In Vitro Cellular & Developmental Biology - Animal

, Volume 45, Issue 1, pp 44–54

Culture of endodermal stem/progenitor cells of the mouse tongue


  • Xiaoyan Luo
    • Department of Cell BiologyDuke University Medical Center
  • Tadashi Okubo
    • Center for Integrative BioscienceNational Institute of Natural Sciences
  • Scott Randell
    • Cystic Fibrosis/Pulmonary Research and Treatment CenterUniversity of North Carolina
    • Department of Cell BiologyDuke University Medical Center

DOI: 10.1007/s11626-008-9149-2

Cite this article as:
Luo, X., Okubo, T., Randell, S. et al. In Vitro Cell.Dev.Biol.-Animal (2009) 45: 44. doi:10.1007/s11626-008-9149-2


The tongue represents a very accessible source of tissue-specific epithelial stem cells of endodermal origin. However, little is known about the properties of these cells and the mechanisms regulating their proliferation and differentiation. Foxa2, an endodermal marker, is expressed throughout the tongue epithelium during embryonic development but becomes confined to a minority of basal cells and some taste bud sensory cells in the adult tongue. Using a previously described line of transgenic mice in which enhanced green fluorescent protein (eGFP) is expressed under the control of a human keratin 5 promoter region (Krt5-eGFP), we have isolated a subpopulation of cells in the basal epithelial layer of the mouse tongue with a high efficiency of generating holoclones of undifferentiated cells in culture with a feeder layer. Krt5-GFPhi cells can both self renew and give rise to differentiated stratified keratinized epithelial cells when cultured on an air–liquid interface.


TongueEndodermal basal stem cellKrt5-eGFPColony forming efficiencyDifferentiation


The mammalian tongue offers a versatile and accessible model system for studying both organogenesis and epithelial stem cell biology (Iwasaki 2002). The epithelial layer of the tongue is of endodermal origin and gives rise to at least three different specialized cell types—keratinocytes, taste receptor sensory cells, and secretory cells. The keratinocytes make up the thousands of posteriorly pointing filiform papillae that cover the dorsal surface of the tongue, as well as the support cells of the less numerous fungiform papillae (Hume and Potten 1976). Each filliform papilla consists of an asymmetric cone of epithelial cells capping a peg of mesenchymal cells and blood vessels. It is made up of morphologically distinct subcompartments. The differentiated keratinocytes within the anterior aspect of the cone contain dense keratohyalin granules and express ‘soft’ orthokeratins, while those in the posterior aspect lack keratohyalin and express “hard” orthokeratins (Hume and Potten 1976; Knapp et al. 1995; Jonker et al. 2004). The taste receptor cells are clustered into taste buds (TBs) in the center of each fungiform papilla, while the serous and mucous secretory cells are the major components of the minor lingual salivary glands localized within the mesoderm of the posterior tongue. It is thought that both the filiform papillae and the TBs are maintained by stem cells and their progeny (Hume and Potten 1976; Stone et al. 2002; Hamamichi et al. 2006; Miura et al. 2006). Given that the tongue is of endodermal origin, studies of these cells may provide unique insights into epithelial stem/progenitor cells other than those of the more intensively studied ectodermally derived epidermis.

During embryonic development, the tongue arises from the floor of the pharyngeal cavity, more specifically from branchial arches I–IV. Initially, the organ consists of a simple outer layer of epithelial cells attached to a basal lamina overlying the mesoderm and precursors of the voluntary muscle, stroma and blood vessels. The development of the filiform and fungiforma papillae, and the differentiation of specialized cell types within them, involves dynamic intercellular interactions, mediated by conserved signaling pathways, including those involving Bmp, Shh, Notch, and Wnt ligands (Jung et al. 1999; Miura et al. 2001; Mistretta et al. 2003; Seta et al. 2003; Liu et al. 2004; Miura et al. 2004; Zhou et al. 2006; Iwatsuki et al. 2007; Liu et al. 2007). Studies on the early morphogenesis of epidermal appendages such as hair follicles have shown reciprocal interactions between the embryonic epithelium and the mesenchyme (Fuchs 2007). To date, most of the signaling pathway genes examined in the tongue are expressed in spatial domains of the epithelium. However, Fgf10 is expressed exclusively in the embryonic mesenchyme (Alappat et al. 2005). Among transcription factors, the transcription factor Pax9, expressed in the epithelium, is required for normal development of asymmetric filiform papillae (Jonker et al. 2004). In addition, recent work has shown that Sox2, also expressed in the epithelium, plays a critical dose-dependent role in the decision of the embryonic tongue endoderm to give rise to keratinocytes versus taste bud cells (Okubo et al. (2006)).

