Transcription Factor Runx1 Activates Opn to Promote Tumor Progression in Head and Neck Cancer

Background:Metastatic progression remains a major burden for head and neck squamous cell carcinoma (HNSCC). Runt-related transcription factor 1 (RUNX1)has been reported to be associated with an aggressive phenotype in several cancers. However, the precise roles of RUNX1 underlying the metastaticprogression of HNSCC remain largely unknown. Methods:RUNX1 expression levels in HNSCC cells and tissues were detected by quantitative real-time PCR (qPCR), Western blottingand immunohistochemistry (IHC). In vitro and in vivo assays were performed to investigate the function of RUNX1 in the metastatic phenotype and the tumorigenic capability of HNSCC cells. Luciferase reporter and chromatin immunoprecipitation (ChIP)-qPCR assays were performed to determine the underlying mechanism of RUNX1-mediated HNSCC aggressiveness. Results:RUNX1 was increased with disease progression in patients withHNSCC.Furthermore, we found that silencing ofRUNX1 signicantly decelerated the malignant progression of HNSCC cells and reduced Osteopontin (OPN) expression in vitro, and weakened the tumorigenicityof HNSCC cells in vivo. Mechanistically, we demonstrated that RUNX1 played an important role in activating MAPK signaling by directly binding to the promoter of OPN. Conclusions: Our results provide new insight into the mechanisms underlying the facilitate metastasisability of RUNX1and reveal the therapeutic potential of targeting RUNX1 in HNSCC.


Background
Metastatic progression remains a major burden for head and neck squamous cell carcinoma (HNSCC) and is associated with eventual resistance to prevailing therapies [1]. Signi cantly, complex molecular transcriptional programs and downstream signaling pathways have been implicated in the development of HNSCC from premalignancy and progression to invasion, metastasis, and treatment resistance [2]. Runt-related transcription factor 1 (RUNX1), also known as acute myeloid leukemia 1 protein (AML1), is a member of RUNX family of transcription factors (RUNX1, RUNX2 and RUNX3), and is an essential master transcription factor implicated in basic cellular and developmental processes, stem cell biology and tumorigenesis [3,4]. According to previous studies, RUNX1 plays a pivotal role in de nitive hematopoiesis, is one of the most frequently mutated genes in a variety of haematological malignancies [5]. Furthermore, gene expression pro les of metastatic adenocarcinomas (lung, breast, prostate, colorectal, uterus and ovary cancers) revealed that RUNX1 is a gene whose expression pattern predicts metastasis [6]. In particular, RUNX1 was found overexpressed in oral SCC, and was essential for the growth and survival [7].
Although much is known about the roles of RUNX1 in different cancers, the mechanisms in the regulation of genes that are intimately associated with tumor progression, invasion and metastasis should be of greater concern. These genes include osteopontin, bone sialoprotein and matrix metalloproteinase 13, all of which have been implicated in metastasis [8,9].
Identifying transcription factor that contribute to the modulation of OPN expression is of interest for therapy that aims to control the OPN-mediated metastatic phenotype. Several known cis-acting transcription factors have been identi ed, such as Myc, AP-1, Oct-1, Wnt/β-catenin/APC/GSK-3b/Tcf-4, Ras/RRF, TF53 [16]. In addition, a previous study showed that RUNX1 has played a key role in the upregulation of the OPN gene in metastatic tumor cells, such as lung, melanoma, leukemia, colon, breast cancer cell lines [17]. However, it is not yet known whether a similar transcriptional regulation occurs between RUNX1 and OPN in HNSCC.
In this study, we set out to elucidate the function of RUNX1 in HNSCC and the role of RUNX1 in transcription regulation of OPN. The results con rmed the clinical relevance of RUNX1 and OPN in HNSCC. Our nude mice xenograft models veri ed that RUNX1 expression increases concomitant with disease progression. Moreover, in vitro studies established that RUNX1 is associated with higher migration and invasion ability. Signi cantly, we found RUNX1 has the binding ability in the OPN regulatory sequence, and OPN is known to in uence the malignant phenotype of tumor cells. Thus, this evidence suggested that, in HNSCC, RUNX1 play an important role in the migration and invasion of tumor cells to control their migratory properties. Its active role in activating the OPN gene in tumor cells may help these cells to show metastatic phenotypes.

