Identification and expansion of a unique stem cell population from adult mouse gallbladder§


  • Rohan Manohar,

    1. McGowan Institute for Regenerative Medicine, Department of Pathology, University of Pittsburgh School of Medicine, Pittsburgh, PA
    Search for more papers by this author
  • Junji Komori,

    1. McGowan Institute for Regenerative Medicine, Department of Pathology, University of Pittsburgh School of Medicine, Pittsburgh, PA
    Search for more papers by this author
  • Lynda Guzik,

    1. McGowan Institute for Regenerative Medicine, Department of Pathology, University of Pittsburgh School of Medicine, Pittsburgh, PA
    Search for more papers by this author
  • Donna B. Stolz,

    1. Center for Biological Imaging, Department of Cell Biology and Physiology, University of Pittsburgh School of Medicine, Pittsburgh, PA
    Search for more papers by this author
  • Uma R. Chandran,

    1. Department of Biomedical Informatics, University of Pittsburgh School of Medicine, Pittsburgh, PA
    2. University of Pittsburgh Cancer Institute, Hillman Cancer Center, Pittsburgh, PA
    Search for more papers by this author
  • William A. LaFramboise,

    1. University of Pittsburgh Cancer Institute, Hillman Cancer Center, Pittsburgh, PA
    2. Clinical Genomics Facility, University of Pittsburgh School of Medicine, Pittsburgh, PA
    Search for more papers by this author
  • Eric Lagasse

    Corresponding author
    1. McGowan Institute for Regenerative Medicine, Department of Pathology, University of Pittsburgh School of Medicine, Pittsburgh, PA
    • McGowan Institute for Regenerative Medicine, University of Pittsburgh, 450 Technology Drive, Room 329, Pittsburgh, PA 15219
    Search for more papers by this author
    • fax: 412-624-5228

  • Potential conflict of interest: Nothing to report.

  • This work was supported, in part, by the Pittsburgh Tissue Engineering Initiative, the Commonwealth of Pennsylvania, and by National Institutes of Health grant R01 DK085711.


The identification of resident stem cells in the mouse gallbladder is, to date, unexplored. In addition, the relationship between adult gallbladder stem cells and intrahepatic bile duct (IHBD) cells is not well understood. The aim of this study was to isolate stem cells from an adult mouse gallbladder and determine whether they were unique, compared to IHBD cells. By limiting dilution analyses and index sorts, we found that an EpCAM+CD49fhi epithelial cell subpopulation from primary gallbladder is enriched in colony-forming cells, compared to EpCAM+CD49flo cells. EpCAM+CD49fhi cells expressed cluster of differentiation (CD)29, CD133, and stem cell antigen-1, but were negative for lineage markers CD31, CD45, and F4/80. Using a novel feeder cell-culture system, we observed long-term (>passage 20) and clonal expansion of the EpCAM+CD49fhi cells in vitro. In a matrigel differentiation assay, EpCAM+CD49f+ cells expanding in vitro underwent organotypic morphogenesis forming ductular structures and cysts. These structures are similar to, and recapitulate a transport function of, primary gallbladder. EpCAM+CD49f+ cells also engraft into the subcutaneous space of recipient mice. We compared primary gallbladder and IHBD cells by flow cytometry and found phenotypic differences in the expression of CD49f, CD49e, CD81, CD26, CD54, and CD166. In addition, oligonucleotide microarrays showed that the expanded EpCAM+CD49f+ gallbladder cells and IHBD cells exhibit differences related to lipid and drug metabolism. Notable genes that were different are cytochrome P450, glutathione S-transferase, Indian hedgehog, and solute carrier family genes. Conclusion: We have isolated an epithelial cell population from primary mouse gallbladder with stem cell characteristics and found it to be unique, compared to IHBD cells. (HEPATOLOGY 2011)

Understanding the resident stem cell populations of the biliary system has great importance for basic biology and biliary diseases. The biliary tree is divided into the intra- and extrahepatic biliary systems. The latter consists of the gallbladder, cystic duct, and the common bile duct.1 The biliary system is a conduit for bile to be transported from the liver to the intestine. The gallbladder, in turn, stores the bile and regulates its content and concentration, playing an important role in the digestive process.2, 3