Once they have been formed, both the filiform papillae and TBs undergo turnover and regeneration (Beidler and Smallman 1965; Potten et al. 1977; Potten et al. 2002a, b). Relatively little is known about the stem/progenitor cells of the TBs. One model identifies them as undifferentiated basal cells in the bud (Hamamichi et al. 2006), but it is not known if this is a self-renewing population or whether it is continuously derived from supporting cells in the surrounding papilla (Beidler and Smallman 1965). In support of this latter idea, recent work suggests that TB neurosensory cells are derived from Krt14+ve cells in the surrounding epithelium (Asano-Miyoshi et al. 2008); Tadashi Okubo, unpublished results). In the case of the filiform papillae, pioneering work, largely by Potten and colleagues, has illuminated the kinetics of epithelial cell turnover and differentiation in the mouse tongue (Hume and Potten 1976; Potten et al. 1978; Potten et al. 2002a, b, c, d). These studies revealed that the proliferative population is restricted to the basal layer, and in the mouse tongue has a marked circadian rhythm, with the peak of DNA synthesis approximately in the middle of the dark cycle. However, a similar marked circadian rhythm does not occur in humans. Pulse chase studies with H3 thymidine have led to a model in which each filiform papilla is made up of four distinct “epithelial proliferative units” (EPUs). According to this model, each unit is supported by a stem cell (SC) that divides asymmetrically to give a SC and a transit amplifying (TA) cell. The TA cells proliferate, move up the papilla along the basal lamina, and generate suprabasal cells that differentiate as a column. Kinetic analysis suggests that TA cells spend about 2–4 d in the basal layer and take 2–5 d to move through the stratified layer (Potten et al. (2002c)). Significantly, the putative SCs of the filiform papillae are not quiescent but appear to divide with a cell cycle time of about 24 h.

One reason for our interest in progenitor cells of the tongue is that they represent an easily accessible population of endodermal cells for basic studies as well as for potential therapeutic manipulations. For example, primary cultures of epithelium from the human oral cavity have been used for autologous corneal grafting (Nishida et al. 2004). However, as outlined above, surprisingly little work has been done on the characterization of tongue basal cells either in vivo or in vitro. To begin to address these questions, we have exploited a transgenic mouse line expressing enhanced green fluorescent protein (eGFP) under the control of a 6-kb human Keratin 5 regulatory region (Schoch et al. (2004)). Previous studies have shown that this transgene is expressed in basal epithelial cells throughout the body including a subset of undifferentiated basal cells in the mouse trachea. Krt5-eGFPhi cells of the adult trachea sorted by flow cytometry have a higher efficiency of forming large colonies in vitro than Krt5-eGFPlo cells, suggesting that they represent a subpopulation of stem or committed progenitor cells with high progenitorial capacity. We show here that Krt5-eGFPhi basal cells isolated from the adult mouse tongue have the potential, as single cells, to give rise to colonies of self-renewing, undifferentiated Trp63+ve epithelial cells. They can also differentiate in vitro into stratified layers of keratinized epithelial cells overlying Trp63, keratin 14, and Foxa2 positive basal cells.

