The development of oral squamous cell carcinoma (OSCC) is a multistep process that requires the accumulation of genetic alterations. To identify genes responsible for OSCC development, we performed high-density single nucleotide polymorphism array analysis and genome-wide gene expression profiling on OSCC tumors. These analyses indicated that the absent in melanoma 2 (AIM2) gene and the interferon-inducible gene 16 (IFI16) mapped to the hematopoietic interferon-inducible nuclear proteins. The 200-amino-acid repeat gene cluster in the amplified region of chromosome 1q23 is overexpressed in OSCC. Both AIM2 and IFI16 are cytoplasmic double-stranded DNA sensors for innate immunity and act as tumor suppressors in several human cancers. Knockdown of AIM2 or IFI16 in OSCC cells results in the suppression of cell growth and apoptosis, accompanied by the downregulation of nuclear factor kappa-light-chain-enhancer of activated B cells activation. Because all OSCC cell lines have reduced p53 activity, wild-type p53 was introduced in p53-deficient OSCC cells. The expression of wild-type p53 suppressed cell growth and induced apoptosis via suppression of nuclear factor kappa-light-chain-enhancer of activated B cells activity. Finally, the co-expression of AIM2 and IFI16 significantly enhanced cell growth in p53-deficient cells; in contrast, the expression of AIM2 and/or IFI16 in cells bearing wild-type p53 suppressed cell growth. Moreover, AIM2 and IFI16 synergistically enhanced nuclear factor kappa-light-chain-enhancer of activated B cells signaling in p53-deficient cells. Thus, expression of AIM2 and IFI16 may have oncogenic activities in the OSCC cells that have inactivated the p53 system. (Cancer Sci 2012; 103: 782–790)
Oral squamous cell carcinoma is commonly found in low-income communities. This cancer mainly affects older men; 90% of cases are in men over 45 years old who have been exposed to risk factors including tobacco and/or alcohol (International Agency for Research on Cancer [IARC] 2004). OSCC is the sixth most common cancer worldwide and affects approximately 270 000 people each year. The incidence and rate of mortality from OSCC are rising in several regions of Europe, Australia and Asia, including Japan. Despite recent progress in OSCC diagnosis and therapy, the 5-year survival rate has not improved in more than two decades.
Oral carcinogenesis is a multifactorial cascade involving numerous genetic changes that affect the activity of oncogenes, tumor suppressor genes and other classes of disease-related genes. Chronic exposure to carcinogens, such as tobacco, causes genetic changes in the epithelial cells of the oral mucosa. The activation of the COX-2, epidermal growth factor receptor, and cyclin D1 oncogenes and the inactivation of the p16 and p53 tumor suppressor genes have also been reported in OSCC.[5-7] In addition to tobacco smoke exposure, chronic alcohol use and chronic inflammation can both induce genetic alterations. The causative agent of cervical cancer, HPV is also reportedly associated with head and neck cancers, including OSCC. Compared to HPV-negative cases, HPV-positive OSCC have an intact p16 gene and wild-type p53, and harbor frequent genetic alterations of the p16 and p53 genes.[9, 10] The HPV oncoproteins E6 and E7 exploit the ubiquitin-proteasome system to degrade and functionally inactivate negative cell-regulatory proteins, including members of the p110 (Rb) family and p53; this process may primarily contribute to HPV-induced carcinogenesis.
The innate immune system provides nonspecific protection and enhances the adaptive immune response against a variety of pathogens, including HPV. The IFI16 and AIM2 proteins were recently found to be innate immune sensors for cytosolic dsDNA. Upon sensing dsDNA, the IFI16 protein induces the expression of IFN-β, whereas the AIM2 protein forms an inflammasome that promotes the secretion of interleukin-1β. Both IFI16 and AIM2 belong to the HIN-200 gene family found on human and mouse chromosome 1; they are positively regulated by type I and II INF and have been described as regulators of cell proliferation, differentiation, apoptotic and inflammatory processes. The overexpression of IFI16 in cells inhibits cell proliferation by potentiating the p53/p21- and Rb/E2F-mediated inhibition of cell-cycle progression, and IFI16 downregulation contributes to oncogenesis.[15, 16] Also, AIM2 expression suppresses cell proliferation and tumorigenicity of human breast cancer cells. Therefore, it has been proposed that AIM2 and IFI16 function as tumor suppressor genes.
