Purpose: Squamous cell carcinoma of the oral cavity is one of the most common human neoplasms, and prevention of these carcinomas requires a better understanding of the carcinogenesis process and a model system in which cancer chemoprevention agents can be tested. We have developed a mouse model using the carcinogen 4-nitroquinoline 1-oxide (4-NQO) in the drinking water to induce tumorigenesis in the mouse oral cavity.
Experimental Design: 4-NQO was delivered by tongue painting or drinking water to two mouse strains, CBA and C57Bl/6. The incidences of oral cavity carcinogenesis were then compared. In addition, we examined the expression of some of the molecular markers associated with the process of human oral cavity and esophageal carcinogenesis, such as keratin (K) 1, K14, p16, and epidermal growth factor receptor, by immunohistochemistry.
Results: After treatment with 4-NQO in the drinking water, massive tumors were observed on the tongues of both CBA and C57Bl/6 female mice. Pathological analyses indicated that flat squamous dysplasias, exophytic papillary squamous tumors (papillomas), and invasive squamous cell carcinomas were present. Immunohistochemistry analyses showed that 4-NQO changed the expression patterns of the intermediate filament proteins K14 and K1. K14 was expressed in the epithelial suprabasal layers, in addition to the basal layer, in tongues from carcinogen-treated animals. In contrast, control animals expressed K14 only in the basal layer. Moreover, we observed more bromodeoxyuridine staining in the tongue epithelia of 4-NQO-treated mice. Reduced expression of the cell cycle inhibitor, p16, was observed, whereas 4-NQO treatment caused an increase in epidermal growth factor receptor expression in the mouse tongues. Interestingly, similar features of carcinogenesis, including multiple, large (up to 0.5 cm) exophytic papillary squamous tumors and invasive squamous cell carcinomas, increased bromodeoxyuridine staining, and increased K14 expression, were also observed in the esophagi of 4-NQO-treated mice. However, no tumors were observed in the remainder of digestive tract (including the forestomach, intestine, and colon) or in the lungs or livers of 4-NQO-treated mice. These results indicate that this murine 4-NQO-induced oral and esophageal carcinogenesis model simulates many aspects of human oral cavity and esophageal carcinogenesis.
Conclusions: The availability of this mouse model should permit analysis of oral cavity and esophageal cancer development in various mutant and transgenic mouse strains. This model will also allow testing of cancer chemopreventive drugs in various transgenic mouse strains.
Squamous cell carcinoma (SCC) of the oral cavity is one of the most common neoplasms in the world. Approximately 350,000 people are diagnosed annually worldwide with SCC (1) . Although some primary tumors can be treated, many patients will develop second primary tumors, suggesting multifocal tumor development. Two major etiological factors in oral cavity SCC are the use of tobacco and alcohol (2) . Similarly, human esophageal cancer occurs worldwide and is the eighth most prevalent cancer (3) . The 5-year survival of patients diagnosed with esophageal SCC is around 30% (4) . Clearly, the development of cancer chemopreventive agents for oral cavity and esophageal cancer is an important research priority (3 , 5, 6, 7) , and for this, animal models are needed.
Previously, several animal models for oral SCC development were generated, including hamster, rat, and mouse models. In the past, the most commonly used model was the carcinogen 7,12-dimethylbenzanthracene-induced hamster cheek pouch tumor model (8) . However, as discussed by Nauta et al. (9) and MacDonald et al. (10) , the 7,12-dimethylbenzanthracene-induced cheek pouch tumors are dissimilar to human oral cavity tumorigenesis in many respects. The 7,12-dimethylbenzanthracene-induced hamster cheek tumors develop during the treatment period. Thus, there is a problem in distinguishing between the transient and reversible changes of the epithelium caused by direct contact with the carcinogen and true premalignant transformation (11) . Moreover, the tumors caused by 7,12-dimethylbenzanthracene do not possess many of the histological features of differentiated SCCs and do not closely resemble early human lesions (9) .