Materials and Methods


All mice were maintained under a 12-h light/dark cycle and handled under Institutional Animal Care and Use Committee-approved protocols. Control mice were outbred ICR. The Krt5-eGFP transgenic mouse line, in which a 6-kb fragment of the human basal cell-specific KRT5 promoter drives the expression of EGFP, has been described (Bruen et al. 2004; Schoch et al. 2004). This line was originally derived on the (C57Bl/6xC3H) hybrid background and is maintained by interbreeding.

Tissue immunostaining.

Adult mouse tongues and tongue epithelial cell cultures were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) overnight at 4°C or 10 min at room temperature, respectively, and processed for paraffin or O.C.T embedding and sectioning at either 7 μm (paraffin sections) or 12 μm (cryosections). Antigen retrieval was performed by boiling 10 min in 10 mM sodium citrate (pH 6), followed by cooling to room temperature and washing with PBS. The following primary antibodies were used in combination, where appropriate, with VECTOR M.O.M blocking reagent (Vector Laboratories, Burlingame, CA BMK-2202): Keratin-14 (mouse monoclonal antibody, Clone LL002, Neomarker, Fremont, CA), Trp63 (mouse monoclonal antibody, 4A4, Santa Cruz Biotechnology, Inc., Santa Cruz, CA), involucrin (rabbit antibody, kindly provided by Dr. Fiona Watt, Cambridge, UK), Foxa2 (rabbit antibody, Sasaki and Hogan 1994), GFP (rabbit polyclonal antibody, ab290, Abcam, Cambridge, MA), and anti-cytokeratin19 (rat antibody, Developmental Studies Hybridoma Bank, Iowa City, IA). Secondary antibodies were biotin-conjugated anti-mouse (Vector Laboratories) and anti-rabbit(Jackson ImmunoResearch Laboratory, Inc., West Grove PA), Alexa 488-conjugated anti-rabbit (Molecular Probes, Eugene, OR), and Cy3- conjugated anti-rabbit (Jackson ImmunoResearch Laboratory, Inc.). VECTASTAIN ABC kits (Vector Laboratories, PK-4000) were used for immunostaining of paraffin sections.

Cell isolation and flow cytometry.

Tongues dissected from 4- to 6-wk-old male and female Krt5-eGFP mice were washed with Dulbecco’s modified Eagle medium (DMEM; GIBCO) containing 3% fetal bovine serum (FBS). The anterior region, excluding the circumvallate and tissue posterior to it, was cut into eight to ten pieces and incubated in Dispase (BD Biosciences, Bedford, MA 40 U/ml) in DMEM at room temperature for 1 h. The dorsal epithelial layer was peeled off with forceps and incubated at 37°C in 15 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer (pH 7.4), 0.15 M NaCl, 0.05% trypsin, 0.5 mM EGTA, 0.4 µM ethylenediaminetetraacetic acid (EDTA), and 1% polyvinylpyrrolidone for 30 min. After gentle pipeting, any undissociated stratified squamous cells were removed by filtration through a 40-µM cell strainer (BD Biosciences) and the single cells collected by centrifugation.

Dissociated single epithelial cells from two tongues (approximately 9 × 105cells) were resuspended in 1 ml PBS containing 1% bovine serum albumin and labeled on ice for 30 min with biotin-hamster anti-rat CD29 (2.5-5.0 μg/ml) and streptavidin–peridinin chlorophyll-α protein-conjugated secondary antibody (BD Biosciences) at a dilution of 1:200 (0.1 µg/ml). Propidium iodide (Sigma, St Louis, MO), 100 ng/ml solution, was added before sorting cells on a Vantage SE flow cytometer (BD Biosciences). Cells gated as ß1-integrinhi were then sorted based on eGFP expression. Data were analyzed with DiVa software (BD Biosciences). Single cell sorting cells were seeded on 96-well plate coated with irradiated 3T3 J2 feeders at a density of 2.5 × 104 cells/cm2.

Tongue epithelial cell culture.