Patient samples and Cell lines
This study was approved by the Ethics Committee of BenQ Medical Center, the A liated BenQ Hospital of Nanjing Medical University (Nanjing, China), and patients consented to the use of the tissue specimens for research purposes. HNSCC tissues that were surgically resected were obtained from the department of pathology, BenQ Medical Center. The specimens were classi ed according to the 2018 NCCN criteria and TMN staging system. Histologic classi cation and tumor stage were reviewed by two pathologists. The HNSCC cell lines FaDu and SCC-9 were purchased from Guangzhou Cellcook Biotech Co., Ltd (Guangzhou, China). The cells were authenticated by STR pro ling. FaDu cells were maintained in RPMI-1640 (Gibco, USA) medium with 10% fetal bovine serum (FBS, Gibco, USA) and 1% Penicillin/Streptomycin (Gibco, USA), SCC-9 cells were maintained in Dulbecco's modi ed Eagle's medium (DMEM/F12, Gibco, USA) with 10% FBS and 1% Penicillin/Streptomycin, supplemented with sodium pyruvate (CellCook, China) and hydrocortisone (CellCook, China). 293T cells were maintained in Dulbecco's modi ed Eagle's medium (DMEM/F12, Gibco, USA) with 10% FBS and 1% Penicillin/Streptomycin, All cells were cultured at 37 °C in a humidi ed 5% CO 2 atmosphere.

Lentiviral Transduction
The overexpression and knockdown of RUNX1 and OPN were performed using a lentiviral packaging system. Cells were grown to 60% con uency and infected with lentiviral vectors targeting RUNX1 and OPN (Shanghai Genechem Co., Ltd. Shanghai, China), or with negative control vectors in the presence of 10 mg/mL Polybrene (hexadimethrine bromide). The supernatant was removed after 12 hours and was replaced with complete culture medium. After 72 hours of transduction, the cells were collected for further experiments.

Cell Proliferation Assay
A Cell Counting Kit-8 (CCK-8) (Dojindo, Japan) was used to determine cell viability. 1 × 10 3 cells were plated in 96-well plates and incubated for 12 hours. Then medium containing CCK-8 solution (10 mL CCK-8 in 100 mL medium) was added to each well at the same time every day for 3 days. The OD values were detected at absorbances of 450 nm and 630 nm. All experiments were performed in triplicate.

Colony Formation Assay
For clonogenicity analysis, 500 viable cells were seeded in 6-well plates. Culture medium was changed every two days. After 10 days of incubation, colonies were xed with 4% paraformaldehyde and stained with giemsa (Solarbio, China). The cells were photographed and the numbers of colonies were scored.

Migration Assay
For the scratch assays, 4 × 10 5 cells were seeded in 6-well plates and allowed to adhere overnight until they reached 95% ~ 100% con uency. Cells were serum-starved for 8 hours prior to beginning the assay. Subsequently, a scratch was made across the cell layer using a 100 µl pipette tip, and cell migration was monitored by recording images after 24 hours. The area of the scratch was quanti ed using the MiToBo plug-in for ImageJ software and plotted as a percentage of total area. For the transwell migration assay, 5 × 10 4 cells were seeded in triplicate in migration chambers (BD Biosciences, Bedford, MA, USA) in serum-free medium. Cells were allowed to migrate through 8 µm pores toward medium containing 10% FBS for 48 hours. Non-migrating cells were removed and cells that migrated through the membrane were xed with 4% paraformaldehyde. Fixed cells were stained with giemsa (Solarbio, China) solution and photographed. The number of cells on the membrane was determined by counting under a microscope.

Invasion Assay
For the invasion assay, 5 × 10 4 cells were seeded in triplicate in Matrigel invasion chambers (BD Biosciences, San Jose, CA, USA) in serum-free medium. Cells were allowed to invade through the Matrigel and 8 µm pores toward medium containing 10% FBS for 48 hours. Non-invading cells were removed using a cotton swab and cells that invaded through the membrane were xed with 4% paraformaldehyde. The procedure for transwell migration described above was then followed to complete the experiments.