Though there has been a lot of recent interest in the liver stem cell field,4 there is still a paucity of data regarding gallbladder stem cells. The biliary system, hepatocytes, and ventral pancreas develop from the ventral foregut endoderm.5, 6 Histological evidence suggesting that both intra- and extrahepatic systems originate from the hepatic diverticulum has led to the hypothesis that they descend from the same progenitor cell. However, the cell-intrinsic factors that result in their specification have heretofore been unclear. Recently, it has been shown that the progenitor cells that give rise to each system separate out during development.7 Using a Pdx1-Cre mouse, Spence et al.7 demonstrated that hepatocytes and intrahepatic bile duct (IHBD) cells derive from Pdx1- cells, whereas the extrahepatic bile duct (EHBD) cells and ventral pancreas derive from Pdx1+ cells. Sox17 controls the specification of the EHBD and pancreatic cells. Sox17 loss-of-function embryos exhibit gallbladder agenesis and the presence of ectopic pancreatic tissue in the extrahepatic bile duct. Conversely, Sox17 gain of function results in ectopic ductal tissue in the developing pancreas. In both cases, the intrahepatic system is not affected. It appears that the IHBD and EHBD cells descend from separate progenitor cells governed by separate transcriptional cascades. It is, therefore, possible that adult IHBD and EHBD cells could be distinct, as well.

The aims of this study were to isolate and characterize stem cells from the adult mouse gallbladder and compare their phenotypic and expression profiles with IHBD cells. In addition to basic biology, an understanding of gallbladder stem cells would be vital to the study of gallbladder carcinoma, a rare, but poorly understood, malignancy8 and congenital diseases involving biliary dysmorphogenesis, such as biliary atresia.9 It would also elucidate the ontogeny of cells in the biliary system.

Stem cells are defined as undifferentiated cells that can self-renew at the single-cell level and form lineage-committed progeny.10 In this study, we used colony-forming and single-cell assays, along with a morphogenesis assay, to characterize an EpCAM+CD49fhi epithelial subpopulation within primary mouse gallbladder that has stem cell characteristics. The gallbladder stem cells can be propagated invitro through long-term passage (>passage 20) and can engraft in the subcutaneous space of recipient mice. Last, gallbladder stem cells and IHBD cells have distinct expression profiles. These data represent one of the first reports to isolate and characterize the resident stem cell population in the adult mouse gallbladder.


CD, cluster of differentiation; CFU, colony-forming unit; CK19, cytokeratin 19; CYP, cytochrome P450; EGF, epidermal growth factor; EHBD, extrahepatic bile duct; EpCAM, epithelial cell adhesion molecule; ELDA, extreme limited dilution analysis; FACS, fluorescence-activated cell sorting; GFP, green fluorescent protein; GST, glutathione S-transferase; IFN, interferon; IHBD, intrahepatic bile duct; ITS, insulin-transferrin selenium; LDA, limiting dilution analysis; MDR, multidrug resistance; Rh-123, rhodamine 123; SAM, significance analysis of microarray; Sca1, stem cell antigen-1; SE, standard error; TEM, transmission electron microscopy; TGF, transforming growth factor; 3D, three-dimensional; UBC, ubiquitin C.

Materials and Methods

Gallbladder Cell Isolation and Culture.

Gallbladder cells were isolated from C57BL/6-Tg ubiquitin C/green fluorescent protein (UBC-GFP) 30Scha/J mice (Jackson Laboratory, Bar Harbor, ME). For further details, see Supporting Methods.

Fluorescence-Activated Cell Sorting Analysis.

Single-cell suspensions were stained with appropriate antibodies (Supporting Table 1) at 1e6 cells/tube and analyzed on the BD FACSCanto or BD FACSAriaII (BD Biosciences, Woburn, MA). For further details, see Supporting Materials.

Oligonucleotide Microarrays.

Expanded gallbladder and IHBD cells were stained with EpCAM-biotin and eluted through two sequential MS MACS separation columns (Supporting Fig. 1). For further details, see Supporting Materials.


EpCAM Is a Gallbladder Epithelial Marker.

Gallbladder cells were isolated from GFP donor mice, and epithelial cells were separated by flow cytometry. EpCAM, an epithelial surface marker, is expressed on simple epithelial cells, such as keratinocytes and thymic epithelial cells,11 as well as on IHBD cells, but not hepatocytes, mesenchymal, or hematopoietic cells.12 Analysis of mouse gallbladder showed that most epithelial cells are EpCAM+ (Fig. 1A). No expression was detected on mesenchymal cells. To confirm epithelial identity, we performed colocalization studies with EpCAM and cytokeratin 19 (CK19), a pan-biliary marker.13 Epifluorescence and confocal microscopy performed on acetone-fixed sections show that most CK19+ cells were EpCAM+ (Fig. 1B). Therefore, EpCAM marks most gallbladder epithelial cells.

Figure 1.

