Oncogenic Function and Early Detection Potential of miRNA-10b in Oral Cancer as Identified by microRNA Profiling

1. miRNA signature of oral cancer was determined.

2. The oncogenic function of miR-10b in oral cancer was first demonstrated.

3. The potential of miR-10b as a circulating biomarker for the early detection of oral cancer was presented.

Oral cancer is one of the most common cancers worldwide, with an incidence estimated to be more than 500,000 annually (1, 2). The disease is more prevalent among males than females (1, 2). Epidemiologic studies have shown a strong association between oral cancer and environmental carcinogens, especially the use of tobacco, alcohol, and betel quid (3, 4). The 5-year mortality rate for oral cancer has not been altered significantly in the last few decades, despite advances in treatment modalities. Even if there is a good treatment response, patients with advanced disease often suffer from substantial functional and cosmetic morbidity, which decreases their quality of life. The genetic alterations leading to oral carcinogenesis remain to be identified.

The miRNAs are endogenous, small noncoding RNAs (18–25 nucleotides long) that negatively regulate gene expression at the translational level by base pairing to the 3′-untranslated region of target mRNAs (5, 6). There are now more than 700 human miRNAs annotated in the Sanger miRBase database, and it has been predicted that there are more than 1,000 total human miRNAs (6, 7). It is estimated that approximately one third to one half of human genes are regulated by miRNAs, and each miRNA is predicted to target several hundred transcripts, making miRNAs one of the biggest family of gene regulators. Because miRNAs participate in a variety of biologic processes, including cell proliferation, differentiation, apoptosis, and migration, it is conceivable that the dysregulation of miRNAs is associated with malignant transformation. Recently, large-scale miRNA screening that shows unique expression profiles has been conducted for several types of cancer. In head and neck cancer, miRNA expression also has been screened (8–12); however, few overlapping molecules were found between profiling studies, indicating that the complexity and underlying mechanisms of miRNA function have not been solved. For example, sample profiling that was derived from a specific anatomic subsite (11) or examination using a different design or methodology (8–11) might lead to a different outcome. Moreover, as head and neck cancer is strongly associated with environmental carcinogens, reports from different geographic areas may lead to different results (10–12).

In the present study, we profiled and compared the miRNA of oral cancer in areas prevalent for betel quid chewing using 6 oral cancer lines and 5 normal keratinocyte cell lines. We identified 23 miRNAs and confirmed differential expression in 10 of them. We further examined the cellular function and plasma levels of miR-10b and showed its highly diagnostic potential for early detection in the plasma of oral cancer.

A total of 6 oral cancer lines, SCC25 (13), SAS (14), OECM1 (15), OC3 (16), CGHNC8, and CGHNC9, and 5 normal keratinocyte cell lines, CGHNK2, CGHNK4, CGK1, CGK5, and CGK6, were used. SCC25 cell line was purchased from Food Industry Research and Development Institute, Hsinchu, Taiwan. SAS, OECM1, and OC3 cell lines were kindly distributed by Professors S.C. Lin and K.W. Chang, Yang Ming University, Taipei, Taiwan. CGHNC8 and CGHNC9 cell lines were established within recent 2 years, derived from oral cancer squamous cell carcinomas of patients from Chang Gung Memorial Hospital, Taoyuan, Taiwan. The normal keratinocytes CGHNK2, CGHNK4, CGK1, CGK5, and CGK6 were primary culture cells from tissue biopsies of grossly normal oral mucosa in recent 2 years, with (CGHNK2, CGHNK4) or without (CGK1, CGK5, and CGK6) human papilloma virus (HPV) immortalization (Chang Gung Memorial Hospital). All the cell lines established in Taiwan (OECM1, OC3, CGHNC8, CGHNC9, CGHK2, CGHNK4, CGK1, CGK5, and CGK6) were derived from patients with oral cancer with the habits of areca chewing and smoking, except two without areca chewing (CGHNK2 and CGK1) and one without smoking (OC3). Cell lines SCC25 (13), SAS (14), OECM1 (15), and OC3 (16) were originally authenticated using the experiments described before. All the cells used were retested on the basis of viability, growth status, morphology, and the status of HPV infection (17). All the cells were used within 3 months after resuscitation of the frozen aliquots, with lower than 20 passages in each experiment. For HPV status, all the cell lines were negative for HPV-16 and HPV-18, except that SAS and OC3 were weak positive for HPV-18. The normal keratinocyte cells were maintained in keratinocyte serum-free medium (KSFM; Gibco BRL). OECM1 cells were grown in RPMI-1640 medium, and all other cell lines were grown in Dulbecco's Modified Eagle's Media supplemented with 10% FBS and 1X antibiotics-antimycotics (Gibco BRL). All cells were cultured at 37°C in a humidified atmosphere of 5% CO₂ air.