Immunohistochemistry
Both the human tissue and mouse tissue are the same immunohistochemical procedure. Following depara nization and rehydration, antigen retrieval was performed using DAKO Target

Western Blotting
Whole cell protein lysates were generated using RIPA buffer; nuclear and cytoplasmic protein lysates were generated using the NE-PER nuclear and cytoplasmic extraction kit (Thermo Scienti c, USA). Both protein isolation reagents were supplemented with 25 µM MG132 and complete protease inhibitor cocktail (Roche Diagnostics, USA). Lysates were separated on a 10% acrylamide gel and immobilized on PVDF membranes (Millipore, USA). Blots were blocked using 5% bovine serum albumin (BSA, HyClone, China) before being incubated overnight at 4 °C with the following primary antibodies: anti-RUNX1 ((1:1000, Abcam, USA), anti-OPN (1:1000, Abcam, USA) and anti-GAPDH (1:1000, Proteintech, USA). GAPDH was used as loading controls. Secondary antibodies conjugated to HRP (Proteintech, USA) were used to detect proteins in conjunction with an enhanced chemiluminescent reagent (ECL; Millipore, USA) in a Bio-Rad Image Lab system. The bands were quanti ed using Imaging J analysis software.

Luciferase Reporter Assay
At 12-24 hours before transfection, 293T cells were plated at 2 × 10 5 cells per well in a six-well cell culture plate. A 2 µg measure of various DNA constructs and 0.3 µg of Renilla construct (Promega, USA) were mixed with 10 µl of lipo3000 (Qiagen, German). The mixture was incubated at room temperature for 20 minutes. After washing the cells with PBS, the DNA/ lipo3000 mixtures were transferred to the cells and incubated at 37 °C in a CO 2 incubator for 6 hours. The supernatant was replaced with complete culture medium. When cotransfection with constructs expressed various transcription factors, appropriate control plasmids pGL3 was also transfected into separate cultured cells as controls, and DNA was maintained in equal amounts. Subsequently, the transfected cells were washed with PBS and cultured for an additional 48 hours. At the end of 48 hours incubation, the transfected cells were lysed with reporter lysis buffer (Promega, USA). The enzymatic activity was measured for re y and Renilla luciferase using Dual-Luciferase Reporter assay System (Promega, USA) and a luminometer. All luciferase assays were carried out in duplicates or triplicates and experiments were carried out at least twice.

Chromatin Immunoprecipitation Analysis
ChIP assays were performed using a ChIP Kit (Thermo Scienti c, USA) according to the manufacturer s instructions. Brie y, 5 × 10 6 cells were treated with 1% formaldehyde to cross-link chromatin-associated proteins to DNA. The cell lysates were subjected to ultrasound to shear the DNA into fragments. 10% of each sample was used as input reference control. Chromatin protein suspensions were then incubated with 10 µl of anti-RUNX1 antibody (Abcam, USA) or control anti-IgG antibody (CST, USA). All the above chromatin supernatants were incubated with 20 µL magnetic protein A/G beads overnight at 4 °C with rotation. The protein-DNA complexes were reversed and puri ed for pure DNA, and then were submitted to SYBR green (Applied Biosystems) qPCR analysis with an ABI Prism 7900HT Fast RealTime PCR system. The following primers were used for quantitative ChIP-PCR Analysis of RUNX1 expression and patient survival using public data sets RUNX1 expression and survival in HNSCC patients was analyzed using the Cancer Genome Atlas (TCGA) database (https://www.r-project.org/) and Ginos Head-Neck Statistics based on Oncomine database (http://www.oncomine.com) [18].

Statistical analysis
The SPSS 23.0 (IBM, USA) and GraphPad Prism (version 7.01, USA) statistical analysis software programs were used for statistical analysis of the experimental data. Each experiment was repeated at least three times. The differences in mean values among groups were evaluated and expressed as the mean ± SEM. A P-value less than 0.05 was considered statistically signi cant (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Student's t-test was used to analyze the expression of cell protein markers, cell viability, relative mRNA levels, migrated cells and invaded cells.

