Cigarette smoking promotes keratinocyte malignancy via generation of cancer stem-like cells

Objectives: Cigarette smoking is involved in the pathogenesis of head and neck squamous cell carcinoma (HNSCC). However, the underlying molecular mechanisms of cigarette smoking-induced HNSCC carcinogenesis are unclear and may involve cancer stem-like cell generation. We examined the effects of cigarette smoke condensate (CSC) on the formation of cancer stem-like cells, which are rich in octamer-binding transcription factor (OCT)-4, inhibitor of differentiation 1 (ID1), nuclear factor (NF)-κB, and B lymphoma Mo-MLV insertion region 1 homolog (BMI-1). Materials and Methods: We used in vitro, in vivo, and archival human HNSCC tissue analysis to evaluate the effects of CSC on cancer stem-like cell formation. Results: We found that CSC regulated OCT-4 expression, which subsequently regulated ID1 and NF-κB, at the promoter, mRNA, and protein levels in vitro. Furthermore, OCT-4 knockdown with siRNA reduced ID1 expression. ID1 and NF-κB synergistically increased the expression of BMI-1 and stimulated keratinocyte sphere generation. In vivo, ID1 and NF-κB acted together to generate malignant xenograft tumors, which were aggressive locally and systemically metastatic. Clinical data confirmed that ID1- and NF-κB-positive patients had poor clinical outcomes and 5-year disease-free survival. Conclusion: Our data suggest that smoking cigarettes promoted cancer stem-like cell generation in the head and neck area via the OCT-4/ID1/NF-κB/BMI-1 signaling pathway.


Introduction
Tobacco use is considered a significant risk factor for the development of head and neck squamous cell carcinoma (HNSCC) [1] and remains a leading risk factor affecting early death and disability worldwide [2]. Indeed, the majority of patients with HNSCC report a history of cigarette smoking [3].
Statistical analysis has demonstrated that smoking is the most important individual risk factor for many cancers [4] and that cigarette smokers are at a 3-12 times higher risk for developing HNSCC compared with nonsmokers [1]. However, how tobacco use is linked to HNSCC is still unclear. Thus, understanding Ivyspring International Publisher the signaling pathways through which tobacco use promotes HNSCC development is essential for discovering novel potential targets for the treatment of this disease.
Inappropriate activation of developmental transcription factors can stimulate developmental pathways out of context [5]. Such changes in developmental transcription factors are related to the generation of stem-like cells that are able to initiate cancer growth. Octamer-binding transcription factor (OCT)-4 is one such transcription factor capable of inducing pluripotent stem cells (iPSCs) from differentiated somatic cells [6], including keratinocytes [7]. For example, primary keratinocytes transfected with OCT-4, SRY-box-containing protein 2 (SOX2), Kruppel-like factor 4, and c-MYC exhibit production of iPSCs [7]. Moreover, OCT-4 transfection in adult neural stem cells yields embryonic stem-like cells [8]. These iPSCs are tumorigenic and cause teratoma formation. Thus, OCT-4 may participate in the generation of immature cells from differentiated keratinocytes in HNSCC patients.
We recently demonstrated that inhibitor of differentiation 1 (ID1) contributes to the tumorigenesis of keratinocytes via regulation of survivin and the phosphatidylinositol 3-kinase (PI3K)/Akt signaling pathways [9]. Alternatively, high survivin and PI3K/Akt activity is also observed in cancer and embryonic stem cells [10]. However, ID1 alone does not cause metastasis [9].
Nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) has been implicated in control of cell proliferation and oncogenesis in many cancers and, as such, has been identified as a therapeutic target [11]. Based on the contribution of ID1 to the dedifferentiation of somatic cells [12] and the concomitant effects of NF-κB on the proliferation of cells and induction of epithelial-mesenchymal transition (EMT), it is highly plausible that these pathways may function synergistically in HNSCC development and progression. Thus, we hypothesize that the synergy of ID1 and NK-κB may promote the generation of cancer stem-like cells in keratinocytes.
In this study, we examined whether cigarette smoke condensate (CSC) was associated with HNSCC via induction of OCT-4, and whether such induction promoted the development of HNSCC by increasing the expression of ID1, NF-κB, and the cancer stem cell markers, B lymphoma Mo-MLV insertion region 1 homolog (BMI-1) [13] and CD44 [14]. We then assessed whether these signaling pathways affected the development and progression of HNSCC in nude mice.

