sorting of CD49f-positive cells and Thy1-positive cells only was performed to characterize these two fractions.
Cell Sorting and Immunocytochemistry.
CD49f-positive cells and Thy1-positive cells were sorted using FACSVantage, cultured in vitro, and subjected to subsequent immunocytochemical analysis. At day 1 after sorting, CD49f-positve cells appeared morphologically uniform and cuboidal in shape. In addition, these cells possessed large nuclei and highly granulated cytoplasm. On immunocytochemical staining, CD49f-positive cells were homogeneously stained for AFP and were heterogeneously stained for ALB and CK19, but did not stain for desmin or α-SMA. ALB was expressed more strongly toward the inside of the colony, whereas CK19 was expressed more strongly toward the periphery of the colony. Some double-positive cells were detected. These results were similar to those of a previous report.15
At day 1 after sorting, the Thy1-positive cell population appeared morphologically to comprise two distinct cell types. One was spindle-like and possessed small nuclei. The other had more highly granulated cytoplasm (Fig. 3A). On immunocytochemical staining, both cell types stained for α-SMA (Fig. 3B), but only the latter cell type stained for desmin (Fig. 3C). At day 5, round cells with large nuclei appeared in colonies of the first cell type and proliferated (Fig. 3D). This cell type did not stain for either α-SMA or for desmin (Fig. 3E, F). None of these cell types incorporated acetylated low-density lipoprotein or latex microspheres, and none exhibiting AFP, ALB, or CK19 staining.
Further Characterization of Thy1-Positive Cells.
We further examined the Thy1-positive cells using RT-PCR and FACS analysis. RT-PCR demonstrated that the Thy1-positive cells expressed desmin, α-SMA, and vimentin mRNA, but did not express the endothelial cell and Kupffer cell markers VE-cadherin, PECAM, Flk-1, and CD16 (Fig. 4A). Supporting the RT-PCR data, FACS analysis showed that Thy1-positive cells did not express the endothelial cell and Kupffer cell markers CD31, CD34, Flk1, and CD16. Thy1-positive cells were CD29+, CD44±, CD105+, CD106±, CD71±, and Sca-1± (Fig. 4B). Further fractionation of Thy1-positive cells by surface antigen expression was difficult.
Morphological and Functional Analyses of the Interaction Between CD49f-Positive Cells and Thy1-Positive Cells.
To examine the interaction between CD49f-positive cells and Thy1-positive cells, we co-cultured both cell fractions. When CD49f-positive cells were co-cultured with Thy1-positive cells (method 1), AFP-producing CD49f-positive cells had increased intracellular granularity at day 3 (Fig. 5A), proliferated until day 7, and then piled up from the periphery of the colonies where they were in contact with Thy1-positive cells (Fig. 5B). At day 14, the piled-up area was widely expanded in CD49f-positive colonies, and some binucleated cells were detected in the piled-up area. (Fig. 5C). In contrast, when CD49f-positive cells were cultured alone or were cultured together with Thy1-positive cells, but without any direct contact (method 2), the CD49f-positive cells in the periphery of the colonies did not maintain their morphologic appearance and began to decrease in number at day 3, failing to show any signs of maturation. Supporting these morphological findings, a number of cells in the CD49f-positive colonies were positive for PAS staining at day 10, when CD49f-positive cells were cocultured with Thy1-positive cells (Fig. 5D). However, CD49f-positive cells cultured alone or separately with Thy1-positive cells were negative for PAS staining even at day 10 (Fig. 5E, F). Additionally, in cocultured cells, the level of TAT and TO mRNA expression was significantly increased (Fig. 6B, C), whereas AFP mRNA was decreased, as assessed by real-time RT-PCR (Fig. 6A). In contrast, no significant increase in TAT and TO mRNA expression was observed in cultures of CD49f-positive cells either alone or in separate cultures together with Thy1-positive cells (Fig. 6B, C).
Transmission Electron Microscopy.
To confirm further the putative maturation of the co-cultured CD49f-positive and Thy1-positive cells, we used transmission electron microscopy to compare the ultrastructures of the cells just after their separation and after 30 days of coculture. The CD49f-positive cells just after separation had large nuclei with developed nucleoli and a number of fat droplets, but had few intracytoplasmic organelles such as mitochondria and peroxisomes (Fig. 7A). The Thy1-positive cells just after separation appeared to comprise two cell types distinguished by their having either clear or dark cytoplasm. This difference was thought to be the result of the number of intracytoplasmic ribosomes. Both cell types had large nuclei, similar to CD49f-positive cells, a large amount of open rough endoplasmic reticulum, and many microfilaments in the cytoplasm (Fig. 7B). In contrast, the cocultured cells had no fat droplets; contained many mitochondria, peroxisomes, and tight junctions with desmosomes; and formed biliary canaliculi with microvilli (Fig. 7C). All of these features are compatible with those of mature hepatocytes.
