Experimental injury is useful to induce tissue stem cells, which may exist in small numbers under normal conditions. The salivary glands originate from the endoderm and consist of acinar and ductal epithelial cells, which have exocrine function. After salivary gland duct ligation, acinar cells disappear as a result of apoptosis, and duct epithelium subsequently proliferates. In this study, we analyzed the tissue stem cells induced by salivary gland duct ligation in mice using immunohistochemistry and flow cytometry. We sorted the Sca-1+/c-Kit+ fraction from adult mice salivary glands by way of fluorescence-activated cell sorting. The sorted cells were apparently homogeneous and were designated mouse salivary gland–derived progenitors (mSGPs). mSGP cells differentiated into a hepatic lineage when cultured in matrigel. In spherical culture in the presence of glucagon-like peptide-1 (GLP-1), these cells differentiated into a pancreatic endocrine lineage. When spheroidal bodies of mSGP, 20 to 30 μm in diameter, were transplanted into liver via the portal vein, the cells integrated into hepatic cords and expressed albumin and α1-antitrypsin, suggesting that they had differentiated into hepatic-type cells. Moreover, ductlike structures formed by mSGP cells also appeared, epithelial cells of which were positive for cytokeratin 19. In conclusion, fluorescence-activated cell sorting (FACS) based on histologic evidence is efficient in isolating adult tissue stem cells of the salivary gland. Tissue stem cells of endodermal origin (e.g., hepatic oval cells, pancreatic epithelial progenitor cells, and salivary gland progenitor cells) have similarities in their molecular markers and tissue location. Our findings suggest the existence of common tissue stem cells in endoderm-derived organs. (HEPATOLOGY 2004;39:667–675.)
When tissue damage occurs under certain conditions, tissue stem cells proliferate, differentiate, and regenerate tissue.1, 2 The liver is an organ of endodermal origin; liver damage usually triggers the proliferation of mature hepatocytes.3 However, if DNA metabolism is disturbed—for example, in the presence of 2-acetylaminofluorene (2-AAF)—the regeneration process is altered. When liver damage to the rat is caused by partial hepatectomy or CCL4 administration during constant exposure to 2-AAF, oval cells appear in portal areas. These cells are hepatic stem cells that are postitive for both α-fetoprotein (AFP) and cytokeratin 19; they can differentiate into hepatocytes or bile duct epithelium.4–6 Petersen and colleagues reported the presence of bone marrow–derived cells in the oval cell regions of the liver of 2-AAF–treated rats with bone marrow transplants who have had partial hepatectomy.7 Oval cells are known to express hematopoietic stem cell markers such as Thy-1, c-Kit, and Sca-1.5, 8, 9 These findings raise the possibility that oval cells originate from hematopoietic stem cells; however, Rao and colleagues recently showed that this is not so.10
Loss of pancreatic acinar cells and simultaneous proliferation of cytokeratin 19–positive, ductlike epithelial cells occur in copper-depleted rats, a representative model of pancreatic tissue damage.11 When such rats are refed copper, albumin-positive, hepatocytelike cells arise in the interstitium of the pancreas (pancreatic hepatocytes).11 However, oval cells have been shown to differentiate into insulin- and glucagon-secreting pancreatic endocrine cells in vitro.12 These findings suggest the existence of a common progenitor of the liver and pancreas.
As experimental models, tissue damage is useful in inducing proliferation of stem cells, which may exist in small numbers under normal conditions. Ligating bile and pancreatic ducts create models of tissue injury in the liver and pancreas, respectively. In rats with ligated bile ducts, regeneration of bile ducts occurs;13 the proliferated epithelial cells that form ducts are immature cells that express AFP and c-Kit.14 In the pancreas of adult rats with ligated pancreatic dusts, the tail part of the pancreas lost atrophies and shows a loss of acinar cells and growth of small duct and islet structures.15 A proportion of the cytokeratin 20–positive cells that make up the small duct structures express insulin or glucagon.16 Thus pancreatic endocrine cells may differentiate from progenitors present in the ductal areas after tissue injury.