Tobacco contains many carcinogens that cause carcinogenesis through the formation of DNA adducts (carcinogen-DNA adducts) in many human tissues (12) . 4-Nitroquinoline 1-oxide (4-NQO) is a water-soluble quinoline derivative that can cause DNA adduct formation, resulting in adenosine substitution for guanosine (13, 14, 15) . 4-NQO also can undergo redox cycling to produce reactive oxygen species that result in mutations and DNA strand breaks (16 , 17) . 4-NQO has been shown to induce oral cavity SCCs in rats (18) and mice (11 , 19 , 20) . In the rat, the 4-NQO model has been used to test cancer chemopreventive agents (21, 22, 23, 24, 25) . 4-NQO treatment has generally been performed by two different methods, either painting 4-NQO on mouse tongues, using a dose of 5 mg/ml (11 , 19) , or dissolving 4-NQO in the drinking water for rats (18 , 26) . Administration of 4-NQO produces a temporal carcinogenesis progression model that demonstrates multiple dysplastic, preneoplastic, and neoplastic lesions after long-term treatment (11) . These sequential changes in the epithelial cells mimic human oral cavity neoplastic transformation. Additionally, 4-NQO-induced mouse oral cavity cancer is similar to human head and neck SCCs in that the expression of many genes related to tumorigenesis in humans is affected (27 , 28) . However, the 4-NQO tongue painting method is laborious and difficult in terms of applying the same dose from animal to animal, and the yield of tumors is relatively low (11) .
Recently, a genetic mouse model of human oral-esophageal cancer was generated by overexpressing cyclin D1 in the oral-esophageal epithelium of p53-null mice (29) . Whereas this model is useful to address some research questions, the overexpression of cyclin D1 may not completely reflect the multiple genetic changes that occur during the process of carcinogen-induced oral cavity carcinogenesis in humans (30 , 31) .
Esophageal cancer occurs worldwide. More than 90% of esophageal cancers in the world are squamous carcinomas, although adenocarcinomas are more prevalent in the United States (3) . Approximately 14,000 new cases of esophageal cancer are expected to be diagnosed in 2003 (32) . There is no commonly used mouse model for human esophageal SCC, although there are rat models for esophageal SCCs in which tumors are induced by a carcinogen and food contaminant, N-nitrosomethylbenzylamine (3 , 33 , 34) . Compared with other animal models, mouse models of oral and esophageal SCC have obvious benefits because various transgenic and knockout lines can be examined in such models. We hypothesized that 4-NQO in the drinking water would induce carcinogenesis not only in the mouse oral cavity but also in the esophagus because the esophageal epithelium would be highly exposed to the carcinogen.
In this study, we compared 4-NQO painting of tongues and delivery of 4-NQO in the drinking water in two mouse strains and found that the latter method, at the doses of 4-NQO used, resulted in a much higher incidence of oral cavity carcinogenesis. We also showed that 4-NQO in the drinking water resulted in a high incidence of esophageal tumors. Moreover, we assessed the expression of some of the molecular markers associated with the process of human oral cavity and esophageal carcinogenesis in this model. When some of the biomarkers of human oral cavity and esophageal carcinogenesis, such as keratin (K) 1, K14, p16, and epidermal growth factor receptor (EGFR), were assayed, we found that 4-NQO induced morphological and pathological neoplastic transformations of mouse tongues and esophagi that resembled the process of tumorigenesis in the oral cavity and esophagus seen in humans (7 , 30 , 35, 36, 37, 38) .
MATERIALS AND METHODS
Animals and Carcinogen Treatment.
Six-week-old female CBA and C57Bl/6 mice, purchased from The Jackson Laboratory, were used for the studies with 4-NQO. The experiments were carried out under controlled conditions with a 12-h light/dark cycle. Animals were maintained on a normal chow (Lab-diet with constant nutrition; Lab-Diet Co., Bretment, MO). The treatments were carried out in two different ways, based in part on the work of Hawkins et al. (11) , with modifications. The carcinogen 4-NQO (Sigma, St. Louis, MO) stock solution was prepared weekly in propylene glycol at 5 mg/ml and stored at 4°C. For the painting method, the tongues were painted with the 4-NQO stock solution three times a week, the same dose that Hawkins et al. (11) used. For the drinking water method, the 4-NQO stock solution was diluted in the drinking water for mice, and the water was changed once a week (see Tables 1⇓ 2⇓ 3⇓ for concentrations). Mice were allowed access to the drinking water at all times during the treatment. Each strain of mice was randomly divided into an experimental group in which the drinking water contained 4-NQO and a control group in which the drinking water contained no 4-NQO, only the same volume of propylene glycol. After an 8- or 16-week carcinogen treatment, mice were analyzed for precancerous and cancerous lesions in the oral cavities and esophagi at different times for up to 12 weeks or until signs of sickness or weight loss.