RUNX1 expression is increased with disease progression in HNSCC.
To investigate the clinical relevance of RUNX1 in HNSCC progression and metastasis, we initially conducted a study to determine RUNX1 protein expression in HNSCC tissues as well as RUNX1 RNA expression in a relevant publicly available dataset. As shown in Fig. 1A, Fig. S1A and Table, in cancerous tissues where RUNX1 was present, the percentage of RUNX1 positive cells was substantially more than in the normal head-neck tissues. Notably, the percentage of RUNX1 positive cells was as the tumor pathology grade advanced. To determine the changes in RUNX1 expression at the primary site of cancer initiation, xenograft tumor from mice injected with the FaDu cells were sacri ced at 1, 2, 4, and 8 weeks of age. These time points represent early through late stages of disease progression. RUNX1 expression in xenograft tumor increased concomitant with disease progression, and with strong nuclear expression evident at 8 weeks ( Fig. 1B and Fig. S1B). This observation was consistent with the results of clinical tissue samples.
We therefore addressed whether there is a clinical relationship of RUNX1 expression in HNSCC patient tumors with survival using publically available datasets. We analyzed the differential protein and mRNA expression of RUNX1 between cancer samples and normal samples, and the relationship of expression levels of RUNX1 in HNSCC samples with patient survival. In this analysis ( Fig. 1C and Fig. S1C-D), the RUNX1 protein levels were signi cantly upregulated in most clinical HNSCC tissues (15 normal and 29 tumor samples, p < 0.01) and in 3 pairs of fresh samples. As shown in Fig. 1D and Fig. S1E, The RUNX1 mRNA expression was signi cantly elevated in HNSCC samples relative to that in normal tissue samples based on TCGA dataset (31 normal and 91 tumor samples, p < 0.05) and oncomine dataset (13 normal and 41 tumor samples, p < 0.0001) [18]. The Kaplan-Meier analysis in the TCGA dataset indicated no signi cant correlation (p = 0.608) between RUNX1 expression levels and patient survival time (Fig. S1F).

Runx1 Loss Decelerates The Malignant Progression Of Hnscc
To study the biological function of RUNX1 in HNSCC, we established RUNX1-silenced cell lines using a lentiviral vector carrying shRNA in FaDu and SCC-9 cells, and showed a signi cant decrease in RUNX1 mRNA and protein expression compared to the expression levels in control cells ( Fig. 2A-B and Fig. S2A-B).
In cell growth and metastasis, the HNSCC cells treated with shRUNX1 were analyzed. The cell proliferation rate was signi cantly decreased in cells with knockdown of RUNX1 compared with control ( Fig. 2C-D). As shown in Fig. 2E, markedly slower wound closure was observed in cells transfected with shRUNX1 by a wound healing assay. Moreover, the results from transwell assays revealed that knockdown of RUNX1 signi cantly decreased the migration and invasion abilities of HNSCC cells (Fig. 2F). Together, these data showed that knockdown expression of RUNX1 could impede HNSCC cell growth and metastasis.
To address experimentally whether RUNX1 loss decelerates HNSCC tumor growth, FaDu shRUNX1 and FaDu sh−NC cells were injected subcutaneously into the ank of each mouse (Fig. 2G). The results indicated that the ability of tumor proliferation was inhibited in the shRUNX1 group compared to that of sh-NC groups, as indicated by the nal xenograft tumor volume and weight (Fig. 2H). Together, these ndings veri ed that deregulation of RUNX1 in HNSCC might reduce the disease progression by decreasing cell proliferation, migration and invasion.