The salivary gland originates from the endoderm and consists of acinar cells and epithelial duct cells.17 Salivary glands secrete amylase into the digestive tract. These features resemble the pancreatic exocrine system. Ligation of a salivary duct also creates a model of tissue injury similar to the pancreatic model.18 After duct ligation, acinar cells disappear by apoptosis and proliferation of duct epithelium occurs.19, 20 Reopening of the obstructed duct leads to repopulation of acinar cells.18 During this regeneration process, the duct cells proliferate and differentiate into acinar cells within five days after reopening the duct.18 Intercalated duct cells may be precursors of both duct epithelial and acinar cells.21
We have recently demonstrated that progenitor cells induced by salivary gland duct ligation in the rat (SGP-1) have the capacity to differentiate into hepatocytes, a cell lineage of the endoderm.22 SGP-1 cells can be induced in vitro to differentiate into hepatic or pancreatic phenotypes.22 Moreover, when SGP-1 cells were transplanted into the liver, they integrated with recipient hepatocytes and became albumin-producing cells.22
In this study, we analyzed the tissue stem cells induced by salivary gland duct ligation in mice using immunohistochemistry and flow cytometry. Based on analytical data, we isolated the tissue-specific stem cells from adult mouse salivary gland using fluorescence-activated cell sorting (FACS). Analysis of these cells revealed features similar to those of rat salivary gland SGP-1 cells; such cells are a potential source of cells for transplantation in the management of hepatic and pancreatic disease.
C57BL/6 mice were purchased from Charles River Japan, Inc. (Osaka, Japan). C57BL/6 background ROSA26 mice were purchased from The Jackson Laboratory (Bar Harbor, ME). The protocol was approved by the Center for Animal Resources and Development of Kumamoto University. Animal care was as outlined in the Guide for Care and Use of Laboratory Animals. Five-week-old male mice were anesthetized by inhaled ether and held in the supine position with the neck extended. An incision was made in the midline of the neck and both submandibular glands were exposed. After double ligation of the main secretory ducts, the neck was sutured. Six days later, the submandibular glands were excised, minced, and incubated in 30 mL of ethylene glycol-bis(β-aminoethyl ether)-N,N,N',N'-tetraacetic acid (EGTA) buffer at 37°C for 20 minutes, then centrifuged at 100g for 5 minutes at room temperature. Pellets were suspended in 40 mL of digestion buffer containing D-MEM/F12 1:1 (Invitrogen Corp., Carlsbad, CA), 1.67 mg/mL collagenase (Invitrogen Corp.), and 1.33 mg/mL hyaluronidase (Nacalai Tesque Inc., Kyoto, Japan); the suspension was incubated for 40 minutes at 37°C. The partly digested tissues were further treated with dispersion buffer containing D-MEM/F12 (Invitrogen Corp.) and 1.67 mg/mL dispase (Invitrogen Corp.) and incubated for 60 minutes at 37°C. The cell suspensions were then passed through a stainless filter and centrifuged at 100g for 5 minutes at 4°C. The pellets were suspended in 10 mL Dulbecco's modified Eagle medium/F12 1:1 medium and washed three times with Williams' E medium (Invitrogen Corp.) containing 10% fetal bovine serum (Invitrogen Corp.).
Long-Term Culture of Isolated Cells.
