MiR‐139‐5p inhibits the tumorigenesis and progression of oral squamous carcinoma cells by targeting HOXA9

Abstract Our study sought to clarify the effects of microRNA‐139‐5p (miR‐139‐5p) in the tumorigenesis and progression of oral squamous cell carcinoma (OSCC) by regulating HOXA9. MiR‐139‐5p and HOXA9 expression in OSCC tissues, tumour adjacent tissues, OSCC cells and normal cells were tested by qRT‐PCR. SAS and CAL‐27 cell lines were selected in among four OSCC cell lines and then transfected with miR‐139‐5p mimics, pEGFP‐HOXA9 and cotransfected with miR‐139‐5p mimics + pEGFP‐HOXA9. We used MTT, colony formation, transwell and wound healing assays to analyse cell viability, proliferation, invasion and migration. The target relationship between miR‐139‐5p and HOXA9 was verified by luciferase reporter assay and Western blot, respectively. MiR‐139‐5p was down‐regulated, whereas HOXA9 was up‐regulated in OSCC tissues and cells. The proliferation, invasion and migration ability of SAS and CAL‐27 cells in miR‐139‐5p mimics group were significantly weaker than those in the control group and the miR‐NC group (P < 0.01). MiR‐139‐5p can negatively regulate HOXA9. The proliferation, invasion and migration of SAS and CAL‐27 cells in the miR‐139‐5p mimics + pEGFP‐HOXA9 group were not significantly different from those in the blank control and negative control groups (P > 0.05). Our results indicated that miR‐139‐5p could directly inhibit HOXA9, which might be a potential mechanism in inhibiting the proliferation, invasiveness and migration of OSCC cells.


Introduction
Oral cancer originates from the epithelial cells of the oral cavity, which is widely known as one of the most frequently developed malignant tumours among head and neck carcinomas [1]. Pathologically, over 90% of all patients suffering from oral cancer are diagnosed as OSCC with well or moderately differentiation [2]. When it comes to the treatment of OSCC, we often give priority to surgery with adjuvant radiotherapy or systemic chemotherapy based on the pathological type, the disease stage, patients' physical state and other correlated factors [3]. In spite of advanced therapy, the 5-year survival rate of OSCC remains unsatisfactory and even lower than 50% when diagnosed during the middle or late stage of the disease [4]. These following cancers are malignant, and have a tendency to attack locally, they are relative with a high risk of recurrence [5].The poor prognosis of OSCC can be attributed to the comprehensive factors including the late diagnosis, radiation resistance, recurrence at the primary site and frequent distant metastasis after treatment [6,7]. The mechanism that is responsible for tumour cells migration and invasion is complicated, which remains unknown. Therefore, more efforts need to make in accounting the potential molecular mechanism for OSCC tumorigenesis.
MicroRNAs (miRNAs) are a large family of non-coding endogenous RNA molecules, consisting of 18-22 nucleotides [8]. MiRNAs execute functions in RNA silencing by binding to the 3 0 -untranslated region (UTR) of the target messenger RNAs (mRNAs) [9]. In the past decades, emerging evidence has demonstrated that miRNAs play important roles in diverse biological processes including cell proliferation, metastasis, survival, autophagy and differentiation [10,11]. A large sum of miRNAs has been described as cancer genes and their deregulations have been measured in a series of tumours, including OSCC [12]. Furthermore, recent studies have revealed that miRNAs serve as either onco-miRNAs or tumour suppressors during cancer progression and tumorigenesis [13,14]. Previous research has demonstrated that a variety of miRNAs that are abnormally expressed in OSCC, such as miR-301b, miR-96, miR-194, miR-221 and miR-139-5p [15][16][17][18][19]. MiR-139-5p has been reported to be down-regulated in various carcinomas, such as bladder cancer, colorectal cancer, prostate cancer and breast cancer [20][21][22][23]. Besides, it was documented to participate in regulating cell proliferation and invasion of human non-small cell lung cancer [24], oesophageal squamous cell carcinoma [25], hepatocellular carcinoma [26], glioblastoma multiforme [19] and so on. However, studies about the regulatory role of miR-135-5p for OSCC cells remained limited; therefore, the clinical value and prognostic potential of miR-139-5p for OSCC still needed to be further investigated. The targeted relationship of miR-139-5p and HOXA9 has been confirmed in other disorders, such as acute myeloid leukaemia [27], and HOXA9 has been indicated to be involved in development of OSCC [28]. Moreover, HOXA9 also participated in the development of other carcinomas. For instance, Li et al. detected that HOXA9 had essential oncogenic effect in leukaemia [29]. Besides, Liu et al. found that the expression of HOXA9 could be enhanced by miR-196b in chronic myeloid leukemogenesis [30]. Mammalian HOX genes are organized in four clusters (HOXA, HOXB, HOXC and HOXD), each containing nine to 11 genes [31]. HOXA9, 180-nucleotide in length, is a member of mammalian HOX family [32]. HOXA9 is required to maintain the development of proper limb, skeletal, mammary gland, urogenital tract and kidney [33]. HOXA9 also plays a role in normal myeloid, erythroid and lymphoid haematopoiesis [34]. The overexpression of HOXA9 remarkably expands haematopoietic stem cells, suggesting its function in early haematopoiesis and thus, it might affect the angiogenesis during the tumorigenesis process [32].
Here, we quantified the expression of miR-139-5p and HOXA9 mRNA in 40 OSCC tissue samples by qRT-PCR. MiR-139-5p was significantly down-regulated in both tissues and cells, whereas HOXA9 was completely opposite. Dual luciferase reporter assay verified that miR-139-5p directly targeted HOXA9. Furthermore, cellular assays were, respectively, performed in vitro to detect the role that miR-139-5p plays in a series of cell activities. In all, we found that miR-139-5p severed as an anti-oncogenic role in OSCC and suppressed cell proliferation, invasion and migration through inhibiting HOXA9.

