Akt Activation Synergizes with Trp53 Loss in Oral Epithelium to Produce a Novel Mouse Model for Head and Neck Squamous Cell Carcinoma

Head and neck squamous cell carcinoma (HNSCC) is the sixth most common type of cancer worldwide. Although new therapeutic approaches, including fractionated radiotherapy, targeted chemotherapy, and concurrent radiotherapy and chemotherapy (1–4), have been recently evaluated, the improvement in overall survival in patients with HNSCC is still low. The term HNSCC comprises epithelial tumors that arise in the oral cavity, pharynx, larynx, and nasal cavity, with the main risk factors being alcohol and/or tobacco abuse (5). HNSCC results from the accumulation of numerous genetic and epigenetic alterations that occur in a multistep process. The molecular alterations displayed by human HNSCC affect several pathways that influence cell proliferation, apoptosis, differentiation, angiogenesis, inflammation, immune surveillance, and metastasis. The major pathways involved in HNSCC development include the pRb- and p53-dependent pathways, epidermal growth factor receptor (EGFR), signal transducer and activator of transcription 3 (Stat3), nuclear factor-κB (NFκB), and transforming growth factor β (TGFβ; reviewed in refs. 6, 7). Moreover, the initiation, growth, recurrence, and metastasis of HNSCC, as in many other solid epithelial cancers, have been related to the behavior of a small subpopulation of “tumor-initiating” or cancer stem cells (8, 9). In spite of the fact that the molecular mechanisms of HNSCC are not completely understood, several candidate genes of potential therapeutic relevance are now being validated through in vitro analyses (6, 10, 11); however, these studies cannot recapitulate the complex nature of HNSCC tumors in vivo. Thus, animal models of HNSCC will become essential tools, providing relevant insights of the molecular perturbations of these tumors. Nonetheless, there are few suitable genetically defined mouse models in which to study the progression of this type of tumor under preclinical settings (6) and that fully recapitulate the molecular characteristics of human HNSCC. Here we present a new HNSCC transgenic mouse model based on the expression of constitutively active Akt kinase combined with the ablation of Tp53 gene in stratified epithelia, which phenocopies the molecular alterations previously found in human HNSCC. The characteristics described here make this model an excellent and unique preclinical tool for the therapeutic management of HNSCC at different steps.

Mice and histologic procedures. The generation of Bk5myrAkt and Trp53^F/F;K14cre mice and the protocols for genotyping have been previously described (12–15). These mice were in an immunocompetent mixed C57/Bl6×DBA/2J×FVB/n background. All the animal experiments were approved by the Animal Ethical Committee (CEEA) and conducted in compliance with Centro de Investigaciones Energéticas, Medioambientales y Tecnológicas (CIEMAT) guidelines. For histologic analysis, oral samples were fixed in formalin and embedded in paraffin before sectioning, or snap frozen in liquid nitrogen and embedded in optimum cutting temperature (OCT) compound for cryosectioning. Sections were stained and processed as described (12, 13). Antibodies were used as follows: 1:50 dilution of anti–Ser473 phosphorylated Akt (Cell Signaling), anti-bromodeoxyuridine (BrdUrd) (Roche), 1:100 dilution of anti-K15 (NeoMarkers), anti-CD31 (Pharmingen), and anti-CD34 (eBioscience); 1:50 dilution of anti–cyclin D1 (NeoMarkers), 1:200 dilution of p63 (Santa Cruz Biotechnology and Abcam), and 1:50 dilution of anti–Thr58/Ser42 phosphorylated c-Myc (Cell Signaling); 1:250 dilution of anti-Foxo3a (Upstate Biotechnology), 1:50 dilution of anti-CD44 (Pharmingen) and anti-K13 (Abcam); and 1:500 dilution of anti-hemagglutinin (HA) tag (Covance), 1:500 dilution of anti-K5 (Covance) and anti-p53 (Novocastra). Biotin-conjugated secondary antibodies for immunohistochemistry and FITC- or Texas red–conjugated secondary antibodies for immunofluorescence were purchased from Jackson ImmunoResearch Laboratories and used at 1:1,000, 1:50, and 1:500, respectively. For immunohistochemistry, signal was amplified using avidin-peroxidase (ABC elite kit Vector) and peroxidase was visualized using diaminobenzidine as a substrate (DAB kit, Vector). Control slides were obtained by replacing primary antibodies with PBS or preimmune sera (data not shown).