For generating feeder layers, we used 3T3 J2 cells kindly provided by Dr. James Rheinwald (Harvard Medical School) and cultured in DMEM (pyruvate-free, Cellgro, Herndon, VA) supplemented with 10% calf serum (Hyclone) 2 mM glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin until near confluence. Cells were harvested using 0.05% trypsin, 0.02% EDTA (GIBCO, Carlsbad, CA), irradiated with 3000R Celsium-137 γ-radiation and plated at 2.5 × 104 cells/cm2 on tissue culture dishes precoated with 0.1% gelatin (Chemicon, Temecula, CA). Adult tongue epithelial cells were seeded on the feeder layer at a density of 3–4 × 103 cells/cm2 in DMEM/Ham’s F-12 (Gibco) 1:1, 10 µg/ml insulin, 5 µg/ml transferring, and 6.7 ng/ml selenium (GIBCO), 25 ng/ml EGF (BD Biosciences), 30 µg/ml bovine pituitary extract (Invitrogen), 5% FBS (GIBCO), 4 mM l-glutamine (GIBCO), 100 U/ml penicillin, and 100 µg/ml streptomycin (tongue basal cell culture medium). For colony-forming assays, cells were seeded at a density of 15–20 × 103 per 35 mm plate on feeder cells under the same conditions.

Induction of cell differentiation.

Sorted adult Krt-eGFP tongue epithelial cells were cultured in 12-well cell culture inserts (8 µm pore size, BD Biosciences) on a layer of rat tail type I collagen (BD Biosciences) containing 3T3 J2 cells. Formation of the type I collagen-3T3 J2 cell matrix was modified from a previous report (Costea et al. 2003). Briefly, seven volumes rat tail type I collagen (3.13 mg/ml), two volumes of reconstitution buffer pH 8.15 (2.2 g NaHCO3, 0.6 g NaOH, 4.766 g HEPES in 100 ml dH2O), and one volume of Dulbecco’s modified Eagle’s medium (10×; GIBCO) were mixed together on ice. Irradiated 3T3 J2 cells were suspended in one volume of FBS and added to the collagen mix at a final concentration of 0.5 × 106/ml. The collagen-3T3 J2 cell matrix (150–200 µl) was loaded into each insert and incubated 1 h at 37°C for polymerization. Both top and bottom chambers were filled with 3T3 J2 culture medium and incubated at 37°C for 24 h before use. The medium was then changed to tongue basal cell culture medium and approximately 5–8 × 103 sorted cells seeded into each well. After 4–5 d of culture, air–liquid interface (ALI) culture conditions were imposed by removing the culture medium from the top chamber and filling the bottom chamber with a serum-free medium composed of DMEM/Ham’s F-12 (3:1), supplemented with 2 mM glutamine, 5 µg/ml insulin, 10 µg/ml transferrin (GIBCO), 50 µg/ml l-ascorbic acid (Sigma), 0.4 mg/ml hydrocortisone (Sigma), 1 mg/ml linoleic acid albumin (Sigma), 10 ng/ml EGF (BD Biosciences), 10 ng/ml FGF7 (R&D System, Minneapolis, MN), 100 U/ml penicillin and 100 µg/ml streptomycin. The medium was changed every 2 d, and the culture continued for 7 d. To examine the effects of FGF10 on epithelial cell culture, 50 ng/ml FGF10 (R&D system) was added into the medium throughout the culture period.


Expression of the Krt5-eGFP transgene in the epithelium of the embryonic and adult mouse tongue.