Patients and samples
Forty cases of OSCC tissues and corresponding adjacent tissues were obtained from OSCC patients who underwent OSCC resection without receiving preoperative chemotherapy or radiation at Henan Provincial People's Hospital from January 2014 to January 2016. Among the patients, there were 24 men and 16 women, and the average age of the patients was around 49.8 AE 5.3 years old. The OSCC tissues were all pathologically identified as OSCC, and the numbers of high-differentiated, medium-differentiated and poor-differentiated cases were, respectively, 13, 8 and 19. According to the tumour node metastasis (TNM) staging, the included OSCC tissues were classified as stage I (n = 5), stage II (n = 10), stage III (n = 15) and stage IV (n = 10). Tissue samples were immediately frozen in liquid nitrogen and stored at À80°C until analysis. The study was approved by the ethics committee of Henan Provincial People's Hospital. All patients included in this study have signed consent forms.

Cell culture
The normal oral keratinocyte NOK cell line and human OSCC cell lines SAS, TCA8113, KON were supplied by Henan Provincial People's Hospital cell centre and CAL-27 cell line was purchased from ATCC. All cells were all cultured in 10% foetal bovine serum (FBS) dulbecco's modified eagle medium (DMEM) at 37°C in a humidified atmosphere with 5% CO 2 .

Quantitative real-time PCR (qRT-PCR)
Total RNA was extracted from the tumour tissues and adjacent tissues or cells using Trizol RNA extraction kit. Then, we reverse-transcribed the RNAs to cDNAs using RT Kit (Fermentas, Waltham, MA, USA), and cDNAs were amplified using PCR Kit (Invitrogen, Carlsbad, CA, USA). We listed the used primers in Table 1. The mRNA expressions were the relative expression compared with U6 and GAPDH expressions in the same samples. U6 served as the internal control for miR-139-5p, and GAPDH served as the internal control for HOXA9. And the 2 ÀMMct method was used to calculate their relative expressions.