Proliferation measurements in the sections were done by double immunofluorescence using anti-K5 (to detect epithelial cells) and anti-BrdUrd monoclonal antibody (mAb) essentially as described (12, 15–18). For the quantification of BrdUrd, at least five different samples of each type (normal oral epithelium, dysplasia, and squamous cell carcinoma) from at least five different mice of each genotype (only nonlesional oral epithelium was observed in control mice) were scored and represented as mean ± SD.

The analysis of blood vessels and microvessel counting was done as described (15, 19). Briefly, frozen tumor samples embedded in OCT compound (Tissue Tek, Sakura) and sectioned (6 μm thick) were fixed in acetone at −20°C for 5 min and the blood vessels visualized using rat anti-mouse CD31 (Pharmingen; diluted 1:30). Double immunofluorescence was done with anti-K5 to detect the epithelial component as above. Microvessel counting was done in the five areas of greatest vascularization (double blind analysis) in each sample and expressed as an average. For each type of sample (normal oral epithelium, dysplasia, and squamous cell carcinoma), the means of four to six individual experiments from at least three mice of each genotype (only nonlesional oral epithelium was observed in control mice) were scored.

Cell culture. Oral tissues (oral mucosa, tongue, and palate) from adult mice of the different genotypes were digested with 1 unit/mL dispase (Roche) for 3 h at 37°C. Oral epithelium was separated and digested with trypsin (5 min, 37°C) and subsequently inactivated and centrifuged. Cell pellet was resuspended in high-calcium medium and left overnight at 37°C, and then CNT-24 culture medium (Genicell) was added. When cells reached 80% confluence, they were trypsinized and passaged. Experiments with these cells were done when culture had already stabilized and cells were passaged every 3 to 4 d. Proliferation curves were done by seeding 10⁴ cells and counting cell number every 24 h for 5 consecutive days. The experiment was done in triplicate, and mean values and SD were calculated for each time point. Luciferase assays were done on transient transfection of the indicated plasmids using Superfect (Qiagen) according to the manufacturer's recommendations, and cells were harvested 48 h after transfection using Promega Dual-Luciferase Assay Kit. Firefly luciferase values were standardized to Renilla luciferase values to account for differences in transfection efficiency. Transfections were done in triplicate, normalized to the control nontransgenic samples, and the mean and SD were calculated for each condition.

Western blot analysis. Western blot analyses of oral keratinocytes were done as previously described (12, 13).

Senescence assay. For senescence assessment, senescence-associated β-galactosidase staining kit (Cell Signaling) was used following the manufacturer's protocol. Briefly, cryosections were fixed in fixing solution for 15 min, washed with PBS, and then incubated overnight with a staining solution including 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside. Sections were counterstained with diluted eosin. In the case of the cultured keratinocytes, cells were let to grow until they were subconfluent and were then fixed and stained as described above. Cells positive for the marker per field (×10 magnification) were counted, and at least 10 fields per genotype were scored. The mean and SD were calculated for each genotype.

In vivo imaging (positron emission tomography). 2-[¹⁸F]Fluoro-2-deoxy-d-glucose (FDG), 300 μCi (11.1 MBq) in 0.2 mL, was i.v. injected after isoflurane anesthesia (2 min, 3% in 100% oxygen) in mice, which were fasted 3 to 4 h before treatment. Mice were kept awake in cage on a heating pad 30 min before FDG injection and during the uptake period before image acquisition. Small animal positron emission tomography (microPET) imaging scans were obtained by using a microPET scanner (ARGUS RC equipment, SUINSA) and started 30 min after FDG injection, with mice anesthetized by isoflurane inhalation (1.5% in 100% oxygen; Fluovac device, Harvard Apparatus). Imaging acquisition was whole body static during a period of 30 min. Images were reconstructed by 3D-OSEM analysis with the rPET_CT 4.7 Build 441 software (SUINSA and HGUGM Biomedical Research Foundation).

