ORIGINAL ARTICLE

MicroRNA-9 inhibits the proliferation of oral squamous cellcarcinoma cells by suppressing expression of CXCR4via the Wnt/b-catenin signaling pathwayT Yu1, K Liu1, Y Wu2, J Fan1, J Chen1, C Li1, Q Yang1 and Z Wang1

Aberrant expression of microRNAs (miRNAs) has been involved in the development and progression of malignancy. MicroRNA-9 (miR-9) has been confirmed to be underexpressed in many types of cancers. However, the relationship between miR-9 and the Wnt/b-catenin signaling pathway in oral squamous cell carcinoma (OSCC) remains largely unknown. Here we showed that the miR-9 wasunderexpressed in patients with OSCC and several OSCC cell lines. Lentivirus-mediated miR-9 overexpression in highly aggressive(Tca8113 and SCC-9) tumor cells significantly inhibited proliferation of the two cell lines in vitro and in vivo. Furthermore, we foundthat the CXC chemokine receptor 4 (CXCR4) gene was a direct target of miR-9. RNA interference silencing of CXCR4 proved that miR-9underexpression led to constitutive activation of b-catenin through activation of CXCR4 expression in OSCC cells. Finally, we alsoanalyzed the possible relationship between miR-9 and the genes downstream of the Wnt/b-catenin pathway in OSCC developmentand progression. These results provide new evidence of miR-9 as a promising tumor gene therapeutic target for OSCC patients.

Oncogene (2014) 33, 5017–5027; doi:10.1038/onc.2013.448; published online 21 October 2013

Keywords: microRNA-9; CXC chemokine receptor 4; lentivirus; cell proliferation; oral squamous cell carcinoma; wnt/b-cateninpathway

INTRODUCTIONOral squamous cell carcinoma (OSCC) is one of the most commonmalignant tumors in southeast Asia, affecting approximately300 000 patients worldwide each year.1,2 It is identified as asignificant public health threat worldwide because its treatmentoften produces dysfunction and distortions in speech, masticationand swallowing, dental health and even in the ability to interactsocially.3 Despite improvements in these therapies, the 5-yearsurvival rate has not improved significantly and remains at about50%.4 Previous studies have suggested that the formation of OSCCis related to the rates of cell proliferation and apoptosis.5,6

Therefore, understanding the main regulatory mechanism of thismalignancy is the key to develop novel and effective therapeuticstrategies for OSCC.

The discovery of non-coding RNA in the human genome was animportant conceptual breakthrough in the post-genome sequen-cing era. MicroRNAs (miRNAs) are short, highly conserved smallnon-coding RNA molecules of about 20–22 nucleotides in lengththat regulate gene expression by binding to the 30 untranslatedregion (30-UTR) of the complementary mRNA sequence, resultingin translational repression and gene silencing.7,8 An increasingbody of evidence has suggested that the aberrant expressionof miRNAs may lead to the development and progressionof malignancy,9,10 including OSCC. MiR-21 is overexpressedand displays oncogenic activity in various carcinomas.11

By multivariate analyses, miR-21 expression is indicated as anindependent predictor of poor survival for patients with OSCC.12

Another study supports a strong association of high levels ofmiR-21 and significantly decreased five-year survival in patients

with HNSCC.13 In addition, miR-100, -125b, -133a, -133b, -137 and -193a are frequently downregulated in OSCC.14 Reduced levels ofmiR-138 and -222 have been proved to be related to highmetastatic potential of OSCC.15 MicroRNA-9 (miR-9), initiallydemonstrated to function in neurogenesis, has been confirmedto be underexpressed in many types of cancers, includingnasopharyngeal carcinoma,16 colon cancer,17 breast cancer,18 andmelanoma,19 all of which is indicative of a tumor suppressorpotential, whereas miR-9 is overexpressed in hepatocellularcarcinoma,20 brain cancer21 and Hodgkin’s lymphoma,22

suggesting oncomir activity for miR-9 in these cancers.Chemokines, a superfamily of small cytokine-like proteins, can

bind to and activate a family of seven transmembrane G-protein-coupled receptors, the chemokine receptors. CXC chemokinereceptor 4 (CXCR4) is expressed on multiple cell types includinglymphocytes, hematopoietic stem cells, endothelial and epithelialcells, and cancer cells.23 In fact, CXCR4 and its ligand stromal cell-derived factor-1, also known as CXCL12, have been involved intumor progression, angiogenesis, metastasis and survival.24 Inaddition, Wnt/b-catenin signal transduction is a pathway thatresults in the cytoplasmic protein b-catenin entering the nucleusto modulate transcription. When the pathway is not activated, b-catenin is subject to a ‘futile cycle’ of continual synthesis anddestruction by the b-catenin destruction complex, comprised ofthe scaffold proteins Axin and APC and the kinases GSK3 andcasein kinase-1.25

Based on previous research on the role of miR-9, CXCR4 and theWnt pathway, we hypothesized that miR-9 regulates CXCR4expression in OSCC progression through the Wnt/b-catenin

1Department of Head and Neck Oncology Surgery, Sichuan Cancer Hospital, Chengdu, People’s Republic of China and 2State Key Laboratory of Oral Diseases, West China Collegeof Stomatology, Sichuan University, Chengdu, People’s Republic of China. Correspondence: Dr T Yu or Dr Z Wang, Department of Head and Neck Oncology Surgery, SichuanCancer Hospital, No. 55, Sec. 4, Renminnan Road, Chengdu, Sichuan 610041, People’s Republic of China.E-mail: vince8181@163.com or zhaohuiwang512@163.comReceived 3 June 2013; revised 20 August 2013; accepted 13 September 2013; published online 21 October 2013

Oncogene (2014) 33, 5017–5027& 2014 Macmillan Publishers Limited All rights reserved 0950-9232/14

www.nature.com/onc

pathway. To explore the roles of miR-9 in OSCC, in the presentstudy, we first investigated the expression level of miR-9 in OSCCcells, and tested its effects on cell growth, apoptosis and colonyformation in vitro and tumorigenesis in vivo. Second, we identifiedthat the CXCR4 gene was a target of miR-9 in OSCC. Finally, we alsoanalyzed the possible relationship between miR-9 and thedownstream genes of the Wnt/b-catenin pathway in OSCCformation and progression. Our study may provide a betterunderstanding of OSCC tumorigenesis. This study may providenew evidence for miR-9 as a promising gene therapeutic target forOSCC.

