Conditional expression of the dominant-negative TGF-β receptor type II elicits lingual epithelial hyperplasia in transgenic mice

Authors


Correspondence to: Feng Li, School of Medicine, Shanghai Jiao Tong University, 280 South Chongqing Road, Shanghai 200025, People's Republic of China. E-mail: lifengwest@gmail.com or

lifengwest@yahoo.com

Abstract

Background: The transforming growth factor-β (TGF-β) signaling pathway is generally believed to be a potent inhibitor of proliferation. However, many epithelia lacking the essential Tgfbr2 gene still maintain normal tissue homeostasis. Here, transgenic mice expressing rtTA from the human keratin 14 (K14) promoter were used to generate an inducible dominant-negative TGF-β receptor type II (Tgfbr2) mutant model, which allowed us to distinguish between the primary and secondary effects of TGF-β signaling disruption by Doxycycline treatment in K14+ epithelial stem cells. Results: We showed that in mice lacking TGF-β signaling in K14+ cells, invasive carcinomas developed on the ventral surface of the tip of the tongue, while filiform papillae on the dorsal surface showed different pathological changes from the tip to the posterior of the tongue. In addition, acetylation levels of histone H4 and histone H3 rapidly increased, while pMAPK activity was enhanced and Jagged2 inactivated in lingual epithelia after disruption of TGF-β signaling. Conclusions: Our results contribute to the understanding of TGF-β signaling in regulating homeostasis and carcinogenesis in lingual epithelia. Developmental Dynamics 242:444–455, 2013. © 2013 Wiley Periodicals, Inc.

INTRODUCTION

The mouse tongue is known to contain four types of papillae, filiform, fungiform, circumvallate (CV), and foliate, with the filiform papillae being the most abundant type (Cameron, 1966; Delay et al., 1986; Kinnamon et al., 1985), but the mechanism of papillae formation remains unclear. Recent studies have implicated multiple signaling pathways in the regulation of epithelial stem cell proliferation, such as the Wnt (Blanpain et al., 2007; Lowry et al., 2005), Notch (Fre et al., 2005), and Bone Morphogenetic Protein (BMP) pathways (Andl et al., 2004). Moreover, it has also been reported that components of the TGF-β (Nakamura et al., 2009), Notch (Seta, 2005), and β-Catenin/Wnt (Liu et al., 2007; Schneider et al., 2010) signaling pathways are expressed in lingual epithelia and/or taste buds (TB). Several investigations have provided evidence for an important role of TGF-β in this orchestration of signals (Bierie and Moses, 2006; Blokzijl et al., 2003; Falk et al., 2008; Shi and Massagué, 2003; Siegel and Massagué, 2003).

TGF-β, which signals through activation of TGF-β receptors type I and II (Tgfβ r1 and Tgfβ r2) and phosphorylation of the signaling mediators Smad2 and Smad3, elicits multiple cellular responses, including a cytostatic effect on various cell types (Bierie and Moses, 2006; Siegel and Massagué, 2003). In cultured epithelial cells, TGF-β molecules act as potent inhibitors of proliferation and show tumor suppressor characteristics (Cameron et al., 1989; Cameron et al., 1997). Given the ascribed effects of TGF-β signaling on growth inhibition, many normal epithelia are still observed when conditional targeting of the Tgfbr2 gene is performed (Bhowmick et al., 2004; Cheng et al., 2005; Forrester et al., 2005; Guasch et al., 2007). However, it is surprising that ablation of Tgfbr2 in stromal fibroblasts generates spontaneous invasive cell carcinomas (SQCC) of the forestomach epithelium (Nolasco et al., 1987). Other strategies used to analyze the function of TGF-β signaling are dominant-negative mutants and an overexpression transgenic mouse model. However, results have often been conflicting, and the interplay between extrinsic and intrinsic factors may contribute to some of these disparate results (Amendt et al., 1998, 2002; Cui et al., 1995, 1996; Frugier et al., 2005; Wang et al., 1997). Recent studies have further suggested several functions of TGF-β signaling in vivo through conditional targeting of the Tgfbr2 gene, such as shortening of the cell cycle without changing cell fate (Falk et al., 2008).