Fluorescence-activated cell sorting analysis. For fluorescence-activated cell sorting (FACS) analysis, oral keratinocytes from the different genotypes were incubated in PBS-EDTA 2 mmol/L for 30 min at 4°C, then treated with trypsin-EDTA for 5 min at 37°C. Cells were then washed in FACS buffer (PBS-1% fetal bovine serum, 0.09% azide) and incubated for 20 min with the indicated primary antibodies or the appropriate control antibodies. When fluorochrome-conjugated antibodies were used, cells were then washed and fixed in 2% paraformaldehyde. Otherwise, cells were washed, incubated with the appropriate fluorochrome-conjugated secondary antibody, and then fixed. Cells were subsequently analyzed in an EPICS XL flow cytometer (Coulter Electronics). Primary antibodies used were rat mAbs anti-CD34 and anti-CD44 (Pharmingen) and rabbit polyclonals anti-CD133 (Abcam) and anti-ΔNp63 (Abcam).

The Akt activation has been previously involved in the development and progression of HNSCC (20–23). To further confirm this, the expression of phosphorylated Akt was studied in tissue microarrays containing 84 HNSCC and 59 oral dysplasias from human patients (not shown; examples are provided in Supplementary Fig. S1). We found a significant number of dysplasias (42 of 59) and tumors (41 of 84) showing the increased Akt activity, suggesting that Akt activation is an early event during human HNSCC development.

To analyze the functions of deregulated Akt activity in vivo, we have recently generated transgenic mice expressing myrAkt in the basal layer of stratified epithelia (12, 13). Several founders and all mice corresponding to one of the established myrAkt lines, all of them characterized by displaying the highest levels of Akt activity (13), developed oral cavity lesions that were easily detectable as oral leukoplakia and erythroplakia in the palate, cheeks, and lips (arrows in Fig. 1A). Microscopic analyses showed hyperplasia of the tongue and palate glands with signs of adenocarcinomatous conversion (Fig. 1A, compare with control in Fig. 1A), in situ carcinomas of the oral mucosa (Fig. 1A), and lip trichoepithelioma (Fig. 1A). We confirmed the expression of the transgene and phosphorylated Akt, indicative of increased Akt activity, in the basal layer of the nonlesional oral epithelia of myrAkt mice (Fig. 1B), which remains in oral dysplasias (Fig. 1B), trichoepithelioma (Fig. 1B), and oral tumors (Fig. 1B). BrdUrd incorporation revealed a mild increase in cell proliferation of myrAkt nontumoral oral epithelia compared with nontransgenic mice (Fig. 2A and B), but we did not find further increase in dysplasias and tumor samples from myrAkt compared with nontumoral tissue (Fig. 2A and B). With respect to the process of epithelial differentiation, which is affected by deregulated Akt activity (12, 13, 22, 24), we detected an altered expression of keratins, with expansion of K5-expressing cells from the basal location into suprabasal compartment and the suprabasal coexpression of K5 and K13 in myrAkt oral epithelia compared with controls (Fig. 2C). Overall, all myrAkt mice develop pretumoral lesions in the oral cavity with age.

Notably, although all the myrAkt mice develop oral lesions; few of them progress into aggressive squamous cell carcinoma (9 of 33). Because deregulated Akt activity may lead to premature senescence (25, 26), we analyzed whether such mechanism might explain the lack of malignant conversion in myrAkt mice. We observed that the areas of malignant tissue display a strong senescence-associated β-galactosidase activity (SA-βgal; “T” in Fig. 3A, n = 10), suggestive that in these regions premature senescence occurred, whereas the hyperplastic and dysplastic areas were predominantly SA-βgal negative (“hyp/dysp” in Fig. 3A). In primary oral keratinocyte cultures, a significant number of enlarged (Fig. 3B) and SA-βgal–positive cells (Fig. 3B) were detected in myrAkt keratinocytes compared with control (Fig. 3B), and myrAkt cultures also displayed a reduced growth rate (Fig. 3C). Akt-induced senescence has been previously associated with the activation of p53-dependent mechanisms (26, 27); in agreement, Western blot and luciferase assays (Fig. 3D) revealed the increased expression of p53 and p53-dependent genes in primary oral myrAkt keratinocytes in parallel with the phosphorylation of Akt and Foxo3a and increased expression of β-catenin and ΔNp63 (Fig. 3D). The mechanisms leading to premature senescence and thus p53 activation may involve overexpression of β-catenin (28), the induction of Pml (29–31), and/or the production of reactive oxygen species due to reduced expression and/or inhibition of Foxo3a (25, 32). Analysis of reactive oxygen species by FACS revealed no major differences between control and myrAkt oral keratinocytes (data not shown). On the other hand, Western blot analysis showed that the increased Akt activity of myrAkt cells is associated with the up-regulation of Pml and β-catenin expression (Fig. 3D), thus suggesting that, in this system, Akt can mediate a premature senescence that is associated with the increased expression of these two proteins.