Dispersed cells derived from submandibular glands were fractionated using flow cytometry, and each subpopulation was sorted using a FACS VantageSE (BD Biosciences, Franklin Lakes, NJ). Sorted cells were plated and cultured on type I collagen (Asahi Techno Glass, Tokyo, Japan) at a density of 1 × 104 cells/cm2. Our standard culture medium is a Williams' E medium supplemented with 5% fetal bovine serum, 20 ng/mL mouse epidermal growth factor (Sigma Chemical, St. Louis, MO), 10−8 mol/L insulin (Invitrogen Corp.), 10−6 mol/L dexamethasone (Sigma Chemical), 100 U/mL penicillin G and 100 μg/mL streptomycin (Invitrogen Corp.), 1× MEM Non-Essential Amino Acids Solution (Invitrogen Corp.), 1× Insulin-Transferrin-Serenium-X (Invitrogen Corp.), and 10 mmol/L nicotinamide. For induction of hepatic lineage, 1.5 mL/well of 1:3-thick Matrigel (BD Biosciences Falcon, San Diego, CA) diluted with serum-free Williams' E medium was poured into a 12-well dish, then plated with 1 × 105 cells/well of the aforementioned mouse salivary gland progenitors (mSGPs) incubated in standard medium. For pancreatic lineage induction, 2 × 102 cells/well were plated in a SUMILONCELLTIGHT Spheroid 96U (SUMITOMO BAKELITE Co., Ltd., Tokyo, Japan) and cultured in the standard medium containing 20 nmol/L of glucagon-like peptide-1 (GLP-1, Sigma Chemical). Medium was renewed every 3 days.
Immunostaining of Tissues and Cultured Cells.
After the salivary glands were excised, they were fixed in 10% formaldehyde and embedded in paraffin. Sections were cut 4 μm thick for hematoxylin-eosin and periodic acid-Schiff (PAS) staining. Other tissues were fixed with periodate-lysine-paraformaldehyde, embedded in optimal cutting temperature compound, and 5- to 7-μm cryostat sections were obtained. Immunostaining was undertaken according to the manufacturer's instructions. Cultured cells on a glass-based dish (Asahi Techno Glass) were washed three times with phosphate-buffered saline (PBS), fixed in 4% paraformaldehyde (Nacalai Tesque, Inc.) for 20 minutes at 4°C, then washed in PBS containing 0.2 % polyoxyethylene 20 sorbitan monolaurate (Tween 20) (Wako Pure Chemical Inc., Osaka, Japan). To inactivate internal peroxidase, samples were incubated in methanol containing 0.3% of hydrogen peroxide for 30 minutes. Nonspecific binding was blocked with either nonimmune serum of the species from which secondary antibodies had been obtained or a nonspecific staining blocking reagent (DAKO Cytomation, Glostrup, Denmark). The sections were then incubated with primary antibodies. The primary antibodies used were anti-CD49f, anti-CD29, anti–Sca-1 (BD Bioscience PharMingen), anti-CK19 (TROMA3, a kind gift from Dr. Rolf Kemler), anti–c-Kit (Chemicon International, Inc., Temecula, CA), anti-laminin (DAKO), anti-albumin (Inter-Cell Technologies, Inc., Hopewell, NJ), anti-AFP (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), anti-insulin (Biogenesis Ltd., Poole, United Kingdom), anti-glucagon (AFFINITI Research Products Ltd., Exeter, United Kingdom), and anti–β-galactosidase (Biogenesis). The secondary antibodies used were Alexa488-labeled anti-rabbit immunoglobulin G (IgG), Alexa594-labeled anti-mouse IgG, Alexa594-labeled anti-goat IgG, Alexa594-labeled anti-rat IgG (Molecular Probes, Inc., Eugene, OR), and peroxidase-conjugated anti-rat IgG (IMMUNOTECH, Marseilles, France). Cells were viewed under a confocal laser-scanning microscope FV500 (Olympus Optical Ltd., Tokyo, Japan).
FACS Analysis and Sorting.
Surface antigens of cells were analyzed using a FACS caliber (BD Biosciences). Cells were washed three times with staining buffer (3% fetal bovine serum (FBS)/PBS), and then incubated with surface antigen antibodies for 20 minutes. Primary antibodies used were FITC-conjugated anti-CD34 (Santa Cruz Biotechnology, Inc.), anti-CD49f, anti-CD49c, anti-CD29, anti-CD51, anti-CD44, anti-CD45, biotin-conjugated CD90, 1/Thy-1, PE-conjugated anti-CD117/c-Kit, anti-Sca-1, and platelet-derived growth factor receptor α chain (BD Bioscience PharMingen). Samples were then washed and incubated with secondary antibodies (FITC-conjugated anti-rat IgG-1/2a, IgG-2b, streptavidin-APC; BD Bioscience PharMingen) for 20 minutes. Twenty microliters of VIAPROBE (BD Bioscience PharMingen) was added to each sample after washing. Cells were sorted using FACS Vantage SE (BD Biosciences). Data were analyzed using CELLQuest (BD Biosciences).