Ethics approval and consent to participate
The surgical samples and clinical data involved in this study were collected according to the IRB at the University of Minnesota and Sun Yat-sen University. The clinical data and materials are available at the Departments of Otolaryngology, Head & Neck Surgery, University of Minnesota and Sun Yat-sen University. The consent form was signed when surgical samples being collected. These clinical data and materials are appropriate for publication.
Fifty-five HNSCC specimens from the Department of Otolaryngology, Sun Yat-sen University (Supplementary Table S1) were used. Twenty-two HNSCC specimens and 12 control tissues (obtained from regions near HNSCC tissues) from the Department of Otolaryngology, University of Minnesota Hospitals and Clinics were used after obtaining written informed consent from patients for research purposes. All specimens and clinical data in this study were procured, handled, and maintained according to the protocols approved by each Institutional Review Board (IRB#1111A07101).

Induction of xenograft tumors in nude mice with Rhek-IP cells
Cells stably transduced with an empty vector, ID1, NF-κB (p65), or ID1+NF-κB p65 (IP) for up to 6 months were sorted using a FACSAria cell sorter (BD Biosciences). Then, cells expressing high levels of green fluorescent protein were selected and expanded in culture. Athymic nude mice (approximately 16-18 g, n = 6/group) were subcutaneously injected with 1 × 10 6 cells in their bilateral flanks. After injection, tumor volumes were measured weekly for up to 24 weeks (average: 23.4 weeks). Xenograft tumors in nude mice were harvested, and their sizes, volumes, and weights were measured. Luciferase-positive xenografts were detected with a bioluminescence detector (Xenogen, IVIS; Caliper Life Sciences, Alameda, CA) using standard protocols. Similarly, OTC-4-transduced cells and CD44-positive cells were injected into nude mice to test whether molecules up-and downstream of ID1 and p65 promoted the growth of xenograft tumors. Animal experiments were performed according to a protocol approved by the Institutional Animal Care and Utilization Committee (IACUC ID# 1402-31329A).

Luciferase assays
Construction of the ID1 reporter was performed as follows: the sequence for the human ID1 promoter (-1,000 to -1,024 bp including both the KpnI endonuclease site at the 5′-end and the HindIII endonuclease site at the 3′-end) was amplified from human genomic DNA by polymerase chain reaction (PCR) using the following primer pair: 5′-atggccGGTACCgaccagtttgtcgtctccatggcg-3′ and 5′-gacaagctgtggctccgcactctcAAGCTTggcgag-3′. The PCR-amplified product was subcloned into pGL4 vectors (Promega, Madison, WI, USA) according to the manufacturer's instructions. CD44 and MMP-9 reporters were constructed using a method similar to that described above. The NF-κB and OCT-4 promoters were gifts from Dr. Frank Ondrey at the University of Minnesota.
Cells were transduced the next day with the empty vector or OCT-4 plasmids at 1.4 μg/mL and then cotransduced with ID1, NF-κB, CD44, and MMP-9 reporters at 1.4 μg/mL for 16 h in transfection medium. A β-galactocidase reporter was used as a control for transfection efficiency. Cells were harvested for luciferase assays, as previously described [9].

Statistical analysis
Student's t-tests were used for evaluation of differences between controls and experimental conditions in vitro. The Kaplan-Meier survival test was used for evaluation of disease-free survival times according to ID1 and p65 protein expression status. Results with two-sided p-values of less than 0.05 were considered significant.

OCT-4 was extensively expressed in HNSCC cell lines and HNSCC specimens in association with ID1 and CD44
CSC regulated OCT-4 expression in Rhek-1A and CA9-22 cell lines in vitro (Supplementary Figure  S1). To evaluate whether OCT-4 was expressed in HNSCC, 22 clinical specimens and five HNSCC cell lines were analyzed via RT-PCR, immunohistochemistry, and enzyme-linked immunosorbent assays (ELISAs). Notably, OCT-4 mRNA transcripts were highly expressed in HNSCC cell lines and weakly expressed in noncancerous cell lines ( Figure 1A). Similarly, OCT-4 mRNA was detected in clinical HNSCC specimens but was absent in normal tissues ( Figure 1B). Immunohistochemistry showed that 18 of 22 HNSCC specimens (82.0%) were positive for nuclear OCT-4 (active form), whereas seven of 12 normal tissues (58.3%) were positive for cytosolic OCT-4 (inactive form, Figure 1C). Moreover, OCT-4 was found to be active in the nuclei of HNSCC cells, yet inactive in the cytosol of normal control cells (Supplementary Figure S2). OCT-4 protein was significantly upregulated in HNSCC tissues compared to control tissues, as determined by ELISA ( Figure 1D). Note that other stem cell markers (Sox2, Nanog, and BMI-1) were rarely expressed in HNSCC tissue samples.