To identify genes involved in OSCC tumorigenesis, OSCC tumors were submitted to genomic analysis by high-density SNP array analysis. A number of amplified or deleted genomic regions in OSCC cells were identified, and a series of genes in the genetically altered regions were selected by expression profile analysis. Of these genes, the NIH-200 gene family locus on chromosome 1q23 was frequently amplified, and AIM2 and IFI16 in the NIH-200 gene family were highly expressed in most OSCC tumors. Although AIM2 and IFI16 were reported to be tumor suppressors, the expression of AIM2 and IFI16 enhanced the cell growth of OSCC cell lines. Here we describe a mechanism by which AIM2 and IFI16 may be functioning as oncogenes in OSCC.
Materials and Methods
Materials and Methods are given in Data S1 and Data S2 in the supporting information.
Higher expression of AIM2 and IFI16 with frequent amplification at 1q23 in OSCC.
To identify novel genes responsible for tumorigenesis in OSCC, we performed high-density SNP array analysis on 28 OSCC tumor samples using an Affymetrix Human Mapping 250K Sty Array (Affymetrix, Santa Clara, CA, USA). The most frequent gains involved segments of chromosomes 1, 3q, 5p, 6p, 7p, 8q, 9q, 14, 15, 16, 17, 19, 20 and 22, whereas the most frequent losses involved segments of chromosomes 3p, 4, 5q, 8p, 10p, 18q and 21q (Fig. S1). Of these, 77 were amplified and ranged in size from 0.3 to 49.3 Mb, and four were found to be deleted and ranged in size from 0.2 to 0.6 Mb, in more than 14 of 28 OSCC tumor samples (Table S1).
To select candidate genes within the regions with altered copy numbers, we analyzed a data set from the NCBI Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/), which contained the gene expression data of four oral tissue samples from healthy volunteers and 16 tumor samples from OSCC patients. Of the genes in the amplified regions, 27 were expressed at more than two-fold higher levels (P < 0.01) (Table S1-1). However, no genes that were downregulated more than two-fold were identified in the deleted region (Table S1 2). Interestingly, 15 of the 27 candidates were previously reported to be cancer-related genes. The genes linked to OSCC included keratin 19, lectin, galactoside-binding, soluble, 1 and IFI16.[19-21] Furthermore, five of the 27 upregulated genes were also found to be IFI genes, including IFI6, IFI35, AIM2, IFI16 and bone marrow stromal cell antigen-2. Among them, the AIM2 and IFI16 genes are located within the HIN-200 gene cluster on 1q23 and have been identified as a new family of innate immune DNA sensors for intracellular DNA called AIM2-like receptors. Because chronic inflammation and infection contribute to the development of several types of cancer, we analyzed the HIN-200 gene cluster locus on 1q23.
Based on the SNP array analysis from 28 OSCC cases, four members of the HIN-200 family, along with other 19 genes, were located in the 1.3-Mb common region of amplification at 1q23 (Fig. 1a). The expression profiles showed that IFI16 and AIM2 are highly expressed in OSCC, but no significant differences in the expression levels of the other 21 genes, including myeloid cell nuclear differentiation antigen (MNDA) and pyrin and HIN domain family, member 1 (PYHIN), were observed between the OSCC and control oral tissues (Fig. S2a–c). Using semi-quantitative and quantitative real-time PCR, we confirmed statistically significant higher expression of AIM2 and IFI16 in tumor samples from OSCC patients and OSCC cell lines (P < 0.05) (Figs 1b–e, S2d). Moreover, the expression of AIM2 was significantly higher in the group with metastasis (N1) than in that without metastasis (N0) (P < 0.05). To confirm the relationship between genomic amplification and mRNA expression levels of AIM2 and IFI16, a scatter plot was used to evaluate the correlation between the two variables in 20 OSCC tumor samples and eight cell lines. As shown in Figure 1(f), we found weak but positive correlations between the DNA copy numbers and mRNA expression levels for the two genes. Although a few cases did not show gene amplification or overexpression of the AIM2 and IFI16 genes, the majority of cases presented gene amplification and high expression of these two genes. Thus, in the HIN-200 family of genes, AIM2 and IFI16, are overexpressed in OSCC, and this overexpression is frequently accompanied by gene amplification.
