RBM24 Mediates Lymph Node Metastasis and Epithelial-Mesenchymal-Transition in Human Hypopharyngeal Squamous Cell Carcinoma Via Regulating Twist1

Background: Hypopharyngeal squamous cell carcinoma (HSCC) has the worst prognosis among head and neck tumours, and Lymph node (LN) metastasis mainly accounts for the poor prognosis. RBM24 (RNA Binding Motif Protein 24) regulates target RNA as an RNA binding protein involved in several cancers. However, its role in HSCC remains completely unknown. Here we attempt to explore the effects of RBM24 on HSCC. Methods: RNA sequencing was conducted to nd the differentially expressed genes in tumour tissues from HSCC patients with LN metastasis and without LN metastasis in our previous study. Expression of RBM24 in HSCC tissues was analyzed by qRT-PCR, western blot and immunohistochemistry. Cell proliferation was tested by CCK8 assay as well as Colony formation analysis. Cell migration and invasion capacity were estimated by transwell assay. The wound healing assay was also carried out to evaluate the motility of FaDu cells. QRT-PCR, western blot and immunouorescence assays were conducted to detect the process of EMT. A popliteal lymph node metastasis model was constructed to explore the effect of RBM24 on HSCC in vivo. Results: RBM24 was remarkably down-regulated in HSCC patients with LN metastasis, and low expression of RBM24 was inextricably linked to the poor prognosis. Knockdown of RBM24 facilitated the proliferation, migration and invasion of RBM24, whereas overexpression of RBM24 showed the opposite effects and suppressed the epithelial-mesenchymal-transition (EMT) process. Overexpression of Twist1 could reverse the inhibitory effects of RBM24 on motility and invasion of FaDu cells. The inhibitory effects of RBM24 on tumour growth and lymphatic metastasis in HSCC were demonstrated by the in vivo experiment as well. Conclusions: These results indicated RBM24 was a suppressor gene and might inhibit EMT and LN metastasis in HSCC regulating Twist1.


RBM24 Mediates Lymph Node Metastasis and Epithelial-Mesenchymal-Transition in Human
. Although rare, it has the worst prognosis compared with other kinds of head and neck tumours, and 5-year relative survival rate is 25%-45% 2-5 . Ho, A. S. et al. expound the number of metastatic lymph nodes is a principal independent factor closely related to mortality risk in hypopharyngeal cancer 6 . Despite the great improvements of diagnostic methods and therapeutic strategies, HSCC patients have developed advanced-stage disease and distant metastasis or cervical lymph node metastasis at the initial diagnosis, and LN metastasis occurs in 65%-80% of patients 7,8 . Hence, it is vital to elucidate the molecular mechanisms of cancer malignant progression in HSCC with LN metastasis to search for biomarkers for early screening and improve the accuracy of diagnosis. RNA binding proteins (RBPs) are almost involved in regulating the entire life of target RNA substrates, and they can be the core of gene expression network 9 . A census of human RBDs has uncovered 7.5% of the 20,500 protein-coding genes in humans participate in gene expression regulation 10 . RBPs have been regarded as participants in tumorigenesis already 11 . Conventional RBPs regulate transcripts by forming ribonucleoprotein (RNP) complexes with their RNA targets via RNA-binding domains (RBDs) such as the RNA recognition motif (RRM), the hnRNP K homology domain (KH domain), the double-stranded RNAbinding domain (dsRBD), S1 domains and others 12 . RBM24 (RNA Binding Motif Protein 24), an RNAbinding protein containing a highly conserved RRM domain, comprises two submotifs, RNP1 and RNP2 13 . Previous researches have veri ed RBM24 is able to bind to G(A/U)GUG motif of target RNA 14,15 .
Researches have shown RBM24 can promote differentiation of embryonic stem cells to cardiomyocytes by regulating alternative splicing 16 and enhance myogenic differentiation by regulating the stability of myogenin mRNA transcripts via binding to its 3'-untranslated region (3'-UTR) mRNA 17 in mice. RBM24 has also been found to play a tumour suppressor role in nasopharyngeal carcinoma(NPC) 18 , colorectal cancer (CRC) 19 and liver cancer 20 , meanwhile the exact function of RBM24 in HSCC tumorigenesis remains unclear.
Epithelial-mesenchymal transition (EMT) is known as a vital process that participates in tumour metastasis and is closely associated with tumour progression. It endows tumour cells with the ability to invade adjacent tissues and metastasize 21  with HSCC who met the criteria were involved in this study. The study was performed in accordance with the Declaration of Helsinki, and the ethics committee of the rst a liated hospital of Chongqing Medical University approved this study. Written informed consent forms were obtained from the patients before they underwent surgery.