OCT-4 increased the promoter activities of ID1 and NF-κB in immortalized keratinocytes and HNSCC cell lines
OCT-4 cDNA and an ID1 reporter gene were constructed to evaluate the effects of OCT-4 on ID1. OCT-4 transfection in Rhek-1A cells increased the transcription of OCT-4 compared to cells transfected with the empty vector (Supplementary Figure S3). Additionally, OCT-4 transient transfection in Rhek-1A and SCC11A cells significantly increased the promoter activity of ID1 compared to cells transfected with empty vector (Figure 2A, B).
Next, to examine whether ID1 mRNA was upregulated, RT-PCR was performed on 3-day cultures after transient transfection. OCT-4 transfection in Rhek-1A cells increased the expression of ID1 mRNA transcripts, which was associated with an increase in BMI-1 transcription ( Figure 2C). FACS analysis showed that OCT-4 transfection for 4 days significantly increased the proportions of ID1-and NF-κB-positive CA9-22 cells ( Figure 2D) and HOK16B cells ( Figure 2E; p < 0.05).

OCT-4 and ID1 regulated CD44 expression in Rhek-1A and HNSCC cells
To examine whether OCT-4 regulated CD44, an HNSCC stem cell marker, luciferase assays were performed. OCT-4 significantly increased the promoter activity and mRNA expression of CD44 in Rhek-1A, CA9-22, and NA cells (Supplementary Figure S4a). Similar results were observed in Rhek-1A cells by qPCR (Supplementary Figure S4b). ID1 significantly increased the promoter activity and expression of CD44 in cells, as demonstrated by luciferase assays and FACS analysis, respectively (Supplementary Figure S4c).

OCT-4 siRNA inhibited the expression of ID1 in NA cells
To verify that OCT-4 regulated ID1 expression, OCT-4 was knocked down using siRNA in NA cells. OCT-4 siRNA reduced the proportion of OCT-4-positive NA cells from 80.3% to 56.7% ( Figure  3A) and simultaneously reduced the proportion of ID1-positive cells from 89% to 69% ( Figure 3B). RT-PCR confirmed that siRNA (A-B) specifically knocked down OCT-4 mRNA transcripts in NA cells ( Figure 3C).

IP synergistically induced the migration of keratinocytes and activity of MMP-9 and BMI-1 in vitro
To determine whether IP regulated the migration of keratinocytes, Matrigel assays were performed. Notably, IP markedly increased the migratory activity of keratinocytes in Matrigel compared to cells transfected with the empty vector ( Figure 3D). To evaluate whether this process was associated with an increase in MMP-9 activity, gelatin zymography was performed. We found that IP enhanced MMP-9 activity compared to cells transfected with the empty vector ( Figure 3E), and this phenotype was associated with an increase in MMP-9 mRNA ( Figure 3F). Furthermore, luciferase assays demonstrated that IP significantly increased the promoter activity of MMP-9 on days 2-3 compared to cells transfected with empty vector ( Figure 3G). FACS analysis revealed that IP upregulated CD104, yet did not alter CD24 and CD133 expression ( Figure 3H). In addition, analysis of cell migration on chamber slides using scratch assays demonstrated that transient transfection with IP for 30 h resulted in nearly full closure of the cell monolayer scratch compared to cells transfected with an empty vector ( Figure 3I).

ID1 and NF-κB p65 synergistically induced the generation of naïve keratinocyte spheres in vitro and metastatic xenograft tumors in nude mice
To assess the importance of IP in the generation of naïve keratinocyte spheres, IP was stably transfected into Rhek-1A cells for evaluation of their synergistic effects on the formation of cellular spheres, the expression of self-renewal markers, and the growth of tumors in animal models. The results showed that IP synergistically increased the formation of keratinocyte spheres in Rhek-1A cells compared to cells transfected with empty vector ( Figure 4A, left panel). FACS analysis showed that IP significantly increased the proportion of BMI-1-positive cells compared to cells transfected with the empty vector, ID1, or NF-κB ( Figure 4A Figure 4C).
To visualize the metastatic process, Rhek-1A cells were stably transfected with ID1 and NF-κB and simultaneously labeled with bioluminescence (luciferase). The cells were then injected into the flanks of four nude mice. The results showed that IP-transfected cells grew xenograft tumors in three of the four nude mice, and two of these mice developed metastatic tumors and malnutrition (Supplementary Figure S6b).