CD49f is heterogeneous in primary gallbladder epithelium. (A) Paraffin sections of mouse gallbladder were stained with EpCAM. EpCAM stained only gallbladder epithelial cells. Basolateral membranes of epithelial cells were EpCAM+ (arrowheads). (B) Acetone-fixed sections were stained with CK19 and EpCAM. Epifluorescent (left panel) and confocal (right panel) micrographs indicate that CK19 costains with EpCAM. (C) Flow cytometric analyses on primary gallbladder indicate CD49f heterogeneity (labeled percentages are frequencies of parent population). Crosshairs on the flow plot indicate autofluorescence of unstained cells determined by the control population. (D) Confocal micrographs of acetone-fixed sections stained with EpCAM and CD49f. White arrows: EpCAM+CD49fhi cells. White arrowheads: EpCAM+CD49flo cells. Scale bars: 100 μm.

CD49f Is Heterogeneously Expressed on Primary Gallbladder Epithelial Cells.

Because there is a paucity of cell-surface markers for gallbladder cells, we began screening primary gallbladder for general markers of stem and progenitor cells (Supporting Table 1). Of the 38 markers we considered, three markers—cluster of differentiation (CD)49f, Dolichos biflourus agglutinin, and stem cell antigen-1 (Sca1)—were heterogeneously expressed on primary gallbladder epithelial cells (Fig. 1C; Supporting Fig. 2). However, we were only able to separate functionally distinct populations—EpCAM+CD49fhi and EpCAM+CD49flo—with CD49f. Function, in this case, was defined by a colony-forming assay (see below). Heterogeneous expression of CD49f was confirmed by immunohistochemistry (Fig. 1D). Various reports have identified CD49f, integrin α-6, as a stem cell marker in fetal and adult liver14-16 and other ductal epithelial tissue, such as the breast.17, 18 EpCAM+CD49fhi cells expressed markers associated with epithelial stem cells, such as CD29, CD133, and Sca1, but not mesenchymal or hematopoietic markers CD31, CD45, and F4/80 (Supporting Table 1). These data led us to hypothesize that CD49f is a candidate gallbladder stem cell marker.

Gallbladder Cells Expanded In Vitro Are CD49f+.

Gallbladder cells were cultured invitro in conditions that select for epithelial cell growth.19 Briefly, total cell isolate from primary gallbladder was plated on irradiated rat mammary tumor cell line LA7 that served as feeder cells. Transmission electron microscopy (TEM) and flow cytometric analyses indicated that there was no fusion between the gallbladder and feeder cells (Supporting Fig. 3). Because stem cells have the capacity for self-renewal, we predicted that expansion invitro would enrich for primitive or stem cells. Flow cytometry analyses of cells after one expansion (p0) showed that only epithelial cells (EpCAM+) expand on the feeders (Fig. 2A). EpCAM cells that were sorted from primary gallbladder did not proliferate (data not shown). Importantly, we found that all gallbladder cells at p0 were CD49f+ (Fig. 2B), supporting the notion that invitro expansion selects for EpCAM+CD49f+ primitive epithelial cells.

Figure 2.

CD49f expression enriches for gallbladder stem cells. Gallbladders (n ≥ 5) from male and female mice were pooled and analyzed or sorted. (A) Flow cytometric analyses on primary and expanded mouse gallbladder cells show that only epithelial cells expand in the LA7 culture system. d0, primary cells; p0, cells at first expansion. Feeder cells (EpCAMGFP) have been gated out of EpCAM versus CD45/TER119 plot for the cells at first expansion. (B) Flow cytometric analyses of expanded gallbladder cells (p0) show that all EpCAM+ cells are CD49f+. LA7 feeder cells are the EpCAMGFP cells in middle plot. (C) LDAs were carried out on EpCAM+CD49fhi and EpCAM+CD49lo cells from primary gallbladder. CFU frequency ± SE (L-Calc) indicates that CD49f enriches for stem cells. Pearson's chi-square statistic was calculated on the 49fhi and 49flo groups by extreme limited dilution analysis (ELDA). Exp#1: chi-square: 17.8; P value, 2.43e-5; Exp#2: chi-square: 42.9; P value, 5.79e-11. Crosshairs on the flow plot in (A-C) indicate where the control population was present in the lower left-hand corner. (D) Index sorts from primary gallbladder. Single EpCAM+ cells were sorted into each well of three 96-well plates. (i) Data from three representative wells that grew showing CD49f expression and subsequent colony morphology. (ii) Table showing index-sort results. Eleven of twelve (91.7%) of the wells that grew originated from CD49fhi cells. Scale bars: 100 μm.

CD49f Enriches for Gallbladder Stem Cells.