MTT assay
Cells of each group were plated into 96-well plates at 5 9 10 3 cells/ well. Then after incubation of the cells for 24, 48 and 72 hrs, respectively, we added 20 ll MTT solutions (5 mg/ml) to each well. After incubated for 4 hrs, we added 150 ll DMSO to each well in order to promote the dissolution of crystals. Cell viability was detected at 0, 24, Table 1 The sequences of primers used in qRT-PCR

Primer
Sequence

Colony formation assay
Cells of each group were plated into six-well plates at a concentration of 500 cells/well after transfection for 24 hrs. After the cells adhered to the wall, 0.1% DMSO was applied to act on the cells for 10 days. After being washed with PBS, the cells were fixed by 4% paraformaldehyde, and stained with 5% crystal violet. Then the numbers of colonies were recorded.

Transwell assay
Cells of each group were plated into the upper Transwell chamber at a concentration of 1 9 10 5 cells/ml, and 600 ll 10% FBS DMEM was added to the lower chamber. After incubation for 24 hrs, the membrane was fixed by 70% ethanol and stained with 0.1% crystal violet. Finally, the number of cells that passed across the membrane was counted in six randomly selected fields under the microscope.

Wound healing assay
Cells of each group were plated in 35 mm 2 culture dish at 8 9 10 5 cells/dish. When the cells grew to a degree that 90% cells were merged, lines were drawn at the bottom of dish with a mark as markers, and scratched at the cell layers of each group with a 200 ll sterile pipette. After cells were incubated for 0, 24 or 48 hrs later, photographs were taken, respectively. The intersection was considered as the observation point, and the open wound rate was measured using a microscope.

Western blotting
After culturing the transfected cells for 48 hrs, the total protein was extracted. The electrophoretic separation with 10% dodecyl sulphate, sodium salt-polyacrylamide gel electrophoresis (SDS-PAGE) was conducted, and each lane was loaded with 50 lg of total proteins. The total proteins were then transferred to polyvinylidene fluoride (PVDF) membrane which was subsequently blocked for 1.5 hrs using 5% skim milk. Primary antibodies of HOXA9 and GAPDH (Abcam, Cambridge, MA, USA) diluted at 1:1000 were added onto the membrane and stained overnight at 4°C. After the membrane was washed three times using TBST, goat anti-rabbit HRP horseradish peroxidase-labelled antibodies (1:1000) were added and incubated. electro-chemi-luminescence (ECL) was used to develop the film.

Data analysis and statistical methods
Receiver operating characteristic (ROC) curve analysis was performed using SPSS 15.0 (IBM, Armonk, New York, USA) to calculated for evaluated the predictive ability of miR-139-5p to distinguish OSCC patients from normal patients. Statistical analyses were conducted using SPSS 21.0 and GraphPad Prism 6.0 (GraphPad Software Inc., San Diego, CA). Measurement data were recorded as mean AE S.D. The comparison between two groups was analysed using t-test, whereas one-way ANOVA test was used for the analyses among more than two groups, followed by post hoc Bonferroni's test. P < 0.05 was statistically significant.

MiR-139-5p was under-expressed in OSCC tissues and diverse OSCC cells
The miR-139-5p level in OSCC tissues and adjacent tissues was measured by qRT-PCR, which indicated that miR-139-5p in OSCC tissues was remarkably decreased compared to adjacent tissues (P < 0.01, Fig. 1A). MiR-139-5p expression was also remarkably downregulated in four OSCC cell lines SAS, TCA8113, KON and CAL-27 compared with the normal cell line (P < 0.01, Fig. 1B). The succeeding experiments were conducted in SAS cell line in which miR-139-5p expression was the lowest. The expression of HOXA9 mRNA in OSCC tissue samples and adjacent tissues was compared by qRT-PCR and results demonstrated a remarkable increasing of HOXA9 mRNA in OSCC tissues compared to adjacent tissues (P < 0.01, Fig. 1C). HOXA9 mRNA expression level was remarkably up-regulated in four OSCC cells SAS, TCA8113, KON and CAL-27 compared with the normal cell line (P < 0.01, Fig. 1D). Further analysis in Table 2 showed that miR-139-5p expression was strongly associated with tumour size and clinical stage, whereas HOXA9 level was correlated to pathological differentiation according to Fisher's exact test (n = 40). Probably Table 2 The baseline characteristics of OSCC patients included (n = 40)