Tissue Dissection and Sectioning.
The tongues and esophagi of mice were dissected immediately after cervical dislocation. Mice were then subjected to autopsies; whole esophagi and stomachs were opened longitudinally, and tumors with a diameter of >1–2 mm were counted. Cross-sections of some of the tissues were cut, and tissues were fixed in freshly made 4% paraformaldehyde overnight at 4°C, embedded in paraffin, and sectioned into 5-μm sections. Multiple sections of tissues from several mice in each treatment category were stained with various antibodies, and control sections were always stained at the same time and under the same conditions as sections from 4-NQO-treated mice.
Gross lesions were identified and photographed. The histological determination of squamous neoplasia was done by a pathologist (S. T.) on the sectioned tissue samples. The sections from the tongues and esophagi were deparaffinized, rehydrated, and stained with H&E for histopathology. The lesions observed were classified into four types: epithelial hyperplasia; dysplasia; papilloma; and invasive carcinoma. Hyperplasia was defined as thickened epithelium with prominent surface keratinization and with or without elongated rete ridges. Dysplasia was defined as loss of polarity in the epithelial cells, nuclear pleomorphism and hyperchromasia, abnormal single cell keratinization (dyskeratosis), and increased or abnormal mitoses. Lesions with such changes involving the entire thickness of epithelium were considered as carcinoma in situ. Papilloma was defined as noninvasive exophytic growth of neoplastic cells, and invasive carcinoma was defined as a lesion with invasion into the subepithelial tissues.
The intermediate filaments and other SCC cancer markers in mouse epithelium were investigated by immunohistochemistry. The anti-cytokeratin 14 (K14, catalogue number NCL-002) antibody was obtained from Novocastra Laboratories (Newcastle, United Kingdom; Ref. 39 ), and anti-cytokeratin 1 (K1, catalogue number PRB-165P) antibody was from BabCo/Covance (Denver, PA). Anti-p16 (catalogue number sc-1661) and anti-EGFR (catalogue number sc-03) antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). Slides were deparaffinized and rehydrated, and antigen retrieval was performed in Antigen Unmasking Solution (Vector Laboratories, Burlingame, CA) in a pressure cooker for 2 min. After quenching endogenous peroxidase with 3% H2O2, the tissue sections were blocked with 1.5% goat serum or Vector special blocking reagent (when the primary antibody was mouse IgG). Then K14 (1:20) and p16 (1:200) antibodies were incubated with tissues for 30 min at 22°C, whereas K1 (1:200) and EGFR antibodies (1:200) were incubated for 1 h at 22°C, followed by incubation with secondary antibodies (1:200; antimouse or rabbit IgGs; Vector Laboratories). As a control, the sections were stained without incubation with primary antibody. Sections were stained by using the Vectastain Elite ABC kit (Vector Laboratories), based on a peroxidase detection mechanism (manual of the product), and 3,3′-diaminobenzidine was used as the substrate. Samples of tongues and esophagi were taken from all of the mice for sectioning; several sections from four to six mice were examined from a given treatment, and immunohistochemical staining was performed on control versus 4-NQO-treated tissues at the same time with the same lot and concentration of antibody to obtain semiquantitative immunohistochemical data.
Bromodeoxyuridine (BrdUrd) Labeling and Immunohistochemistry.
Mice were inoculated with BrdUrd labeling reagent (catalogue number 00-0103; Zymed, South San Francisco, CA) i.p., according to the protocol of the manufacturer (1 ml/100 g body weight). After 5 h, the mice were sacrificed by cervical dislocation, and the tissues were quickly dissected, cut into cross-sections, fixed in freshly made 4% paraformaldehyde, and embedded in paraffin. The BrdUrd immunohistochemistry was carried out with a BrdUrd Detection Kit (catalogue number 93-3943; Zymed). Tissue sections were deparaffinized and rehydrated in graded ethanols. After quenching endogenous peroxidase with 3% H2O2, antigen retrieval was performed by digestion with trypsin for 10 min at 37°C. The DNA in tissue samples was subsequently denatured with denaturing solution from the BrdUrd Detection Kit. Tissues were then incubated with blocking solution for 30 min, followed by a 1-h incubation with biotinylated mouse monoclonal anti-BrdUrd antibody at room temperature. Slides were incubated with streptavidin-peroxidase for 10 min at room temperature. The incorporation of BrdUrd into the genomic DNA of proliferating cells was detected by staining with 3,3′-diaminobenzidine tetrahydrochloride. Finally, slides were counterstained with hematoxylin. Negative slides were not incubated with anti-BrdUrd antibody.