To evaluate CD49f as a gallbladder stem cell marker, we performed limiting dilution analyses (LDAs) and index sorts. The LDA quantifies the frequency of a specific subpopulation of cells with a biological activity20 and was key to the isolation of hematopoietic21 and neural22 stem cells. In the evaluation of stem cells, biological activity is typically defined as the ability to form a colony and the LDA serves to quantify stem and progenitor cells. We separated EpCAM+CD49fhi and EpCAM+CD49flo cells from primary gallbladder and performed LDAs. EpCAM+CD49fhi cells exhibited a significantly higher enrichment in colony-forming unit (CFU) frequency (ranging from 1 of 15 to 1 of 4), compared to EpCAM+CD49flo cells (1 in 71 to 1 in 62) (Fig. 2C). Chi-square tests confirmed that ranges in CFU frequency ± standard error (SE) were significantly different between EpCAM+CD49fhi and EpCAM+CD49flo cells (P < 0.001).

We then performed index sorts to confirm these data. An index sort records the phenotype and well number of each deposited cell during a single cell sort. In this manner, the specific surface-marker profile of cells that form colonies can be determined retrospectively. In our experiment, 288 single EpCAM+ cells were sorted and the CD49f profile of each sorted cell was recorded. Retrospective analyses indicated that 11 of 12 (91.7%) of the colonies that formed originated from CD49fhi cells (Fig. 2D). These data, together with the LDA results, definitively demonstrate that CD49f enriches for candidate gallbladder stem cells.

Expanded Gallbladder Cells Exhibit Morphological Heterogeneity.

We observed the formation of two distinct types of colonies in EpCAM+CD49f+ gallbladder cultures at p0. The first type consisted of large colonies with an undifferentiated phenotype comprising small cells with a large nuclear-cytoplasmic ratio (Fig 3A,B, red arrowheads). We termed these the “flat colonies.” The second type was smaller, more organized colonies called “glandular colonies” with an organotypic phenotype consisting of cells organized around a lumen (Fig. 3A,B, white arrowheads). Flat colonies were more numerous than glandular ones. TEM on the flat colonies revealed a single layer of cuboidal epithelial cells (Fig. 3C). These cells have defined apical-basolateral polarity, apical microvilli, and appear to secrete basement membrane at their basolateral surface. They also have interdigitating lateral membranes and junctional apparatus typical of gallbladder epithelial cells.

Figure 3.

EpCAM+CD49f+ gallbladder cells exhibit morphological heterogeneity in vitro. (A) At first passage (p0) invitro, gallbladder cells exhibit two distinct morphologies. Flat colonies (red arrowheads) and glandular colonies (arrowheads) are observed. Only flat colonies are observed at later passages (p1, p6, and p17). (B, C) Light and electron micrographs showing flat colonies and glandular colonies. Flat colonies consist of cuboidal cells with defined apical to basolateral polarity, show putative fluid transport (arrows), and have junctional apparatus typical of gallbladder epithelial cells. Glandular colonies consist of columnar cells organized around a central lumen. Secreted product (*) is observed in the lumen. (D) To determine whether glandular colonies could expand past first passage, colonies were picked from first expansion (p0) culture. Only flat colonies expanded past p0. MV, microvilli; RER, rough endoplasmic reticulum; LY, lysosome; M, mitochondria; S, secretory granule; N, nucleus; BM, basement membrane; Arrowheads: tight junctions. *Glycocalyceal substance. Unless specified otherwise, scale bars: 100 μm.

Conversely, the glandular colonies consist of columnar epithelial cells organized around a central lumen (Fig. 3C) and exhibit junctional apparatus. Unlike flat colonies, numerous secretory granules are seen in their apical cytoplasm, and secretory products are present in their lumen (Fig. 3C). The flat and glandular colonies are distinct by morphology and ultrastructure. Importantly, only the flat colonies are observed at late passages (Fig. 3A), indicating that the glandular colonies are not capable of long-term self-renewal (>p3). To test this hypothesis, we passaged single colonies from p0 cultures. None of the glandular colonies could be successfully repassaged (Fig. 3D). This suggests that serial passage of the gallbladder cells past the first expansion enriches for EpCAM+CD49f+ cells that form flat colonies. Because we found no additional markers to further purify gallbladder stem cells, we hypothesized that the cells past the first expansion are candidate stem cells. To determine their stemness, we tested whether the expanded EpCAM+CD49f+ gallbladder cells could satisfy the stem cell criteria of clonogenic self-renewal and lineage commitment.

EpCAM+CD49f+ Cells Differentiate into Gallbladder-Like Structures In Vitro .

We developed a novel invitro differentiation assay by utilizing the basement membrane extracellular matrix, Matrigel. Matrigel has been shown to promote or maintain the differentiation or three-dimensional (3D) morphogenesis of numerous cell lines and primary cells, including hepatocytes and IHBD cells.23-25 In our assay, expanded EpCAM+CD49f+ gallbladder cells (>p1) were mixed with serum-free media and layered above with Matrigel (Fig. 4A). Within 1 week, we noticed the formation of two distinct morphogenetic structures—ductular structures that adhered to the plastic (Fig. 4B) and cysts that were suspended in the Matrigel (Fig. 4C). Both structures persisted for over 6 weeks in culture. The Matrigel was removed from the plastic to confirm that the cysts were suspended in it. Similar morphogenesis was observed with primary gallbladder cells (data not shown).