The stereotypic morphogenesis of the mouse tongue epithelium proceeds from embryonic day 12 (E12) to 17 (Hall et al. 1999; Kim et al. 2003). Using immunohistochemistry of frozen sections, Krt5-eGFP transgene expression is detected from E12.5 (data not shown) and by E16.5 is uniform within the basal layer of the epithelium (Fig. 1A–C). Expression continues in the basal cells of the adult tongue (Fig. 1D–K). In the fungiform papillae, Krt-eGFP is high in most of the basal cells and in the cells immediately surrounding the taste buds (Fig. 1D–H). In the asymmetric, posteriorly pointing, conical filiform papillae, expression is more variable. Highest Krt5-eGFP expression is seen predominantly, but not exclusively, in the basal cells of the anterior wall (Fig. 1I–K). According to the model of Hume and Potten (Hume and Potten 1976), these are stem and transit amplifying cells for the column of anterior squamous keratinocytes that express ‘soft’ orthokeratin. Other studies have shown that this subset of basal cells expresses higher levels of TopGal, a reporter for canonical Wnt signaling, and Lef1, a downstream component of Wnt signaling, than posterior cells (Okubo et al. 2006; Rhee et al. 2006). By contrast to the rather variable levels of expression of the Krt5-eGFP transgene in the filiform papillae, other basal cell-specific proteins such as Keratin14 and ß4-integrin appear more uniformly expressed (Fig. 1I–M). The transcription factor Trp63 was also uniformly expressed in the basal layer, as well as in a few suprabasal cells (data not shown).
Figure 1.

Keratin-5 eGFP transgene expression in the embryonic and adult tongue. (A–C) E16.5 Krt5-eGFP tongue. (A) DIC image of cryosection. (B) GFP immunostaining (green) merged with DIC image; insert shows that Krt5-eGFP is expressed throughout the epithelial layer at this stage. Nuclei labeled with DAPI (blue). (C) Whole mount fluorescence microscopy image of live Krt5-eGFP tongue. (D–M) Cryosections of adult Krt5-eGFP tongue with posterior to the right. (D) DIC image. E GFP immunostaining merged with DIC image showing Krt5-eGFP-positive cells are confined to the basal layer. (F, G) Section through a single fungiform papilla. (F) DIC image. (G) GFP immunostaining merged with DIC image showing strong Krt5-EGFP transgene expression in basal cells and in cells immediately surrounding the taste bud. (H) whole mount image of dorsal surface of live tongue showing strong expression in fungiform papillae. (I) In the filiform papillae, GFP expression is strongest in the anterior basal cells (asterisk shows anterior). (J) By contrast, K14 immunostaining shows more uniform expression. (K) merge of GFP and K14 immunostaining. (L) DIC image. (M) CD104 (ß4-integrin) immunostaining merged with DIC. Scale bars 50 µm (A–C; I–M); 20 µm inset in (B); 100 µm (D–E); 20 µm (F–G); 200  µm (H).

Foxa2 is expressed in a subset of adult tongue epithelial cells.

Foxa2 is a member of the Forkhead box (Fox) transcription factor family. It was previously reported that Foxa2 is expressed in the dorsal epithelium of the developing mouse tongue and in adult serous gland cells but was absent from the remainder of the adult tongue (Besnard et al. 2004). We confirmed these findings for embryonic stages (Fig. 2A and data not shown for serous glands). However, in the adult, we did find strong Foxa2 staining in a subset of basal cells of the filiform and fungiform papillae (Fig. 2B–E). In addition, Foxa2-positive cells were located within the TB, with variable levels of expression. Some cells in this location were negative for Krt-19, a marker of sensory cells (Knapp et al. 1995), while others co-expressed Foxa2 in the nucleus and Krt-19 in the cytoplasm (Fig. 2F–I). Given the more restricted expression of Foxa2 in the adult tongue, we asked whether expression levels change during the marked circadian cycle of proliferation seen in the mouse tongue. We confirmed that this rhythm was occurring by following BrdU incorporation during 2-h periods for 24 h, observing a peak of proliferation around 4 a.m., as previously reported by others (Potten et al. 2002a; data not shown). However, there was no fluctuation in Foxa2 expression in the epithelium during the same 24-h period.
Figure 2.

Foxa2 expression in embryonic and adult tongue. (A) anti-Foxa2 immunostaining of E14.5 Krt5-eGFP tongue. Inset shows that most cells are positive at this stage. (B–I) Expression in adult tongue. (B) DIC image of filiform papillae. (C) Anti-Foxa2 staining shows a few positive cells in the base of the papillae (arrowheads). (D) DIC image of fungiform papilla. (E) Anti-Foxa2 staining labels cells in the basal layer and in the TB (arrowheads). (F) DIC image of TB and surrounding cells; (G) Anti-Foxa2 immunostaining; (H) anti-Krt-19 immunostaining of TB sensory cells; (I) merge of (F, G) and (H) showing that some Foxa2 positive cells do not express Krt-19 (arrowhead), while others do. Scale bars: 100 µm (A); 20 µm for (A) inset, and (B–C) 50 µm for D–I.