Cells with dark brown-stained nuclei by 3,3′-diaminobenzidine were identified as cells with BrdUrd incorporation. The BrdUrd labeling index (percentage) was determined by the number of BrdUrd-positive nuclei in a total of 300 (for tongues) or 100 (for esophagi) epithelial cell nuclei in a certain area. The data were analyzed statistically using Student’s t test.
The Carcinogen 4-NQO in Drinking Water Causes Tumor Formation on Tongues and in Esophagi of Mice.
We carried out several rounds of 4-NQO treatment of female CBA and C57Bl/6 mice, using different doses and durations of carcinogen treatment. We also varied the manner of 4-NQO delivery, using either tongue painting or delivery via the drinking water (Table 1)⇓ .
For the 4-NQO tongue painting protocol, we followed the method of Hawkins et al. (11) . Female CBA mice (30 mice) were treated with 4-NQO for 16 weeks. All mice survived the treatment period, and all survived the 12 posttreatment weeks. With this delivery method, 4-NQO treatment caused a much lower level of tumor formation (Table 1)⇓ .
In the studies using 4-NQO in the drinking water, all control (propylene glycol only) and 4-NQO-treated CBA and C57 B1/6 mice survived the 16-week 4-NQO (100 μg/ml) treatment, and about 80% mice survived for 12 weeks after carcinogen treatment (Table 1)⇓ . No gross lesions were visible at the end of the 16-week 4-NQO treatment. We examined the tongues for lesions at different time points after carcinogen treatment and found that gross lesions developed over the following 4–12 weeks. 4-NQO exposure induced many precancerous and cancerous lesions on tongues and in esophagi of both CBA and C57Bl/6 mice. This carcinogenic process was multifocal, i.e., there were several lesions at different sites on one tongue or one esophagus. At the time of sacrifice, these lesions covered almost all of the surface of the tongues; therefore, it was difficult to quantify the lesions. In the esophagi, the lesions were also multifocal. No differences in the numbers of lesions or in the severity or distribution of the lesions between the upper and lower esophagus were seen. Some of the lesions were exophytic [protruding from the surface of the tongue into the oral cavity (papillomas)], and others were endophytic [invading the deeper layers of the tongue through the basement membrane (invasive carcinomas)]. We also observed dysplasias in tongue sections from grossly nontumorous areas in the 4-NQO-treated animals, although the focus of our investigation was primarily on grossly identifiable tumor formation. The tumor incidence (papilloma and/or carcinoma) was 100% by 12 weeks after this dose (100 μg/ml) of 4-NQO treatment. Massive SCCs were observed on the dorsal sides of the tongues (Table 1⇓ , Fig. 1⇓ ) and in the esophagi (Table 2)⇓ of 4-NQO-treated CBA and C57Bl/6 mice. Gross lesions were not observed in other areas of the oral cavities, most likely because of the manner in which mice drink water. The upper surface of the tongue is the region of the oral cavity that has the greatest exposure to the water.3 The palate, buccal, and floor of mouth regions were not examined in detail for hyperplasia or dysplasia, nor were molecular markers examined in these regions. Some of the mice treated with the highest dose of 4-NQO showed some body weight loss (about 5–10% of body weight), compared with the control group, at 10–12 weeks after the end of the 16-week 4NQO treatment. No gross lesions were observed in the lungs, livers, colons, or forestomachs of any of the 4NQO-treated animals. No lesions were found in the control (propylene glycol-treated) mice.
Different doses (20, 50, and 100 μg/ml) of 4-NQO in the drinking water and various lengths of 4-NQO treatment (8 and 16 weeks) were also tested. Doses of both 50 and 100 μg/ml resulted in significant lesions on the tongues; however, only the 100 μg/ml treatment for 8 or 16 weeks induced significant numbers of esophageal lesions (Table 3)⇓ .