Total RNA from cells was isolated with TRIzol reagent (Gibco BRL) following the manufacturer's instructions. The concentration, purity, and amount of total RNA were quantified using the Nano-Drop ND-1000 Ultraviolet Spectrophotometer. Agilent's miRNA microarray system (G4470A, Agilent Technologies Inc.), which contains 470 human miRNAs, was used for expression profile analysis. For each microarray analysis, 1 μg of total RNA was used according to manufacturer's instructions. Slides were scanned on a microarray scanner (model G2565A, Agilent) with a high dynamic range setting, and the Feature Extraction software (Agilent) was used for data extraction.

GeneSpring GX software (version 7.3.1, Agilent) was used to analyze the expression level of miRNAs from the microarray assay. Data were filtered by Agilent present/absent flags, and the intensity level was set to exclude weak signals. The ANOVA with the Benjamini and Hochberg correction for false-positive reduction was used to find differentially expressed miRNAs with a false discovery rate (FDR) < 0.1. Those miRNAs showing more than a 2-fold change between the cancer and the normal group were selected by an n-fold change filter tool. Hierarchical cluster analysis was applied to average linkage using Pearson correlation as a measure of similarity between sample groups.

Total RNA was isolated from cells with TRIzol reagent (Gibco BRL). Reverse transcription was conducted as previously described (18), except for the use of miRNA specific stem-loop RT primers. TaqMan miRNA assays kits (ABI) were used to examine specific miRNA expression by reverse transcription quantitative PCR (RT-qPCR), according to the manufacturer's suggested protocol. The real-time PCR results, recorded as threshold cycle numbers (C_t), were normalized against an internal control (U6 RNA), and the comparative threshold cycle method (ΔΔC_t) was used to determine miRNA expression.

Cells were seeded for transfection at a density of 5 × 10⁵ in a 100-mm dish and cultured for 16 hours. When 50% confluence was reached, cells were transfected with various concentrations (75–900 μmol/L) of miR-10b antagomir or scramble oligonucleotides, by using Lipofectamine 2000 reagent (Invitrogen) in OPTI-MEM medium (Invitrogen) for 16 hours; then, the medium was replaced with fresh complete medium. The nucleotide sequence of miR-10b antagomir was 5′-CAC AAA TTC GGT TCT ACA GGG TA-3′. The scramble oligonucleotides contained the same nucleotides as the miR-10b antagomir but in random sequence, 5′ CTG TTC GCA CAG CTT GGT TAA AA-3′.

To determine colony-forming ability, cells were transfected with either antagomir or scramble oligonucleotides. A total of 1,000 cells were seeded in 6-well plates and allowed to grow for 7 days without disruption. After 70% ethanol fixation, the plate was stained with 0.5% crystal violet, and cell colonies were counted.

Cell migration was evaluated by an in vitro wound-healing assay. After transfection with either miR-10b antagomir or scramble oligonucleotides, 70 μL of 5 × 10⁵ cell/mL transfectants was seeded in an ibidi culture insert (Applied BioPhysics, Inc.) on top of a 6-well plate. After 8 hours of incubation, the culture insert was detached to form a cell-free gap in a monolayer of cells. After changing to culture medium with 1% fetal calf serum, the cell migration status toward the gap area was photographed every 6 hours.

The cell invasion assay was conducted using BioCoat Matrigel (Becton Dickinson Biosciences) and Millicell invasion chambers (Millipore) similar to those previously described (19). Briefly, Matrigel was coated onto the membrane of the Millicell upper chamber with a pore size of 8 μm in a 24-well plate for 12 hours at 37°C. After transfection with either antagomir or scramble oligonucleotides, cells were seeded into the upper chamber at a density of 1 × 10⁵ cells per well in 0.2 mL of 1% FBS medium. The lower chamber contained complete culture medium, which included 10% FBS to attract invading cells. Cells were incubated at 37°C for 24 hours, and the number of cells that invaded through the Matrigel-coated membranes was counted and compared with the number of cells that passed through the membrane in the control chambers. The invaded cells on lower side of membrane were fixed, stained with crystal violet, and photographed.