High expression of AIM2 and IFI16 enhanced cell growth by preventing apoptosis in OSCC cells.
An important inflammasome component, AIM2 senses potentially dangerous cytoplasmic DNA and regulates caspase-1 activation. Cytoplasmic overexpression of AIM2 also reportedly reduces cell proliferation and increases susceptibility to cell death in transfected murine fibroblasts. To determine whether the high expression levels of AIM2 or IFI16 have an effect on OSCC cell growth, we introduced an shRNA expression vector against AIM2 (shAIM2), IFI16 (shIFI16) or luciferase (shLuc) as a control into the human OSCC cell line SAS. SAS cells expressing the shRNA for either AIM2 (SAS/shAIM2) or IFI16 (SAS/shIFI16) had decreased growth rates relative to control-transfected cells (SAS/shLuc) (Fig. 2a,b). Notably, downregulation of both AIM2 and IFI16 expression had the most significant effect on growth inhibition. Similar effects were observed in the HSC4-OSCC cell line (Fig. S3). Next, apoptosis and cell cycle were investigated by flow cytometry with propidium iodide (PI) and Annexin V, respectively. The cell cycle profiles of SAS/shAIM2 and SAS/shIFI16 cells were not significantly different from those of the control SAS/shLuc cells, but the SAS/shAIM2 and SAS/shIFI16 cells exhibited a higher percentage of cell death (sub-G1 population) than the SAS/shLuc cells (Figs 2c, S4). In the SAS/shAIM2 and SAS/shIFI16 cells, the percentage of cells that bound Annexin V increased approximately 10- and 3-fold, respectively, compared with the SAS/shLuc cells (Fig. 2d). These data suggest that high expression of AIM2 and IFI16 enhances cell survival by preventing OSCC cells from entering apoptosis.
Activation of NF-κB signaling by AIM2 and IFI16 in OSCC.
Once bound to the DNA in the cytoplasm, AIM2 activates both NF-κB and caspase-1, and cytosolic DNA also triggers NF-κB activation by IFI16. To clarify the mechanisms by which constitutive expression of AIM2 and/or IFI16 prevent apoptosis in OSCC cells, we studied caspase-1 and NF-κB for constitutive activation in OSCC. Initially, the SAS-OSCC cell lines transfected with the shLuc, shIFI16 or shAIM2 vectors were examined for the expression of cleaved caspase-1 by immunoblot analysis using a cleaved caspase-1 specific antibody. Cleaved caspase-1 was not detected in any of these cell lines but was detected in the human acute monocytic leukemia cell line (THP-1) transfected with poly deoxyadenylic-deoxythymidylic acid (poly[dA:dT]) as a positive control (Fig. S5a). The cleaved form of caspase-1 was also not detected in seven of the eight OSCC cell lines except for HSQ89 (data not shown). In addition, the treatment of SAS-OSCC cells with poly(dA:dT) had no effect on caspase-1 cleavage (Fig. S5b), suggesting that dsDNA could not trigger the formation of the AIM2 inflammasome in OSCC cells.