Examination of tissue sections from mice treated with 4-NQO in the drinking water at the highest dose (Tables 1⇓ and 2⇓ ) revealed pathological evidence of carcinogenesis. All tongues and esophagi from CBA and C57Bl/6 mice exhibited multiple lesions, including large (up to 0.5 × 0.5 × 0.3 cm) papillomas and/or carcinomas (Fig. 2, A–L)⇓ . Tongues and esophagi from control mice showed a very clearly defined epithelium with a single layer of basal cells. 4-NQO-treated mouse tongues and esophagi showed focal increases in thickness (hyperplasia); a loss of organization of the epithelium (dysplasia/in situ carcinoma); exophytic, mostly papillary, neoplastic squamous lesions (papillomas); and invasive SCCs identified by the invasion of neoplastic epithelial cells into the subepithelial tissues (Fig. 2, A–L)⇓ .
4-NQO Induced Alterations in K14 and K1 Expression in Mouse Tongues and Esophageal Epithelia during Carcinogenesis.
Keratins belong to the family of the intermediate filaments in epithelial cells. There are about 30 different keratins, with different cell type- and differentiation-specific expression patterns (40) . The keratin genes expressed in mouse soft epithelia are K1–K20 (41) . Keratins K5 and K14 are expressed in the proliferating basal layer cells, whereas keratins K1 and K10 are expressed in the suprabasal layers of many types of epithelia (42) . Evidence that keratin alterations are related to oral cavity and esophageal carcinogenesis has been accumulating. Kannan et al. (43) discovered a correlation between the expression levels of K10/K11, K19, and K14 and the stages of tumorigenesis of oral mucosa. The studies of Cintorino et al. (44) and Fong et al. (39) demonstrated that increased K14 is a biomarker in human and mouse esophageal carcinogenesis. In addition, high K14 expression throughout the epithelium was found in the majority of cases of human SCC, regardless of origin or degree of differentiation (45) .
Immunohistochemical analyses showed that the expression pattern of K14 was altered in mouse tongue epithelia by 4-NQO delivered via the drinking water (Fig. 3A)⇓ . In CBA mice, after treatment with 4-NQO, K14 was expressed in the suprabasal layers in addition to the basal layer of the tongues. In contrast, the control mice exhibited K14 expression only in the basal layer. The expression pattern of K14 was altered after 4-NQO treatment not only in the sections of carcinoma, but also before the grossly visible lesions appeared (Fig. 3A)⇓ .
Esophageal epithelia of 4-NQO-treated CBA mice were much thicker than those of control mice. The lumens of the esophagi became very narrow. The expression of K14 was also changed in the esophageal epithelia of CBA mice treated with 4-NQO; K14 staining was found in both basal and suprabasal layers, whereas in control mice, K14 was only clearly detected in the basal layer (Fig. 3A)⇓ . Additionally, the results from different doses and different durations of 4-NQO treatment showed that 50 and 100 μg/ml 4-NQO caused a dramatic increase in the level and area of the expression of K14 in the mouse tongues and esophageal epithelia. Only the lowest dose (20 μg/ml 4-NQO for 8 weeks) did not induce such changes in K14 (Fig. 3B)⇓ .
Aberrant expression of K1 has been observed in precancerous lesions of the human oral cavity (46) . 4-NQO treatment resulted in an increase in the level of cytokeratin 1 (K1) expression and an expansion in the area of expression in mouse tongue epithelium (Fig. 3C)⇓ . Similar to the 4-NQO induced changes in K14 expression level in mouse esophagi, K1 expression was elevated in the epithelia of esophagi (Fig. 3C)⇓ . Both CBA and C57Bl/6 mice showed similar changes in K1 and K14 expression in tongues and esophagi after 4-NQO treatment (data not shown).
The Expression of p16 Was Decreased in the Epithelia of Mouse Tongues and Esophagi after 4-NQO Treatment.