Figure 4.

Expanded EpCAM+CD49f+ cells form in vitro gallbladders. (A) Schematic of invitro differentiation assay. EpCAM+CD49f+ gallbladder cells (>p1) were plated on tissue-culture–treated dishes in serum-free media and layered on top with Matrigel. Two different forms of morphogenesis were observed: ductular structures (red arrow) and cysts (white arrow). Also, colonies of cells were observed growing on the plastic (white arrowhead). (B) Ductular structures are luminal, consisting of long interconnecting ducts lined by a single or double layer of cells. (i) Light micrographs showing interconnecting ducts. *Lumen. (ii) Confocal micrograph of live ductular structure insitu in the Matrigel. (C) Cysts similarly consist of a single or double layer of cells surrounding a hollow lumen. (i) Light and confocal micrographs of live cysts insitu in the Matrigel, showing a hollow lumen lined by cells. (ii) Cysts were isolated from Matrigel and stained with toluidine blue. (D) Electron micrographs of cysts (21 days postplating) and primary gallbladder, indicating similar ultrastructure. Both consist of columnar epithelial cells with apical to basolateral polarity, basal nuclei, microvilli, and junctional apparatus. (E) Cysts were incubated with 100 μM of Rh-123 ± 10 μM of verapamil. Confocal micrographs showing time-lapse images of optical sections at indicated times. Rh-123 accumulates in lumen over 1 hour. Transport is blocked with verapamil. MV, microvilli, M, mitochondria, N, nucleus. Unless specified otherwise, scale bars: 100 μm.

The ductular structures consisted of ball-shaped interconnecting ducts. Confocal microscopy on ductular structures in Matrigel showed that they are hollow (Fig. 4B, ii). The cysts similarly consisted of the hollow, ball-shaped structures, but lacked interconnecting ducts and had much larger lumen (Fig. 4C). They appeared early on in culture—approximately 2 or 3 days postplating—and expanded over time (Supporting Fig. 4). TEM studies show that the cysts exhibit similar ultrastructure to primary mouse gallbladder (Fig. 4D).

We then tested whether the ductular structures and cysts represent two different morphogenetic programs. Ductular structures and cysts were separated and LDAs were performed, where the cells were sorted back into Matrigel or on LA7 feeder cells. When sorted back into Matrigel, ductular structures could reform ductular structures and cysts, and cysts were able to reform both structures, as well (data not shown). In addition, both expanded equally well on the feeders and no differences in LDA were observed (data not shown). Last, we performed the same assay in borosilicate dishes that inhibit cell attachment. We found that only cysts formed, which, when passaged, could form both cysts and ductular structures. Therefore, ductular structures and cysts do not represent separate morphogenetic programs. Their appearance might be a function more of their microenvironment—attached to plastic versus suspended in the Matrigel—than intrinsic differences.

The physiological function of the gallbladder is to concentrate the bile and regulate its content by secretory processes.2, 3 These functions are, in part, the result of multidrug resistance (MDR) proteins. Rhodamine 123 (Rh-123), an MDR substrate, has been shown to accumulate in the lumen of cysts formed by a hepatic progenitor cell line grown in Matrigel.23 We reasoned that such a transport assay would also be indicative of function for gallbladder cells. Rh-123 was added to media of Matrigel cultures, and confocal images were taken at various time points. We observed the steady accumulation of dye in the lumen of cysts over 1 hour (Fig. 4E). This transport was blocked by the addition of verapamil, an MDR inhibitor. These data indicate that cysts transport dye from their basal side into the lumen, thereby recapitulating a transport function of the gallbladder.

EpCAM+CD49f+ Gallbladder Cells Can Self-Renew and Differentiate From Single Cells.

To test clonogenic self-renewal of EpCAM+CD49f+ gallbladder cells, we sorted single cells into 384-well plates seeded with LA7 feeders and imaged every well to confirm the presence of the single cell (Fig. 5A). In this manner, five clonal gallbladder cultures were generated. Further analyses were carried out with Clone B21 and Clone N12 (Fig. 5A). Surface-marker analyses indicated that both Clone B21 and N12 have the same phenotype as the parent gallbladder cell population (Supporting Table 1), suggesting that the clones are able to recapitulate the heterogeneity of the parent. In addition, both Clone B21 and N12 exhibit the same morphogenesis in Matrigel as the parent (Fig. 5B).

Figure 5.