Given the reported effects of TGF-β signaling on growth inhibition and cell cycle, it is very surprising that K14-Cre/Tgfbr2 conditional knockout mice are viable and appear phenotypically normal through early adulthood. Interestingly, it is rectal and genital epithelia, not skin epithelia, that develop SQCC with age in these mice (Guasch et al., 2007). Thus far, no reports have shown the effect of TGF-β signaling on the proliferation of lingual epithelia and papillae in the adult tongue. In the current study, we used a keratin 14 (K14) promoter to conditionally disrupt TGF-β signaling. After crossing with a transgenic mouse line carrying an EGFP-pBi-DeltaTgfbr2 construct (PTR) (Frugier et al., 2005), the progeny express rtTA under the control of a cell-type-specific promoter, which leads to cell-type-specific expression of a dominant-negative TGF-β receptor. In a previous study using this model, we showed that K14 is differentially expressed from the tip to the posterior along the surface of the tongue, and a cell migration stream was revealed in lingual epithelia and filiform papillae (Li and Zhou, 2012). Here, we further showed that mice lacking TGF-β signaling in K14-positive cells developed invasive carcinoma in the ventral surface of the tongue tip. By contrast, Tgfbr2 mutant lingual epithelia on the dorsal surface of the tongue were phenotypically normal, although filiform papillae showed different pathological changes from the tip to the posterior of the tongue. We also investigated changes in epigenetic modifications and signaling pathways when TGF-β signaling was disrupted in an effort to understand the intrinsic and extrinsic factors affecting the development/differentiation of lingual stem cell/progenitors.

RESULTS

Disruption of TGF-β Signaling Promotes Invasive Carcinoma on Tip of Tongue After 35 Days of TGF-β Signaling Disruption

Multiple stem cell populations have been described in a previous cell genesis model of the tongue, including K14+ and Sox2+ progenitor cells (Okubo et al., 2006, 2009). K14+ cells continuously give rise to both mature TB receptor cells and surrounding keratinocytes (Okubo et al., 2009). Many studies have reported that the normal epidermal phenotype is still present after targeted and conditional expression of the Tgfbr2 gene in K14- or K5-positive cells, and oncogenic transformation with Ha-Ras is required to promote invasive SQCC (Guasch et al., 2007; Lu et al., 2006). In order to further investigate the nature of K14+ progenitor cells in lingual epithelia, we crossed K14-rtTA (X-linked K14-rtTA transgene) mice with PTR mice (Xie et al., 1999). The double-transgenic K14-rtTA-PTR mice specifically express K14 in the lingual epithelia and also co-express EGFP and the dominant-negative ΔTgfbr2 genes upon treatment with Doxycycline (Dox) (see Supp. Fig. S1, which is available online) (Frugier et al., 2005).

Adult K14-rtTA-PTR mice were exposed to Dox and sacrificed at 5 hr, 9 hr, 1 day, 3 days, 7 days, 35 days, and 75 days after induction (Supp. Fig. S1). As the rtTA protein is retained in daughter cells originating from K14+ progenitors due to a shortened cell cycle after disruption of TGF-β signaling (Falk et al., 2008; Li and Zhou, 2012), Dox induction should continually induce GFP expression in those daughter cells (Supp. Fig. S1). With extended exposure to Dox, mutated cells should gradually be found in the lingual epithelia and papillae. By observing the morphology of lingual epithelia and papillae, we were able to investigate the role of TGF-β signaling on maintaining the homeostasis of lingual epithelia and filiform papillae.

First, the expression levels of keratin 14 (K14) and Tgfbr2 were observed by immunostaining. Immunostaining with anti-K14 showed that K14 was expressed in lingual epithelia covering the fungiform papillae, TB, and connective tissue of filiform papillae and fungiform papillae. It should be noted that K14 expression was detected along the basal membrane (Supp. Fig. S2A and B). As expected, Tgfbr2 expression was also detected in lingual epithelia, papillae, and TB. However, expression of the Tgfbr2 gene was obviously different from that of K14, which was confined to suprabasal cells (Supp. Fig. S2C and D).

Our previous study revealed that GFP expression (representing cells with disrupted TGF-β signaling) was obviously distributed throughout the dorsal surface and in both papillae- and non-papillae-containing areas of the ventral surface of the tongue. Further analysis showed that GFP expression was variable, depending on the location of the tongue, papillae, and the time of Dox induction (Li and Zhou, 2012). Meanwhile, we observed GFP expression in the back skin of the mice, as expected (Supp. Fig. S3A). Another study has shown that constitutive expression of activated ErbB2 driven by the K14 promoter induces hair follicle abnormalities and severe skin hyperplasia in transgenic mice (Xie et al., 1999). The mice in this study had bleeding skin from day 30 (Supp. Fig. S3B), and their condition had deteriorated to such a degree with necrotizing ears and bleeding skin that ethical concerns mandated sacrificing them by day 75.