Total RNA was extracted from cultured cells and their spheroid bodies using ISOGEN kits (Nippon Gene, Tokyo, Japan). Complementary DNA was prepared from 1 μg of total RNA using Superscript II (Invitrogen Corp.) with oligo-dT primers (Invitrogen Corp.) according to the manufacturer's instructions. The resulting complementary DNA was amplified using GeneAmp PCR 9700 (Perkin-Elmer Corp., Norwalk, CT) with the following sets of primers:23, 24
Polymerase chain reaction products were electrophoresed in 4% NuSieve GTG agarose (FMC Bioproducts, Rockland, ME) and visualized using ethidium bromide (Nacalai) staining.
Transplantation of Salivary Gland-Derived Cells.
For transplantation, two types of cells were prepared. One was dispersed cells cultured on type I collagen and the other was spheroidal bodies cultured in matrigel as previously mentioned. Two-thirds partial hepatectomy was undertaken as previously described.25 Using a 30G needle and syringe, the cells were injected into the portal veins of mice that had undergone partial hepatectomy. Two weeks after cell transplantation, the livers were infused with saline by way of the inferior vena cava and removed.
For whole-mount β-galactosidase staining, livers were fixed for 30 minutes using 2% formaldehyde and 0.2% glutaldehyde in PBS.26 Livers were incubated in staining medium containing 1 mg X-gal (TAKARA BIO Inc., Shiga, Japan) in a buffer of 5 mM potassium hexacyanoferrate (III), 5 mM potassium hexacyanoferrate (II) trihydrate, 2 mM MgCl2, 0.02% Nonidet P-40 (Nacalai), and 40 mM Hepes (Sigma Chemical) overnight at 37°C.26 Stained livers were washed twice and examined under a stereomicroscope (Leica Microsystems AG, Wetzlar, Germany).
Histologic Changes in the Mouse Submandibular Gland After Duct Ligation.
Salivary gland tissue consists of PAS-positive acinar cells and PAS-negative duct epithelial cells (Fig. 1A). Five to six days after ligation of the main secretory duct, almost all acinar cells had disappeared and were replaced by proliferating small duct structures, as occurs in rats (Fig. 1B). The small-duct epithelial cells expressed cytokeratin 19 and CD49f (Fig. 1C, D). CD29-positive cells were distributed over the same area as CD49f-positive cells (data not shown).
In some models of tissue damage, cells expressing hematopoietic stem cell markers are known to appear in regenerating tissue. As mentioned above, when the bile duct is ligated in rats, the cells that arise during a regenerative response are known to express c-Kit.14 Hepatic oval cells express surface antigens Thy-1, c-Kit, and Sca-1.5, 8, 9 In mice, mammary gland progenitor cells express Sca-1 antigen.27 In duct-ligated submandibular glands, Sca-1–positive and c-Kit–positive cells were observed among clusters of small cells in the periductal areas (Fig. 1E, F). Some of these cells were positive for both markers (Fig. 1G–I). In nonligated glands, cells positive for Sca-1 or c-Kit could be seen occasionally, but no double-positive cells were found (data not shown). These findings indicate that Sca-1 and c-Kit may be key markers for identifying progenitor cells in the salivary gland.
Flow Cytometric Fractionation of Duct-Ligated Mouse Submandibular Gland Cells.