Expression of ID1 and NF-κB in patients with HNSCC was associated with poor clinical outcomes
To verify the importance of IP expression in the clinical setting, we employed an independent set of 55 HNSCC specimens (Supplementary Table S1), stained for IP expression by immunohistochemistry, and evaluated the correlations between IP expression and clinical parameters/outcomes. Using Log-rank (Mantel-Cox) analysis, we observed significant differences in disease-free survival (χ 2 = 2.66, p = 0.077) and lymphatic node metastasis (p < 0.05) between IP + and IPpatients.

Discussion
In this study, we demonstrated for the first time, that CSC was linked to the expression of OCT-4, BMI-1, and CD44, which are expressed in either immature keratinocytes [16,17] or HNSCC stem-like cells [18]. After overexpression of OCT-4, keratinocytes expanded rapidly and formed spheres in vitro but were unable to trigger xenograft tumor growth in nude mice. Thus, other signals or cofactors from the cancer stem cell niche that modify or affect transactivation of NF-κB or ID1 are needed to transform keratinocytes. Indeed, CSC alone did not transform keratinocytes, but rather potentiated the transformation when combined with other carcinogens. However, when OCT-4 downstream molecules (e.g., ID1 and NF-κB) were overexpressed, cells were capable of initiating xenograft tumors in nude mice and metastasizing in vivo.
Malignancy occurs in non-tumorigenic Rhek-1A cells via two developmental transcription factors (ID1 and NF-κB), resulting in cell dedifferentiation coupled with the expression of BMI-1 and CD44. Consistently, ID1 has been shown to be involved in cell dedifferentiation [9], and NF-κB has been shown to be involved in cell proliferation and EMT [18]. Importantly, both ID1 and NF-κB are highly regulated in 65-75% of HNSCC cases with a history of smoking [19]. In this study, we found that ID1 and NF-κB were important regulators of stem cell markers, such as BMI-1 and CD44, in keratinocytes. The former is considered a self-renewal marker for many stem cells [20]. In contrast, CD44 is a specific HNSCC cancer stem cell marker [14].
Experimentally, the synergistic effects of ID1 and NF-κB expression yielded metastatic xenograft tumors in nude mice. This clearly suggested that smoking cigarettes may promote the generation of cancer stem-like cells in the head and neck area via regulation of multiple stem-like cell markers. Moreover, animal studies demonstrated that within 2 months, cells stably transfected with ID1 and NF-κB caused aggressive growth of xenograft tumors in nude mice and induced severe malnutrition. Furthermore, approximately 50% of animals exhibited wasting syndrome (cachexia). Apparently, the synergistic effects of ID1 and NF-κB ultimately duplicated the clinical process of disease and mimicked the malignant behaviors of HNSCC. Coincidently, approximately 50% of patients with HNSCC suffer from recurrence and metastasis [14].
In addition to the upregulation of CD44 and BMI-1, ID1, and NF-κB synergistically regulated the enzymatic activity of MMP-9, which is highly expressed in HNSCC [21] and may be involved in the metastasis of HNSCC [20]. Thus, synergy between ID1 and NF-κB may be responsible for the malignant behaviors of HNSCC. Indeed, OCT-4, CD44, ID1, and NF-κB alone contribute to the carcinogenesis of HNSCC [14]. However, none of these targets alone were able to induce metastasis of xenograft tumors in nude mice, as shown in this study and a prior study [9]. In this study, OCT-4 was not capable of replacing the combined effects of ID1 and NF-κB owing to its dual functions in keratinocyte dedifferentiation or differentiation.
As shown in this study, ID1 and NF-κB alone induced the growth of xenograft tumors that were limited in size and were nonmetastatic in nature. However, synergy between ID1 and NF-κB induced the aggressive growth of xenograft tumors and promoted metastasis of tumors in nude mice. Hence, ID1 and NF-κB together transformed keratinocytes in such a way that the cells became naïve and metastatic. The metastasis of transformed keratinocytes may be related to the expression of MMP-9, which was found to be synergistically upregulated by ID1 and NF-κB. MMP-9 promotes cell migration by degrading collagens and other extracellular matrix proteins in the matrix. However, owing to the small number of animals and differences between xenograft tumors and in situ human tumors, further human studies, especially the concentration of CSC in the patients' blood, are warranted to verify our observations.
In summary, our data suggest that CSC promoted the generation of cancer stem-like cells in the head and neck through OCT-4 signaling. Subsequently, OCT-4-induced synergistic action of ID1 and NF-κB triggers the expression of important head and neck cancer stem cell markers, BMI-1 and CD44.