In order to further investigate the effect of TGF-β signaling on the differentiation/development of lingual epithelial cells, we focused on their morphological changes and that of papillae throughout the tongue in the current study. The tongue was separated into three parts: (1) tip of tongue, ∼1.0 mm; (2) middle of tongue and posterior of tongue including the “V” region; (3) posterior of tongue near CV papillae. Histological sections were made in three different planes to separate the three parts: (1) frontal section for the tip of the tongue; (2) sagittal section along the mid-line for the middle-posterior tongue; (3) horizontal section for the posterior tongue containing CV papillae. The morphological changes of filiform papillae and lingual epithelia were observed at five sites (sites 1–5) (Supp. Fig. S3C) and are described below in turn from anterior to posterior after disruption of TGF-β signaling over time.

The frontal section of the tongue tip (site 1) was first observed. It is well known that abundant papillae cover the surface of the tongue tip. Two main types, fungiform and filiform papillae, could be observed on the dorsal surface (Fig. 1A and S4A), ventral surface (Supp. Fig. S4B), and side surface (Supp. Fig. S4C). After 24 hr of TGF-β signaling disruption, the ordered arrangement of the hook-shaped filiform papillae was lost on the dorsal surface (Supp. Fig. S4D) as well as on the ventral surface (Supp. Fig. S4E). After 3 days (Supp. Fig. S4G–I) or 7 days (Fig. 1C and Supp. Fig. S4J–L) of TGF-β signaling disruption, abnormal filiform papillae continued to be observed. After 35 days of TGF-β signaling disruption, the formation of filiform papillae was significantly blocked on the dorsal surface (Supp. Fig. S4M), while the invasion of lingual epithelia was observed on the ventral surface (Fig. 1D and Supp. Fig. S4N). The current results collectively showed that 24 hr of TGF-β signaling disruption interfered with the formation of filiform papillae, which recovered by day 3, while 35 days of TGF-β signaling disruption severely inhibited the formation of filiform papillae on the dorsal surface and promoted development of invasive carcinoma on the ventral surface of the tongue tip.

Figure 1.

Tongue tip histology. Hematoxylin-stained section from the tongue tip (frontal section) of animals without Dox treatment (A) or with Dox treatment for 24 hr, 7 days, and 35 days (B–D, respectively). B: After 24 hr, abnormal filiform papillae were found on the dorsal surface and ventral surface. C: After day 7, abnormal filiform papillae continued to be seen on the dorsal surface and ventral surface. D: After day 35, abnormal filiform papillae were still found in the dorsal surface, and epithelial cells had invaded into the connective tissue on the ventral surface. Scale bar = (A–C) 200 μm; (D) 500 μm.

Hyper-Proliferation of Lingual Epithelia Occurs From the Middle to Posterior of Tongue After TGF-β Signaling Disruption

According to traditional Chinese medicine, observations of the tongue surface (dorsal surface) can provide strong visual indicators of an individual's overall harmony or disharmony (Liu et al., 2003; Wu et al., 2005). Our previous study revealed that K14 expression varies along the tongue within 3 days of Dox induction (Li and Zhou, 2012). Given the above hypothesis, one can speculate that different pathological changes may be observed along the tongue after disruption of TGF-β signaling over time in the same Dox induction model (Li and Zhou, 2012).

Along the dorsal surface of the tongue, a large number of filiform papillae were observed in control mice (Fig. 2A,B and Supp. Fig. S5A). After 5 hr of TGF-β signaling disruption, hyper-proliferation of lingual epithelia was seen in the “V” region of the posterior tongue (Fig. 2C and D). After 24 hr of TGF-β signaling disruption, the hyper-proliferation of lingual epithelia was still present (Fig. 2E and F). Meanwhile, many keratinized fila accumulated on the surface of the tongue (Supp. Fig. S5B and C). After 3 days of TGF-β signaling disruption, the hyper-proliferation of lingual epithelia ceased, and we also failed to observe the orderly arrangement of filiform papillae in the “V” region of posterior tongue, which normally points posteriorly (Fig. 3G). In the middle tongue, thickened epithelia on the dorsal surface was observed after disruption of TGF-β signaling (Fig. 3B–G), compared with control mice (Fig. 3A). As expected, lingual epithelia were significantly thickened after 35 days of TGF-β signaling (Fig. 3E,G). On the other hand, the striated muscle cells under the lingual epithelia were arranged in a disorderly manner after disruption of TGF-β signaling (Fig. 3B–E), compared with control mice (Fig. 3A). In the posterior tongue containing CV papillae (site 5), hyper-proliferation of lingual epithelia was also observed after 35 days of TGF-β signaling disruption (Fig. 4B).