To assess NF-κB activation in OSCC, we measured IκBα protein in eight OSCC cell lines and 10 primary OSCC tumors by immunoblot analysis. We observed significantly higher levels of phosphorylated IκBα and lower levels of total IκBα in most OSCC cell lines and primary tumor samples compared to control gingival tissues (Fig. 3a). This result suggests that NF-κB signaling is often activated in OSCC. To confirm this hypothesis, four OSCC cell lines (HSQ89, HSC3, HSC4 and SAS) were treated with various concentrations of the NF-κB inhibitor Bay 11-7082 for 48 h and examined for cell viability. In three of the four cell lines (not HSQ89), cell viability was inhibited by Bay 11-7082 treatment, although the effect varied among the cell lines (Fig. 3b). To determine whether the observed cell death was due to apoptosis, two cell lines, SAS and HSC3, treated with or without Bay 11-7082 were stained with Annexin V/PI and analyzed by flow cytometry. Over 90% of the OSCC cell lines underwent apoptosis 48 h after treatment with 10-μM Bay11-7082 (Figs 3c, S6). Moreover, there was a dose-dependent increase in the total IκBα and decrease in the phosphorylated IκBα levels after exposure to 1–10-μM Bay11-7082 (Fig. 3d). To determine whether high expression of AIM2 and/or IFI16 contributes to the constitutive NF-κB activation in OSCC, SAS cells were transfected with either shIFI16 and/or shAIM2 vectors and analyzed for IκBα expression. As expected, the SAS cells treated with shIFI16, shAIM2 or both shRNA significantly increased protein levels of phosphorylated-IκBα and reduced the total IκBα levels compared with control cells transfected with shLuc and parental cells (Fig. 3e). The reduction of NF-κB activation by shIFI16 or shAIM2 was also confirmed by NF-κB-dependent luciferase reporter activity assay (Fig. 3f). The overexpression of IFI16 and AIM2 may enhance IκBα kinase activity and promote the degradation of IκBα and NF-κB activation, leading to the acceleration of cell growth in OSCC cells.
Restoration of p53 function inhibits constitutive NF-κB activation in OSCC cells.
Studies have reported that the overexpression of HIN-200 proteins can decrease cell proliferation and block cell cycle progression at the G1-S phase transition. It has been shown that IFI16-mediated growth arrest is partly dependent on the function of p53. Because a high frequency of mutations in p53 was noted in OSCC, we determined whether p53 dysfunction results in abrogation of the growth suppressive effects of AIM2 and IFI16 in OSCC cells. Initially, we determined the expression and genomic alterations of p53 in eight OSCC cell lines. All cell lines showed p53 point mutations (Table S2-1, Doc. S2) and five (Ca9-22, HSC4, HSQ89, SAS and Sa3) expressed a detectable level of p53 protein, including a truncated form of p53 (Fig. 4a).[20, 21] In primary OSCC samples, five out of 11 tumors had abnormally high expression levels and/or point mutations of p53 (data not shown) (Table S2-2). We introduced the wild-type p53 expression vector into the SAS cell line and confirmed that the expression of wild-type p53 decreased the growth rate of SAS cells (Fig. 4b,c). The percentage of Annexin V-positive cells significantly increased in wild-type p53-transfected cells (Fig. 4d), indicating that this p53-mediated growth suppression of SAS cells is associated with apoptosis.
It has been reported that p53 regulates glucose metabolism through the NF-κB pathway, and both basally expressed and genotoxicity-activated p53 inhibits the activity of IκB kinase and the transcriptional activity of NF-κB. Therefore, we determined whether p53 could inhibit NF-κB activation in OSCC cells. A significant increase in the IκBα protein level and a decrease in phosphorylated IκBα were observed when SAS cells were transfected with wild-type p53 (Fig. 4e). Wild-type p53 also inhibited the basal NF-κB reporter activity of SAS cells (Fig. 4f). These data suggest that constitutive activation of the NF-κB pathway in OSCC is partly due to a loss of p53 function.
The overexpression AIM2 and IFI16 synergistically promotes cell growth only in the absence of functional p53.