Cells, either transfected with antagomir or scramble oligonucleotides, were seeded at 1 × 10⁵ cells per well in 6-well plates for 8 hours. For determination of chemosensitivity, cells were treated with various doses (0–80 μg/mL) of cisplatin and continuously cultured for 2 days. The number of surviving cells was counted and compared with the number of surviving untreated cells. For determination of radiosensitivity, the 1,000 antagomir or scramble oligonucleotide transfected cells were seeded into a 6-well cell culture plate. After incubation for 8 hours, cells were exposed to a range of radiation doses (0–6 Gy) and continuously cultured for 7 days, followed by the calculation of the surviving colonies. The survival fraction was calculated as the number of colonies divided by the number of cells seeded times plating efficiency, as previously described (15).

This study was approved by the Institutional Review Board of Chang Gung Memorial Hospital, and written informed consent was obtained from all participants. The EDTA plasma samples were collected from 54 patients with oral cancer in the week before receiving operation and/or chemoradiotherapy. The characteristics of these patients with oral cancer are summarized in the Table 1. These included 51 (94%) males and 3 (6%) females, with a mean age of 52.0 years. The EDTA plasma samples from 36 age- and sex-matched normal individuals and 7 patients with oral precancer leukoplakia were also obtained to compare. The miRNA of each specimen was purified from 200 μL plasma using the miRNeasy Mini Kit (Qiagen Inc.) and dissolved in 20 μL of RNase-free water. The miRNA levels were determined by RT-qPCR using the method described above.

To determine whether plasma miR-10b levels could serve as a diagnostic marker, we established a xenograft oral cancer SAS tumor in BALB/C nude mice. Five-week-old male BALB/c null mice, cared for according to institutional guidelines, were used for the study. For the xenograft in situ study, mice were injected subcutaneously in the upper portion of the hind limb with 1 × 10⁶ SAS cells. Mice were then monitored for the tumor growth by calculating the size of the tumor's length × width × height after 6 weeks. The EDTA plasma was drawn before cancer cell injection and 6 weeks after xenografting. The level of plasma miR-10b was determined by RT-qPCR as described above.

The χ²t test was used to determine whether miRNA levels were different between the two groups of samples. A P value of less than 0.05 was considered statistically significant. To evaluate whether RT-qPCR data could separate differences between normal individuals and patients with cancer, receiver operating characteristic (ROC) analysis was conducted, and the area under curve (AUC) was used to measure the level of separation between the 2 groups.

An miRNA microarray (Agilent Technology) was used to globally profile miRNAs that are differentially expressed between 6 oral cancer cells and 5 lines of normal keratinocytes. The mean signal intensity of the cancer cell lines versus normal keratinocytes is shown in Fig. 1A. After normalization and filtering to exclude weak signals, 190 miRNAs were selected for clustering analysis. Unsupervised hierarchical clustering analysis and the ANOVA were used to find differentially expressed miRNAs. With an FDR < 0.1 and more than a 2-fold change between the cancerous and the normal group, 23 miRNAs were found of which 19 were upregulated and 4 downregulated in the cancer samples (Fig. 1B, Table 2). The expression trends of these miRNAs were homogenous across almost all samples in the same group, with low P values (all P < 0.05, except miR-196a and miR-638, which were marginal). miR-10b, miR-9*, miR-196a, and miR-196b showed high overexpression (>10-fold). These results suggest that these 23 miRNAs may play important roles in oral carcinogenesis.

To further assess the significance of these miRNAs in oral cancer, RT-qPCR assays were independently examined between 6 oral cancer cells and 5 lines of normal keratinocytes. For each miRNA, the results are summarized in the Supplementary Table S1. Figure 2A shows representative miRNAs that were significantly differentially expressed between cancer and normal cells (also see in Supplementary Fig. S1). There were 20 miRNAs confirmed to have more than a 1.5-fold change in expression and 10 miRNAs with statistical differences (P < 0.05). To obtain an overall view of differential expression in these miRNAs, Fig. 2B was plotted to show both fold change and statistical P values for each miRNA. As shown, miR-10b, miR-196a, and miR-196b were significantly elevated (>50-fold) in oral cancer cell lines (P < 0.05).