We previously reported a method for enriching mouse fetal HPCs that relies on formation of cell aggregates.15 In this study, we used this system in combination with FACS to isolate CD49f-positive cells and Thy1-positive cells. Approximately 50% of the cell aggregates were composed of CD49f-positive cells, which expressed c-Met but not c-Kit. Immunocytochemical staining of CD49f-positive cells revealed a homogeneous distribution of AFP and heterogeneous patterns of ALB and CK19 staining. Shiojiri et al.16 reported that AFP expression in endodermal cells signaled their commitment toward the liver lineage and that AFP-producing hepatoblasts can differentiate into all possible endodermal epithelial cells present in the adult liver, including hepatocytes and cells that form the intrahepatic and extrahepatic bile ducts. Suzuki et al.17–19 reported that c-Met+ CD49f+(low) c-Kit− CD45− TER119− cells were primitive endodermal stem cells persisting in the fetal mouse liver that comprised 0.3% of the total cells in the fetal liver17, 18 and were a heterogeneous population containing ALB-negative hepatic stem cells and ALB-positive hepatic precursors.19 Therefore, it is likely that the CD49f-positive cells were AFP-producing primitive hepatic endodermal cells and that the heterogeneous staining pattern of ALB and CK19 reflect different populations of hepatic stem or progenitor cells at various stages of differentiation present among the CD49f-positive cells. The antigenic profile of the primitive hepatic endodermal cells we isolated was CD49f+ c-Met+ c-Kit− CD45−, which is consistent with a previous report by Suzuki et al.17
On close morphological examination, the population of Thy1-positive cells was found to contain at least three subpopulations. Two of these subtypes stained positive for α-SMA and partially positive for desmin, whereas the remaining cell type did not exhibit either α-SMA or desmin staining. However, none of the cell types stained positive for hepatic endodermal specific markers such as AFP, ALB, and CK19. Furthermore, Thy1-positive cells did not have the characteristics of endothelial cells or Kupffer cells by RT-PCR or FACS analysis. These results indicated that the Thy1-positive cells were not hepatic, endothelial, or Kupffer cells, but were mesenchymal cells.
Desmin is one of the principal intermediate filament proteins expressed in cardiac, skeletal, and smooth muscle cells.20 In the liver, desmin is expressed selectively in HSCs.21, 22 However, the Thy1-positive cells in this study were morphologically different from adult HSCs, which do not express the Thy1 surface antigen (data not shown). However, they are positive for α-SMA, one of six isoactins expressed in mammalian cells and the isoform that is typical of smooth muscle cells,23 especially those in blood vessels.24 In fibrotic liver and in vitro culture, HSCs change their normal quiescent phenotype to an activated myofibroblast-like phenotype and express α-SMA.25, 26 Charbord et al.27 reported that stromal cells from different developmental sites, including bone marrow and fetal liver, followed a vascular smooth muscle cell differentiation pathway.27, 28 Although the desmin and α-SMA findings suggest that these Thy1-positive cells are of the mesenchymal lineage and comprise a heterogeneous set of cell populations, their complete characterization requires further study.
In this study, colonies of AFP-producing CD49f-positive cells subjected to coculture with Thy1-positive cells morphologically resembled mature hepatocytes; a number of cells contained in these colonies even stored a significant amount of glycogen. Additionally, real-time RT-PCR analysis showed that the cocultured CD49f-positive and Thy1-positive cells expressed increasing amounts of TAT and TO mRNA over time, and transmission electron microscopy confirmed that they had differentiated into mature hepatocytes by day 30. However, the Thy1-positive cells did not morphologically resemble endodermal cells and did not express endodermal cell markers. These results suggest that CD49f-positive cells, but not Thy1-positive cells, were responsible for the expression of TO and TAT observed in the coculture. Therefore, it is likely that the CD49f-positive cells were primitive hepatic endodermal cells with the capacity to differentiate into mature hepatocytes. In contrast, CD49f-positive cells cultured alone or in separated cultures with Thy1-positive cells failed to exhibit signs of morphologic or functional maturation. These results suggest that cell-to-cell contact with Thy1-positive cells is essential for the maturation of CD49f-positive cells in vitro.
Thy1 is a highly conserved protein present in the brain and hematopoietic system of rats, mice and humans, and is thought to play an important role in hematopoiesis.29, 30 It has been reported that Thy1 may function as a receptor for a ligand on other cells, allowing them to contact and adhere onto myogenesis and murine thymocytes.31, 32 In the hematopoietic system, it has been suggested that Thy1 allows the hematopoietic stem cells to recognize and to bind to stromal determinants. In this study, cell-to-cell contact with Thy1-positive cells was essential for the maturation of primitive hepatic endodermal cells. Therefore, similar to its activity in the hematopoietic system, the Thy1 protein may play an important role in allowing Thy1-positive cells to recognize, to adhere to, and to promote the maturation of primitive hepatic endodermal cells in the fetal liver. The pathway by which Thy1 must signal this maturation is not yet known, but probably is mediated via a surface antigen on primitive hepatic endodermal cells.
In conclusion, we have identified two cell populations, one CD49f-positive and the other Thy1-positive, in our enriched mouse fetal HPC cell aggregates. The CD49f-positive cells were primitive hepatic endodermal cells. The Thy1-positive cells are probably of mesenchymal origin and promoted the maturation of CD49f-positive cells by cell-to-cell contact. A large number of CD49f-positive primitive hepatic endodermal cells can be isolated using our cell aggregate method and FACS sorting.
The authors thank Masaya Nagao for assistance with FACSVantage, Makio Fujioka for assistance with the transmission electron microscopy, and Snow Brand Milk Product Co. for providing the deleted form of hepatocyte growth factor.