During the process of carcinogenesis, the dysregulation of the cell cycle is a critical event. Several mitogenic signals activate cyclin D and related cyclin-dependent kinase (CDK) complexes and induce cells to pass the G1-S-phase transition and enter S phase (47) . One of cell cycle inhibitors is p16, also known as MTS-1 (major tumor suppressor 1), which inhibits CDK4/6. The binding of p16 results in the release of cyclin D1 from CDK4 and CDK6 and its subsequent degradation (48, 49, 50) . p16 is a target commonly involved in esophageal carcinogenesis (51) , and the abrogation of p16 occurs frequently in human oral cancers (52) . The loss of expression of p16 has been observed in oral premalignant lesions (53 , 54) , primary tumors of the oral cavity (55, 56, 57, 58) , and squamous carcinomas of the upper aerodigestive tract (59) .
Therefore, the expression of the CDK inhibitor p16 in the epithelia of tongues and esophagi was investigated. In control CBA mice, most nuclei that stained with anti-p16 antibody were in the basal and suprabasal layers of mouse tongue epithelium (Fig. 4A)⇓ . In the sections from 4-NQO-treated CBA mice, the morphology of the epithelium was altered, and the number of p16-stained nuclei decreased, particularly in the tumors (Fig. 4B)⇓ . Additionally, p16-stained nuclei were detected in the well-organized basal and suprabasal layers of esophagi in control CBA mice, whereas very few p16-positive nuclei were observed in the thickened esophageal epithelium of 4-NQO-treated mice (Fig. 4, C versus D)⇓ .
In addition to the CBA mice, we examined the changes in expression patterns of p16 in the epithelia of mouse tongues and esophagi of C57Bl/6 mice after 4-NQO treatment. 4-NQO caused very similar changes in the expression patterns of p16 in the tongue and esophageal epithelia of these mice (data not shown).
EGFR Protein Staining Increased in the Dysplastic Lesions and Tumors in Mouse Tongues and Esophagi.
The EGFR is a transmembrane protein with an intracellular tyrosine kinase domain that regulates cell growth when bound to its ligands, epidermal growth factor (60) or transformation growth factor α (61) . The EGFR is overexpressed in many types of cancer, including stomach, breast, and brain tumors (62 , 63) , and therapies have been designed to target the EGFR (64 , 65) . Overexpression of the EGFR in head and neck SCC (66, 67, 68, 69, 70, 71) and esophageal SCC (72) has also been described.
We examined the EGFR levels in the epithelia of tongues and esophagi in control and 4-NQO-treated CBA mice. Samples from 4-NQO-treated mice demonstrated that hyperplasias in mouse tongue epithelia displayed elevated EGFR levels, compared with the samples from control mice (Fig. 4, G and H)⇓ . In control mice, the EGFR was expressed primarily in the basal layer of the tongue epithelium, whereas in 4-NQO-treated mice, the expression of the EGFR was seen throughout the entire thickness of hyperplastic epithelium. In addition, the levels of EGFR were higher in sections of carcinoma than in sections from mild hyperplasia (data not shown). These data are similar to the observations in human head and neck cancer (69) .
The esophageal epithelia of CBA mice showed higher EGFR levels and expanded regions of EGFR expression after 4-NQO treatment than control esophageal epithelia (Fig. 4, I and J)⇓ . In the esophageal epithelia of 4-NQO-treated mice, the normal regions adjacent to the tumor areas did not display the increase in level and the expansion of EGFR expression (data not shown).
We also assayed for the effects of 4-NQO on the expression of the EGFR in the epithelia of tongues and esophagi of C57Bl/6 mice. C57Bl/6 mice also showed increased areas of EGFR expression in the esophageal epithelia after 4-NQO treatment (data not shown).
4-NQO Enhanced Cell Proliferation in the Epithelia of Mouse Tongues and Esophagi.
The incoporation of BrdUrd (a thymidine analog) into the DNA is used as an index of cells in S phase (73) . Compared with the tongue epithelia of control CBA mice, the number of BrdUrd-labeled cells was greatly increased in the tongue epithelia of 4-NQO-treated mice. Moreover, the BrdUrd-stained nuclei were present not only in the basal layer, but some staining was also observed in the suprabasal layers (Fig. 5A⇓ ; see Table 3⇓ for quantitative data).
All doses of 4-NQO caused an increase in the BrdUrd labeling in the mouse tongue epithelia, despite the fact that 20 μg/ml 4-NQO for 8 weeks did not induce obvious gross cancerous lesions (Fig. 5B⇓ ; Table 3⇓ ). Moreover, with increasing doses and durations of 4-NQO treatment, the BrdUrd-stained nuclei were present in both the basal and suprabasal layers of epithelia of the tongues and throughout the tumor tissue (Fig. 5B⇓ , a versus d and e).
