MicroRNA-211 Enhances the Oncogenicity of Carcinogen-Induced Oral Carcinoma by Repressing TCF12 and Increasing Antioxidant Activity

Exposure to carcinogenic substances or viruses is the major etiologic factor of head and neck squamous cell carcinoma (HNSCC) including oral squamous cell carcinoma (OSCC; refs. 1–3). Many factors, such as arecoline, are oxidative inducers (3). The five-year survival of OSCC remained low and this malignancy tended to relapse after treatment (4). Therefore, it is important to specify the molecular dysregulation in the regulatory axis contributive to OSCC pathogenesis to develop therapeutic modalities (5). miRNAs are small, noncoding RNAs of 19–25 nucleotides, which regulate physiologic process or pathogenesis by targeting the mRNA to cause transcriptional repression or mRNA degradation (6, 7). A number of disruptions in miRNA–target gene regulatory axes in OSCC have been discovered (8–13).

miR-211 promoted OSCC oncogenicity and served as an indicator for poor OSCC survival (14). We also showed that miR-211 targets TGFßRII, which is able to upregulate c-myc expression in HNSCC (15). miR-211 also functions as an oncogenic miRNA in colorectal cancer by targeting CHD5 (16). In addition, miR-211 promotes cell growth by targeting the p53-induced loc285194 long noncoding RNA (lncRNA) in colorectal cancer (17). A recent study has also specified that miR-211 prevents the ER induction by transcriptional repression of C/EBP homologous protein (CHOP) to modulate the survival of cells (18). Despite that miR-211 is a tumor suppressor in melanoma and some other cancers (19–21), the function of miR-211 as an oncogenic molecule has become more evident in a fraction of malignancies (15–18).

Transcription factor 12 (TCF12; ref. 22), belongs to class I helix-loop-helix (HLH) protein family known as E protein, which can bind to DNA E-box site (23, 24). TCF12 was shown to regulate the differentiation of lymphocytes or the development of neural or mesenchymal tissues (25–28). Recurrent mutations in TCF12 gene or the translocation fusion of a fragment of TCF12 with other molecules contribute to craniosynostosis or mesenchymal malignancies (29, 30). In colorectal cancer, TCF12 expression correlated with metastasis by repression of E-cadherin (22). TCF12 has been reported as targeted by miR-154 and miR-211 in melanoma cells and other types of normal cells (20, 31, 32). Nevertheless, the roles of TCF12 in OSCC and other cancers are still obscure.

Reactive oxygen species (ROS) are intracellular chemical species that contain oxygen. Accumulated ROS causes oxidative stress and may induce cytotoxicity in cells (33), antioxidants are considered to be suppressors against cancers, as cancer cells seem to possess a higher tolerance of ROS than normal cells. However, recent evidence showed that the upregulation of antioxidant protein Nrf2 might promote survival and resistance to therapies in cancer cells (34). A subpopulation of HNSCC cells carrying low ROS may exhibit more stemness and chemoresistance properties (35).

Peroxiredoxin (PRX) and thioredoxin (TRX) are two related families of antioxidant proteins. PRXs uptake H₂O₂ to change into oxidized form. The oxidized PRXs are then reduced by TRXs (36, 37). Family with sequence similarity 213, member A (FAM213A) was discovered as one of the members in the PRX-like subfamily. Moreover, it also possesses an essential domain for the activation of TRX (36). FAM213A was originally identified during fetal liver development and was activated in M-CSF–stimulated monocytes (38). It was later found to protect cells from oxidative stress and modulate osteoclast differentiation (39). A recent study indicated that FAM213A could be one of the candidate antioxidants beneficial for high-altitude adaptation in Andean people (40).

4-nitroquinoline 1-oxide (4NQO) is a water-soluble carcinogen, which breaks DNA and induces ROS (41, 42). The murine 4NQO tongue carcinogenesis has become a powerful model to address oral carcinogenesis (13, 15). To address the oncogenic roles of miR-211 in the animal model, we generated miR-211 Tg mouse lines driven by the K14 promoter, which were tagged with enhanced GFP (EGFP). We identified the enhancement of OSCC progression by 4NQO induction in this mouse model. Furthermore, we identified transcription factor TCF12 as a new target of miR-211 in OSCC cells. The enhanced FAM213A expression mediated by the repression of TCF12 through miR-211 expression reinforces OSCC oncogenesis and protects cells from the oxidative damages.

