Squamous cell carcinomas (SCCs) originate in stratified epithelia, with a small subset becoming metastatic. Epithelial stem cells are targets for driver mutations that give rise to SCCs, but it is unknown whether they contribute to oncogenic multipotency and metastasis. We developed a mouse model of SCC by targeting two frequent genetic mutations in human SCCs, oncogene Kras G12D activation and Smad4 deletion, to mouse keratin 15-expressing (K15 + ) stem cells. We show that transgenic mice developed multilineage tumors, including metastatic SCCs. Among cancer stem cell-enriched (CSC-enriched) populations, those with increased side population (SP) cells correlated with epithelial-mesenchymal transition (EMT) and lung metastasis. We show that microRNA-9 (miR-9) contributed to SP expansion and metastasis, and miR-9 inhibition reduced the number of SP cells and metastasis. Increased miR-9 was detected in metastatic human primary SCCs and SCC metastases, and miR-9-transduced human SCC cells exhibited increased invasion. We identified α-catenin as a predominant miR-9 target. Increased miR-9 in human SCC metastases correlated with α-catenin loss but not E-cadherin loss. Our results demonstrate that stem cells with Kras G12D activation and Smad4 depletion can produce tumors that are multipotent and susceptible to EMT and metastasis. Additionally, tumor initiation and metastatic properties of CSCs can be uncoupled, with miR-9 regulating the expansion of metastatic CSCs. IntroductionSquamous cell carcinomas (SCCs) are derived from stratified epithelia present within the skin and oral cavity. A subset of aggressive SCCs become metastatic and lead to metastasis-associated death. The rate of metastasis in skin SCCs ranges from 0.1% to 10% (1), with poorly differentiated tumors and those with greater vertical tumor thickness having an increased risk of metastasis (2). Genetic alterations and intrinsic tumor cell properties controlling SCC metastasis are largely unknown. Genetically engineered mice provide a powerful tool for dissecting driver mutations that contribute to SCC initiation and metastasis. To date, very few genetic mutations causing spontaneous SCC formation and metastasis have been found, particularly metastasis to the lung, which is the leading cause of SCC-associated death (3). Mice with a Smad4 deletion in stratified epithelia develop spontaneous SCCs in the skin, oral cavity, and forestomach (4-6). Among these models, oral SCCs metastasize to lymph nodes (4), whereas skin and forestomach SCCs do not metastasize (5, 6).Because stratified epithelia undergo constant self-renewal and rapid turnover, it is believed that driver mutations for SCCs must initially occur in resident stem cells that renew these epithelia throughout life. In mouse skin, the hair follicle bulge harbors

Because increased transforming growth factor-␤ (TGF␤) production by tumor cells contributes to cancer progression through paracrine mechanisms, identification of critical points that can be targeted to block TGF␤ production is important. Previous studies have identified the precise signaling components and promoter elements required for TGF␤ induction of TGF␤1 expression in epithelial cells (Yue, J., and Mulder, K. M. (2000) J. Biol. Chem. 275, 30765-30773). To determine how regulation of TGF␤3 expression differs from that of TGF␤1, we identified the precise signaling pathways and transcription factor-binding sites that are required for TGF␤3 gene expression. By using mutational analysis in electrophoresis mobility shift assays (EMSAs), we demonstrated that the c-AMP-responsive element (CRE) site in the TGF␤3 promoter was required for TGF␤-inducible TGF␤3 expression. Electrophoresis mobility supershift assays indicated that CRE-binding protein 1 (CREB1) and Smad3 were the major components present in this TGF␤-inducible complex. Furthermore, by using chromatin immunoprecipitation assays, we demonstrated that CREB-1, ATF-2, and c-Jun bound constitutively at the TGF␤3 promoter (؊100 to ؉1), whereas Smad3 bound at this site only after TGF␤ stimulation. In addition, inhibition of JNK and p38 suppressed TGF␤ induction of TGF␤3 transactivation, whereas inhibition of ERK and protein kinase A had no effect. Small interfering RNA-CREB1 and small interfering RNA-Smad3 significantly inhibited TGF␤ stimulation of TGF␤3 promoter reporter activity and TGF␤3 production. Our results indicate that TGF␤ activation of the TGF␤3 promoter CRE site, which leads to TGF␤3 production, is required for TGF␤RII, JNK, p38, and Smad3 but was independent of protein kinase A, ERK, and Smad4.

No abstract