In esophagi, 20 μg/ml 4-NQO did not cause an increase in BrdUrd staining. 4-NQO at 50 μg/ml for 8 weeks caused esophageal lesions only in one of three mice, and this treatment regimen slightly increased the incorporation of BrdUrd. Only the 8-week and 16-week treatments with 100 μg/ml 4-NQO caused significant increases in BrdUrd incorporation in the esophageal epithelia, and both treatments resulted in large tumorigenic lesions in mouse esophageal epithelia (Fig. 5, A and B⇓ ; Table 3⇓ ).
Carcinogenesis Induced by 4-NQO in Drinking Water in Mouse Tongue and Esophageal Epithelia.
Hawkins et al. (11) and Steidler et al. (19) reported that painting 4-NQO on mouse tongues could induce oral cavity SCC. However, only a low incidence of SCC was seen by Hawkins et al. (11) . Steidler et al. (19) observed that all animals that received 4-NQO for 16 weeks by painting 3×/week developed SCCs, but not until 50 weeks. Our current study demonstrates that 4-NQO in the drinking water causes premalignant and malignant lesions in the oral cavities, particularly in the tongues, and in the esophagi of mice as well. Furthermore, the incidence of carcinogenesis is higher when 4-NQO is delivered in the drinking water than when 4-NQO is painted on the tongues (Table 1)⇓ . Delivery of the 4-NQO in the drinking water at the doses we used was associated with a carcinogenic process that occurred more rapidly than that observed by Steidler et al. (19) . We observed that 100% of the mice exhibited tongue or esophageal neoplasms at the highest 4-NQO doses by 24–28 weeks after the start of the experiment (Tables 1⇓ 2⇓ 3)⇓ .
The mouse tongues did not exhibit any gross abnormal morphological changes in the epithelia or inflammation during the 4-NQO treatment, and the neoplastic lesions developed over time after the 4-NQO treatment was completed. These results indicate that in this model, we can eliminate the possibility that the premalignant lesions observed are transient and regress when the carcinogen is removed. Because 4-NQO is a carcinogen that induces mutagenesis (74) , the morphological changes and the precancerous and cancerous lesions in mouse tongues that develop after the period of 4-NQO treatment most likely result from the accumulation of genetic mutations. The lesions progressed over time after 4-NQO treatment, from hyperplasia, dysplasia, and papilloma to invasive carcinoma. In the esophagus we noted squamous carcinomas and not adenocarcinomas. Our data demonstrate that the mouse model we have developed is a useful model for the analysis of both oral cavity and esophageal squamous carcinogenesis.
Tanuma et al. (18) and Tanaka et al. (26) observed that 20 μg/ml 4-NQO in drinking water induced SCC in rat oral cavities. They treated rats with 4-NQO for 8 weeks and then maintained these rats for another 22–24 weeks (30–32 weeks total) after stopping the treatment. In our study, we treated mice with 20, 50, and 100 μg/ml 4-NQO in drinking water for 8 weeks, and after another 16 weeks, the 50 and 100 μg/ml 4-NQO treatments caused large lesions in mouse tongues, whereas the 20 μg/ml dose did not. Similar to our results, Ide et al. (75) reported that a low dose of 4-NQO (0.001%, i.e., 10 μg/ml for 50 weeks) did not induce tumors on wild-type mouse tongues. However, it is possible that after a longer time period, lesions would have been seen in the mice we dosed with 4-NQO at 20 μg/ml for 8 weeks because we did observe some changes at the molecular level, such as an increased BrdUrd labeling index in tongues from these mice. Thus, our findings are in many respects similar to those reported for rats given 4-NQO in their drinking water, although we examined our 4-NQO-treated mice at 24 weeks, whereas the rats were analyzed at 30–32 weeks.
After 4-NQO treatment, not only the tongues but also the esophagi of mice exhibited signs of neoplastic transformation. Interestingly, no visible, gross lesions or carcinomas were found in the stomach, intestine, or colon. Although they did not assay for carcinogenesis, von Pressentin et al. (76) reported that a 2-week 4-NQO treatment (20 μg/ml in the drinking water) caused mutagenesis in mouse tongues and esophagi, but not in colon.