Characteristic
Case. MiR-139-5p inhibited the viability, proliferation, invasion and migration of SAS and CAL-27 cell lines Firstly, we generated ROC curves and found that miR-139-5p exhibited AUC of 0.985, HOXA9 exhibited AUC of 1 ( Fig. 2A), indicating sufficient power to distinguish OSCC tissues from adjacent tissues. Then, results of qRT-PCR suggested that the miR-139-5p expression in miR-139-5p mimics group is significantly higher than that in control and miR-NC groups in both cell lines(**P < 0.01 compared with control in SAS cell line, ## P < 0.01 compared with control in CAL-27 cell line, Fig. 2B). Next, results of MTT assay showed that cells of miR-139-5p mimics group had significantly lower viability than control and miR-NC groups (**P < 0.01 compared with control in SAS cell line, ## P < 0.01 compared with control in CAL-27 cell line, Fig. 2C), suggesting that miR-139-5p inhibited the viability of SAS and CAL-27 cell lines Colony formation assay was used to detect the colony forming ability. Similar to the results of MTT assay, as shown in Figure 2D, both cell lines in miR-139-5p mimics group had fewer colony formations than control groups (**P < 0.01 compared with control in SAS cell line, ## P < 0.01 compared with control in CAL-27 cell line), indicating that miR-139-5p could inhibit SAS and CAL-27 cells proliferation. Wound healing assay was used to assess the migration of SAS and CAL-27 cells. The results suggested that open wound rate of cells in the control and miR-NC groups surpassed that in the miR-139-5p mimics group (**P < 0.01 compared with control in SAS cell line, ## P < 0.01 compared with control in CAL-27 cell line, Fig. 3A). The results revealed that miR-139-5p inhibited the migration of SAS and CAL-27 cells. And the invasion of SAS and CAL-27 cells was measured by Transwell assay. The results of transwell assay indicated that cells in the miR-139-5p mimics group had weaker invasiveness than cells in the control groups (**P < 0.01 compared with control in SAS cell line, ## P < 0.01 compared with control in CAL-27 cell line,  HOXA9 is the target of miR-139-5p HOXA9 was predicted to be a target of miR-139-5p with the possible binding sequence ACUGUAGA in HOXA9 3 0 UTR by the online database TargetScan Human 7.0 (Fig. 4A). The luciferase activity of HOXA9-wt 3 0 UTR + miR-139-5p mimics decreased significantly (P < 0.01, Fig. 4B), while that of HOXA9-mut 3 0 UTR + miR-139-5p mimics was not significantly different from the control groups. It suggested that miR-139-5p can directly inhibit HOXA9.
The suppression effect of miR-139-5p on HOXA9 was evaluated at protein levels by Western blot. HOXA9 protein level was downregulated in SAS and CAL-27 cells transfected with miR-139-5p mimics in comparison with the control groups, further verifying the inhibition of miR-139-5p on HOXA9 (**P < 0.01 compared with control in SAS cell line, ## P < 0.01 compared with control in CAL-27 cell line, Fig. 4C).