RESULTSExpression levels of miR-9 in OSCC clinical specimens and cell linesThe four oral squamous cell carcinoma cell lines (Tca8113, SCC-4,SCC-9 and SCC-25) were compared with human normal oralkeratinocytes. The miR-9 expression in Tca8113, SCC-4, SCC-9 andSCC-25 cells was clearly downregulated by 2–10-fold (Figure 1a).The Tca8113 and SCC-9 OSCC cell lines, which possess the lowestlevels of miR-9 expression among all tested OSCC cell lines, wereselected for further studies. In addition, we analyzed theexpression levels of miR-9 in 16 clinical tongue oral squamouscell carcinoma specimens and their adjacent non-cancerousspecimens. The results indicate that the expression levels of miR-9were significantly lower in the 16 clinical tongue OSCC specimensthan those in the corresponding adjacent non-cancerous speci-mens (P¼ 0.0035) (Figure 1b). In addition, to identify the role ofthe expression levels of miR-9 transcript and CXCR4 protein onOSCC malignant progression, we performed immunohisto-chemistry (IHC) for CXCR4 protein expression and real-time RT–PCR for miR-9 expression in metastasis (specimens from patientswith neck lymph nodes metastasis) and non-metastasis (speci-mens from patients without neck lymph nodes metastasis) OSCCcells. Supplementary Figure 1 demonstrated that the miR-9 levelwas significantly decreased in metastasis OSCC specimens,compared with that in non-metastasis OSCC specimens(P¼ 0.0059); however, the CXCR4 level was significantly increasedin metastasis OSCC specimens, compared to that in non-metastasis OSCC specimens.

Effects of miR-9 overexpression through lentivirus vector on theproliferation and invasion of Tca8113 and SCC-9 cells in vitroFirstly, we expressed miR-9 recombinant lentivirus vector in thehighly malignant Tca8113 and SCC-9 cell lines (Figure 1c). Todetermine whether the recombinant vectors were transducedinto tumor cells, green fluorescent protein (GFP) expression ofTca8113 and SCC-9 cells in mock, miR-control and miR-9 LVgroups was assessed using fluorescence microscope and flowcytometry. The results indicate that the delivery efficiency of miR-control and miR-9 LV was more than 95%, and no GFP expressionwas observed in Tca8113 and SCC-9 mock groups, as expected.After miR-control (MOI¼ 50) and miR-9 LV (MOI¼ 50) transduc-tion for 72 h in Tca8113 and SCC-9 cells, the expression levels ofmiR-9 were substantially increased 8–9-fold in miR-9 LV expres-sing Tca8113 and SCC-9 cells, respectively, compared to theirparental cells (Figures 1d–g). To investigate whether miR-9overexpression affects Tca8113 and SCC-9 cells invasion, weconducted three-dimensional cell migration assays using trans-well chambers and invasion assays with matrigel-coated millicellchambers. The results indicate that miR-9 overexpressiondecreased cell invasion by B80% compared with controls(Supplementary Figure 2).

The XTT assay demonstrated that cell proliferation wassignificantly inhibited in miR-9 transfectants in comparison withthe mock or miR-control transfectant cells. Specifically, weobserved the following growth, expressed as a percentage of

the control: (1) Tca8113-mock, 100.0±8.4; miR-control, 91±3.2;miR-9, 20.2±4.2; and (2) SCC-9 mock, 100.0±11.2; miR-control,104.3±13.6; miR-9, 34.8±3.5; with Po0.01 for both (Figures 2aand b). Following this, we also used the plate colony formationassays to assess the effect of miR-9 expression on proliferation ofOSCC cell lines in vitro. Elevated miR-9 expression significantlydecreased the Tca8113 and SCC-9 cells colony formation (Po0.01)(Figures 2c–e).

The miR-9 overexpression promotes cell cycle arrest and apoptosisof OSCC cells in vitroTo detect the effects of miR-9 expression on OSCC cell proliferationand apoptosis, we firstly examined the cell cycle distribution ofcells treated with miR-9 in the Tca8113 and SCC-9 cell lines. Aftertransduction with miR-9 (MOI¼ 50) and miR-control (MOI¼ 50) for72 h in Tca8113 and SCC-9 cells, both of the cell lines showeddecreased accumulation in S and G2/M phases and increasedaccumulation in G1 and Sub-G1 phase. After 72 h of transductionwith miR-9, the percentage of cells in the S phase decreased fromB29 to 19% in Tca8113 and from 36 to 20% in SCC-9 cells. Thepercentage of cells in the G2/M phase decreased from B19 to 10%in Tca8113 and from 18 to 11% in SCC-9 cells. Conversely, thepercentage of cells in Sub-G1 phase increased from B4 to 18% inTca8113 and from 2% to 21% in SCC-9 cells. Interestingly, thepercentage of cells in G1 phase increased from B49 to 53% inTca8113 and from 44 to 48% in SCC-9 cells (Figures 3a–c).The results suggested that CXCR4 downregulation might promoteOSCC cells apoptosis.

Following this, we used an Annexin V-FITC kit to determine thepercentage of cells that underwent apoptosis. After transductionwith miR-9 (MOI¼ 50) for 72 h, Annexin V/PI staining of Tca8113and SCC-9 cells showed that B26% and B34% of cells wereannexin V positive, respectively. These data suggested that theinhibitory effect on the expression of miR-9 indeed promoteTca8113 and SCC-9 cell apoptosis (Figures 3d–f).

Inhibition of tumor cell proliferation by miR-9 overexpressionin vivoTo investigate whether miR-9 expression caused inhibitory effecton preestablished tumor growth in nude mice, we observed andmeasured tumor sizes in 36-nude mice in a 35-day follow-upperiod. On day 42, we found a rapid increase in the Tca8113tumors volume in the mock (2139±369 mm3) and miR-control(2568±732 mm3) groups when compared with the miR-9 LV(423±98 mm3) group (Po0.01), and the same tendency wasfound in SCC-9 tumors. The volume of tumors in mock(2310±496 mm3) and miR-control (2489±735 mm3) groups weresignificantly increased when compared with the miR-9 LV(512±88 mm3) group (Po0.01). However, there were no differ-ences in tumor size and growth tendency between mock and miR-control groups either in Tca8113 or in SCC-9 tumors (Figures 4a–c).To determine whether the vectors were transduced into tumorcells, GFP expression of tumor tissues was assessed by counting10-random fields of the frozen sections under an invertedfluorescence microscope (FL) � 100 (Figure 4d). This vectorindependently expresses a green fluorescent protein that allowsdirect monitoring of the delivery efficiency of the gene-silencingconstruct. The delivery efficiency ofmiR-control and miR-9 LVgroups was approximately 75%, and no GFP expression wasobserved in the mock group, as expected.