Figure 2.

Histology of the middle-posterior tongue (low magnification). Hematoxylin-stained section from the middle-posterior tongue (sagittal section) of animals without Dox treatment (A, B) or with Dox treatment for 5 hr (C, D), 24 hr (E, F), and 3 days (G, H), respectively. In the posterior tongue from control mice, many filiform papillae were observed in lingual epithelia (B). After 5 hr (D) or 24 hr (F) of TGF-β signaling disruption, hyper-proliferation of filiform papillae was observed in the posterior tongue. After 3 days of TGF-β signaling disruption, hyper-proliferation of filfiorm papillae disappeared (H). Scale bar = 500 μm.

Figure 3.

Histology of the middle tongue (higher magnification). Hematoxylin stained section from the tongue tip (saggital section) of animals without Dox treatment (A) and with Dox treatment for 5 hr (B), 9 hr (C), 24 hr (D), 3 days (E), and 35 days (F). Note the pathological changes in the connective tissues under the lingual epithelia (A–E, arrows). Arrows indicate the connective tissue and muscle tissue under the dorsal surface of tongue, which show the pathological changes since 5 hr Dox induction. G: The thickness of lingual epithelia in the middle tongue significantly increased after 5 hr of TGF-β signaling disruption (a, b: P < 0.05; c, d: P < 0.05). Scale bar = 200 μm.

Figure 4.

Histology of the posterior tongue. Hematoxylin-stained section from the posterior tongue (horizontal section) containing CV papillae of animals without Dox treatment (A) or with Dox treatment for 35 days (B). Hyper-proliferation of lingual epithelia was observed in the posterior tongue containing CV papillae (B). Arrows indicate thickened lingual epithelia. Scale bar = 500 μm.

In essence, the hyper-proliferation of epithelia was observed on the dorsal surface of the middle-posterior tongue, including the area containing the CV papillae after 35 days of TGF-β signaling disruption. Unexpectedly, the disruption of TGF-β signaling quickly (<24 hr) induced the hyper-proliferation of lingual epithelia in the “V” region of the dorsal surface.

Acetylation Levels of Histone H4 and Histone H3 Rise Rapidly in Lingual Epithelia After Disruption of TGF-β Signaling

Our previous study suggested the presence of a cell migration stream consisting of multiple stem cell pools and differentiated cell pools in the lingual epithelia. The local microenviroment plays an important role in regulating cell differentiation/development in the lingual epithelia (Li et al., 2012; Li and Zhou, 2012). It has also been suggested that it is the position which determines whether a cell will specialize into one of two different alternate fates, and the final fate of the cell is based purely on the particular microenvironment the cell is forced by chance to occupy in the lingual epithelia. The pressure on the cell originates from the growth and subsequent crowding of the cellular mass (Cameron, 1966). Disruption of TGF-β signaling shortens the cell cycle, without affecting the final fate of mutant cells (Falk et al., 2008; Li and Zhou, 2012). Thus, disruption of TGF-β signaling was speculated to hasten the developmental rate of lingual epithelial cells in the current study and increase the cell density.

Epigenetic repression and derepression are important in controlling the balance between epidermal stem/progenitor cell proliferation and differentiation (Frye et al., 2007; Gu et al., 2010). It has been reported that acetylation of histone H4-K16 increases gene transcription both in vitro and in vivo, controls chromatin structure, and regulates cellular lifespan (De Gobbi et al., 2011; Ellis et al., 2009; Kanwal and Gupta, 2010; Kouzarides, 2007; Magklara et al., 2011). Thus, we attempted to trace the process of cellular development using antibodies against acetylated histone H4 (AcH4) and H3 (AcH3) after TGF-β signaling disruption.