OSCC cell lines SAS, OECM1, HSC3, FaDu, OC4, and SCC25; 293T cells and phoenix package cells; and six primary OSCC cells isolated from different tumors were used. Cell lines were achieved from ATCC or JCRB cell banks or derived according to previous protocols during 2012–2014 (11, 13, 15). All cell lines were authenticated by short tandem repeat analysis. The cultivation conditions are described in Supplementary Table S1. SAS-miR-211, OECM1-miR-211, and control cells expressing GFP were established previously (14). The treatment conditions of miR-211 mimic/inhibitor/control (Applied Biosystems) were 60 or 30 nmol/L for 48- or 72-hour treatment. The dose of siTCF12 (Dharmacon) and siFAM213A (Santa Cruz Biotechnology) oligonucleotides were 60 nmol/L for 30 or 48 hours (Supplementary Table S2). The dose of 4NQO treatment was 5 μmol/L. All chemicals were purchased from Sigma-Aldrich. siRNAs and TaqMan assay probes are described in Supplementary Tables S2 and S3.

Phenotypes including proliferation, migration, wound healing, invasion, anchorage-independent growth (AIG), and ALDEFLUOR assay followed protocols previously published (11, 13, 43, 44).

The murine pri-miR-211 sequence and EGFP were cloned to establish Tg mouse lines in C57BL/6 (Supplementary Table S4; ref. 13). For genotyping, genomic DNA isolated from mouse tail tip was used for PCR and Southern blot analysis. The RNA isolated from mouse ear was used for gene transcription analysis (15). Other details are described in Supplementary Methods. 4NQO (100 μg/mL) was added in drinking water of 6- to 8-week-old mice for 16 weeks. Mice were sacrificed at the time point when body weight loss >1/3, or at the defined time points (13, 41).

SAS cells (3 × 10⁵) were injected into the central portion of the tongue of BALB/c athymic mice (National Laboratory Animal Center, Taipei, Taiwan). The mice were sacrificed at the third week after inoculation. The primary tongue tumors and neck region were photographed under the Illumatool Bright Light System (LT-9500; TLS) to visualize the positive tumor and nodes with green fluorescence. Tongue and neck tissues were subjected to histopathologic evaluation. SAS cells (5 × 10⁵) were injected into the flank of BALB/c athymic mice. The tumors were measured every week and the mice were sacrificed at the sixth week after inoculation. The tumor volumes were calculated by the formula = 0.5 × a × b² using parameters measured by microscale under light microscopy or gauge (8). a, the longest diameter; b, the shortest diameter. All animal studies were done in accordance with the guidelines of National Yang-Ming University Institutional Animal Care and Use Committee (IACUC).

The TCF12 coding sequence (CDS), TCF12 CDS plus 3′UTR (WT), and TCF12 CDS plus mutated 3′UTR were cloned into the pBabe-puro vector to produce retroviral constructs (Supplementary Table S4). The TCF12 expression cells were designated as CDS, WT, and MUT, together with VA (vector alone) control. Short hairpin shTCF12 constructs (Supplementary Table S5) packed in lentiviruses were purchased from the RNA interference consortium (Academia Sinica, Taipei, Taiwan). Cell exhibiting the knockdown of TCF12 were designated b4 and b5 together with shLuc (control).

Primary OSCC tumors together with paired noncancerous matched tissues (NCMT; Supplementary Table S6) were collected for qRT-PCR and Western blot analysis. The OSCC TMAs encompassing paired NCMT/OSCC tissue cores or OSCC tumor cores only (Supplementary Table S7) were fabricated to carry out IHC and in situ hybridization (ISH) analysis (45). Detailed methods are described in Supplementary Methods and Supplementary Table S8. This study was approved by the Institutional Review Board with approval no. 2013-11-011B and 12MMHIS177. Written informed consents were obtained from participants.

GeneChip Human Genome U133 Plus 2.0 arrays (Affymetrix) were used. Qualified RNA was submitted to National Yang-Ming University VGH Genome Research Center (VYMGC) for GeneChip analysis. The accession number is GSE70186.