To functionally show that p53-mediated senescence was responsible for the reduced malignant conversion of oral lesions in myrAkt mice, we bred them with a mouse line bearing p53 loss in stratified epithelial tissues (Trp53^F/F; K14Cre; refs. 14, 15). Whereas somatic ablation of the Trp53 gene in stratified epithelia leads to spontaneous tumor development (15) affecting primarily the epidermis and the oral cavity to a much lesser extent (14 of 57), the Trp53^F/F;K14Cre;myrAkt mice displayed overt malignant tumors in the oral cavity (Fig. 4A), which also affected the palate (Fig. 4A), the ventral side of the tongue (Fig. 4A), and the external and internal sides of the lips (Fig. 4A). Tumor incidence studied in a large cohort of Trp53^F/F;myrAkt (n = 33), Trp53^F/+;K14Cre;myrAkt (n = 20), and Trp53^F/F;K14Cre;myrAkt mice (n = 29) revealed that the complete absence of Trp53 leads also to earlier tumor appearance (Fig. 4B).

The use of mouse models as preclinical tools to asses the efficiency of targeted therapies requires that the tumors generated can be followed by in vivo imaging in individualized manner. This is of a particular relevance in the case of tumors arising in places where they can only be detected after necropsy, such as the internal parts of the oral cavity. Moreover, this may allow a more defined quantification of the tumor growth properties and the presence of secondary tumors and/or metastasis. We thus studied whether the Trp53^F/F;K14Cre;myrAkt mouse tumors can be monitored by positron emission tomography (PET) imaging. On FDG administration, tumors can be easily detected in oral area (arrow in Fig. 4C) and, importantly, the tumor progression at different time points can also be followed in vivo in the same animals (Fig. 4C). In addition, in vivo PET imaging studies also revealed the development of secondary proximal (denoted by yellow arrows in Fig. 4C) and distant tumors (denoted by white arrow in Fig. 4C) in mice with time. Histopathologic examination revealed that these distant tumors corresponded to lymph node metastases (Supplementary Fig. S2). These metastases can be detected only when the primary tumors display undifferentiated highly aggressive characteristics (Supplementary Fig. S2A) at advanced stages. The metastatic cells, detected by the expression of keratin K5, affected primarily the proximal lymph nodes located at the submandibular region (Supplementary Fig. S2B′–D). In addition, K5-positive cells were also detected infiltrating distal mesenteric lymph nodes (Supplementary Fig. S2E) and forming micrometastatic nodules in the lungs (Supplementary Fig. S2F, F′).

We have also generated two other transgenic mouse models. The first one expresses myrAkt and is deficient for pRb in stratified epithelial tissues (Rb^F/F;K14Cre; ref. 18). The somatic loss of Rb1 in stratified epithelia does not lead to spontaneous tumor formation (18, 33). Neither the pathologic characteristics of the oral lesions that developed in myrAkt mice nor the tumor development rate was affected in the absence of pRb in myrAkt mice (Supplementary Fig. S3). The second model bears the simultaneous ablation of Pten (34) and Trp53 suppressor genes in stratified epithelia (Pten^F/F; Trp53^F/F;K14Cre). In contrast to the corresponding parentals, mice lacking both Pten and p53 displayed a progressive weakening and died of yet unknown causes or had to be sacrificed due to ethical reasons (Supplementary Fig. S4A). However, all these animals developed tumors in the oral cavity or lips with complete penetrance by 18 weeks of age (Supplementary Fig. S4B). In some instances, only sporadic tumors, primarily localized in the mouth margins, were observed in mice lacking Pten or p53 (Supplementary Fig. S4C and D). Overall, the gross appearance (Supplementary Fig. S4E), localization, and histologic characteristics of the tumors (Supplementary Fig. S4F–H) in double-deficient mice were indistinguishable from those observed in Trp53^F/F;K14Cre;myrAkt mice. These results show that the activation of Akt, either by myrAkt expression or by Pten ablation, in conjunction with Trp53 (but not Rb1) loss, leads to the development of malignant carcinomas in the oral region in mice.