The interrelationship between RUNX1 and OPN.
To search for the putative correlation between RUNX1 and OPN, we analyzed using clinical samples data, further corroborated by TCGA and oncomine databases. Firstly, we found the protein expression levels of OPN were signi cantly upregulated in most clinical HNSCC samples (15 normal and 29 tumor samples, p < 0.0001) (Fig. 3A). Similarly, the mRNA expression of OPN were signi cantly elevated in HNSCC samples on TCGA database (31 normal and 91 tumor samples, p < 0.05) and Oncomine database (13 normal and 41 tumor samples, p < 0.0001) (Fig. 3B and Fig. S3A). Next, clinical data showed a positive correlation between the expression of RUNX1 and OPN (R = 0.38, P < 0.05) (Fig. 3C), and the TCGA dataset showed a similar result (R = 0.36, P < 0.001) (Fig. 3D). In the Ginos head-neck dataset from oncomine [18], we found a signi cant co-expression of RUNX1 and OPN (correlation coe cient = 0.767) (Fig. S3B). Finally, the Kaplan-Meier analysis in the TCGA dataset indicated no signi cant correlation (p = 0.6301) between OPN expression levels and patient survival time (Fig. S1F).
Prompted by the data mining results, we next assessed the inter-regulation of RUNX1 and OPN in vitro and in vivo. In studying the subcellular localization of RUNX1 and OPN, we learned that a small portion of OPN is localized in the nucleus, and then turned our attention to the underlying mechanism by which RUNX1 and OPN regulate each other in the nucleus. Interestingly, we observed a concurrent loss of OPN mRNA and protein upon RUNX1 deletion in nucleus and cytoplasm of HNSCC cells (Fig. 3E-F and Fig.  S3D-G). However, RUNX1 mRNA and protein levels were not affected by OPN deletion (Fig. 3G-H and Fig.  S3H-K). Moreover, in vivo experiments also con rmed that the expression of RUNX1 did not change signi cantly in xenograft tumors treated with different OPN conditions (Fig. 1F). These results motivated us to study the regulation of OPN by RUNX1.

Runx1 Has Direct Roles In Transcription Regulation Of Opn
RUNX1 was predicted to bind to the OPN DNA promoter at three possible binding sites by the Jaspar database (http://jaspar.genereg.net/), which is an open-access database for eukaryotic transcription factor binding pro les (Fig. 4A and Fig. S4A). ChIP-qPCR experiments using anti-RUNX1 antibody revealed that only one of the putative binding sites of the RUNX1 protein to the OPN promoter in FaDu cells showed signi cant binding (Fig. 4B-C), and the remaining two are not obviously bounding (Fig. S4B-C).
To better understand the role played by RUNX1 in the regulation of OPN, we generated a reporter construct containing the 1,000 bp region upstream of the transcriptional start site of OPN upstream of luciferase. We transfected 293T cells with RUNX1 overexpression or the respective controls together with the OPN reporter construct and analyzed luciferase expression. The results showed that the overexpression of RUNX1 signi cant increased OPN promoter activity (Fig. 4D). Thus, these data suggest that RUNX1 stimulates OPN expression at the transcriptional level.

RUNX1 mediates HNSCC cell metastasis and MAPK pathway activation via stimulating OPN
We further investigated whether RUNX1-mediated cancer metastasis is related to OPN, Both sh-RNA and the overexpression lentivirus vector were used to knockdown and overexpress of RUNX1 and OPN in FaDu cells. The effects of RUNX1 and OPN on HNSCC cells were investigated by adding shOPN lentiviral vector to RUNX1 overexpressed stable strain for further transfection. RUNX1-overexpress signi cantly increased cell proliferation, which could be attenuated by further OPN-knockdown in FaDu cells (Fig. 5A). The similar patterns were observed in wound healing and Transwell experiments. Markedly faster wound closure was observed in RUNX1-overexpress cells. Moreover, overexpression of RUNX1 signi cantly increased the migration and invasion abilities of FaDu cells. Next, our results showed that the migration and invasion abilities were antagonized in cells with further OPN downregulation (Fig. 5B-E).
Since it has been shown that OPN could promote tumor progression through the MAPK pathway [19,20], we sought to examine the protein markers expression levels of the MAPK pathway, in OPN-knockdown HNSCC cells with the situation of RUNX1-overexpress. Under the overexpression of RUNX1, the expression levels of p-MEK and p-ERK were enhanced. On the contrary, with the downregulation of OPN, this expression level is signi cantly weakened ( Fig. 5F and Fig. S5), suggesting that RUNX1 might activates the MAPK pathway accompanied by OPN stimulation.