MiR-139-5p inhibited the viability of SAS and CAL-27 cells by suppressing HOXA9
Firstly, the results of qRT-PCR demonstrated that the miR-139-5p expression in miR-139-5p mimics + pEGFP-HOXA9 group was remarkable higher than that of other groups (**P < 0.01 compared with control in SAS cell line, ## P < 0.01 compared with control in CAL-27 cell line, Fig. 5A). It turned out that the expression of miR-139-5p was not affected by the overexpression of HOXA9. Then, in Figure 5B and C, HOXA9 mRNA expression and protein level were significantly up-regulated in pEGFP-HOXA9 group compared with the pEGFP-NC1 group (**P < 0.01 compared with control in SAS cell line, ## P < 0.01 compared with control in CAL-27 cell line), while HOXA9 protein level was almost identical in miR-139-5p mimics + pEGFP-HOXA9 and pEGFP-NC2 groups. The results showed that miR-139-5p inhibited the expression of HOXA9.
In MTT assay, the overexpression of HOXA9 (pEGFP-HOXA9 group) significantly increased cell viability, whereas the viability of SAS and CAL-27 cells in miR-139-5p mimics + pEGFP-HOXA9 group were not significantly different from that in corresponding controls (P > 0.05, Fig. 5D). These findings indicated that miR-139-5p could eliminate the increased cell viability of OSCC cells induced by the overexpression of HOXA9.

MiR-139-5p inhibited the proliferation, invasion and migration of SAS and CAL-27 cells by suppressing HOXA9
The cell proliferation of miR-139-5p on HOXA9 was assessed by colony formation assay. Compared with the control and pEGFP-NC groups, the pEGFP-HOXA9 group formed more colonies while the colony formation of SAS and CAL-27 cells in miR-139-5p mimics + pEGFP-HOXA9 group were not different from that in corresponding control groups (**P < 0.01 compared with control in SAS cell line, ## P < 0.01 compared with control in CAL-27 cell line, Fig. 5E).
The cell migration of SAS and CAL-27 cells was detected by wound healing assay. As shown in Figure 6A, the open wound rate of cells in the pEGFP-HOXA9 group was higher than that in the control groups. No significant difference in the open wound rate was observed among the miR-139-5p mimics + pEGFP-HOXA9 group, the the control group and the pEGFP-NC group (**P < 0.01 compared with control in SAS cell line, ## P < 0.01 compared with control in CAL-27 cell line).
Then Transwell assay was conducted to confirm the suppression effect of miR-139-5p on HOXA9 in cell invasion. The pEGFP-HOXA9 group had stronger invasiveness in comparison with the control group and the pEGFP-NC group. There was no remarkable difference in the invasiveness of SAS and CAL-27 cells among the miR-139-5p mimics + pEGFP-HOXA9 group, the control group and the pEGFP-NC group (**P < 0.01 compared with control in SAS cell line, ## P < 0.01 compared with control in CAL-27 cell line, Fig. 6B).