Because miR-10b consistently showed highest level of overexpression in oral cancer cells in both microarray and RT-qPCR assays, this molecule was selected for further functional investigation. An in vitro loss-of-function analysis was applied, which silenced the miRNAs using antagomir oligonucleotides. Results showed that anti-10b caused substantial levels of inhibition, with approximately 90% reduction compared with the scramble control at day 1 (Supplementary Fig. S2A). The 1-day treatment using 150 μmol/L of anti-miR10b was applied in subsequent cellular studies.

The potential effect of miR-10b on cell growth was examined in 2 oral cancer lines, OECM1 and SAS. Treatment with the specific miR-10b antagomir had no effect on cell growth or colony formation in either cell line (Supplementary Fig. S2B). The potential effect of miR-10b on chemo-/radiosensitivity was also examined. Also, treatment with the specific miR-10b antagomir had no differential effect on cell survival in response to drug or irradiation (Supplementary Fig. S2C).

Cell migration and invasion were determined using in vitro wound-healing and Matrigel invasion assays in 2 oral cancer cell lines. For the migration assay, miR-10b antagomir transfectants showed slower migration toward the gap area than the controls (Fig. 3A). At 12 hours, the miR-10b antagomir transfected cells showed 48% and 38% reduction in OECM1 and SAS cells, compared with the controls (Fig. 3B). For the invasion assay, cells were seeded in the upper chamber and allowed cells invading to the outer surface of the Matrigel. As shown, the invaded cells of miR-10b antagomir transfectants were dramatically reduced compared with the controls (Fig. 3C), with reduction to 58% and 52% at 24 hours, respectively, in OECM1 and SAS cells (Fig. 3D). Apparently, suppression of miR-10b inhibited the invasion and migration ability of oral cancer cells.

Several target genes of miR-10b have been reported (20–24), so that we investigated whether the alterations of cell functions by miR-10b in oral cancer cells may through these genes. They are HOXD10, the prometastatic gene, that leads to RhoA/RhoC upregulation (20–22); Tiam1, the Rac-activating protein (23); and the KLF4 cancer stem cell–associated gene (24). To clarify whether these molecules were associated with miR-10b, we have examined the expression levels of these potential target genes (HOXD10, RhoA/RhoC, Tiam1, and KLF4) in the cell lines versus normal keratinocytes and in antagomir-expressing cell lines. Results are shown in the Supplementary Fig. S3. Unfortunately, none of these genes showed convincingly increased expression in antagomir-treated cells (Supplementary Fig. S3A and S3B). Furthermore, results of HOXD10 were opposite of what was expected, which was increased in cancer cell lines compared with normal keratinocytes. Therefore, the downstream regulatory pathway of miR-10b leading to cell invasion in oral tissues may not be through these molecules and is awaited to be further investigated.

To assess the association of miR-10b in cancer formation, xenografted SAS tumors in mice were established. A total of 1 × 10⁶ cells were subcutaneously injected into the upper area of the hind limb, and tumors were allowed to form for 6 weeks. Plasma samples were collected before and 6 weeks after tumor injection, and miR-10b expression levels were determined. Results are shown in Fig. 4A. Although varying in size, all the tumors were more than 1,000 mm³ in volume in the 4 xenografted mice, ranging from 1.1 cm³ to 3.1 cm³. Plasma miR-10b was significantly elevated in all mice, after tumor formation, with an average increase of approximately 20-fold (3.54 to 71.63; P = 0.043).

To further assess potential clinical applications of miR-10b, its expression was determined in 93 plasma samples, including 36 from normal donors, 7 from patients with precancer, and 54 from patients with oral cancer. As shown in Fig. 4B, the level of miR-10b in patients with cancer was significantly higher than in normal subjects, with approximately 4-fold of elevation in average (P < 0.0001). The expression level of miR-10b in plasma of patients with precancer was also shown with a striking result that average value of precancer was equal to that of patients with oral cancer, significantly higher than in normal subjects. However, it showed no significance in different T stage (P = 0.455), N stage (P = 0.450), and overall stage (P = 0.417). Using ROC analysis, this molecule yielded an AUC of 0.932 and 0.967 in oral cancer and precancer samples, respectively (Fig. 4C and D). A logistic regression model with the best prediction rate yielded 94.4% sensitivity and 80.0% specificity in distinguishing patients with oral cancer from the healthy controls. These results indicate that miR-10b has potential as an early detection marker for oral cancer.