In control mice, we only observed the higher level AcH3+ cells in the TB (Fig. 5A and B). After 5 hr of TGF-β signaling disruption, higher level AcH3+ cells were observed in the basal cell layer, connective tissue, and muscle tissue (Fig. 5C and D). In control mice, higher level AcH4+ cells were seen in the basal cell layer and connective tissue (Fig. 5E and F). After 5 hr of TGF-β signaling disruption, many cells with the higher expression of AcH4 were observed in the basal/superbasal cell layer, filiform papillae, and fungiform papillae (Fig. 5G and H). Meanwhile, we also found fish-net cells with the higher level of AcH4 expression in the lingual epithelia (Fig. 5H, arrow). After 5 hr of TGF-β signaling disruption, the number of cells significantly increased in the lingual epithelia (Fig. 5I). More importantly, the number of AcH3+ and AcH4+ cells also increased in the lingual epithelia (Fig. 5J).

Figure 5.

Acetylation of histone H4 and histone H3 in lingual epithelia rapidly increases after 5 hr of TGF-β signaling disruption. Higher level AcH3+ cells were only observed in TB of control mice, shown here at low magnification (A) and high magnification (B). C,D: After 5 hr of TGF-β signaling disruption, higher levels of AcH3+ cells were seen in the basal cell layer, connective tissue, and muscle tissue. In control mice, the higher level AcH4+ cells were observed in the basal cell layer and connective tissue, shown here at low magnification (E) and high magnification (F). After 5 hr of TGF-β signaling disruption, many cells with the higher level of AcH4 were observed in the basal/superbasal cell layer, filiform papillae, and fungiform papillae. Fish-net cells with the higher level of AcH4 were found in the lingual epithelia. After 5 hr of TGF-β signaling disruption, the number of cells significantly increased in the lingual epithelia (I, a b: P < 0.05). After 5 hr of TGF-β signaling disruption, the number of AcH4+ (J, a b: P < 0.05) and AcH3+ cells (J, c d: P < 0.05) significantly increased, shown here at low magnification (G) and high magnification (H). B, basal cell layer; CT, connective tissue; FuP, fungiform papillae; FP, filiform papillae. Scale bar (A,C, E, G) = 80 μm; (B, D, F, H) = 50 μm.

Further analysis of AcH4+ cells was carried out to show their distribution along the control tongue. In the sagittal section, the higher level of AcH4+ cells was only detected in the TB at the anterior tongue (between sites 1 and 2) (Fig. 6A). The high level of AcH4+ cells was still detected in the TB at the middle of the tongue (site 3) (Fig. 6B), while they were found in the basal cell layer at the posterior tongue (between sites 4 and 5) (Fig. 6C). After 24 hr of TGF-β signaling disruption, higher levels of AcH4+ cells were observed in lingual epithelia and papillae at the anterior tongue (between sites 1 and 2) (Fig. 6D) and in the lingual epithelia covering the fungiform papillae at the middle of the tongue (site 3) (Fig. 6E). Increased AcH4+ cells were still found in filiform papillae at the posterior tongue (between sites 4 and 5) (Fig. 6F). After quantifying the relative proportion of AcH4+ cells in the lingual epithelia in control mice, an obvious distribution pattern was found: lowest in the tip, higher in the middle, and highest in the posterior of the tongue. After 24 hr of TGF-β signaling disruption, the AcH4+ cells significantly increased in the tip and middle of the tongue (Fig. 6G).

Figure 6.

Distribution of AcH4+ cells along the tongue. In the sagittal section, the higher acetylation level of histone H4 was only detected in the TB in the anterior tongue (between sites 1 and 2) (A) and also in the TB at the middle tongue (site 3) (B), but it was detected in the basal cell layer at the posterior tongue (between sites 4 and 5) (C). After 24 hr of TGF-β signaling disruption, the higher acetylation level of histone H4 was observed in lingual epithelia and papillae at the anterior tongue (between sites 1 and 2) (D), in the lingual epithelia covering the fungiform papillae at the middle tongue (site 3) (E), and in the filiform papillae at the posterior tongue (between sites 4 and 5) (F). G: Quantification of AcH4+ cells showed an obvious differential distribution from the tip (low) to the posterior (high) of the tongue in control mice (a, c, e: P < 0.05). After 24 hr of TGF-β signaling disruption, the number of AcH4+ cells in the tip (a, b: P < 0.05) and middle (c, d: P < 0.05) of the tongue significantly increased but did not change in the posterior tongue (e, f: P > 0.05). B, basal cell layer; CT, connective tissue; FuP, fungiform papillae; FP, filiform papillae. Scale bar = 50 μm.