We next studied whether p53 loss can overcome the myrAkt-induced premature senescence. The expression of SA-βgal was almost negligible in tumors from Trp53^F/F;K14Cre;myrAkt mice (n = 15; Fig. 5A) compared with the expression of this senescence marker in myrAkt tumors (see Fig. 3). Moreover, the proliferative arrest observed in myrAkt keratinocytes did not take place in Trp53^F/F;K14Cre;myrAkt cells (Supplementary Fig. S5). In agreement, proliferation was increased in Trp53^F/F;K14Cre;myrAkt compared with myrAkt oral tumors (Fig. 5A; Supplementary Fig. S6), and Trp53^F/F;K14Cre;myrAkt primary oral keratinocytes did not display significant SAβgal staining (Fig. 5B) or reduce growth rate (not shown). Luciferase experiments also showed that, as expected, the induction of p53-dependent targets was negligible in Trp53^F/F;K14Cre;myrAkt keratinocytes (Supplementary Fig. S7). These data also reinforce the suggestion that deregulated Akt contributes to a premature senescence that prevents further malignant conversion and increased proliferation, and that this process can be evaded by the loss of p53.

Because deregulated Akt activity in keratinocytes can also modulate angiogenesis resulting in increased epithelial tumorigenicity (19), we have also analyzed this process in normal oral mucosa, dysplasia, and tumor samples in myrAkt and Trp53^F/F;K14Cre;myrAkt mice. CD31 staining of normal oral epithelia of control and myrAkt mice revealed increased formation of blood vessels by the expression of constitutively active Akt (Supplementary Fig. S8 and data not shown). CD31 staining also revealed that the number blood vessel was enhanced in dysplasias and tumors compared with normal oral epithelia in myrAkt mice (Supplementary Fig. S8 and data not shown). However, we did not detect significant differences in the number of blood vessels in Trp53^F/F;K14Cre;myrAkt compared with myrAkt samples (Supplementary Fig. S8 and data not shown). Similarly, the altered differentiation of oral epithelia promoted by myrAkt expression (Fig. 2C) was not further aggravated by the subsequent p53 loss (Supplementary Fig. S9). These results suggest that the augmented tumor susceptibility and malignancy mediated by ablation of the Trp53 gene in myrAkt mice is not primarily mediated by altered differentiation, nor did it augment angiogenesis.

Among the most common alterations found in human HNSCC are the activation of EGFR and the subsequent activation of the NFκB and Stat3 pathways (35–37). We found by Western blot analysis (Fig. 5C) a mild increase in EGFR expression in myrAkt and Trp53^F/F;K14Cre;myrAkt keratinocytes compared with control cells. On the other hand, activation of EGFR (as observed by Tyr-phosphorylated EGFR; Fig. 5C) was only observed in transgenic samples, although no significant differences were found between myrAkt and Trp53^F/F;K14Cre;myrAkt cells. With regard to the NFκB pathway, we observed decreased expression of IκBα and increased expression of IκB kinase (IKK)-γ, IKKα, p50, and p65/RelA, suggesting increased NFκB activity. This was further corroborated by luciferase experiments (Fig. 5C). With respect to Stat3 activation, no significant changes were observed in Tyr-phosphorylated, active Stat3 in myrAkt oral keratinocytes compared with controls or in luciferase experiments (Fig. 5C). In contrast, Trp53^F/F;K14Cre;myrAkt keratinocytes showed a remarkable increase in both NFκB and Stat3 pathways (Fig. 5C). Indeed, the expression of IκBα was further decreased, whereas the expression of p50, p65/relA, phosphorylated p65, and Tyr-phosphorylated Stat3 were increased in Trp53^F/F;K14Cre;myrAkt compared with myrAkt cells (Fig. 5C). As above, these findings were further corroborated by luciferase experiments in primary oral keratinocytes using the same Stat3 and NFκB responding elements (Fig. 5C). Together, these data indicated that activation of NFκB and Stat3 takes place in the Trp53^F/F;K14Cre;myrAkt cells independently of EGFR activation, which showed similar levels in Trp53^F/F;K14Cre;myrAkt and Trp53^F/F;myrAkt keratinocytes.