To determine whether a lack of functional p53 is associated with the growth-promoting effect of AIM2 and IFI16 in OSCC cells, we transfected human lung cancer H1299 cells, which lack endogenous p53, with various combinations of the AIM2, IFI16 and p53 expression vectors. The transfection of AIM2, IFI16, or p53 alone had no significant effect on the growth rate of H1299 cells (Fig. 5a,b). However, co-transfection of AIM2 and IFI16 strongly promoted cell proliferation with increased IκBα phosphorylation. Strikingly, the growth-promoting effect of AIM2 and IFI16 was abrogated in the presence of wild-type p53. To further confirm these results, we performed cell proliferation assays using the human mammary tumor cell line MCF-7 expressing wild-type p53, which has been used to study the role of IFI16 in p53-dependent apoptosis. Transfection with either or both of the AIM2 and IFI16 expression vectors retarded the growth rate of MCF-7 cells. The shRNA inhibition of p53 expression caused a slight increase in cell growth rate (Fig. 5c,d). Importantly, co-transfection of AIM2 and IFI16 with a shRNA expression vector for p53 led to an approximately two-fold higher proliferation rate with significant upregulation of phosphorylated IκBα. Thus, the simultaneous high expression of AIM2 and IFI16 confers a proliferative advantage in cells with functionally inactive p53, in part, through the activation of NF-κB signaling.
Finally, we examined whether co-expression of both AIM2 and IFI16 can activate NF-κB in an OSCC cell line. Very low expression levels of AIM2 and IFI16 and a low level of NF-κB activation were found in HSQ89 with p53 mutation. Although transfection of either AIM2 or IFI16 alone did not have a significant effect on NF-κB activation and cell growth, co-transfection of AIM2 and IFI16 resulted in accelerated cell proliferation, a decrease in total IκBα, and an increase in phosphorylated IκBα (Fig. 5e,f). This result suggests that co-expression of AIM2 and IFI16 activates NF-κB signaling in OSCC cells. Taken together, these results suggest that co-expression of AIM2 and IFI16 can promote cell proliferation through activation of NF-κB signaling pathway in the absence of p53.
The oral cavity contains some of the most varied and extensive flora in the entire human body. A member of a group of DNA-based viruses, HPV infects the skin and mucous membranes within the human body. Infection with HPV 16 and 18 increases the risk of oral cavity cancer and oropharyngeal cancer. Because the interferon-inducible AIM2 and IFI16 genes act as innate immune sensors for cytosolic double-stranded DNA,[12, 13] the expression of AIM2 and/or IFI16 might be activated by recurrent infections in the oral cavity. In this study, we showed that constitutive high expression levels of AIM2 and IFI16, along with other interferon-inducible genes, were associated with genomic alterations in OSCC. Upregulation of a series of interferon-inducible genes and enhancement of interferon-signaling pathways has previously been reported in OSCC by protein expression analysis using tandem mass spectrometry or by mRNA expression profile analysis using a DNA microarray.[26, 27] The expression of the interferon-inducible gene is an important characteristic of OSCC. Moreover, OSCC cells were shown to become resistant to IFN-β-mediated inhibition of cell growth. The constitutive expression of the interferon-inducible genes may affect the development of OSCC, and the expression of a subset of the interferon-inducible genes may arise as a result of genomic alterations.
Although the AIM2 and IFI16 genes were thought to be tumor suppressors in a series of cancers, high expression levels of AIM2 and IFI16 enhanced the cell growth and prevented apoptosis in OSCC. In normal aged human cells, increased levels of the IFI16 protein reportedly contribute to senescence-associated cell growth arrest partly through the p53/p21CIP1 and Rb/E2F pathway and upregulation of IFI16 expression by p53. This is a part of a component of the positive feedback loop between p53 and IFI16. Because the loss of p53 and/or IFI16 function in cells has been suggested to contribute to defects in cellular senescence-associated cell growth arrest, we speculate that dysregulated IFI16 function together with the loss of p53 function can contribute to the development of OSCC. This study demonstrated that the co-expression of AIM2 and IFI16 synergistically activates NF-κB signaling, and the activation of NF-κB in OSCC is also dependent on p53 inactivation. Future studies will determine the precise activation mechanisms of NF-κB signaling and/or the transcriptional activity of IFI16 in OSCC with inactivated p53.
The authors have no conflict of interest to declare.
absent in melanoma 2
hematopoietic interferon-inducible nuclear proteins with a 200-amino-acid repeat
inhibitor of kappa B alpha
nuclear factor kappa-light-chain-enhancer of activated B cells