In vitro colony forming potential of adult tongue epithelial cells.

To examine the in vitro proliferative and differentiation potential of adult mouse tongue epithelial cells, single cells were plated on a feeder layer of irradiated 3T3 J2 cells and cultured for 12 d in medium containing serum, EGF, bovine pituitary extract and other supplements (see Materials and Methods). After about 5 d, three different types of colonies were identified, based on classical studies with skin epidermal cells (Barrandon and Green 1987); holoclones, consisting of small, densely packed undifferentiated cells (Fig. 3A), paraclones, composed of large, flattened cells, and meroclones with an outer rim of small cells surrounding a central area of larger, flatter cells (Fig. 3B–D). Over time, holoclones consistently gave rise to larger colonies than meroclones or paraclones. When cells were cultured from Krt5-eGFP tongues neither the paraclones nor the central cells of the meroclones expressed high levels of eGFP. By contrast, all the cells of most of the holoclones and the cells in the outer rims of the meroclones, were Krt5-eGFPhi (Fig. 3C). However, in some colonies scored as holoclones by phase contrast microscopy, the central cells nevertheless expressed lower eGFP than the outer cells, even though they were morphologically small and undifferentiated (data not shown).
Figure 3.

Adult mouse tongue epithelial cells cultured on irradiated 3T3 J2 feeder cells. (A) Phase contrast image of typical holoclone showing closely packed, small cells that pile up around the edge of the colony. (B) Typical meroclone with small cells around the edge of the colony and larger, flatter cells in the center. (C) Holoclone derived from single Krt5-eGFPhi cell showing transgene expression in most outer cells but lower levels in some central cells. (D) Phase contrast image of the same colony as in C. Scale bars 100 µm for (A), 50 µm for (B), 200 µm for (C–D).

We next compared the colony-forming efficiency of subsets of tongue basal cells sorted using flow cytometry based on their expression of ß1-integrin and different levels of Krt5-eGFP. As in the epidermis, ß1-integrin is preferentially expressed in the basal cells of the oral epithelium (Nishida et al. 2004). ß1-Integrinhi (CD29hi) cells were sorted into four classes (P3-P6) based on level of Krt5-eGFP expression (Fig. 4) and seeded onto feeder layers. The cells with the highest Krt5-eGFP expression (fraction P3) had a higher colony-forming efficiency than cells with the lowest Krt5-eGFP expression (0.98% compared with 0.40%; average of four experiments). Significantly, after 6–8 d culture, about 80% of the colonies generated from P3 cells were scored as holoclones compared with 20% or less for fractions P4-P6 (Fig. 4E). Very similar results were obtained when cells were sorted based only on levels of Krt5-eGFP expression without also sorting for ß1-integrin expression (data not shown).
Figure 4.

Flow cytometry and colony-forming efficiency of Krt5-eGFP cells. Epithelial cells from Krt5-eGFP +ve mice and wild-type littermates were dissociated into single cells for flow cytometry followed by in vitro culture on feeder layers. (A, B) FACS based on level of beta1-integrin labeling (PerCP) and exclusion of propidium iodide. (A) Cells from Krt5-eGFP tongue with no antibody. (B) Cells from Krt5-eGFP transgenic tongue. Beta1-integrin-positive cells were gated (P2) for subsequent Krt5-eGFP expression analysis. (C, D) Cells were gated into four fractions based on the Krt5-eGFP expression levels from highest (P3) to lowest (P6). (C) Control wild-type tongue cells. (D) Krt5-eGFP tongue cells. (E) Percent of total colonies from the different fractions that are either holoclones or meroclones and paraclones (average of four experiments). Overall colony-forming efficiency (both holoclones and mero/paraclones) P3 = 0.98%; P4 = 0.52%; P5 = 0.48% and P6 = 0.40%.