Jaspar (http://jaspar.genereg.net), an open-access database of the transcription factors' binding preferences in multiple species (46), was used to predict potential transcription factor–binding sites.

The data were shown as mean ± SE. t-test, Mann–Whitney test, X² test, two-way ANOVA test, linear regression analysis, and Kaplan–Meier survival analysis were used to compare the differences among variants.

We generated the mmu-miR-211–based Tg mouse model. The schematic diagram (Fig. 1A, a) illustrates the K14-EGFP-miR-211 transgene construct, which results in constitutive expression of EGFP and miR-211 driven by K14 promoter in squamous cells. The characterization of these Tg mouse lines is shown in Supplementary Fig. S1. Increase in the thickness of cell layers, as well as increased expression of Ki67 and Bcl-xL are seen in the squamous epithelium of Tg mice (Supplementary Figs. S2 and S3). We induced tumorigenesis by adding 4NQO in the drinking water (Fig. 1A, b). The treatment successfully induced tumors on the tongue surface, in the esophagus, and occasionally on the palate or buccal mucosa. The tumors with intensified green fluorescence on dorsal tongue were identified easily in Tg mice. Moreover, the tumors in unopened esophagi were discernable rather readily due to their intensive fluorescence (Fig. 1A, c). The incidence of tongue tumor number and tumor size were significantly higher in Tg mice than Wt mice (Fig. 1A, d). Histopathologic evaluation showed epithelial hyperplasia or dysplasia in normal looking tongue mucosa (Fig. 1B, a, top left and middle panels). Tissue sections of exophytic lesions showed pathogenesis varied from squamous cell papilloma (SCP), moderate to severe epithelial dysplasia (Dys), squamous cell carcinoma with submucosal invasion (SCC-S) to SCC with muscle invasion (SCC-M; Fig. 1B, a). The quantitation revealed increased severity of squamous pathogenesis after increasing the transgene dosage (Fig. 1B, b). Tg mice exhibited shorter survival than Wt mice for 1.7 weeks (Fig. 1B, c). The results suggest that the K14-EGFP-miR-211 Tg mice have higher susceptibility to 4NQO for oral tumor induction than Wt mice. The esophageal tumor induction is shown in Fig. 1A, c and Supplementary Fig. S4.

Orthotopic xenograft model of SAS-miR-211 was further adopted to address primary tumorigenesis and locoregional metastasis in nude mice. The xenograft tumors in tongue and neck metastatic lesions were demonstrated by fluorescence image and histopathologic analysis (Fig. 1C, a, b and Supplementary Fig. S5). miR-211 expression was associated with higher tumor growth and higher percentage of nodal metastasis (Fig. 1C, c, left and middle). In the primary tumors subset exhibiting a size <10 mm³, the locoregional metastasis of SAS-miR-211 xenografts were also more potent than controls (Fig. 1C, c, right). miR-211 expression rendered higher neck metastasis of OSCC.