Discussion
More attention is paying to the explosion of potential molecular mechanism responsible for OSCC due to its increased incidence in both developed and developing countries [35][36][37][38]. Among previous studies, most abnormally expressed miRNAs were discovered in the OSCC. For instance, Stuart Hunt et al. found that miR-124 suppressed OSCC motility by targeting ITGB1 [39]. Shiah et al. demonstrated that down-regulated miR-329 and miR-410 promoted the proliferation and invasion of OSCC cells by targeting wnt-7b [40]. Wang et al. detected that miR-433 inhibits OSCC cell growth and metastasis by targeting HDAC6 [41]. Of note, a number of target genes can be regulated by a single miRNA instantly. For example, it has been reported that miR-139-5p suppresses the non-small cell lung cancer cell proliferation and induces apoptosis via inhibiting c-Met [42]. Besides, miR-139-5p is able to repress the activation of AMFR and NOTCH1 in colorectal cancer [43]. Also, miR-139-5p functions as a master regulator of glioblastoma metastasis through targeting ZEB1 and ZEB2 [19]. Furthermore, the increased expression level of miR-139-5p can lead to suppressed cell viability and proliferation in various cancers through cell cycle arrest, which induces more cells arrested at G0/G1 or S phase [44,45]. Besides, cell cyclin involving E1, D1, MMP2 and MMP9 is significantly changed in cell transfected with miR-139-5p mimics [25,46]. Furthermore, a decrease in the apoptotic protein such as Bax [47,48] and an increase in the anti-apoptotic protein including Bcl-2 as well as Mcl-1 [47,49], regulated by miR-139-5p, also participates in the tumour process. In the present study, miR-139-5p was found down-regulated in the OSCC sample tissues, and its up-regulation suppressed the viability, proliferation, migration and invasiveness of OSCC cells. Notably, emerging evidence has highlighted that miR-139-5p served as a tumour suppressor in various carcinomas, including bladder cancer, colorectal cancer, prostate cancer and breast cancer [20][21][22][23]. Besides, miR-139-5p has also been confirmed as a prognostic element for progression of breast cancer [50]. Therefore, the suppressor role of miR-139-5p in OSCC proposed in our study was convincing, and we should explore it further.
To better understand the precise molecular mechanism that is responsible for the antitumour effects of miR-139-5p in OSCC, we searched potential target genes of miR-139-5p using TargetScan (http://www.targetscan.org). HOXA9 was predicted to harbour one highly conservative miR-139-5p binding sites in the sequence of 3 0 UTR. With the use of dual luciferase reporter assays, HOXA9 was eventually identified as a direct target for miR-139-5p. The increase in miR-139-5p expression was accompanied by the down-regulated HOXA9 expression. QRT-PCR results revealed that HOXA9 was upregulated in SAS cells, which was inconsistent with some of the previous studies. Ruiz et al. found that HOXA9 was under-expressed in cervical cancer cells and its restoration decreased the proliferation, migration and expression of epithelial-to-mesenchymal transition (EMT) genes [51]. Hwang JA et al. detected that HOXA9 inhibits lung cancer cells migration and its hypermethylation is closely linked with the recurrence in non-small cell lung cancer [52]. Miso et al. found that the HMGA2-TET1-HOXA9 signalling was coordinately regulated in breast cancer. Both TET1 and HOXA9 suppressed breast cancer development in nude mice [53]. On the contrary, HOXA9 was believed to simulate ovarian cancer progression by inducing peritoneal macrophages to acquire an M2 tumour-promoting phenotype [54]. Similarly, Ko et al. asserted that the HOXA9 expression in epithelial ovarian cancer cells promised a comfortable environment for tumour growth [55]. The work performed by Pojo et al. discovered that HOXA9 promoted human glioblastoma initiation and aggressiveness [56]. In terms of the promoter role of HOXA9 during OSCC development and progression, our results verified that miR-139-5p suppressed OSCC cell viability, proliferation, invasion and migration by directly inhibiting HOXA9. We also found that the effects of HOXA9 could be deteriorated by exogenous miR-139-5p. But whether HOXA9 contributed to the alteration of tumour microenvironments need further proved.
Still, there are several limitations in our study. Due to the resource constraint of clinical samples, only 40 pairs of human OSCC tissues and matched adjacent tissues were involved in the present study. Such a small sample size might result in minor errors in the relative expression of miR-139-5p. We intended to future research to verify the above conclusions. Besides, the correlated signalling pathway might help explain the molecular mechanism of anti-oncogenic effect of miR-139-5p on OSCC cells. Moreover, we verified the anti-oncogenic effect of miR-139-5p in SAS OSCC cell line in vitro, while the other OSCC cell lines as well as the experiments in vivo (i.e. in established animal models) were lacking. In addition, the role of miR-139-5p and HOXA9 in other mechanisms, such as EMT, also needed to be further investigated, which could provide evidence for discovering novel targets for OSCC. As a result, further researches should pay attention to resolving the above limitations.

Conclusion
In summary, the present study demonstrated that miR-139-5p, functioning as a suppressor in the tumorigenesis process, suppressing OSCC cell mobility by targeting HOXA9 directly. The relative expression of miR-139-5p and HOXA9 was quantified by qRT-PCR. Dual luciferase reporter gene assay verified that HOXA9 was a target for miR-139-5p. In vitro assays performed with OSCC cells further detected the role of miR-139-5p played during tumorigenesis. Combined with previous studies on miR-139-5p, we believe that miR-139-5p might be a crucial molecular that participates in the OSCC tumorigenesis. Anyway, we hope that the results might provide a therapeutic strategy for the treatment of OSCC.