Expanded EpCAM+CD49f+ cells undergo clonogenic self-renewal. (A) Schematic of single-cell clonogenic assay. EpCAM+CD49f+ gallbladder cells at first passage (p0) were sorted into 384-well plates at 1 cell/well. In this manner, eight clones were generated. Representative images of CloneB21 and CloneN12 are depicted. (B) Clonal gallbladder cultures were expanded further, and differentiation assays were carried out. Clones exhibit similar morphogenesis, compared to parent cultures. *Lumen, Scale bars: 100 μm.

Expanded EpCAM+CD49f+ Cells Can Engraft In Vitro .

We then determined whether the expanded EpCAM+CD49f+ cells could survive and engraft invivo. An ideal location for engraftment would be the native gallbladder. However, because there currently are no protocols that allow for the injection and maintenance of cells in the gallbladder, we attempted engraftment at an ectopic location. Okumura et al.26 have reported the long-term engraftment of invitro explants of human gallbladder in the subcutaneous space of athymic nude mice. We injected the expanded EpCAM+CD49f+ cells mixed with Matrigel into the subcutaneous neck region of immunodeficient mice. We observed the formation of cyst-like structures in the subcutaneous space 1 week postinjection (Fig. 6A). These cysts consisted of cells organized around a central lumen. Seven of seven (100%) mice injected formed cysts. However, engraftment was short term. Only 1 of 3 (33%) mice exhibited cyst formation at 2 weeks. Similar results were obtained with clonal cultures.

Figure 6.

Expanded EpCAM+CD49f+ cells engraft in vivo. Parent or clonal gallbladder cultures were mixed with Matrigel and injected into the subcutaneous (subq) space of Rag2−/−γC−/− mice. Cysts were observed at 1 and 2 weeks postinjection. (A) Macroscopic view (left panel) of Matrigel isolated at 1 week from mouse injected with parent EpCAM+CD49f+ cells (right panel). Sections of same Matrigel stained with EpCAM and GFP, showing that cysts are hollow and EpCAM+. Cysts were seen in at 1 week in 7 of 7 and 2 of 3 mice injected with parent and clonal cultures, respectively. Also, 1 of 5 and 1 of 3 mice, respectively, exhibited engraftment at 2 weeks. (B) Clonal gallbladder cells remain EpCAM+CD49f+invivo. Expanded gallbladder cells from CloneB21 were injected subq. At 2 weeks, the Matrigel was removed and cells were reisolated. Flow cytometric analysis shows that reisolated cells are EpCAM+CD49f+. These cells were expanded invitro on feeder cells. Flow cytometric analysis at p1 (second passage) showing that they remain EpCAM+CD49f+. Crosshairs on the flow plot in (A-C) indicate where the control population was present in the lower left-hand corner. Debris and cell aggregates were gated out in each plot. In the middle plot, GFP cells were gated out, as well. *Lumen. Unless specified otherwise, scale bars: 100 μm.

We then isolated cells from cysts invivo 2 weeks postinjection and cultured them invitro to test their ability to reinitiate cultures with stem cell properties. Flow cytometry analyses showed that cells isolated from cysts invivo were EpCAM+CD49f+. These cells reexpanded in vitro on feeder cells, forming colonies morphologically identical to parent and clonal cultures (data not shown), and remained EpCAM+CD49f+ (Fig. 6B). These data indicate that expanded EpCAM+CD49f+ cells survive and engraft invivo while retaining their proliferative ability in vitro.

Gallbladder Cells Are Unique Compared to IHBD Cells.

There is evidence indicating that intra- and extrahepatic bile duct cells develop separately.7 To date, there are no reports of the molecular differences—if any—between IHBD and gallbladder cells. We first screened primary IHBD cells with the same antibody panel used for primary gallbladder cells (Supporting Table 1). Most IHBD cells express EpCAM12 and we used EpCAM expression to separate IHBD cells from other liver cells. Briefly, after liver perfusion of GFP+ mice, the high spin fraction was separated and used to isolate IHBD cells.

Interestingly, we found differences in integrin expression, including CD49f (Fig. 7A). Other notable markers that showed differences between IHBD and gallbladder cells were CD49e, CD81, CD54, CD26, and CD166 (Fig. 7A). We then determined whether expanded gallbladder cells and IHBD cells were different. IHBD cells are capable of expansion on LA7 feeders, and feeder cells select for EpCAM+ cells (Fig. 7B). In addition, IHBD cells form flat colonies similar to gallbladder cells. The phenotypic profiles of IHBD cells and gallbladder cells converged in culture, and we did not detect any differences using the foregoing panel of antibodies. Consequently, we used oligonucleotide microarrays to test for expression differences between the two cell types.

Figure 7.