To examine whether oncogenic factors can stimulate the miR-211 expression, we treated OSCC cells with 4NQO or arecoline. 4NQO treatment for 48 hours resulted in miR-211 upregulation in OSCC cells (Fig. 2A, a, left). The induction was mediated by the increase of pri-miR-211 transcript (Supplementary Fig. S6). With the treatment of various doses of arecoline for 24 hours, miR-211 expression was also upregulated in SAS cell (Fig. 2A, a, right). mRNA expression and reporter activity of potential miR-211 targets CDH4, HAS2, KLLN, and UBA2 predicted by in silico modules and TCF4, TCF12, and TGFβRII known targets were assayed (Fig. 2A, b and c). The mRNA expression and reporter activity of HAS2, KLLN, TCF4, TCF12, and TGFβRII were decreased in OSCC cells with exogenous miR-211 expression. However, 4NQO treatment only consistently downregulated TCF12 in OSCC cells (Fig. 2A, d). Downregulation of TCF12 mRNA and protein expression were found in SAS cells with exogenous miR-211 expression (Fig. 2A, e). FaDu, HSC3, and OECM1 cells had lower TCF12 expression than SAS cell, but higher miR-211 expression than SAS (Fig. 2B, a). Supplementary Table S9 illustrates the complementarity between the TCF12 3′UTR sequence and the miR-211. We generated a wild-type TCF12 3′UTR reporter (wt) and a mutant reporter (mut). Reporter assays in OSCC cells indicated that miR-211 repressed the reporter activity of TCF12 by directly targeting the wild-type 3′UTR sequence, and that the mutation relieved the repression (Fig. 2B, b). Downregulation of TCF12 expression in the squamous epithelium of Tg mice relative to the Wt mice was noted (Fig. 2B, c). In oral epithelium from Wt to Tg/+, then to Tg/Tg mice, progressive increase of miR-211 expression was noted. However, the nuclear TCF12 expression gradually decreased (Fig. 2B, d and Supplementary Fig. S7). In OECM1, the CDS and the MUT had higher TCF12 expression than the WT. The WT had slightly higher TCF12 expression than the control. Exogenous miR-211 expression had more profound repression of TCF12 in the WT than the MUT (Fig. 2C, a and b). HSC3 had the highest miR-211 expression and lowest TCF12 expression. The CDS had much higher TCF12 expression than WT. (Fig. 2C, a). The results implicated that the 3′UTR sequence can create an opportunity for miR-211 repression, which substantiated the targeting of miR-211 on TCF12. The efficacy of shTCF12 constructs was validated in SAS (Fig. 2C, c). Stable shTCF12 cells b4 and b5 were established in SAS and OECM1 (Fig. 2C, c, d). To further confirm the repression of TCF12 by miR-211, we treated SAS cells with miR-211 inhibitor. The upregulation of TCF12 protein mediated by miR-211 inhibition was rescued by the knockdown of TCF12 (Fig. 2D, a, top). The downregulation of TCF12 expression in SAS-miR-211 cell was also rescued by miR-211 inhibitor (Fig. 2D, a, bottom). The analysis of nuclear extract and the cytosolic fraction indicated that the vast majority of cellular TCF12 localized in nuclei and its levels corresponded to the fluctuation of miR-211 levels being modulated by miR-211 mimic or inhibitor (Fig. 2D, b and Supplementary Fig. S8). Therefore, the nuclear TCF12 immunoreactivity in tissue is likely a true signal. As the percentages of TCF12 in nucleus did not change conspicuously after miR-211 modulation, miR-211 expression was unable to regulate the translocation of TCF12.

In OECM1 cells, neoplastic activities including proliferation, migration, invasion, and AIG were decreased in TCF12 WT cell, and such activities further reduced in MUT cell (Fig. 3A, a). However, except for AIG, other oncogenic activities in CDS were similar to WT. It appeared that OECM1 CDS was rather resistant to TCF12-driven oncogenic suppression (Supplementary Fig. S9). In HSC3 cells, the neoplastic activities of CDS were lower than those in WT (Fig. 3A, b). The increased migration in OECM1-miR-211 was repressed by TCF12 expression (Fig. 3A, c). OECM1 with the knockdown of TCF12 exhibited increased neoplastic activities (Fig. 3B, a). Apart from the in vitro oncogenic modulation, SAS with the knockdown of TCF12 increased subcutaneous tumorigenicity in vivo (Fig. 3B, b). The clues indicate that TCF12 mediates suppressor activity against OSCC in general. The reduced neoplastic activities resulted from the miR-211 inhibition were rescued by the knockdown of TCF12 (Fig. 3C). Thus, miR-211 induces OSCC oncogenicity by targeting the TCF12 tumor suppressor.

Downregulation of TCF12 mRNA expression was found in 78% (39/50) of human OSCC tumors as compared with their paired NCMTs (Fig. 3D, a). TCF12 protein expression was unequivocally lower in OSCC tumors relative to the NCMTs (Fig. 3D, b). It appeared that miR-211 expression in the NCMTs was higher than the corresponding OSCC tumors, and there was no correlation between miR-211 expression and TCF12 mRNA expression in OSCC tumors (Supplementary Fig. S10). To clarify this ambiguity, miR-211 staining and nuclear TCF12 staining were performed in human OSCC. It indicated a reverse association between miR-211 expression and TCF12 expression (Fig. 3D, c and Supplementary Fig. S11). Besides, the increase of miR-211 staining in tongue epithelium from Wt to heterozygous, and to homozygous Tg mice was noted. 4NQO treatment enhanced such increase. On the contrary, there was progressive decrease of nuclear TCF12 staining in tongue epithelium from Wt to Tg mice. 4NQO treatment enhanced such decrease during the murine multistep carcinogenesis (Supplementary Figs. S7 and S12). The relatively high miR-211 expression in NCMTs revealed by qRT-PCR could be a result of the intensive submucosal miR-211 expression (Supplementary Fig. S11A).