Immunohistochemistry and scoring
The para n-embedded tissue sections (4µm) were depara nized in xylene and rehydrated in graded ethanol. Sections were heated in citrate buffer for 15 minutes at 100℃ for antigen retrieval. Next, the slides were rinsed in phosphate buffer saline (PBS) and were treated with Peroxidase Block at 37℃ for 10 minutes. After being blocked with 10% normal goat serum at 37℃ for 15 minutes to minimize nonspeci c staining, slides were incubated with the primary RBM24 antibody (1:200, ab94567, Abcam, UK) overnight at 4℃. The next day, slides were treated with HRP-conjugated polyclonal goat secondary antibody for 15 minutes, and antigen-antibody reaction signals were detected with Diaminobenzidine (DAB) reagents (ZSGB-BIO, China) at room temperature for 1 minute. Finally, slides were counterstained with hematoxylin for 20 seconds. The slides using PBS instead of primary antibody were negative control. And all of the slides were estimated by two experienced pathologists. IHC staining scores depended on both the proportion of RBM24-positive tumour cells and the staining intensity. The proportion of RBM24-positive tumor cells were scored as 0 (0-5%), 1 (5-25%), 2 (25-50%), 3 (50-75%), and 4 (75-100%). The staining intensity was scored as 0 (negative staining), 1 (weak staining), 2 (moderate staining), and 3 (strong staining). The nal IHC scores were the multiplication of the two scores. ROC curve was used to analyze IHC scores to distinguish the protein level of RBM24 and the cutoff value was 4. Scores < 4 were de ned as low expression while scores ≧ 4 were considered as high expression.
Cell line and culture The human hypopharyngeal carcinoma cell line FaDu was purchased from the cell bank of Chinese Academy of Sciences (Shanghai, China) and cultured in minimal essential medium (MEM) (Boster, USA) supplemented with 10% fetal bovine serum(PAN, Germarny) and antibiotics(100 U/ml of penicillin and streptomycin) in an incubator containing 5% CO 2 at 37℃.

Lentivirus transductions
To overexpress RBM24, lentiviral particles (OE-RBM24 and Control) containing green uorescent protein (GFP) were obtained from GeneChem Corporation (Shanghai, China). Fadu cells were seeded into a sixwell plate then infected by the lentiviral particles with a multiplicity of infection (MOI) of 10 when the cells were at 20-40% con uency. After 48-72h of lentivirus infection, infected cells were selected with 2 µg/ml puromycin (Beyotime, China). Then the cells were collected and examined by qRT-PCR and Western blot.

Western blot (WB) and antibodies
The total protein extracts of tissues and FaDu cells were obtained using the total protein extraction kit ( Cell Proliferation Assay The cell growth ability was measured using Cell Counting Kit-8 (CCK-8, GlpBio, USA), a cell proliferation reagent. After adding 10 microliters (10 µl) of CCK-8 to each well, then the 96-well plate was incubated at 37℃ for 1 hour. Finally, the absorbance was evaluated at 450 nm to collect uninterrupted data for 96 hours.
Colony Formation Analysis 1,000 FaDu cells were inoculated onto each well of a six-well plate and incubated in a medium containing 10% FBS at 37℃ and 5% CO 2 for 2 weeks. After xing the cells with methanol and staining the cells with 0.5% crystal, colonies composed of at least 50 cells were counted.