MiR-9 expression inhibited the protein expression of CXCR4 andKi-67 in vivoTo further confirm whether miR-9 expression can inhibit OSCCcells proliferation and regulate the CXCR4 expression in tumortissues, we analyzed the protein expression of CXCR4 and

miR-9 inhibits the proliferation of OSCCT Yu et al

5018

Oncogene (2014) 5017 – 5027 & 2014 Macmillan Publishers Limited

Figure 1. microRNA-9 expression in oral squamous cell carcinoma specimens and cell lines. (a) Real-time PCR was performed to quantify miR-9in four OSCC cell lines (SCC-4, Tca8113, SCC-9 and SCC-25) and a human normal oralkeratinocytes. All data were presented as representativeof an average of three independent experiments. (b) Real-time RT–PCR showed that miR-9 was significantly downregulated in OSCC clinicalspecimens compared with normal adjacent specimens (n¼ 16); (c) Tca8113 and SCC-9 tongue squamous cell carcinoma cells were transducedwith miR-9-LV (MOI¼ 50) and miR-control (MOI¼ 50) for 72 h, then photographed under an inverted fluorescence microscope (� 100).(d, e) After transduction with miR-9 LV and miR-control for 72 h in Tca8113 and SCC-9 cells, miR-9 expression levels were detected by real-timeRT–PCR analysis in the two cell lines. (f, g) Densitometric analysis of miR-9 levels, relative to RNU48, determined by real-time RT–PCR analysisand compared with Tca8113 and SCC-9 cells transduced by miR-9 LV. Error bars indicate mean±s.d., n¼ 3 experiments, **, Po0.01.

miR-9 inhibits the proliferation of OSCCT Yu et al

5019

& 2014 Macmillan Publishers Limited Oncogene (2014) 5017 – 5027

Figure 2. MiR-9 overexpression inhibited cell viability and colony formation. (a, b) Cell proliferation activity after transfection of miR-9 wasdetermined by XTT assay; (c) representative photographs of plate colony formation of Tca8113 and SCC-9 cells. (d, e) The quantitative analysisof plate colony formation of Tca8113 and SCC-9 cells. Error bars indicate mean±s.d.; n¼ 3 experiments; **Po0.01.

Figure 3. Cell cycle and apoptosis analysis of Tca8113 and SCC-9 cells were treated with miR-9 LV. (a) After transduction with miR-9 LV(MOI¼ 50) and miR-control (MOI¼ 50) in Tca8113 and SCC-9 cells for 72 h, cell nuclei were fixed, stained with propidium iodide and analyzedby flow cytometry. Representative histograms were shown. (b) The fraction of Tca8113 cells in sub-G1 (black), G1 (2 N; gray), S (white) and G2-M(4N; speckled) are shown. (c) The fraction of SCC-9 cells in sub-G1 (black), G1 (2N; gray), S (white) and G2-M (4N; speckled) are shown. Error barsindicate mean±s.d., n¼ 3 experiments. (d) The apoptosis cells were stained with Annexin V-FITC and were analyzed as describe in the text. PIindicates propidium iodide. (e): the quantitative analysis of positive Annexin V Tca8113 cells in different groups. (f ) The quantitative analysis ofpositive Annexin V SCC-9 cells in different groups. Error bars indicate mean±s.d., n¼ 3 experiments. **Po0.01.

miR-9 inhibits the proliferation of OSCCT Yu et al

5020

Oncogene (2014) 5017 – 5027 & 2014 Macmillan Publishers Limited

proliferation-related gene Ki-67 by immunohistochemistry, andthe results showed that CXCR4 and Ki-67 protein expression wasweaker in the miR-9 LV group than that in mock and miR-control

groups (Figures 4e–f). These results indicate that miR-9 over-expression can significantly regulate CXCR4 protein expressionand inhibit the tumor growth in vivo.

miR-9 inhibits the proliferation of OSCCT Yu et al

5021

& 2014 Macmillan Publishers Limited Oncogene (2014) 5017 – 5027

Figure 4. Reduced Tca8113 and SCC-9 tumor volumes in nude mice injected with miR-9 LV. (a) After 6 times injection of saline solution (0.1ml),miR-control (0.1ml) and miR-9 LV (0.1ml), on day 42, photographs of representative mice and tumors are shown. (b, c) Tumor size wasmeasured using vernier caliper and tumor volume was determined as described in MATERIALS AND METHODS section for 42 days. Each datapoint is the mean value of five-six primary tumors. (d) GFP expression of tumor tissues transduced with miR-9 LV and miR-control. GFPexpression of tumor tissues in different groups were observed, counted and photographed under a inverted fluorescence microscope(FL)� 100. Scale bar ¼ 100 mm; (e) CXCR4 immunohistochemistry (IHC) assay. CXCR4-positive staining was located in the tumor nuclei andcytoplasm, and more positive cells could be seen in Mock and miR-control cells than that in miR-9 LV Tca8113 and SCC-9 cells. f: Ki-67-positivestaining was located in the tumor nuclei, and more positive cells could be seen in Mock and miR-control cells than that in miR-9 LV Tca8113and SCC-9 cells. Error bars indicate mean±s.d.; *Po0.05; **Po0.01.

miR-9 inhibits the proliferation of OSCCT Yu et al

5022

Oncogene (2014) 5017 – 5027 & 2014 Macmillan Publishers Limited

CXCR4 is a target gene of miR-9Using the Target Scan (http://www.targetscan.org), we found that theprotein encoded by CXCR4 was a potential target of post-transcriptional repression by miR-9 (Figures 5a and b). We haveperformed a luciferase reporter assay to determine whether CXCR4mRNA had a target site for miR-9. We used a vector encoding eitherthe partial sequence of the 30-UTR of CXCR4 mRNA, including the

predicted miR-9 target site (positions 402–408), or a vector lacking themiR-9 target site. The results indicate that the luminescence intensitywas significantly reduced by transfection of the wild-type 30-UTR ofCXCR4 while deletion of positions 402–408 blocked the decrease ofluminescence in Tca8113 or SCC-9 cells (Po0.01, Figures 5c–f).