To investigate the regulation of TCF12 on downstream genes, we performed GeneChip analysis using OECM1 knockdown cells and HSC3 expression cells. A total of 31 and 3 presumed oncogenic and suppressor gene spots were identified (Fig. 4A, a and Supplementary Fig. S13). Gene annotation and network analysis are shown in Supplementary Fig. S14. We selected 7 genes, NNMT, TNFFS10, ABCA, CLCA, FAM213A, GDF15, and TRIM29 (Tripartite motif-containing protein 29), which either have been shown in GeneChip for the presence of more than two spots or having the Jaspar score of >50 for qRT-PCR analysis (Fig. 4A, b and Supplementary Fig. S15; Supplementary Table S10). Changes in FAM213A and TRIM29 mRNA expressions across three different cells were consistent (Supplementary Fig. S16). Analyses applied on different OSCC cells showed that relative to controls, FAM213A and TRIM29 were downregulated in cells having exogenous TCF12 expression; while they were upregulated when TCF12 expression was knocked down (Fig. 4B, a). To investigate whether miR-211 regulates FAM213A and TRIM29, SAS cells were treated with miR-211 mimic or inhibitor. The repression of TCF12 induced by miR-211 mimic drastically upregulated FAM213A, but it only slightly upregulated TRIM29. The upregulation of TCF12 induced by miR-211 inhibition resulted in the downregulation of FAM213A and TRIM29 (Fig. 4B, b). The regulation of miR-211–TCF12 axis on FAM213A was more eminent than TRIM29. In a panel of OSCC cells, including 6 cell lines and 5 primary tumor cell cultures, miR-211 expression is positively correlated with FAM213A mRNA expression (Fig. 4C, a). Downregulation of TCF12 and upregulation of FAM213A were noted in the stripped tongue and skin epithelium of Tg mice in relation to Wt mice (Fig. 4C, b). Gene expression data from the NCI-60 database and The Cancer Genome Atlas (TCGA) HNSCC database identified a reverse correlation between TCF12 and FAM213A in cancer cell lines and HNSCCs (Supplementary Fig. S17). OSCC cells treated with 4NQO displayed miR-211 upregulation (Fig. 2A, a, left), accompanied with the downregulation of TCF12 in mRNA (Fig. 4C, c) and protein level (Fig. 4D, a). This TCF12 downregulation was miR-211 associated (Fig. 4D, a). In addition, 4NQO-induced FAM213A upregulation in OSCC cells was related to TCF12 downregulation but was irrelevant with TCF4 or TGFβRII (Figs. 2A, d, 4C, c, and 4D, b). The findings substantiate the existence of miR-211–TCF12–FAM213A regulatory axis in OSCC, which is 4NQO inducible.