Transwell Migration and Invasion Assay
To estimate the cell migration and invasion assays of FaDu cells, transwell assays were performed with 8 µm pore-chambers (Corning, USA), with or without Matrigel (BD Biosciences, Bedford, MA, USA) following the manufacture's instruction. 200 µl serum-free MEM containing 5 × 10 4 were seeded into the upper chamber. 600 µL MEM medium supplemented with 20% FBS were added into the lower chamber. After 48 hours, the cells were xed with methanol and stained with 0.5% crystal violet. Finally, the number of stained cells were counted with a microscope.
Wound healing scratch assay FaDu cells were cultured in six-well plates and 200 µl sterile pipette tips were used to create linear scratches in the con uent monolayers of cells. Cellular debris was washed away using PBS. Then the wound closure was observed with a microscope at 0 hour and 24 hours after injury.
Animal Experiments BALB/c male nude mice (5 weeks old; Tengxin, China) were chosen to establish popliteal lymph node metastasis models and fed under speci c pathogen-free conditions in the Animal Laboratory of Chongqing Medical University. The mice were randomly divided into 2 groups (5 mice per group) and injected with 2×10 6 FaDu-Control or FaDu-OE-RBM24 cells suspended in 30 µl PBS in their footpads. Tumour size and weight of the mice were measured every 5 days using a caliper. Tumour growth and LN metastasis were observed by the small animal in vivo imaging (IndiGo 2.0.5.0 system, Berthold Technologies, Germany) 15 days after cell implantation. All the mice were sacri ced 25 days later, the primary tumour tissues and lymph nodes were removed and para n-embedded. Tumour volumes were calculated in accordance with the formula 0.5×length×width 2  A Student's t-test or the Mann-Whitney U-test was selected to assess signi cant differences between two groups while one-way ANOVA was applied to three or more groups. The correlation between expression of RBM24 in immunohistochemistry and clinicopathological features was evaluated by Pearson's chi-square test or Fisher's exact test. Kaplan-Meier & log-rank analysis was performed to evaluate overall survival (OS). All results are shown as mean± standard error of mean (SEM). P-value < 0.05 was considered statistically signi cant.

Results
RBM24 was down-regulated in HSCC patients with lymph node metastasis.
We conducted RNA sequencing to nd the differentially expressed genes (DEmRNA) in HSCC patients with LN metastasis and without LN metastasis in our previous study 25,26 . And RBM24 was one of the down-regulated genes in the LN metastasis group. Then we expanded the sample size to detect expression of RBM24 at the mRNA level as well as the protein level. Using 8 pairs of tumour tissues (including 8 tissues from patients with LN metastasis and 8 tissues from patients without LN metastasis) and corresponding adjacent normal tissues, we identi ed the lower mRNA level in the LN metastasis group than in the normal group (p < 0.01) as well as than in the HSCC patient group without LN metastasis (p < 0.05) (Fig. 1 a). Correspondingly, we identi ed the protein levels of RBM24 were lower in the LN metastasis group than in the other two groups through WB analysis in 5 pairs of tumour tissues (including 5 tissues from patients with LN metastasis and 5 tissues from patients without LN metastasis) and corresponding adjacent normal tissues) (p < 0.05) ( Fig. 1b). We also conducted IHC to determine the expression of RBM24 in tissues. The different staining of RBM24 in HSCC tissues was exhibited in Fig.  1c. The frequency of low expression of RBM24 in tissues from patients with LN metastasis showed a greater proportion (25/41[61.0%]) than in those without LN metastasis (14/16[12.5%]) ( Table 1). The results indicated RBM24 was down-regulated in HSCC patients with LN metastasis. Furthermore, we assessed the association between the expression of RBM24 and clinical features in 57 HSCC patients. Results showed the expression of RBM24 was connected with tumour stage (p<0.01) and lymph node stage (p<0.001) while was not connected with age, gender or pathological stage (  (Fig. 1d). These results indicated RBM24 was down-regulated in HSCC tissues with LN metastasis and was closely related to poor prognosis.