Consequently, the downregulation of miR-9 expression inTca8113 and SCC-9 cells was concurrent with the upregulation

Figure 5. CXCR4 is a direct target gene of miR-9. (a, b): We used a vector encoding either the partial sequence of the 30-UTR of CXCR4 mRNA,including the predicted miR-9 target site (positions 402–408) or a vector lacking the miR-9 target site. (c, f ) Interaction of miR-9 with CXCR430-UTR. After 24 h transfection with miR-9, miR-control or mocks, a reporter plasmid containing CXCR4 wild type-30-UTR or deletion-30-UTR anda plasmid expressing Renilla luciferase (hRluc) were co-transfected into Tca8113 and SCC-9 cells. Firefly luciferase activity was normalized toRenilla luciferase activity. Relative luciferase activity in miR-9 transfectants was compared with that in mock cultures, which was set at 1, in cellstransfected with wild type-30-UTR or deletion-30-UTR. (g, h): CXCR4 protein levels were assessed post-transfection with miR-9 LV (MOI¼ 50),control (100 nM), or inhibitor (100 nM) in Tca8113 and SCC-9 cells. b-Actin loading control is also shown. (i, j) Densitometric analysis of proteinlevels of CXCR4, relative to b-Actin, determined by western blotting and compared with Tca8113 and SCC-9 cells. Error bars indicatemean±s.d.; n¼ 3 experiments; **Po0.01.

miR-9 inhibits the proliferation of OSCCT Yu et al

5023

& 2014 Macmillan Publishers Limited Oncogene (2014) 5017 – 5027

of CXCR4 protein expression. To ascertain whether miR-9 regulatesCXCR4, we employed a gene overexpression approach to examinethe effect of miR-9 on the CXCR4 protein. The Tca8113 and SCC-9cells were transfected with a miR-9 LV or inhibitor. CXCR4 proteinexpression, determined via western blot, was downregulated inmiR-9 mimics-transfected cells relative to the control cells. Bycontrast, CXCR4 protein expression was upregulated in miR-9inhibitor-transfected cells (Figure 5g–j).

MiR-9 inhibits tumorigenesis by regulating expression of CXCR4 viathe Wnt/b-catenin signaling pathway in OSCC cellsCXCR4 is the promoter of the Wnt/b-catenin signaling pathway, sowe hypothesized that miR-9 might play a role relevant to Wnt/b-catenin signaling. To test our hypothesis, we detected b-cateninprotein levels by western blot in miR-9 LV-transfected cells. Theexpression of b-catenin protein levels was substantially reducedby B80% and 90% in miR-9-LV-expressing Tca8113 and SCC-9cells, respectively, compared to their parental cells. Furthermore,we detected the downstream genes of the Wnt/b-cateninsignaling pathway, including Bcl-2, c-myc and c-Jun. Expressionof these genes was tested using western blot in mock, control-and miR-9 LV-transfected cells. Results indicate that the upregu-lated miR-9 could repress protein levels of Bcl-2, c-myc and p-c-

Jun, but did not affect c-Jun (Figures 6a–d). To determine whethermiR-9 functions by targeting CXCR4, we inhibited CXCR4 expres-sion by CXCR4-LV, and tested the protein levels of b-catenin, Bcl-2,c-myc, p-c-Jun and CXCR4. We observed downregulated expres-sion of these proteins, suggesting that CXCR4 is the target gene ofmiR-9 (Figures 6e–h).

DISCUSSIONDespite considerable advances in cancer treatment, the overallsurvival rate of OSCC patients has not markedly improved.Therefore, it is necessary to pursue new concepts to enhancetreatment results and to develop new therapeutic strategies. Inrecent years, emerging evidence has demonstrated an importantrole for miRNAs in regulating human cancers.26 Functional studieshave directly documented the potent pro- and antitumorigenicactivity of specific miRNAs both in vitro and in vivo.27 ThesemiRNAs are involved in tumor cell processes like proliferation,apoptosis, angiogenesis, migration and metastasis.28 Severalstudies have revealed that miR-9 may target multiple signalingpathways, such as the NF-kB1 and Notch signaling pathway, topromote the development and progression of malignancy.29,30

Aberrant miR-9 levels have been reported in many types of cancer,

Figure 6. MiR-9 regulated b-catenin expression and the downstream genes of the Wnt/b-catenin signaling pathway. (a, b) After transductionwith miR-9 LV and miR-control for 72 h in Tca8113 and SCC-9 cells, b-catenin, Bcl-2, c-myc, c-Jun and p-c-Jun protein levels were detected bywestern blotting analysis in the two cell lines. b-Actin loading control is also shown. (c, d) Densitometric analysis of protein levels of b-catenin,Bcl-2, c-myc, c-Jun and p-c-Jun, relative to b-actin, determined by western blotting and compared with Tca8113 and SCC-9 cells. (e, f ) Aftertransduction with CXCR4-LV and miR-9 LV for 72 h in Tca8113 and SCC-9 cells, b-catenin, Bcl-2, c-myc, p-c-Jun and CXCR4 protein levels weredetected by western blotting analysis in the two cell lines. b-actin loading control is also shown. (g, h) Densitometric analysis of protein levelsof b-catenin, Bcl-2, c-myc, p-c-Jun and CXCR4 relative to b-actin, determined by western blotting and compared with Tca8113 and SCC-9 cells.Error bars indicate mean±s.d.; n¼ 3 experiments; **Po0.01.

miR-9 inhibits the proliferation of OSCCT Yu et al

5024

Oncogene (2014) 5017 – 5027 & 2014 Macmillan Publishers Limited

suggesting that miR-9 is involved in tumor formation orprogression. However, by regulating various mRNA targets, miR-9 may have oncomir activity in some types of cancer cells.31,32 Forexample, miR-9 is overexpressed in human primary brain tumors,Hodgkin’s lymphoma cells and some types of gastric cancercells.21,33,34 MiR-9 knockdown inhibits the proliferation of humangastric cancer cells and overexpression of CDX2, a direct target ofmiR-9, has a similar effect. In contrast, downregulation of miR-9 isalso noted in pancreatic cancer, gastric cancer and ovariancancer.35–37 We have taken the position that identification ofnovel cancer pathways and responsible genes regulated by miR-9is an important first step in understanding OSCC oncogenesis.Based on this view, using the algorithm Target Scan (http://www.targetscan.org), miR-9 has seven nucleotides of the ‘seed’region that are complementary to the CXCR4 30-UTR. We predictedthat the CXCR4 gene is a target of post-transcriptional repressionby miR-9 and performed the functional verification. Our luciferasereporter assay demonstrated that miR-9 directly regulated CXCR4in OSCC cells. CXCR4 is a highly conserved seven-spantransmembrane G-protein-coupled receptor that binds theligand CXCL12.38 To date, CXCR4 is one of the most commonchemokine receptors that has been demonstrated to beoverexpressed in more over 23 human cancers.39 CXCR4selectively binds the CXC chemokine stromal cell-derived factor1 (or CXCL12), which has been found to play an important role intumorigenesis, proliferation, metastasis and angiogenesis incancers.24