Combined analysis from the Jaspar database with manual precision mapping of E-box elements defined 13 E-boxes on sense or antisense strand in the −1000-TSS segment of human FAM213A gene. The region between -608 to -792 seemed to be a hotspot as it contained 7 E-boxes (Fig. 5A, a). To specify that the TSS∼ −1000 segment could possess FAM213A promoter activity, it was analyzed. It showed an increased luciferase activity in the reporter harboring FAM213A −1000-TSS segment with the knockdown of TCF12 in SAS and HSC3 cells (Fig. 5A, b, left). In contrast, OECM1 with TCF12 expression had lower reporter activity (Fig. 5A, b, right). To clarify whether TCF12 directly binds to the hotspot region in the promoter, ChIP assay was carried out to amplify the sequences being precipitated by anti-TCF12 antibody. Comparing with the control, the use of anti-TCF12 antibody yielded amplicons in a dose-dependent manner in HSC3 cell (Fig. 5A, c, left). Moreover, OECM1 expressing TCF12 also had higher binding of TCF12 to this hotspot region relative to controls (Fig. 5A, c, right). The results suggest that TCF12 is able to repress the promoter activity of FAM213A in OSCC cells through binding to this region containing an E-box cluster. Knockdown of FAM213A was carried out in OSCC cells (Fig. 5B, a). FAM213A knockdown significantly repressed the oncogenic phenotypes including proliferation, migration, invasion, and AIG in SAS and OECM1 (Fig. 5B, b). The effects on migration, invasion, and AIG were more profound than proliferation. In OECM1 cell, the migration being induced by the knockdown of TCF12 was repressed when FAM213A was knocked down (Fig. 5B, c). Furthermore, the migration and invasion induced by miR-211 were repressed when FAM213A was knocked down (Fig. 5C). With the knockdown of FAM213A, the ALDH1-postive OSCC cell population decreased by about 50% (Fig. 5D). Strong FAM213A staining was noted in human OSCC tissues (Supplementary Fig. S18A). The progressive increase of FAM213A expression was also noted during the 4NQO-induced multistep carcinogenesis (Supplementary Fig. S18B). Thus, FAM213A, a downstream gene transcriptionally repressed by TCF12, plays an oncogenic role in OSCC.

To stratify that the intracellular ROS was regulated by miR-211–TCF12–FAM213A axis in OSCC cells, cells were challenged with H₂O₂ to evoke ROS. The assays showed that exogenous miR-211 expression and the knockdown of TCF12 decreased the ROS, whereas the knockdown of FAM213A increased the ROS in OSCC cells (Fig. 6A). The effects of miR-211 in reducing ROS were limited. To elucidate the impact of TCF12–FAM213A on the cell migration being modulated by ROS, wound-healing assays were performed. The induction of ROS drastically impeded the migration of SAS cells, while the knockdown of TCF12 attenuated such impedance (Fig. 6B, left). The induction of ROS slightly inhibited the migration of OECM1 cells. Knockdown of TCF12 reverted such inhibition, and the reversion could be further repressed by the knockdown of FAM213A (Fig. 6B, right). Collectively, the results substantiate a role of FAM213A in abrogating ROS-associated deleterious effects on cell migration. The formation of 8-OHdG in nucleic acid and the genesis of carbonyl group in proteins are markers of oxidative stress. The knockdown of FAM213A or the 4NQO treatment increased carbonyl proteins (Fig. 6C). By staining the nuclear 8-OHdG and the carbonyl proteins, we also detected more profound oxidative stress in tongue epithelium of Tg mice than Wt mice (Fig. 6D and Supplementary Fig. S19). Oxidative stress was particularly high in the tumor tissues of Tg mice. Although there was high FAM213A expression in tumors of Tg mice (Supplementary Fig. S18B), its scavenger efficiency was not sufficient for ROS attenuation in epithelial tissue. It is likely that FAM213A contributes to oncogenesis via other activity in addition to ROS scavenger.

The nearly absent miR-211 staining was noted in the human NCMT tissues in TMA (Fig. 7A, top and Supplementary Fig. S11A). There was stronger miR-211 staining in corresponding paired OSCC tissues. Remarkable miR-211 expression was noted in stroma subjacent to the human oral epithelium. There was a complete absence or relatively fainter FAM213A staining in NCMT tissues. However, the cytosolic FAM213A staining in paired and unpaired tumor tissues were much stronger (Fig. 7A, bottom and Supplementary Fig. S18A). Quantitation of the pixel readings showed a significant increase of both miR-211 and FAM213A expression from NCMT to OSCC (Fig. 7B, a and b), which was highly correlated (Fig. 7B, c). Although the FAM213A expression in stage I–III tumors and stage IV tumors were not much different in view of the pixel scoring (Fig. 7B, d), OSCC having strong FAM213A expression exhibited a trend of worse overall and disease-free survival (Fig. 7B, e and Supplementary Fig. S20A). In the OSCC at stage I–III, association between FAM213A expression and worse survival was more evident (Fig. 7B, f and Supplementary Fig. S20B).