In short, the results showed that the cell number significantly increased in epithelia of tongue tip after 5 hr of TGF-β signaling disruption. Meanwhile, AcH4+ and AcH3+ cells rapidly increased in the tip of the tongue after 5 hr of TGF-β signaling disruption. In addition, the acetylation levels of histone H4 varied, depending on the location of tongue from the tip to posterior in control mice. Curiously, AcH4+ cells significantly increased in the first half of tongue. In our previous study, GFP expression was not observed in lingual epithelia after 5 hr of disruption of TGF-β signaling (Li and Zhou, 2012), indicating that those AcH4+ and AcH3+ cells should not have originated from K14+ cells at that time but rather were cells in an earlier stage of maturation. To maintain the normal balance of the cell migration stream, those immature cells located downstream may have been removed when TGF-β signaling was interrupted in this mouse model and disturbed the maturation process. Thus, those immature cells failed to complete modifications such as deacetylation in the final maturation stage due to a shortened cell cycle.

Jagged2 Is Inactivated After 3 Days of Dox Induction

Loss of the Tgfbr2 gene in K14+ cells leads to the activation of intergrin-FAK (focal adhesion kinase)-Src signaling. Recently, it was reported that the WNT and Notch signaling pathways regulate the self-renewal of stem cells (Androutsellis-Theotokis et al., 2006; Falk et al., 2008; Lie et al., 2005). In order to further explore the molecular mechanism underlying the expansion of lingual epithelial stem cells in the absence of TGF-β signaling, we detected components of those pathways in the mouse tongue epithelia by immunostaining.

In the control mouse, Jagged2 was detected in lingual epithelia and filiform papillae (Fig. 7A). After 3 days of Dox induction, Jagged2 expression was sparse in lingual epithelia and filiform papillae (Fig. 7B). It was noted that Jagged2 was also found in muscle tissue (Fig. 7B, arrow). We further checked for the presence of cleaved Notch1 (Val1744). In the control mouse, cleaved Notch1 was detected in the connective tissue of fungiform papillae (Fig. 7C, arrow). After 3 days of Dox induction, we still failed to detect the cleaved Notch1 in lingual epithelia (Fig. 7D), although it was observed in the muscle tissue (Fig. 7D, arrow). In the control mouse, p42/44 MAPK (erk1/2) was found in the connective tissues and lingual epithelia (Fig. 7E). After 3 days of Dox induction, p42/44 MAPK was still detected in the filiform papillae and lingual epithelia, with the stronger signal observed in lingual epithelia covering the surface of papillae compared with control mice (Fig. 7F). In the control, β-catenin was observed in lingual epithelia, filiform papillae, and fungiform papillae (Fig. 8A and B). Many β-catenin+ cells were also observed in the interpapillary epithelia (Fig. 8A and B, arrow). After 35 days of TGF-β signaling disruption, β-catenin was still expressed in lingual epithelia and papillae (Fig. 8C and D).

Figure 7.

Inactivation of Jagged2 after 3 days of Dox induction. A: In the control mouse, we detected Jagged2 in lingual epithelia and filiform papillae. B: After 3 days of Dox induction, Jagged2 was sparsely detected in lingual epithelia and filiform papillae. Note that Jagged2 was detected in muscle tissue (B, arrow). C: In the control mouse, cleaved Notch1 (Val1744) was detected in connective tissue of fungiform papillae (arrow). D: After 3 days of Dox induction, cleaved Notch1 still could not be detected in lingual epithelia but was also observed in muscle tissue (arrow). E: In the control mouse, p42/44 MAPK (erk1/2) was observed in connective tissues and lingual epithelia. After 3 days of Dox induction, p42/44 MAPK was still detected in papillae and lingual epithelia, with the stronger signal in lingual epithelia covering the surface of papillae compared with control mice. CT, connective tissue; FuP, fungiform papillae; FP, filiform papillae. Scale bar (A, B, D, E) = 50 μm; (C, F) = 20 μm.

Figure 8.

β-catenin expression in lingual epithelia and filiform papillae. A,B: In control mice, β-catenin expression was observed in lingual epithelia, filiform papillae, and fungiform papillae, and many β-catenin+ cells were seen in the interpapillary epithelia. C,D: After 35 days of TGF-β signaling disruption, β-catenin expression was still observed in lingual epithelia and papillae. A nickel solution was added into the substrate solution to produce the desired gray-black stain. Arrows indicate the positive gray-black staining in lingual epithelia (A and B). CT, connective tissue; B, basal cell layer; FuP, fungiform papillae; FP, filiform papillae. Scale bar = 50 μm.