Based on our histologic data, we performed a flow cytometric analysis of dispersed cells of submandibular glands. The percentage of cells positive for Sca-1 increased from 1.42 ± 0.84% (mean ± SD) prior to ligation to 27.01 ± 3.37% after (Fig. 2A). c-Kit positive cells increased from 0.97 ± 0.74% to 5.33 ± 4.10 % (Fig. 2B). The cell fraction of Sca-1 and c-Kit double-positive cells was 0.68 ± 0.07% in ligated glands (Fig. 2C). CD49f-positive cells also increased from 10.04 ± 1.96% to 30.01 ± 13.21%. The percentage of cells positive for Thy-1 or CD34 did not change (data not shown).
For cell analysis in vitro, dispersed cells derived from ligated submandibular glands were fractionated into four subpopulations using Sca-1 and c-Kit antibodies (Fig. 2C). Each population was cultured promptly after FACS. Of these four populations, Sca-1+/c-Kit+ cells adhered best to, and proliferated on, type I collagen. Colonies of cells with a small and polygonal epitheliumlike shape became visible; these colonies were harvested and plated on type I collagen-coated 24-well plates for further purification. Cells from these colonies were then purified using limiting dilution on type I collagen-coated 96-well plates. We obtained five subclones and investigated each clone. They were similar to each other in size, appearance, and doubling time and maintained their configuration throughout generations for more than 6 months. We used one of the subclones derived from ROSA26 mice in the present study and designated it mSGP-1.
Characterization of mSGP.
Phase contrast micrographs of mSGP-1 cells cultured on type I collagen are shown (Fig. 3A, B). They adopted a polygonal shape (Fig. 3A) and formed small clusters (Fig. 3B). Like rat SGP-1, mSGP-1 cells seeded at a low density were positive for both CD49f and intracellular laminin (Fig. 3C), and accumulation of laminin around the mSGP colony was observed (Fig. 3D). mSGP-1 cells are also positive for AFP (Fig. 3E) but negative for albumin and insulin (data not shown). When cultured on type I collagen-coated plates, mSGP-1 cells formed small clusters, and cells at the tips were positive for cytokeratin 19 (Fig. 3F). Flow cytometric analysis of mSGP-1 cells cultured on type I collagen (Fig. 2D) demonstrated that the cells were positive for Sca-1, weakly positive for c-Kit and PDGF receptor α chain, and negative for CD34 (data not shown) and Thy-1. With regard to integrins, a proportion of mSGP cells were positive for CD49f, CD29, CD49c, and CD51.
mSGP Cells Differentiate Into Endodermal Lineage.
When cultured in matrigels, fewer than 5% of the mSGP-1 cells aggregeted, began to form spheroidal bodies 36 hours later, and gradually grew. Results of immunofluorescent staining of spheroidal bodies cultured in matrigels for 2 weeks are shown in Fig. 4A–F. Most spheroidal bodies expressed albumin and AFP, indicating a hepatic lineage (hepatic spheroidal body). We performed five independent experiments and evaluated the differentiation of 10 spheroidal bodies in each experiment. Forty-five of 50 spheroids contained albumin-positive cells, and 37 of 50 spheroids contained AFP-positive cells. No spheroidal body containing insulin- or glucagon-positive cells was observed in matrigel culture. Plating these spheroidal bodies on a type I collagen-coated plate led to adhesion and proliferation but loss of albumin expression (data not shown). When cultured in U-bottom plates containing 20 nmol/L of GLP-1–supplemented standard medium, mSGP-1 cells formed spheroidal bodies that expressed insulin and/or glucagon (Fig. 4G–I). In these cultures, most spheroid bodies (>80%) contained insulin- and/or glucagon-positive cells (pancreatic spheroidal body), but none contained albumin-positive cells. In the absence of GLP-1, no pancreatic spheroidal body was observed. Therefore, mSGP cells are capable of differentiating into two endodermal lineages in culture: the liver and the pancreas.