In this study, we profiled oral cancer–specific patterns of miRNA expression using 6 oral cancer cell lines and 5 lines from oral normal keratinocytes with 470 miRNA array analyses. Microarray data revealed that oral cancer cell lines show a distinct pattern of microRNA expression, compared with normal keratinocytes. After clustering analysis, 23 miRNAs were found to be significantly differentially expressed between the 2 groups (Fig. 1, Table 1). The RT-qPCR method independently examined these 23 miRNAs and found 10 miRNAs consistent with the array results that showed significant differential expression (>2-fold) and low P values (<0.05; Fig. 2, Supplementary Table S1); these data suggest that these miRNAs are important in oral cancer formation. Upregulation of miR-15b and miR-128a, and downregulation of miR-31 and miR-503, were concordant with previous reports in head and neck cancer (8, 11, 25, 26). Upregulation of miR-10b, miR-148b, miR-196a, miR-196b, miR-582, and miR-301 and downregulation of miR-503 were newly observed in head and neck cancer. However, these novel miRNAs have been reported to be associated with other cancers. These data support the previously suggested participation of these miRNAs in malignant transformation, as oncomir functions of miR-10b in breast and esophageal cancers (20, 24), miR-196s in colorectal and gastric cancers (27), miR-301 in liver cancer (28), and miR-148b in lung cancer (29).

Although miR-10b has been reported to promote cancer metastasis in several cancers, little is known about this molecule in oral cancer. In agreement with previous reports, we found that silencing miR-10b significantly decreased cellular migratory and invasive ability, suggesting positive regulatory roles for this molecule in oral cancer cells as well (Fig. 3). Several genes have been reported as the potential targets of miR-10b, such as HOXD10, RhoA/RhoC, Tiam1, and KLF4 (20–24). However, we did not found significant associations in any of these molecules with miR-10b expression (Supplementary Fig. S3). Therefore, the downstream regulatory pathway of miR-10b leading to cell invasion in oral tissues is awaited to be further investigated. We further noted that miR-10b has a minimal role in cell growth regulation or the stress response (Supplementary Fig. S2). All these results help elucidate the function of miR-10b in the carcinogenesis of oral cancer.

Recently, several studies showed that miRNAs can be stably expressed in plasma, serum, and other body fluids, indicating a potential for the use of miRNAs as clinical biomarkers (30–32). In our xenografted tumor study, we found that circulating miR-10b dramatically increased (20-fold) in the plasma after tumor development (Fig. 4A), indicating that this molecule may potentially be used as a diagnostic marker. However, the level of miR-10b elevation was not correlated to tumor size. This result may reflect the fact that the miR-10b level was more strongly associated with cell invasion ability and less strongly associated with tumor growth (Fig. 3). It was also supported by a previous report that the administration of miR-10b antagomirs significantly suppressed tumor metastasis but not the growth of the primary tumor (33). Because circulating miR-10b was remarkable increased after xenografted tumor formation, we determined the differential expression of this molecule in patients with oral cancer and precancer and normal volunteers. We found that miR-10b was significantly upregulated in the plasma from patients with oral cancer compared with the controls with high AUC value (0.932, P < 0.0001; Fig. 4C). Remarkably, the elevation of miR-10b in patients with precancer plasma was close to that in patients with cancer, suggesting that miR-10b possesses a good predictive ability for early detection of patients with oral cancer (Fig. 4B–D). In conclusion, we have profiled at least 10 miRNAs significantly associated with oral cancer, including miR-10b. Through promoting cell migration and invasion, miR-10b actively participates in cancer formation. Our study in clinical samples suggests that plasma miR-10b may have a high potential as a novel less-invasive biomarker for the early detection of oral cancer.

No potential conflicts of interests were disclosed.

This study was supported by grants from Chang Gung Memorial Hospital (CMRPD160351-3 and CMRPD190371) and the Department of Health of Taiwan (DOH-99-TD-C-111-006).

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.