DISCUSSION

Previous studies have shown that disruption of Tgfbr2 expression fails to induce significant pathological changes in internal stratified or simple epithelia (Forrester et al., 2005; Nolasco et al., 1987). Meanwhile, transient hyperplasia has been observed in Tgfbr2 null mammary epithelia of neonatal mice, which regresses during adulthood, and tissue homeostasis is maintained (Cheng et al., 2005; Forrester et al., 2005). Similarly in our studies, although tissue homeostasis was maintained in the dorsal surface of the adult tongue when TGF-β signaling was disrupted, filiform papillae showed different changes: inhibition in the tip versus hyperproliferation from the middle to posterior of the tongue. The unexpected greater susceptibility to progression of invasive carcinoma in the lingual epithelium on the ventral surface compared to the dorsal surface and the different proliferation patterns of papillae from the tip to posterior of the tongue are indicative of the complicated pathological changes of lingual epithelia, similar to results reported in other tissues. The current results also provide information for exploring further the molecular mechanism of tongue diagnosis.

In the K5.CrePR1-Tgfbr2 CKO model, no significant pathological changes have been found in the head-and-neck epithelia of adult mice where the Tgfbr2 gene is conditionally disrupted by application of RU486 to the oral cavity daily for 5 days. However, activation of either K-ras or H-ras in combination with the deletion of Tgfbr2 induces squamous cell carcinoma of head-and-neck epithelia (Lu et al., 2006). The K14-Cre-Tgfbr2 CKO mice are viable and appear phenotypically normal through early adulthood, and the loss of Tgfbr2 does not affect back skin epidermis and the appendages. However, adult K14-Tgfbr2 null mice develop SQCCs in their anal and genital regions, displaying visible signs of tumor formation with age. Further histological analyses revealed frequent spontaneous SQCCs arising in this model within transitional zones between two merging but distinct epithelial tissue types (Guasch et al., 2007). We speculate that there may be different connective tissues under the dorsal or ventral surface of the tongue, which contribute to the different pathological changes between these two locations. Indeed, we observed pathological changes of connective tissues under the dorsal surface in the first half of the tongue after TGF-β signaling disruption. Previous tissue recombination studies have also demonstrated that the reciprocal inductive interactions between the mesenchyme and the overlying epithelium are indispensable for the fate induction of various epithelia (Blanpain et al., 2007).

Although the Tgfbr2 gene is conditionally disrupted, many epithelia still develop normally. When carcinogenesis protocols are applied to transgenic mice or CKO mice that disrupt all TGF-β signaling, skin tumorigenesis is promoted, suggesting that loss of Tgfbr2 is not an initiating event in skin tumorigenesis. Furthermore, cancer progression occurs rapidly when oncogenes are activated (often oncogenic Ha-Ras) in Tgfbr2 null epithelial tissues (Guasch et al., 2007; Lu et al., 2006), and invasive squamous cell carcinoma of the forestomach appears when the Tgfbr2 gene is conditionally inactivated in mouse fibroblasts (Bhowmick et al., 2004). Thus, activation of various growth factors is one of the possible mechanisms for stimulation of epithelial proliferation in selected tissues after disruption of TGF-β signaling. Consistent with those studies, the current results revealed the rapid increase of AcH3+ and AcH4+ cells after 5 hr of TGF-β signaling disruption in lingual epithelia. In addition, the AcH3+ cells and AcH4+ cells also showed a different distribution along the whole tongue. It is well known that total cellular histone acetylation is involved in the regulation of gene expression (Ellis et al., 2009; Kouzarides, 2007; Magklara et al., 2011). Several studies have associated the effect of histone deacetylase (HDAC) inhibitors on gene transcription with an increased accumulation of acetylated histones H3 and H4 in total cellular chromatin (Heitman et al., 1983; Jeter and Cameron, 1971; Jiang et al., 2008; Schwartz et al., 2011). Histone acetylation up-regulates p21WAF1 expression in human colon cancer cell lines (Chen et al., 2004). Furthermore, acetylation of histone H3 lysine 56 is involved in the core transcriptional network in human embryonic stem cells (Xie et al., 2009), and acetylation of histone H4 on lysine 16 (H4-K16Ac) regulates chromatin structure and protein interactions (Kouzarides, 2007). In short, the higher levels of acetylated histones H3 and H4 can disturb the cellular gene expression profile, generating an abnormal microenviroment in lingual epithelia after the disruption of TGF-β signaling. The consequence is ectopic expression of the K14 gene, as GFP expression previously has been observed in the filiform papillae after 3 days of Dox induction (Li and Zhou, 2012). Combined with the results from the above epigenetic study, it is reasonable to think that the complex phenotypic changes revealed in this study may be the result of the interplay between the changes in epigenetic modifications and abnormal gene expression.