Reverse-transcriptase polymerase chain reaction analysis was perforemd on messenger RNA obtained from freshly isolated Sca-1+/c-Kit+ cells from submandibular glands, monolayer-cultured mSGP-1 cells, and cells from spheroidal bodies in thick matrigel (Fig. 5). Freshly isolated Sca-1+/c-Kit+ cells from submandibular glands and monolayer-cultured mSGP-1 cells expressed the genes of AFP and cytokeratin 19 but not of albumin or α1-antitrypsin. Hepatic spheroidal bodies expressed the albumin gene but not the genes for α1-antitrypsin, tryptophan 2,3-dioxygenase (data not shown), or tyrosine aminotransferase (data not shown), which are markers of mature hepatocytes. These findings indicate that differentiated mSGP cells in hepatic spheroidal bodies are still immature.
Transplantation of mSGP Cells Into Liver.
To examine in vivo differentiation, mSGP-1 cells were transplanted into the liver by way of the portal vein. Two types of cells were prepared for the transplants: dispersed mSGP-1 cells (2 × 105 cells/mouse) cultured on type I collagen and spheroidal bodies cultured in matrigel (100 spheroids/mouse). The recipients were 8-week-old female mice that had undergone two-thirds partial hepatectomy before transplantation. When mSGP-1 cells were transplanted as dispersed cells, the cells were scattered throughout the recipient liver, but proliferation of the cells was seldom detected. Next, three groups of spheroidal bodies having diameters of 20 to 30, 50 to 70, and 100 to 150 μm, were each transplanted separately. Transplants of spheroid bodies having diameters of 100 to 150 μm caused death in most recipients because the spheroids obstructed vessels in the liver. However, spheroidal bodies having diameters of 20 to 30 μm proliferated and expanded better than spheroids of other sizes. Donor-derived cells were detected by whole-mount β-galactosidase staining of recipient livers, which was undertalen 14 days after transplantation (Fig. 6). The β-galactosidase–positive cells were seen focally or in meshlike form along the surface of the liver (Fig. 6A, B). Liver sections revealed donor cells distributed under the capsule of the liver in a comblike configuration (Fig. 6C).
Results of immunostaining with β-galactosidase antibodies sections of liver from animals that had received spheroidal bodies are shown in Fig. 7A–D. Donor-derived cells (red) could not be distinguished from surrounding recipient cells and were integrated into hepatic cords (Fig. 7A–D). mSGP-derived cells were distributed in hepatic cords from the portal triad toward the central vein (Fig. 7C, D). To quantitate proliferation of transplanted mSGP in the recipient liver, we performed β-galactosidase staining of secrions from right lobes. Three observers assessed 10 random fields of vision from each of five recipient mice at a magnification of ×100 and calculated the percentage of cells that were β-galactosidase–positive in each field. It was estimated that mSGP-1 cells occupied 10.40 ± 2.76 % of the liver. Immunostaining with anti-albumin antibodies demonstrated that the mSGP-derived cells expressed albumin (Fig. 7E, F), indicating their ability to differentiate into hepatic-type cells. Furthermore, they also expressed α1-antitrypsin, a marker of mature hepatocytes, which was not detected during in vitro differentiation (Fig. 7G, H). Therefore, mSGP cells differentiated more into a hepatic lineage in vivo than in vitro. In addition, a proportion of mSGP cells formed ductlike structures, the epithelial cells of which were positive for cytokeratin 19 (Fig. 7I, J), suggesting that they had differentiated into bile ducts. These observations indicate the multipotency of mSGP-1 cells in vivo.