The Wnt/b-catenin canonical Wnt-signaling pathway has beenimplicated in tumorigenesis at several sites, including the colon,rectum, breast and liver.40 Its central component, namely,b-catenin, plays a critical role in this process. A number ofdownstream target genes of Wnt/b-catenin signaling have beenreported that play critical roles in carcinogenesis by affecting cellgrowth, cell cycling, cell survival and invasion. Aberrant activationof Wnt signaling is a significant feature of human pancreaticadenocarcinoma, as revealed by aberrant b-catenin expression ina significant fraction (30–65%) of tumors.41,42 However, therelative mechanism underlying the upstream and downstreammodulation function of stromal cell-derived factor-1/CXCR4 onOSCC progression is poorly understood.

In this study, we demonstrated that miR-9 was significantlydownregulated in four OSCC cell lines, compared to the humannormal oral keratinocytes. In addition, we analyzed the expressionlevels of miR-9 in 16-clinical tongue oral squamous cell carcinomaspecimens and their adjacent non-cancerous specimens. Theresults indicate that the expression levels of miR-9 weresignificantly lower in 16-clinical tongue OSCC specimens than inadjacent non-cancerous specimens. Lentivirus-mediated miR-9overexpression in vitro significantly inhibited the proliferation,colony formation and invasion of OSCC cells, while inducing theircell cycle arrest and apoptosis. After constructing the xenografttumor models using Tca8113 and SCC-9 cells in nude mice, thetumor volume in the miR-9 LV treatment group was significantlylower than those in the mock and miR-control groups on day 42.These findings demonstrated that miR-9 can be classified as atumor-suppressive microRNA in OSCC and might be a promisingtarget for cancer therapeutics. To further identify the molecularmechanism of miR-9 expression on OSCC tumor progressionin vivo, the CXCR4 and Ki-67 protein expression with immuno-histochemistry were examined. The results indicate that theCXCR4 and Ki-67 protein expression significantly decreased intumors treated with miR-9 LV compared with those in the mockand miR-control groups. These data indicate that lentivirus-mediated miR-9 overexpression could inhibit tumor progressionof OSCC cells both in vitro and in vivo. As CXCR4 expression isinvolved in the Wnt pathway43,44 and the wnt/b-catenin signalingpathway has been reported to be activated in OSCC,45 wespeculated that miR-9 might inhibit tumorigenesis by regulating

Wnt/b-catenin signaling via CXCR4. To elucidate the underlyingmechanism, we detected the expression levels of b-catenin andthe genes downstream of the Wnt/b-catenin signaling pathway,including Bcl-2, c-myc, c-Jun and p-c-Jun. The results showed thatmiR-9 underexpression promoted the expression of CXCR4proteins, which activated the effect of CXCR4 on Wnt/b-cateninsignaling. The activated b-catenin was then translocated into thenucleus, where it in turn activated its downstream effectors andfunctionally contributed to tumorigenesis.46 In the present study,we demonstrated for the first time that miR-9 functions as atumor-suppressive miRNA in OSCC via targeting CXCR4 andinhibiting the Wnt/b-catenin signaling pathway.

Taken together, the results suggest that miR-9 has a crucial rolein OSCC progression through its target gene, CXCR4; thisdisruption contributes to tumorigenesis via the Wnt/b-cateninsignaling pathway. Therefore, miR-9 may be a future therapeutictarget for the treatment of OSCC.

MATERIALS AND METHODSCell lines and reagentsHuman tongue squamous cell carcinoma cell lines, including SCC-4, SCC-9,SCC-25, Tca8113, and a human normal oral keratinocyte cell line wereobtained from ATCC and the State Key Laboratory of Oral Diseases atSichuan University, respectively. The primary antibodies to human b-catenin, Bcl-2, c-myc, c-Jun, p-c-Jun, CXCR4, Ki-67 and b-actin, and thesecondary antibodies and miR-9 inhibitor were purchased from Santa CruzBiotechnology (Santa Cruz, CA, USA). Cell proliferation and cytotoxicityassay kit were from Dojindo Laboratories (Kumamoto, Japan). AnnexinV-FITC/PI Apoptosis kit was purchased from KeyGEN BioTECH (Nanjing,China).

Construction of lentiviral vector for overexpression of miR-9 andknockdown of human CXCR4 expressionTo generate lentivirus-encoding miR-9 (LV-miR-9), we first amplified a386-bp DNA fragment carrying pri-miR-9 from genomic DNA using thefollowing PCR primers: miR-9-sense, 50-CGGAGATCTTTTCTCTCTTCACCCTC-30

and miR-9-antisense, 50-CAAGAATTCGCCCGAACCAGTGAG-30 . The constructswere co-transfected with pHelper 1.0 and pHelper 2.0 Packing Plasmid(GeneChem, Shanghai, China) into 293T packing cells using Lipofectamine2000 (Invitrogen, Carlsbad, CA, USA). Forty-eight hours after transfection, thesupernatant was collected, centrifuged at 1000 r.p.m. for 5 min and filteredthrough 0.22mm pore nitrocellulose filters. LV-GFP, a control virus, wassimilarly produced. OSCC cells were transfected with LV-miR-9 or LV-GFP inthe presence of 4mg/ml polybrene (Sigma-Aldrich, St Louis, MO, USA) at amultiplicity of infection of 10 and centrifuged at 1800 r.p.m. for 45 min.Thereafter, cells were cultured for 72 h and analyzed by fluorescence-activated cell sorting (BD Biosciences, Bedford, MA, USA).

siRNA was designed, based on the CXCR4 sequence (NM_001008540):siRNA: 50-GGTGGTCTATGTTGGCGTCTG-30 . Oligonucleotides with asequence predicted to induce efficient RNAi of CXCR4 (containing senseand antisense sequences) were synthesized (siRNA sense: 50-GATCCCGGGTGGTCTATGTTGGCGTCTGGAA GCTTGCAGACGCCAACATAGACCACCTTTTTT-30 , antisense: 50-CTAGAAAA AAGGTGGTCTATGTTGGCGTCTG-CAAGCTTCCAGACGCCAACATAGACCACCCGG-30 ; These oligonucleotideswere annealed in STE buffer at 94 1C for 5 min and cooled gradually. Thedouble-stranded products were cloned downstream to the human U6promoter of the pRNAT/U6 vector and designated as CXCR4-shRNA. Acontrol vector (NC-shRNA) was constructed (50-GAAGCAGCACGACTTCTTC-30) with no significant homology to any mammalian gene sequence and,therefore, served as a nonsilencing control. The resulting lentiviral vectorcontaining two human CXCR4 shRNA was named CXCR4-lentivirus (CXCR4-LV). A negative control lentiviral vector containing NC-shRNA wasconstructed by a similar process (NC-LV).