Cultured primary keratinocytes, obtained from the backskin epidermis of K14-Tgfbr2 null mice, have been shown to display increased β1-intergrin, FAK, Src, and MAPK activities, as well as enhanced migration and invasion (Guasch et al., 2007). In agreement with those results, we also found enhanced pMAPK activity in lingual epithelia after the disruption of TGF-β signaling. Unexpectedly, we also found the loss of Jagged2 expression in lingual epithelia and papillae after 3 days of TGF-β signaling disruption. Although the function of Jagged2 is unknown, activation of Jagged2 by hypoxia in tumor cells induces epithelial-mesenchymal transition (EMT) and also promotes cell survival in vitro (Valsecchi et al., 1997; Xing et al., 2011). Jagged2 also plays an important role in promoting the survival and proliferation of hematopoietic progenitors (Van de Walle et al., 2011). In conclusion, although future studies will be necessary to illuminate the molecular mechanisms underlying the pathological changes of lingual epithelia as we have described, our results provide a mechanistic framework for understanding the role of TGF-β signaling in regulating homeostasis and carcinogenesis in lingual epithelia.

EXPERIMENTAL PROCEDURES

Double Transgenic Mice and Dox Treatment

K14-rtTA and TetO-EGFP-Tgfbr2 (PTR) mouse lines obtained from Jackson Laboratories (Bar Harbor, ME) were bred and maintained at the Monell Chemical Senses Center animal facility. All procedures involving animals were approved by the Monell Chemical Senses Center Institutional Animal Care and Use Committee.

For Dox administration (Sigma, St. Louis, MO), the drug was diluted in 5% sucrose in water to a final concentration of 0.3–0.5 mg/ml and supplied as drinking water. Animals were allowed unlimited access to the Dox-containing water, which was changed every 2–3 days. A single intraperitoneal injection of Dox (10 mg/kg body weight) was also administered while the mice began receiving Dox-containing water.

Histology and Immunostaining Procedure

For immunocytochemistry, mice were perfused transcardially with 2–4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS; pH 7.2–7.4). The tongue tissues were dissected, post-fixed in PFA for 2–12 hr, and cryoprotected in 30% sucrose in PBS at 4°C overnight. After sectioning on a cryostat, 10–12-μm sections were collected onto Superfrost Plus Microscope slides (Fisher Scientific, Pittsburgh, PA). Polyclonal primary antibodies used were specific for GFP (goat, Abcam [Cambridge, MA] ab-5450; rabbit, ab-6556), anti-acetyl histone H4 (rabbit, Millipore, Billerica, MA; 06–866), β-catenin (rabbit, Cell Signaling Technology, Danvers, MA; 9562), cleaved Notch1 (Val1744 rabbit, Cell Signaling Technology, 2421), Jagged2 (goat, Santa Cruz Biotechnology, Santa Cruz, CA; sc-8158), and acetyl histone H3 (rabbit, Millipore 06–599). Monoclonal primary antibodies were against p42/44 MAPK (Cell Signaling Technology, 9106). Staining was performed with the TSA Plus system from Perkin Elmer (Waltham, MA) according to the manufacturer's instructions. Fluorescent images were captured with the Leica TCS SP2 Spectral Confocal Microscope (Leica Microsystems Inc., Mannheim, Germany). Stainings against GFP and anti-acetyl histone H4 were performed with the standard immunocytochemical procedure according to the manufacturer's instructions (Vectastain Elite ABC Kits, Vector Labs, Burlingame, CA).

Standard hematoxylin and eosin staining was used in the current study. Brightfield images of the sections were digitally captured with ImagePro Plus (Media Cybernetics Inc., Silver Spring, MD) in serial sections of the tongue. The thickness of lingual epithelia and ratio of the area of positively stained cells to total lingual epithelia were quantitatively calculated with ImagePro Plus (Media Cybernetics Inc., Silver Spring, MD) in serial sections. These data were analyzed by one-way analysis of variance using SPSS11.5 software. Differences were considered to be significant when P < 0.05.

Ancillary