Another common alteration of human HNSCC is the reduced expression of TGFβ type II receptor (TGFβRII; ref. 38). We also observed by Western blot a reduction in TGFβRII expression in myrAkt and Trp53^F/F;K14Cre;myrAkt oral keratinocytes (Fig. 5C). However, when tumors were studied by immunohistochemistry, we observed that the expression of TGFβRII was reduced in Trp53^F/F;K14Cre;myrAkt compared with myrAkt oral tumors (Fig. 5D). Finally, we have previously shown that human HNSCC displaying increased Akt activation are characterized by several alterations such as increased expression of cyclin D1, c-myc, β-catenin, and ΔNp63 and reduced Foxo3a expression (22). Western blot analysis of primary keratinocytes (Fig. 5C), luciferase experiments, and immunohistochemistry of tumor samples (Supplementary Fig. S10) confirmed these alterations in Trp53^F/F;K14Cre;myrAkt tumors and keratinocytes, although no major differences were found between Trp53^F/F;K14Cre;myrAkt and myrAkt samples. Collectively, these data show that mouse Trp53^F/F;K14Cre;myrAkt tumors and primary cells recapitulate and phenocopy most of the molecular alterations found in human HNSCC.

The initiation, growth, recurrence, and metastasis of solid tumors, including HNSCC, have been related to the behavior of a small subpopulation of tumor-initiating or cancer stem cells. In human HNSCC, these cells are characterized by the expression of specific markers such as CD44 and CD133 (8, 9). On the other hand, we have recently shown that myrAkt expression leads to an increased number of adult epidermal stem cells of the hair bulge characterized by K15 and CD34 expression (13). Consequently, we monitored whether the expression of myrAkt may affect stem cell behavior in the tumors and how this may be modulated by subsequent p53 loss. We observed increased expression of K15 and CD34 in parallel with Akt activity (Fig. 6A) in tumors derived from Trp53^F/F;K14Cre;myrAkt mice compared with myrAkt samples (Fig. 6A). We also noticed that the population expressing ΔNp63 in myrAkt tumors (Fig. 6A) is increased due to the expansion into the suprabasal-like compartments in Trp53^F/F;K14Cre;myrAkt samples (Fig. 6A). Finally, we also observed CD44 and CD133 expression in myrAkt and Trp53^F/F;K14Cre;myrAkt tumors (Fig. 6A). As in the case of ΔNp63, the number of cells expressing these cancer stem cell markers is increased in Trp53^F/F;K14Cre;myrAkt samples compared with myrAkt tumors due to the expansion of positive population into suprabasal-like compartments (Fig. 6A,, insets). To further corroborate these observations, we also carried out FACS analysis of primary oral keratinocytes. These showed an increase in the population of cells expressing CD34, ΔNp63, and CD44 (Fig. 6B) in myrAkt cultures, which in the case of CD34 is strikingly increased in Trp53^F/F;K14Cre;myrAkt cells (Fig. 6B). Nonetheless, no significant changes were observed in the population of cells expressing CD44 o ΔNp63 between myrAkt and Trp53^F/F;K14Cre;myrAkt keratinocyte cultures (Fig. 6B), indicating that, in both cases, the observed increase in tumors (Fig. 6B) is limited to the in vivo situation. With respect to CD133, we observed, in agreement with previous data of human HNSCC lines (8, 9), that the number of cells expressing this cancer stem cell marker is very low, although it is augmented in myrAkt (4-fold) and further increased (10-fold) in Trp53^F/F;K14Cre;myrAkt keratinocytes (Fig. 6B). Collectively, these data indicate that in our myrAkt mouse models there is an increase in the population of cells bearing characteristics of epithelial and tumor stem cells, especially in the case of Trp53^F/F;K14Cre;myrAkt mouse.

The recent improvements of therapies, using fractionated radiotherapy, targeted chemotherapy, and concurrent radiotherapy and chemotherapy (1–4), have led only to a moderate increase in the overall survival in HNSCC patients. Molecularly targeted therapies are promising in HNSCC management, and several candidate molecules of potential therapeutic relevance are now being validated through in vitro analyses (6, 10, 11). Therefore, in vivo systems aimed to the analysis of these therapies are necessary. However, these approaches have been hindered by a lack of appropriate animal models mimicking these tumors at both the pathologic and molecular levels. In recent years, several possible mouse models for HNSCC have been described. Nevertheless, the molecular characteristics of these tumors were not systematically compared with human counterparts. The first report of a genetically engineered mouse model was the targeted expression of cyclin D1 transgene to the oral-esophageal epithelium (L2-cyclin D1 transgenic mice; ref. 39). These mice exhibited dysplasia in the tongue, esophagus, and forestomach by 16 months of age but did not develop tumors (39) unless they were crossbred with p53 heterozygous mice, which resulted in invasive oral-esophageal cancer development in L2-cyclin D1/p53^+/− mice (40). More recently, using inducible systems, it has been reported that the activation of an oncogenic K-rasG12D allele in the oral cavity of the mouse induces oral tumor formation in mice, but these lesions were classified as benign squamous papillomas (41). Finally, the TGFβRII deletion in combination with activation of either K-ras or H-ras in mouse head and neck epithelia caused HNSCC (38), in contrast to the targeted ablation of this gene in all stratified epithelia, which results in anal and genital squamous cell carcinomas (42), indicating that progression to cancers occurs rapidly when the TGFβRII-null epithelial tissues are exposed to activated oncogenes and/or loss of additional tumor suppressors.