Cell culture and treatmentThe cells were cultured at 37 1C in 5% CO2 in DMEM containing 10% fetalbovine serum (FBS, Gibco, Rockville, USA), 2 mM glutamine, 1 mM sodiumpyruvate, 10 mM HEPES, 100 U/ml penicillin G, 100 mg/ml streptomycin(Sigma, St Louis, MO, USA). Tca8113 and SCC-9 cells were infected byaddition of lentivirus into the cell culture at an MOI of approximately 50.

miR-9 inhibits the proliferation of OSCCT Yu et al

5025

& 2014 Macmillan Publishers Limited Oncogene (2014) 5017 – 5027

After culturing for another 72 h, fluorescence-activated cell sorting wasdone for quantitation of infection efficiency.

Real-time quantification of mature miR-9 by stem-loop RT–PCRThe total RNAs from tumor tissues and cells were isolated using TRIzolreagent according to the manufacturer’s instructions. Complementary DNAwas synthesized from total RNA (1mg) in 20ml reactions, using reversetranscriptase (Epicentre, Madison, WI, USA) and the RT primer for miR-9 orRNU48 obtained from Invitrogen. The PCR mixtures were incubated at95 1C for 5 min, and this was followed by 40 cycles of 95 1C for 10 s, 60 1Cfor 20 s, 72 1C for 20 s and 78 1C for 20 s. All PCR reactions were performedin triplicate using a Rotor-Gene 3000 real-time PCR system (CorbettResearch, Mortlake, NSW, Australia). The expression of miR-9 were basedon the 2-delta delta Ct method, using RNU48 as an internal control. Allreactions were performed in triplicate, and included negative controlreactions that lacked cDNA.

Cell proliferation assaysCells transfected with miRNA plated in 96-well plates at 3� 103 cells perwell. After 72 h, cell proliferation was determined with the XTT assay, usingthe Cell Proliferation Kit II (Roche Molecular Biochemicals, Mannheim,Germany) as previously reported.47 Triplicate wells were measured for cellviability in each treatment group. Additionally, we used plate colonyformation assay to evaluate the colony formation ability of tumor cells. Thetumor cells were cultured at 1000 cells/5 ml with DMEM medium and 10%FBS in a six-well plate. After 10 days in culture, the cells were fixed withmethanol for 10 min and stained with 1% crystal violet solution for 20 minto visualize colonies for counting.

Cell cycle assayA total of 1� 106 cells were taken from appropriate samples and stainedwith propidium iodide to stain nuclei. Flow cytometry was done using aFACS Calibur (BD Biosciences) and the fraction of cells in each phase of cellcycle was determined using cell cycle analysis platform in FlowJo software(Tree Star Inc., Ashland, OR, USA).

Apoptosis assayCells (1� 106) were collected, washed and resuspended in phosphate-buffered saline (PBS), annexin V-fluorescein isothiocyanate (FITC; 5 ml/ml;KeyGEN, Nanjing, China) and propidium iodide (PI; KeyGEN) were addedand incubated for 20 min at 4 1C. Cells were analyzed by FACScan flowcytometer (Becton Dickinson) with FlowJo software (Tree Star Inc.).

Cell invasion assayThe cell invasion assay was performed with QCM 24-well Invasion Assay kit(Chemicon International, Temecula, CA, USA). This kit contains 24 insertsand each insert contains an 8-mm-pore-size polycarbonate membranecoated with a thin layer of ECMatrix. Briefly, cells were resuspended inserum-free RPMI 1640 and 3� 105 cells were added to the interior of theinserts, which had been previously rehydrated at room temperature for30 min. Five hundred microliters of RPMI 1640 containing 10% fetal bovineserum was added to the lower chamber as chemoattractant. Cells wereincubated for 48 h at 37 1C in a CO2 incubator (5% CO2). Then, the uppersurface of the chamber was scraped to remove non-invasive cells. Invadedcells were fixed and stained with 0.1% crystal violet and photographedunder a light microscope (� 200). The stained inserts were transferred to aclean well containing 500ml of 10% acetic acid for 15 min at roomtemperature. The optical density of the stained cell was measured at560 nm.

Tumor xenografts model and miR-9 recombinant lentiviral vectortreatmentThe construction of the tumor xenografts model was conducted at theLaboratory Animal Center of Sichuan University (Chengdu, China). Theanimal experiment was authorized by the West China Hospital EthicsCommittees. Thirty-six nude mice (4 weeks of age) were bred in an asepticcondition and kept at a constant humidity and temperature (25–28 1C)with a 12-h light–dark cycle. After 1 week of acclimatization, 0.1 ml ofTca8113 and SCC-9 cell suspension at a concentration of 2� 106 cells/mlwas subcutaneously injected into the right lateral of axilla region. Tumorxenografts were staged for 7 days and reached approximately 150 mm3 in

size. Then, the same volume (0.1 ml) of saline solution (mock), controlvector (miR-control) and miR-9 LV were injected into multiple sitesintratumorally in three different groups, respectively, every 7 days. The sizeof tumors was measured with vernier calipers, and the volume of tumorwas determined using the simplified formula of a rotational ellipsoid(L�W2� 0.5). On day 42, mice were euthanized and the tumors wereremoved for further study.

ImmunohistochemistryFive-micrometer slices of formalin-fixed and paraffin-embedded tissueswere cut onto silanized glass slides, deparaffinized in xylene, rehydrated ingraded ethanol concentrations (100, 95, 70 and 50%) and finallysubmerged in phosphate-buffered saline (PBS). The slices were blockedendogenous pre-oxidase with 3% hydrogen peroxide solution for 15 minand placed in an autoclave with 0.01 M sodium citrate solution at 121 1C for3 min for antigen retrieval. Tissue slices were incubated with the CXCR4, Ki-67 primary antibody overnight at 4 1C and then incubated with biotin-labeled secondary antibody at room temperature for 1 h. Negative controlswere performed by replacing the primary antibody with PBS. DAB(diaminobenzidine tetrahydrochloride) was used as a chromogen. Allslices were then counterstained with hematoxylin and then observed in abright field microscope.