Expanded EpCAM+CD49f+ cells are unique, compared to intrahepatic biliary cells. (A) Flow cytometric analyses of EpCAM+ (epithelial) cells from primary gallbladder and primary liver (see Materials and Methods). d0, primary cells. Phenotypic differences with select cell-surface markers suggest that gallbladder cells and IHBD cells are different. Crosshairs on the flow plot in (A-C) indicate where the control population was present in the lower left-hand corner. (B) IHBD cells expand on LA7 feeders and exhibit similar morphology to expanded gallbladder cells. Flow cytometric analyses (upper panel) of expanded IHBD cells (p2) indicate that all GFP+ cells are EpCAM+. Expanded IHBD cells (lower panel) exhibit a flat colony appearance. (C) Heat map of 53 genes with known annotations and fold changes (≥2) differentially expressed between IHBD cells and gallbladder cells (Supporting Table 3). Data were imported in Cluster and heat maps were generated with TreeView ( Black represents genes whose expression was at the mean intensity; red represents intensities that are greater than the mean; green represents intensities that are less than the mean. Unless specified otherwise, scale bars: 100 μm.

We have shown that the expanded gallbladder cells or EpCAM+CD49f+ cells are capable of self-renewal and lineage commitment. It is possible that the expanded IHBD cells might satisfy these requirements, as well. However, the evaluation of IHBD stem cells belongs to a different study, and we focused on the differences in the transcriptomes of the expanded gallbladder and IHBD cells. Briefly, expanded gallbladder cells and IHBD cells were separated from LA7 feeder cells using magnetic-activated cell sorting (Supporting Fig. 1). Differential gene expression between expanded gallbladder and IHBD cells (fold change, ≥2) were calculated by significance analysis of microarrays (SAMs),27 using a false discovery rate of 10%. In this manner, we found 64 genes to be up-regulated in IHBD cells (Fig. 7C), including those involving lipid metabolism (eight genes), stem cell proliferation (three genes), and drug metabolism (two genes) (Supporting Table 2). Notable genes or groups of genes that were different were cytochrome P450 (CYP), Indian hedgehog, glutathione S-transferase (GST), and solute carrier families 22, 26, 37, and 45 (Supporting Table 3). These differences indicate that the expanded gallbladder cells and IHBD cells have distinct transcriptomes and suggest functional differences as well.


Little is known about the resident stem cells in the gallbladder and the relationship between the stem cells of the hepatobiliary system. Our aim here was to identify and characterize stem cells in the adult mouse gallbladder. We found that an EpCAM+CD49ffhi subpopulation from primary mouse gallbladder can expand from single cells and exhibits morphogenesis in organotypic culture invitro. Both parent and clonal cultures were capable of survival and short-term morphogenesis in an adapted invivo assay. We, therefore, concluded that EpCAM+CD49ffhi gallbladder cells satisfy the stem cell criteria of clonogenic self-renewal and lineage commitment and represent a gallbladder stem cell population. Last, we determined that gallbladder stem cells and IHBD cells expanded in vitro have distinct transcriptomes, suggesting that cells of the IHBD and EHBD systems are different.

This study is the first to describe the identification and prospective isolation of stem cells from an uninjured mouse gallbladder. Previous reports of stem cells in the EHBD system have focused on injury models28 or disease conditions, such as biliary atresia.29, 30 Furthermore, these studies do not distinguish epithelial from nonepithelial cells in their isolation protocols. We used EpCAM to isolate gallbladder epithelial cells, thereby preventing contamination by nonepithelial cells. This is especially important, because we detected EpCAMCD49f+ cells in primary gallbladder by both immunohistochemistry and flow cytometry. The isolation and subsequent expansion of EpCAM+CD49f+ cells thereby allows for the definitive identification of resident epithelial stem cells.

We confirmed CD49f as a gallbladder stem cell marker by LDAs and index sorts from primary gallbladder. EpCAM+CD49fhi cells have a significantly higher CFU readout relative to EpCAM+CD49flo cells. The low enrichment in CFU readout indicates that additional markers are required to further purify stem cells, such that single cells can be isolated and expanded.31 Therefore, expression of EpCAM and CD49f enriches, but does not select for stem cells. All gallbladder epithelial cells expanded invitro were EpCAM+CD49f+. However, these cells exhibited morphological heterogeneity at first expansion, forming flat and glandular colonies. Interestingly, none of the glandular colonies and only a fraction of the flat colonies were capable of serial passage. It appears that the EpCAM+CD49fhi population in primary gallbladder is itself heterogeneous, with only a subpopulation of cells capable of self-renewal. We could not identify any additional markers to select for this specific subpopulation directly from primary tissue and, therefore, characterized the stemness of the EpCAM+CD49f+ cells expanded past p0.