Here we present a novel and plausible mouse model of human HNSCC to study the consecutive steps involved in initiation and progression that will allow us to address questions like the cell of origin and the role of cancer stem cells in the maintenance of these tumors. In this mouse model, constitutively active Akt leads to dysplasia in oral tissues, showing that the activation of Akt is sufficient to initiate oral carcinogenesis. In agreement, we found Akt activation in a vast majority of human oral dysplasias. Consequently, the myrAkt mice could also be used in preclinical intervention and prevention studies. This is also of a particular relevance for studying radiotherapy, one of the current therapies for HNSCC, because the activation of Akt is associated with intrinsic radioresistance, tumor cell proliferation, and hypoxia, the three major radioresistance mechanisms (reviewed in ref. 43).

Importantly, in myrAkt transgenic mice, aggressive HNSCC are developed only sporadically due to the induction of p53-dependent premature senescence. Although the mechanism by which Akt activity induces premature senescence in oral lesions is unknown at present, the overactivation of β-catenin (28) and the induction of Pml (29–31) are likely candidates. Of note, the expression of Pml is sustained in Trp53^F/F;K14Cre;myrAkt cells (Fig. 4D) whereas the transcriptional activity of β-catenin is slightly reduced in Trp53^F/F;K14Cre;myrAkt keratinocytes (Supplementary Fig. S8). On the other hand, this observation suggests that those therapies targeted to sustain p53 activation may help to prevent the development of aggressive squamous cell carcinomas. We also showed that the loss of p53, but not of pRb, in conjunction with constitutive Akt activity abrogated premature senescence, and that myrAkt expression or Pten ablation in mice bearing the p53 loss in epithelia allows the development of aggressive HNSCC. Similar processes of escape to premature senescence have been previously described for PTEN-deficient prostate tumor formation (26). We also showed that several changes taking place in human HNSCC, and affecting EGFR, NFκB, Stat3, and TGFβRII (6, 7), also occur in the mouse tumors. These observations, together with the fact that these tumors can be followed by in vivo imaging using PET and display some of the pathologic characteristics presented by human lesions, strongly support that this represents a suitable preclinical model for intervention studies in which questions with respect to therapy response and resistance can be addressed (discussed in ref. 43).

Our observations, besides reinforcing the current criteria of premature senescence as a tumor suppression mechanism, also suggest the relevance of putative stem cells in the development of tumors. In agreement with our previous findings (13, 22), we observed that the constitutive Akt activity promotes increased expression of putative adult and cancer stem cell markers, which in particular cases are further increased due to Trp53 ablation. In conclusion, we have developed a mouse model of primary HNSCC that will be useful for future studies with translational applications including molecular and genetic analysis of HNSCC, molecular imaging, and testing of new therapeutic agents.

No potential conflicts of interest were disclosed.

Grant support: Ministerio de Educación y Ciencia grant SAF2005-0003, Comunidad Autónoma de Madrid Oncocycle Program grant S2006/BIO-0232, Ministerio de Ciencia e Innovación grant PS-090100-2006-3, and Ministerio de Sanidad y Consumo grant ISCIII-RETIC RD06/0020 (J.M. Paramio) and NIH grant CA37111, National Institute of Environmental Health Sciences Center grant ES00784, and Cancer Center Support Grant CA16672 (J. DiGiovanni) and Ministerio de Educación y Ciencia grant SAF2007-66227 (M. Garín).

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.

We thank Drs. A. Berns and P. Krimpenfort [Netherlands Cancer Institute (NKI), Amsterdam, the Netherlands] for providing different mutant mice, Jesús Martínez and the personnel of the animal facility of CIEMAT for the excellent care of the animals, and Pilar Hernández (CIEMAT) for the histologic preparations.