Plasmid construction and luciferase reporter assayPartial wild-type sequences of CXCR4 30-UTR and those with deleted miR-9target sites (positions 402–408) were inserted between the XhoI–PmeIrestriction sites in the 30-UTR of the hRluc gene in psiCHECK-2 vector(Promega, San Luis Obispo, CA, USA). The synthesized DNA was cloned intothe psiCHECK-2 vector. Tca8113 and SCC-9 cells were transfected with10 ng of vector, 10 nM of miR-9 and 1 ml of Lipofectamine 2000 (Invitrogen)in 100 ml of Opti-MEM (Invitrogen). The activities of firefly and Renillaluciferases in cell lysates were determined with a dual-luciferase assaysystem (Promega). Normalized data were calculated as the quotient ofRenilla/firefly luciferase activities.

Western blottingThe cells were lysed in lysis buffer (PBS containing 1% Triton X-100,protease inhibitor cocktail and 1 mmol/l phenylmethylsulfonyl fluoride) at4 1C for 30 min. Equal quantities of protein were subjected to SDS–PAGE.After transfer to Immobilon-P transfer membrane, successive incubationswith anti-b-catenin, bcl-2, c-myc, c-Jun, p-c-Jun, CXCR4 and b-actinantibody, and horseradish peroxidase-conjugated secondary antibodywere carried out. The immunoreactive proteins were then detected usingthe ECL system. Bands were scanned using a densitometer (GS-700, Bio-Rad Laboratories, Richmond, CA, USA) and quantification was performedusing Quantity One 4.6.3 software (Bio-Rad Laboratories).

Statistical analysisData are expressed as mean±s.d., when normally distributed. Thestatistical significance of differences was determined by Student’s two-tailed t-test in 2 groups and one way analysis of varience (ANOVA) inmultiple groups. Differences at probability of less than 0.05 wereconsidered statistically significant. All data were analyzed with SPSS 15.0software.

CONFLICT OF INTERESTThe authors declare no conflict of interest.

REFERENCES1 Kademani D. Oral cancer. Mayo Clin Proc 2007; 82: 878–887.2 Petersen PE. The World Oral Health Report 2003: continuous improvement of oral

health in the 21st century- the approach of WHO Global Programme. CommunDent Oral Epidemiol 2003; 31: 3–23.

3 Yu T, Wang XY, Gong RG, Li A, Yang S, Cao YT et al. The expression profile ofmicroRNAs in a model of 7,12-dimethyl-benz[a]anthrance-induced oral carcino-genesis in Syrian hamster. J Exp Clin Cancer Res 2009; 13: 64.

4 Kessler P, Grabenbauer G, Leher A, Bloch-Birkholz A, Vairaktaris E, Neukam FW.Neoadjuvant and adjuvant therapy in patients with oral squamous cell carcinoma:long-term survival in a prospective, non-randomized study. Br J Oral MaxillofacSurg 2008; 46: 1–5.

miR-9 inhibits the proliferation of OSCCT Yu et al

5026

Oncogene (2014) 5017 – 5027 & 2014 Macmillan Publishers Limited

5 Hu J, He Y, Yan M, Zhu C, Ye W, Zhu H et al. Dose dependent activation of retinoicacid-inducible gene-I promotes both proliferation and apoptosis signals in humanhead and neck squamous cell carcinoma. PLoS One 2013; 8: e58273.

6 Uchida F, Uzawa K, Kasamatsu A, Takatori H, Sakamoto Y, Ogawara K et al.Overexpression of CDCA2 in human squamous cell carcinoma: correlation withprevention of G1 phase arrest and apoptosis. PLoS One 2013; 8: e56381.

7 Ambros V. The functions of animal microRNAs. Nature 2004; 431: 350–355.8 Bartel DP. MicroRNAs: genomics, biogenesis, mechanism and function. Cell 2004;

116: 281–297.9 Calin GA, Croce CM. MicroRNA-cancer connection: the beginning of a new tale.

Cancer Res 2006; 66: 7390–7394.10 Bandyopadhyay S, Mitra R, Maulik U, Zhang MQ. Development of the human

cancer microRNA network. Silence 2010; 1: 6.11 Selcuklu SD, Donoghue MTA, Spillane C. MiR-21 as a key regulator of oncogenic

processes. Biochem Soc T 2009; 37: 918–925.12 Li JS, Huang HZ, Sun LJ, Yang M, Pan CB, Chen WL et al. MiR-21 indicates poor

prognosis in tongue squamous cell carcinomas as an apoptosis inhibitor. ClinCancer Res 2009; 15: 3998–4008.

13 Avissar M, McClean MD, Kelsey KT, Marsit CJ. MicroRNA expression in head andneck cancer associates with alcohol consumption and survival. Carcinogenesis2009; 30: 2059–2063.

14 Wong TS, Liu XB, Ho ACW, Yuen APW, Ng RWM, Wei WI. Identification of pyruvatekinase type M2 as potential oncoprotein in squamous cell carcinoma of tonguethrough microRNA profiling. Int J Cancer 2008; 123: 251–257.

15 Jiang L, Liu XQ, Kolokythas A, Yu JS, Wang AX, Heidbreder CE et al. Down-regulation of the Rho GTPase signaling pathway is involved in the microRNA-138-mediated inhibition of cell migration and invasion in tongue squamous cellcarcinoma. Int J Cancer 2010; 127: 505–512.

16 Gao F, Zhao ZL, Zhao WT, Fan QR, Wang SC, Li J et al. miR-9 modulates theexpression of interferon-regulated genes and MHC class I molecules in humannasopharyngeal carcinoma cells. Biochem Biophys Res Commun 2013; 431: 610–616.

17 Bandres E, Agirre X, Bitarte N, Ramirez N, Zarate R, Roman-Gomez J et al. Epi-genetic regulation of microRNA expression in colorectal cancer. Int J Cancer 2009;125: 2737–2743.

18 Selcuklu SD, Donoghue MT, Rehmet K, de Souza Gomes M, Fort A, Kovvuru P et al.MicroRNA-9 inhibition of cell proliferation and identification of novel miR-9 tar-gets by transcriptome profiling in breast cancer cells. J Biol Chem 2012; 287:29516–29528.

19 Liu S, Kumar SM, Lu H, Liu A, Yang R, Pushparajan A et al. MicroRNA-9 up-regulates E-cadherin through inhibition of NF-kB1-Snail1 pathway in melanoma.J Pathol 2012; 226: 61–72.