mSGP-1 is one of the multipotent tissue stem cells derived from adult mouse salivary glands. They maintain a polygonal, epithelial-like configuration and the ability to proliferate on type I collagen in culture. Analysis of surface markers, using antibodies to Sca-1 and c-Kit, showed that Sca-1 expression of freshly sorted cells was preserved in almost all cultured cells; some c-Kit–positive cells were present. Some populations of mSGP-1 cells expressed α6 and α3 integrins, which bind to laminin, and αV integrin, which binds to fibronectin. These integrins are known to activate the major signaling pathways for cell cycle progression or remodeling of cytoskeleton. It is, therefore, acceptable that tissue stem cells express such integrins.28
The most striking feature that mSGP shares with rat SGP-1 is the expression of cytosolic laminin. Laminin is a major component of the basement membrane. The cytosolic expression of laminin was found in the rat SGP cell.22 In rat salivary glands, duct ligation led to the formation of small clusters of cells, which expressed laminin in the cytosol in areas of regenerative response.22 In the mouse, salivary gland tissue prior to ligation expresses laminin in the basement membranes of acinar and epithelial duct cells. After ligation, small clusters of laminin-positive cells appeared in periductal areas (data not shown). mSGP-1 cells were a separated Sca-1+/c-Kit+ cell population in glands after duct ligation. These Sca1+/c-Kit+ cells were localized in the ductal and periductal areas of the gland, where a regenerative response occurred. It is assumed that mSGP-1 cells originate in the periductal area of the salivary gland, because they are positive for cytosolic laminin and negative for cytokeratin 19. Tissue stem cells of organs that have exocrine functions, such as the pancreas16 or mammary gland,29, 30 are localized in the ductal or periductal areas of the tissues. It is suggested that such stem/progenitor cells in the salivary gland have a similar localization.
The mSGP cells have multipotency in vitro; they can differentiate into hepatic and pancreatic lineages. To induce hepatic differentiation, it is necessary to culture these cells in matrigel. For pancreatic differentiation, it is necessary to treat spherical cultures of them with GLP-1. Each method of induction is selective, but differentiation is incomplete; more effective methods are necessary. When spheroidal bodies cultured in matrigel for hepatic differentiation were plated on type I collagen, adhesion and proliferation occurred. Expanded spheroids, however, lost expression of albumin (data not shown). Thus, the three-dimensional formation seems to be essential for both the induction and maintenance of differentiation.
Transplants of spheroidal bodies that had been cultured in matrigel and had diameters of 20 to 30 μm proliferated better than other mSGP-1 transplants. When mSGP cells were transplanted as dispersed cells, their proliferation in the recipient liver was rarely observed. For repopulation of transplanted cells in the liver, mSGP-1 cells that have differentiated slightly to a hepatic lineage seem to be preferable to more undifferentiated progenitor cells.
The transplanted mSGP-1 cells completely integrated into hepatic cords. With respect to function, mSGP-1 cells differentiated into a hepatic phenotype; they expressed albumin and α1-antitrypsin in the recipient liver. In addition, the expression of α1-antitrypsin, which did not occur in culture, was detected in transplanted cells; this finding suggests an important role for the micro-environment of the recipient liver in the differentiation of mSGP-1 cells into a hepatic phenotype. mSGP-1 cells also differentiated into cytokeratin 19–positive, ductlike epithelial cells. Thus, mSGP-1 cells exhibit multipotency in vivo.
During embryogenesis, both the liver and ventral pancreas appear to arise from the same cell population located within the embryonic endoderm; whether these cells differentiate into hepatic or pancreatic tissue is determined by their location, local growth factors, and expression of cell adhesion molecules.31 Thus it can be assumed that epithelial cell populations within the pancreas and liver share common stem cell populations. Yang and colleagues reported that hepatic oval cells transdifferentiate into insulin, glucagon, and pancreatic polypeptide-expressing pancreatic endocrine cells in vitro.12 Furthermore, when Dabeva and colleagues transplanted pancreatic epithelial progenitor cells into the liver by way of the spleen or the portal vein, the transplanted cells were integrated into recipient liver, expressed albumin, and differentiated into hepatocytes.32 In addition, Suzuki and colleagues isolated hepatic stem cells from fetal mouse liver using fluorescence-activated cell sorting and transplanted them into the pancreas and duodenal wall of adult mice.24 These cells subsequently differentiated into pancreatic ductal and acinar cells or intestinal epithelial cells, respectively.24 The fact that mSGP-1 cells, the tissue stem cells of salivary glands, can differentiate into hepatic and pancreatic lineages suggests the existence of common tissue stem cells of endodermal origin.
We thank Tatsuko Kubo for assistance with the preparation of tissue for histology and Kaede Yanagida for assistance with writing the manuscript.