4-NQO Induced the Expression of Cytokeratin Proteins in the Epithelia of Mouse Tongues and Esophagi.
Keratins, such as K1 (46) , K14 (39 , 43 , 44) , and K19 (77) , have been used as biomarkers of SCCs. In normal squamous epithelia, the single layer of basal cells expresses K5 and K14; when these cells move upward into the suprabasal layers, they switch their keratin expression to K1 and K10 (42) . In our studies, 4-NQO caused an increase in the level and an expansion in the regions of expression of K14 in the mouse tongues and esophagi, which is consistent with the reports of Cintorino et al. (44) , Kannan et al. (43) , Viaene et al. (78) , and Fong et al. (39) that K14 was present in SCCs. In addition, the expression of K1 was also increased in the epithelia of mouse tongues and esophagi; this result is also similar to a previous study of human samples (46) .
The Effects of 4-NQO on Cell Growth and Cell Cycle Regulators in Mouse Tongue and Esophageal Epithelia.
We investigated the level of cell proliferation in the epithelia of mouse tongues and esophagi by using BrdUrd and p16 antibodies. BrdUrd, a thymidine analog, can be incorporated into DNA in the S phase of cell cycle. Hence, the cells that traversed S phase can be detected with an anti-BrdUrd antibody. After 4-NQO treatment, the mouse tongue and esophageal epithelia showed a greater number of BrdUrd-stained cells than the tongue and esophageal epithelia of control mice (Table 3⇓ ; Fig. 5⇓ ). Also, analyses of different doses and durations of 4-NQO showed that 20 μg/ml 4-NQO for 8 weeks increased the incorporation of BrdUrd into the DNA of epithelial cells, even though there were no obvious lesions in the mouse tongues. This suggests that 4-NQO results in increases in cell proliferation in mouse tongues that are related to the process of carcinogenesis.
p16 is a member of a group of CDK inhibitors (48) . Previous studies reported p16 protein inactivation, resulting from either gene deletion or promoter methylation, in oral cancers (79) , esophageal squamous carcinomas (51) , upper aerodigestive tract (59) , and head and neck squamous carcinomas (80) . In control mice, most p16-stained cells were in the basal layer of the tongue epithelium, whereas in invasive carcinomas of the tongue, only a few cells showed p16 staining (Fig. 4)⇓ . These results agree with the previous studies described above.
4-NQO Treatment Resulted in Increased Expression of the EGFR in Mouse Tongues and Esophageal Epithelia.
Evidence has accumulated that an increase in expression of EGFR family members is involved in human head and neck cancer (Refs. 67 and 69, 70, 71 ; for review, see Ref. 30 ). Thus, inhibition of EGFR expression or activity could result in the suppression of head and neck cancer cell growth (81, 82, 83) . Uhlman et al. (84) reported that laryngeal epithelial lesions with higher EGFR expression levels were more likely to progress to invasive cancer as compared with the epithelial lesions with lower levels of EGFR staining. We found that with 4-NQO treatment, the intensity of EGFR staining in the epithelia of mouse tongues and esophagi increased during the process of carcinogenesis (Fig. 4)⇓ . Our findings suggest that, similar to what is observed in humans, an increase in EGFR protein expression is a marker of carcinogenesis induced by 4-NQO.
Taken together, our data indicate that this new mouse model of 4-NQO-induced oral and esophageal cancer is a useful model in that it mimics many features and molecular events observed human SCC development. This model should be advantageous for studies of oral and esophageal cancer development. Moreover, it should also be useful for cancer chemoprevention studies in a variety of mutant and transgenic mouse strains.
We appreciate the insightful scientific input provided by the members of the Gudas laboratory.
Grant support: NIH Grant R01 DE10389 (to L. J. G.) and supported in part by a postdoctoral fellowship from the American Institute for Cancer Research (to X-H. T.).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Requests for reprints: Lorraine J. Gudas, Department of Pharmacology, Weill Medical College of Cornell University, 1300 York Avenue, New York, New York 10021. Phone: (212) 746-6250; Fax: (212) 746-8858; E-mail:
↵3 Dr. Hai T. Nguyen, personal communication.
- Received July 7, 2003.
- Revision received October 20, 2003.
- Accepted October 21, 2003.