Representative analyses of miR-211 and FAM213A staining in tissues harvested from Wt mice at different time points during the multistep carcinogenesis were illustrated (Fig. 7C, a, b and Supplementary Fig. S18B). The progressive increase of miR-211 and FAM213A expression followed the increased severity of epithelial pathogenesis (Fig. 7C, c and d). The increase of miR-211 and FAM213A expression occurred prior to neoplastic formation. In agreement with this tendency, nuclear TCF12 expression decreased during the multistep carcinogenesis (Supplementary Fig. S12). The findings confirm that 4NQO may modulate miR-211–TCF12–FAM213A axis and contributes to the progression of OSCC. The schema in Fig. 7D depicts our thought on how miR-211 targets TCF12, which then transinactivates FAM213A for oral carcinogenesis.

miR-211 is associated with the pathogenesis of several malignancies including OSCC (14–17, 19–21). Studies have elucidated that miR-211 is versatile in targeting multiple genes in different kinds of cells (10, 15–21, 31, 47). In addition to a miR-211 target being found in melanoma (20, 31), TCF12 is also a miR-211 target in OSCC. Although the oncogenic activity of HAS2 and KLLN appears irrelevant to 4NQO-miR-211 regulation, their pathogenic roles acting as new miR-211 targets need further specification. The repression of TCF12 on the transcription of FAM213A may account for a novel molecular mechanism underlying the OSCC oncogenesis as induced by miR-211. We established K14-EGFP-miR-211 Tg mouse models. These Tg mice have increased thickness, increased prosurvival protein Bcl-xL expression, and endogenous oxidative stress in squamous epithelium, and a higher 4NQO-associated tumor induction. It is worth noting that the miR-211 dosage is also correlated with the tumor burden and aggressiveness. These findings substantiate the contribution of miR-211 for promoting squamous carcinogenesis. Because of the presence of GFP tag, this design of genetic engineering has facilitated the rapid screening of infant mice carrying transgene. With the assistance of fluorescence, some insidious tumors on the tongue surface or embedded in the unopened esophagus, not readily detectable with visible light, were more clearly defined. As mice with advanced tumors and poor health are sacrificed before endpoint, this might underlie the limited tumorigenic enrichment in Tg mice. The assessments of locoregional metastasis are also confounded accordingly. We have demonstrated that miR-211 expression promotes primary OSCC tumor growth and neck nodal metastasis in orthotopic nude mice model (8).

This study identified that 4NQO can upregulate miR-211 in oral keratinocytes both in vivo and in vitro. Therefore, the enrichment of miR-211 expression induced by 4NQO enhances the severity of tumorigenesis in transgenic mice, could be a plausible mechanism. The role of miR-211 in the human carcinogenesis process of esophagus, skin, and cervix is unclear. This animal model would be useful for assessing the chemical- or viral-associated tumorigenesis in squamous epithelium other than oral tongue. Because of the rather strong miR-211 expression in human stromal cells and the versatility of miR-211 in targeting genes, the roles that miR-211 plays in stromal pathogenesis require elucidation.

Our functional assays clarified that miR-211 targets TCF12 both in vitro and in Tg mouse model. TCF12 functions to suppress OSCC oncogenicity. In addition, most miR-211–associated tumor phenotypes are reversed by TCF12. As TCF12 is a negative regulator, the exogenous TCF12 expression in OECM1 cell is limited, and resistance to TCF12-mediated suppression may emerge. TCF12 was conspicuously downregulated in the vast majority of OSCC tumors. As TCF12 has many spliced variants and it is prone to get mutated or translocated in diseases (29, 30), to further define its functional implications and to characterize the genomic abnormalities in a wide variety of malignancies is required. Antibodies detecting nuclear TCF12 activity more specifically are required to facilitate tissue studies. Our approaches pinpoint FAM213A and TRIM29 as potential downstream effectors of TCF12. A fraction of potential TCF12 targets may have been masked in the initial screening according to our criteria. TRIM29 seems to be a key regulator of epithelial–mesenchymal transition (EMT) and tumor invasion, but its roles among neoplasms are controversial. It upregulates CD44 and induces EMT in K-ras–induced pancreatic carcinoma (48). On the contrary, it also suppresses TWIST1 and inhibits EMT in breast carcinoma (49). As miR-211 expression upregulates TRIM29 rather limitedly, unclear factors may exist to confound this regulation in OSCC. Although no effort has been made to further delineate its interplay with TCF12, to resolve oncogenic stimuli that regulate TCF12-TRIM29 would be important for understanding tumor invasion and EMT.