We determined that the expanded EpCAM+CD49f+ cells can self-renew clonogenically. However, defined protocols for gallbladder epithelial cell differentiation do not exist. In the past, researchers have used collagen-gel sandwich culture to observe cyst morphogenesis with rabbit gallbladder epithelial cells.32, 33 The collagen gel is supplemented with exogenous growth factors, such as epidermal growth factor (EGF) and transforming growth factor (TGF)β1. We postulate that our 3D culture system is similar to collagen gel culture, in that Matrigel is an appropriate growth factor containing extracellular matrix that supports morphogenesis. Cyst formation in our culture was similar in morphology and ultrastructure to that observed before.32, 33 We also observed dye transport reminiscent of a transport function of the gallbladder. In addition, we observed similar morphogenesis in vivo after transplantation. We chose an ectopic location, because engraftment in the native gallbladder would be technically challenging and the subcutaneous space has been shown to engraft human gallbladder cells.26 Lee et al.34 have shown that gallbladder cells can engraft into the native liver of severe combined immunodeficiency mice. However, engraftment was significant only with tremendous injury to the liver (e.g., retrorsine and partial hepatectomy or carbon tetrachloride treatment) and required very large numbers of cells. For these reasons, we concluded that subcutaneous, rather than liver, engraftment would be a more apt invivo assay. In our hands, the EpCAM+CD49f+ cells only engraft in the short term (2 weeks post-transplantation). This short-term engraftment might be the result of a lack of growth stimulus in the recipient. Also, under physiological conditions, the rate of cell proliferation in the gallbladder epithelium is low.35 Future studies will determine whether long-term engraftment of EpCAM+CD49f+ cells is possible. For now, these data conclusively show that parent and clonal EpCAM+CD49f+ cells can organize into organotypic structures that mimic the morphology, ultrastructure, and function of the native gallbladder, both invitro and invivo.

Spence et al.7 have recently showed that IHBD cells and EHBD cells develop from separate precursors. However, there are no reports describing their similarities or differences in the adult. We found that expression of CD49f, CD49e, CD81, CD26, CD54, and CD166 was different between primary IHBD cells and gallbladder cells. The aim of our experiment was to evaluate the differences, if any, between gallbladder stem cells and IHBD cells. Expanded EpCAM+CD49f+ gallbladder cells (>p0) represent a purer stem cell population than primary EpCAM+CD49fhi cells. The latter forms both flat and glandular colonies, and only a fraction of the flat colonies can self-renew. Therefore, we ran microarray analyses on expanded EpCAM+CD49f+ cells (>p1) and expanded IHBD cells. The major groups of differentially expressed genes were CYP genes, GST, and the solute carrier family genes. Also, interferon (IFN)-inducible protein 27 was differentially expressed between gallbladder cells and IHBD cells. Interestingly, the expression of CD54 is known to be immunologically mediated.36 The immunologic properties of bile duct cells have long been considered. They are the primary site of damage in inflammatory diseases, such as primary biliary cirrhosis37 and biliary atresia9 and in liver allograft rejection.38 The differential expression of an IFN-inducible protein and CD54 indicates that the immunologic properties of IHBD cells and gallbladder cells could be different. Studies of IHBD cells are hindered by a technical inability to isolate and expand them from primary tissue.2, 39 We circumvented this hurdle by using LA7 feeder cells that allow for a robust expansion of IHBD epithelial (EpCAM+) cells. This expansion assay, along with the 3D Matrigel assay, could serve as interchangeable, technically easy tools to study bile duct cells. In all, the complete elucidation of the differences between IHBD cells and gallbladder cells belongs to another study, as does the evaluation of the stem cell characteristics of expanded IHBD cells. The focus of this article was to isolate and characterize gallbladder stem cells.

We postulate that this study will have important clinical significance. Gallbladder stem cells could be used to treat biliary atresia, as has been noted with hepatic progenitor cells.40 These cells could also be reprogrammed into hepatocytes or endocrine cells. There have been recent reports of the differentiation of gallbladder epithelial cells into hepatocytes,34, 41 and ectopic endocrine cells have been observed in the EHBD cells of Hes1−/− deficient mice.42 Gallbladder stem cells might be capable of such plasticity, which, along with the ready availability of donor tissue, would make them an attractive candidate for cell-based therapy.


The authors thank Mara Sullivan and Ming Sun for tissue processing for electron microscopy; Michael Burger and Christin Sciulli at the Clinical Genomics Facility for sample processing and initial data analysis; and Kelly Quesnelle for help with ingenuity pathway analysis. The authors also thank Aaron DeWard for editorial assistance and Drs. Michalopoulos, Fox, Orwig, Demetris, and Strom for their valuable advice and input.