To test the colony-forming efficiency of single cells, Krt5-EGFPhi cells from fraction P3 (Fig. 4D) were sorted as single cells into 96-well plates pre-coated with 3T3J2 feeder cells. Under these conditions, the colony-forming efficiency was about 0.78% (three colonies out of 384 cells, in two separate experiments). Holoclones were picked, expanded, and resorted into single Krt5-eGFPhi cells that again gave rise to holoclones with the same efficiency. This process of sorting was carried out twice with a total of four passages of the cells. These results demonstrate that K5-eGFPhi cells from the adult mouse tongue have single cell colony-forming ability and undergo self-renewal in culture.

K5-EGFPhi cells can differentiate into stratified keratinocytes.

To test their differentiation potential in vitro, Krt5-eGFPhi basal cells were sorted and fraction P3 plated on a layer of collagen containing irradiated 3T3J2 feeder cells. The cells were cultured for 4 to 5 d in basal cell culture medium and switched to serum-free differentiation medium and air–liquid interface (ALI) conditions for another 10 to 12 d. Figure 5A–D shows that, under these conditions, the Krt-5-eGFPhi cells give rise to a multilayered squamous keratinized epithelium. The basal cells express high levels of keratin14 (Fig. 5B) and Trp63 (Fig. 5C), and thus resemble basal cells of the normal adult mouse tongue. Suprabasal cells express loricrin, a major component of the cornified cell envelope and strongly expressed in the suprabasal cells of the filiform and fungiform papillae (Steinert and Marekov 1995; Kalinin et al. 2001; Fig. 5D). However, no structures resembling filiform or fungiform papillae with TBs were ever seen.
Figure 5.

Differentiation of tongue basal epithelial cells. Sorted K5-EGFPhi cells (fraction P3) were on a collagen-3T3 J2 matrix. After 4 d, samples were switched to differentiation medium and ALI conditions and cultured for another 7 d before analysis. (A) Staining with hematoxylin and eosin; (B) Expression of Krt14 predominantly in the basal cells; (C) All the basal cells are positive for Trp63; (D) Loricin, a marker of differentiated keratinocytes, is expressed in the suprabasal cells; (E, F) In the presence of FGF10 in the culture medium, the epithelium is more folded. (E) shows that some of the basal cells are positive for Foxa2. Scale bar: 50 µm.

Culture of tongue epithelial cells with FGF10.

During embryonic development, part of the posterior region of the dorsal tongue epithelium invaginates into the underlying mesenchyme and gives rise to the folds of the circumvallate papilla and to the serous and mucus-producing lingual glands. Studies on the development of other endodermal glandular structures such as the submucosal glands of the mouse trachea have shown that invagination of the epithelium is promoted by FGF10 expression in the underlying mesenchyme (Rawlins and Hogan 2005). We therefore asked whether adding FGF10 to the culture medium of tongue epithelial cells would induce the formation of gland-like structures. As shown in Fig. 5E,F, cells cultured with FGF10 had fewer layers of keratinized squamous cells than controls, and the basal layer was folded rather than flat. However, no gland-like structures were seen after 7 d culture. When isolated cells were cultured under differentiation conditions with FGF10, some cells in the basal layer were positive for Foxa2 expression (Fig. 5E).


Here, we report for the first time a system to grow and expand adult mouse tongue epithelial cells in culture, using irradiated fibroblasts as a feeder layer. In addition, we exploit a previously described Krt5-GFP transgenic mouse line to isolate subpopulations of basal cells from the adult tongue and test their proliferative and differentiation capacity in vitro. The Krt5-GFP line has been a useful tool for isolating basal cells from the mouse trachea. This has led to the finding that Krt5-eGFPhi tracheal cells are more able than non-basal cells to give rise to large colonies of cells that can differentiate into secretory and ciliated epithelium under air–liquid culture conditions (Schoch et al. 2004).