FAM213A attenuates ROS-associated signals, which then elicits the differentiation of bone marrow monocytes (39). Our approaches further unravel that TCF12 downregulates FAM213A through transcriptional repression (22), and FAM213A depletion decreases oncogenicity. In addition to the fact that endogenous miR-211 and FAM213A are rather synchronized, 4NQO stimulation also upregulates these oncogenic events concordantly. These clues signify a new miR-211–TCF12–FAM231A regulatory axis in OSCC oncogenesis upon carcinogenic stimulation. Despite that FAM213A has weak impact on proliferation relative to other neoplastic activities, FAM213A expression increases from normal mucosa to neoplasm in both human OSCC and murine carcinogenesis process, FAM213A is contributive to the early establishment of OSCC. Our previous study identified the prognostic values of miR-211 expression in OSCC (14), this study suggests the potential of FAM213A expression in determining the survival of patients carrying stage I–III tumors.

miR-211 was reported to reduce ER stress in colorectal cancer (18). miR-204, another member of the miR-211 family, increases the sensitivity to oxidative stress in neuron cells (47). This study specifies that miR-211 upregulates FAM213A, which then reduces ROS in OSCC cells. We also show that FAM213A may protect cancer cells from the oxidative damages. Although the findings of increased oxidative stress in the oral epithelium of Tg mice is to our surprise, the results may implicate a potential linkage between miR-211 expression and the higher cellular injury for tumor susceptibility. As the antioxidative activity of miR-211 seems negligible, it is unable to attenuate the oxidative stress. The upregulation of proteins associated with proliferation or survival in epithelium may be responded to the cell injury in epithelium. FAM213A expression is highly associated with invasion and the increased ALDH1-positive cell population. miR-211 is found to facilitate the neck metastasis of OSCC xenografts in this study. As oxidative stress could hinder the distal metastasis of melanoma cells (50), and whether the ROS attenuation or the stemness induction mediated by FAM213A can profit the metastatic dissemination needs elucidation. As the vast majority of OSCCs have strong FAM231A expression, the therapeutic efficacy of OSCC could be improved by incorporating anti-FAM213A strategy into conventional chemotherapy.

In this study, we specify that miR-211 is an oncogenic regulator of OSCC by targeting the TCF12 tumor suppressor, and TCF12 transcriptionally represses FAM213A oncogenic molecule. 4NQO-miR-211-TCF12-FAM213A tends to be a novel regulatory cascade for oral carcinogenesis. Whereas the activities of FAM213A in enhancing specific oncogenic signals remained to be resolved, its dual properties and high expression in OSCC may implicate targeting values (8, 39). The K14-EGFP-miR-211 Tg mice might be a suitable model to develop regimen for OSCC intervention.

No potential conflicts of interest were disclosed.

Conception and design: Y.-F. Chen, S.-C. Lin, K.-W. Chang

Development of methodology: Y.-F. Chen, S.-C. Lin, K.-W. Chang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y.-F. Chen, C.-C. Yang, C.-J. Liu, S.-C. Lin, K.-W. Chang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y.-F. Chen, C.-C. Yang, S.-C. Lin, K.-W. Chang

Writing, review, and/or revision of the manuscript: Y.-F. Chen, S.-C. Lin, K.-W. Chang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y.-F. Chen, S.-Y. Kao, C.-J. Liu, S.-C. Lin, K.-W. Chang

Study supervision: Y.-F. Chen, S.-C. Lin, K.-W. Chang

Other(manuscript comment): S.-Y. Kao

The authors thank Prof. Tin-Fen Tsai and Ms. Courtney Anne Curtis for their assistance.

S.-C. Lin received grant MOST102-2628-B-010-006-MY3 from Ministry of Science and Technology. K.-W. Chang received grant V103C-070 from Taipei Veterans General Hospital and grant 104AC-P504 from Aim for the Top University Plan from Department of Education. S. -Y. Kao, K.-W. Chang, and C.-C. Yang received grant for Health and Welfare Surcharge of tobacco products number MOHW104-TD-B-111-02 from Ministry of Health and Welfare for Excellence for Cancer Research.

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.