Discussion
As previously demonstrated, the RUNX1 protein acts as either tumor suppressor or oncogene in different cancer types, depending on activate or repress target gene expression in different cellular context [21,22].
Indeed, RUNX1 has displayed dual roles in breast cancer and also in gastric cancer [23][24][25][26]. The key ndings of this study support the oncogenic function of RUNX1 in HNSCC. The clinical samples data were indicative for strong RUNX1 protein overexpression concomitant with higher grade in the HNSCC tissues. Complementary in vivo studies demonstrate that RUNX1 expression increases concomitant with disease progression in the xenografts of mice. Furthermore, we found that the expression of RUNX1 was increased in clinical HNSCC tissues, consistent with the results from TCGA and Oncomine database. Although the TCGA data showed no correlation between RUNX1 expression and patient survival time, in vitro experiment, RUNX1 knockdown in the HNSCC cells results in a marked abrogation of migration and invasion ability than control cells. Moreover, RUNX1 loss weakened the tumorigenesis ability in mice. Our results are in line with a previous study in which a comprehensive analysis of RUNX1 protein in epithelial cancer was performed [7].
The above ndings persuaded us to investigate the functional implication of RUNX1 in mechanisms of invasion and metastasis. Mammalian RUNX1 transcription factors mediate the homeostasis of malignant cells through their ability to promote gene activation [4]. Recent genetic studies have highlighted a potential role for OPN in HNSCC, and a high level of OPN activity has been reported to be related to metastasis and poor outcomes of HNSCC patients [27][28][29]. Our study examined the correlation between RUNX1 and OPN using clinical samples, TCGA and Oncomine database. The results show that RUNX1 is positively correlated with OPN expression. In parallel, our study revealed that RUNX1 dramatically affect OPN expression in HNSCC cells, and nuclear /cytoplasmic expression of RUNX1 was positively correlated with OPN expression. In contrast, RUNX1 expression was not in uenced by OPN deletion, implying a pathological signi cance for RUNX1 regulation of OPN in HNSCC development.
As a transcription factor, RUNX1 may regulate the transcription of target genes in many ways [30]. Our luciferase reporter assay veri ed that RUNX1 stimulated the transcriptional activity of the OPN promoter.
Moreover, this speculation is further supported by the nding that RUNX1 can bind to the OPN promoter in a ChIP-qPCR analysis. Consistent with this result, RUNX1 is a key transcription factor that can directly binds to the OPN regulatory sequence to upregulate the gene expression in metastatic tumor cells [17]. Collectively, our in vitro data strongly suggests that OPN is a RUNX1-target gene.
It is noteworthy that some research has veri ed that RUNX1, as a transcription factor, cooperates with its target genes to promote tumor metastasis [31]. To verify whether RUNX1-mediated cancer metastasis is related to OPN, we evaluated cancer-related phenotypic properties of the cells including proliferation, migration and invasion. RUNX1 overexpress exhibited a signi cantly greater ability to migrate and invade compared with the control cells. Meanwhile, knockdown OPN showed the opposite effect. Notably, when RUNX1 overexpression and OPN knockdown were simultaneously present, the phenotype is neutralized. This nding supports the idea that RUNX1 mediates HNSCC cell metastasis via directly stimulating OPN.
Additionally, we found RUNX1 overexpress was associated with upregulation of typical markers from the MAPK pathway, representing one of the major oncogenic pathways implicated in HNSCC etiology [32].
Similarly, other study has shown a positive correlation between RUNX1 and the MAPK signaling pathway in cancer [33]. Furthermore, we veri ed that these markers also appear to be neutralizing when RUNX1 overexpression and OPN knockdown coexist.

Conclusions
In conclusion, we have shown that RUNX1 transcription factor acts as an oncogene to facilitate metastasis in HNSCC by directly stimulating OPN transcription to activate the MAPK signaling pathway.
These ndings enhance our understanding of HNSCC metastasis. In addition, these results of our study highlight the therapeutic potential of targeting the oncogenic activity of RUNX1 in HNSCC.