Analysis of Krt5-eGFP expression in the adult tongue showed that the transgene is efficiently expressed in the basal cells of the fungiform papillae and in the cells immediately surrounding the TBs. The latter are likely to include the Krt14+ve stem cells that are thought to continuously give rise to sensory cells within the TBs (Asano-Miyoshi et al. 2008; Tadashi Okubo, unpublished observations). The K5-eGFP transgene is also expressed in the basal cells of the conical filiform papillae, with highest levels in the anterior walls. Models of cell turnover in the adult mouse tongue suggest that the long-term stem cells of the filiform papillae reside right at the base of these structures and that most of the basal cells in the walls of the papillae are transit-amplifying cells. However, this model has not yet been confirmed by long-term, single cell lineage tracing experiments of the kind that have recently been carried out in the interfollicular epidermis of the skin (Clayton et al. 2007). It remains possible that all of the basal cells in the walls have the capacity to function as stem cells and can both self renew and give rise to cells that move upward along the basal lamina and out into the superbasal layers. According to this second model, at any particular time, the probability that an individual basal cell will undergo a division that gives rise to another basal cell (self-renewal), or to a basal cells and a TA cell, is variable. However, over the long term, all basal cells can self-renew and behave as stem cells. Distinguishing between these models will require rigorous long-term cell lineage studies in vivo. Meanwhile, our finding that Krt5-eGFPhi cells have a higher efficiency of giving rise to holoclones in culture than Krt-eGFPlo cells is open to at least two interpretations. One possibility is that high Krt5-eGFP expression is an invariable and intrinsic characteristic of a subset of basal cells that are true stem cells. Alternatively, high transgene expression may merely reflect the fact that basal cells in some locations have a higher metabolic activity than others in response to local signals. For example, due to mechanical forces, the anterior region of the filiform papillae may have a higher rate of cell turnover or stress than the posterior. Feedback mechanisms, either through the epithelial layers or the underlying mesenchyme and/or blood vessels may then send stronger proliferative/survival signals to the adjacent basal cells. Further studies are clearly needed to distinguish between these and other possible models.

During embryonic development, the dorsal epithelium of the tongue gives rise to several different cell types—keratinocytes, sensory cells of the taste buds, and secretory cells of the minor salivary glands (lingual glands). The embryonic epithelial cells are therefore multipotent. Cell transplantation and mesenchymal–epithelial recombination experiments in vivo have not yet been done to ask whether basal epithelial cells of the adult tongue also have the ability, in the right environment, to give rise to more than one cell type. In ALI culture, adult tongue basal cells sorted on the basis of Krt5-eGFPhi expression only gave rise to stratified squamous keratinized epithelium. We were unable to induce the formation of secretory glands either by adding FGF10, nor did we observe the formation of taste buds, although this is not unexpected given that regeneration of TBs in vivo, presumably from adjacent K14+ve epithelial cells, requires TB innervation (Cheal and Oakley 1977). To test the full developmental potential of embryonic tongue epithelial cells, it may be necessary to combine the sorted cells with embryonic mesenchyme from different regions of the pharyngeal cavity and to culture them in vitro or as grafts in immunocompromised mice.

In conclusion, we have shown here that undifferentiated basal cells of the tongue, a very accessible source of endodermal epithelial cells, can be cultured in vitro. They undergo self-renewal and can be induced to differentiate into squamous keratinized epithelium. Further studies may lead to strategies for converting these basal cells into other endodermal lineages. For example, it would be attractive to be able to convert the progeny of tongue basal cells into secretory cells for tissue replacement therapy of salivary glands, which do not normally regenerate in humans after irradiation.


This work was supported by a grant from Becton, Dickinson and Company (BD) to BLMH. We thank members of the Hogan lab and BD’s corporate research unit (BD Technologies, Research Triangle Park, NC) for many helpful suggestions and for critical reading of the manuscript.

Copyright information

© The Society for In Vitro Biology 2008