20 Sun Z, Han Q, Zhou N, Wang S, Lu S, Bai C et al. MicroRNA-9 enhances migrationand invasion through KLF17 in hepatocellular carcinoma. Mol Oncol 2013; S1574-7891: 00068–00069.

21 Nass D, Rosenwald S, Meiri E, Gilad S, Tabibian-Keissar H, Schlosberg A et al. MiR-92b and miR-9/9* are specifically expressed in brain primary tumors and can beused to differentiate primary from metastatic brain tumors. Brain Pathol 2009; 19:375–383.

22 Leucci E, Zriwil A, Gregersen LH, Jensen KT, Obad S, Bellan C et al. Inhibition ofmiR-9 de-represses HuR and DICER1 and impairs Hodgkin lymphoma tumouroutgrowth in vivo. Oncogene 2012; 31: 5081–5089.

23 Teicher BA, Fricker SP. CXCL12 (SDF-1)/CXCR4 pathway in cancer. Clin. Cancer Res2010; 16: 2927–2931.

24 Burger JA, Kipps TJ. CXCR4: a key receptor in the crosstalk between tumor cellsand their microenvironment. Blood 2006; 107: 1761–1767.

25 Janda CY, Waghray D, Levin AM, Thomas C, Garcia KC. Structural basis of Wntrecognition by Frizzled. Science 2012; 337: 59–64.

26 Garzon R, Calin GA, Croce CM. MicroRNAs in Cancer. Annu Rev Med 2009; 60:167–179.

27 Garzon R, Marcucci G, Croce CM. Targeting microRNAs in cancer: rationale, stra-tegies and challenges. Nat Rev Drug Discov 2010; 9: 775–789.

28 Esquela-Kerscher A, Slack FJ. Oncomirs - microRNAs with a role in cancer. Nat RevCancer 2006; 6: 259–269.

29 Liu N, Sun Q, Chen J, Li J, Zeng Y, Zhai S et al. MicroRNA-9 suppresses uvealmelanoma cell migration and invasion through the NF-kB1 pathway. Oncol Rep2012; 28: 961–968.

30 Jing L, Jia Y, Lu J, Han R, Li J, Wang S et al. MicroRNA-9 promotes differentiation ofmouse bone mesenchymal stem cells into neurons by Notch signaling.Neuroreport 2011; 22: 206–211.

31 Iorio MV, Ferracin M, Liu CG, Veronese A, Spizzo R, Sabbioni S et al. MicroRNAgene expression deregulation in human breast cancer. Cancer Res 2005; 65:7065–7070.

32 Lehmann U, Hasemeier B, Christgen M, Muller M, Romermann D, Langer F et al.Epigenetic inactivation of microRNA gene hsa-miR-9-1 in human breast cancer.J Pathol 2008; 214: 17–24.

33 Nie K, Gomez M, Landgraf P, Garcia JF, Liu Y, Tan LH et al. MicroRNA-mediated downregulation of PRDM1/Blimp-1 in Hodgkin/Reed-Sternberg cells: apotential pathogenetic lesion in Hodgkin lymphomas. Am J Pathol 2008; 173:242–252.

34 Rotkrua P, Akiyama Y, Hashimoto Y, Otsubo T, Yuasa1 Y. MiR-9 downregulatesCDX2 expression in gastric cancer cells. Int J Cancer 2011; 129: 2611–2620.

35 Omura N, Li CP, Li A, Hong SM, Walter K, Jimeno A et al. Genomewide profiling atmethylated promoters in pancreatic adenocarcinoma. Cancer Biol Ther 2008; 7:1146–1156.

36 Luo H, Zhang H, Zhang Z, Zhang X, Ning B, Guo J et al. Downregulated miR-9 andmiR-433 in human gastric carcinoma. J Exp Clin Cancer Res 2009; 28: 82.

37 Laios A, O’Toole S, Flavin R, Martin C, Kelly L, Ring M et al. Potential role of miR-9and miR-223 in recurrent ovarian cancer. Mol Cancer 2008; 7: 35.

38 Caruz A, Samsom M, Alonso JM, Alcami J, Baleux F, Virelizier JL et al. Genomicorganization and promoter characterization of human cxcr4 gene. FEBS Lett 1998;426: 271–278.

39 Balkwill F. Cancer and the chemokine network. Nat Rev Cancer 2004; 4: 540–550.40 Peifer M, Polakis P. Wnt signaling in oncogenesis and embryogenesis-a look

outside the nucleus. Science 2000; 287: 1606–1609.41 Lowy AM, Fenoglio-Preiser C, Kim OJ, Kordich J, Gomez A, Knight J et al. Dysre-

gulation of beta-catenin expression correlates with tumor differentiation inpancreatic duct adenocarcinoma. Ann Surg Oncol 2003; 10: 284–290.

42 Al-Aynati MM, Radulovich N, Riddell RH, Tsao MS. Epithelialcadherin and beta-catenin expression changes in pancreatic intraepithelial neoplasia. Clin Cancer Res2004; 10: 1235–1240.

43 Jin Z, Zhao C, Han X, Han Y. Wnt5a promotes ewing sarcoma cell migrationthrough upregulating CXCR4 expression. BMC Cancer 2012; 12: 480.

44 Choe Y, Pleasure SJ. Wnt signaling regulates intermediate precursor production inthe postnatal dentate gyrus by regulating CXCR4 expression. Dev Neurosci 2012;34: 502–514.

45 Fracalossi AC, Silva Mde S, Oshima CT, Ribeiro DA. Wnt/beta-catenin signallingpathway following rat tongue carcinogenesis induced by 4-nitroquinoline 1-oxide. Exp Mol Pathol 2010; 88: 176–183.

46 Farr GH, Ferkey DM, Yost C, Pierce SB, Weaver C, Kimelman D. Interaction amongGSK-3, GBP, axin, and APC in Xenopus axis specification. J Cell Biol 2000; 148:691–702.

47 Chiyomaru T, Enokida H, Tatarano S, Kawahara K, Uchida Y, Nishiyama K et al. miR-145 and miR-133a function as tumour suppressors and directly regulate FSCN1expression in bladder cancer. Br J Cancer 2010; 102: 883–891.

Supplementary Information accompanies this paper on the Oncogene website (http://www.nature.com/onc)

miR-9 inhibits the proliferation of OSCCT Yu et al

5027

& 2014 Macmillan Publishers Limited Oncogene (2014) 5017 – 5027