Inhibition of RBM24 Promoted Proliferation, Migration, and Invasion of FaDu Cells
To explore the biological function of RBM24 in Fadu cells, qRT-PCR and western blot were performed to verify the RBM24 expression. FaDu cells were transfected with RBM24-siRNAs (siRNA#1, siRNA#2, and siRNA#3) as well as control siRNA(Si-NC). The qRT-PCR and western blot results showed that siRNA#3 (Si-RBM24) was the most e cient to knock down RBM24 (Supplementary Fig. 1a and b, Fig. 2a and b), thus it was selected for further experiments. We found inhibition of RBM24 promoted the proliferation of FaDu cells through CCK-8 staining and colony formation assays ( Fig. 2c and d). Moreover, knockdown of RBM24 enhanced migration and invasion (Fig. 2e) and wound healing (Fig. 2f) of FaDu cells.  (Fig. 3a) and western blot (Fig. 3b) assays. Results showed overexpression of RBM24 reduced the proliferation (Fig. 3c and d) and inhibited the mobility (Fig. 3e and  f) of FaDu cells. These results revealed RBM24 might negatively regulate proliferative and invasive behaviours of FaDu cells.
Overexpression of RBM24 Down-regulated Twist1 and Inhibited EMT of FaDu Cells EMT is generally considered as an important process in tumour metastasis, and TGFβ has been widely reported as a potent inducer of EMT 27 . Our results indicated that loss of RBM24 may induce the malignant progress of hypopharyngeal squamous cell carcinoma (HSCC), so we turned to explore the association between RBM24 and TGFβ-induced EMT. We induced EMT in FaDu cells with 20 ng/ml TGFβ for 48 hours and then tested the mRNA levels of EMT-inducing transcription factors (EMT-TFs) to nd which was in response to RBM24 overexpression in FaDu cells. Finally, Twist1 was identi ed to be the downstream effector of RBM24, while Snail, Slug or ZEB1 did not change signi cantly (Fig. 4a). We also con rmed RBM24 overexpression also down-regulated the protein level of Twist1 (Fig. 4c). Correspondingly, we also found the mRNA and protein levels of E-cadherin increased while the levels of N-cadherin and Vimentin decreased in the RBM24-overexpression FaDu cells (Fig. 4b and c), indicating RBM24 overexpression restrained TGFβ-induced EMT. Furthermore, overexpression of RBM24 resulted in decreasing E-cadherin and increasing N-cadherin and Vimentin levels evaluated by confocal immuno uorescence assay (Fig.   4d). These results suggested RBM24 could inhibit the process of TGFβ-induced EMT and down-regulate expression of Twist1.
Overexpression of Twist1 Reversed the Inhibiting Effects of RBM24 on Cell Invasion, Migration and EMT of FaDu cell.
Since Twist1 played an essential role in HSCC progression, we determined whether Twist1 was necessary for the functions of RBM24 in HSCC cells by transfecting FaDu-OE-RBM24 cells with Twist1 plasmids ( Fig. 5a and b). The results indicated overexpression of RBM24 suppressed cell migration, invasion and EMT while the inhibition effects were reversed by overexpression of Twist1 (Fig. 5c-e). It demonstrated overexpression of RBM24 could suppress cell motility, invasion capacity and EMT via negatively regulating Twist1.

Overexpression of RBM24 Suppressed HSCC Tumor Growth and Lymph Node Metastasis In Vivo
To explore the effects of RBM24 expression on hypopharyngeal tumour growth and lymph node metastasis in vivo, we injected FaDu cells stably expressing RBM24 (FaDu-OE-RBM24) or an empty vector (FaDu-Control) into the footpads of BALB/c male nude mice (Fig. 6a). As Fig. 6b showed, clear uorescent signals were detected at the footpads in both groups. However, the uorescence signal only emerged in the popliteal fossa of the mouse injected with FaDu-Control cells, representing LN metastasis.
As the results showed in Fig 6c and d, tumours in mice injected with FaDu-OE-RBM24 cells grew more slowly compared with those injected with FaDu-Control cells. And the volume of tumours was lower in the OE-RBM24 group comparatively. The H&E assay showed stable overexpression of RBM24 resulted in fewer metastatic tumour cells in the lymph node (Fig. 6e). Meanwhile, 20% (1/5) of the mice implanted with FaDu-OE-RBM24 cells developed lymph node metastasis, whereas 75% (3/4) developed lymph node metastasis in the FaDu-Control group (Fig. 6f). Moreover, IHC assay showed the expression level of Ecadherin and RBM24 was higher while the level of Vimentin and Twist1 was lower in tumours induced by FaDu-OE-RBM24 cells than that induced by FaDu-Control cells (Fig. 6g). Hence, it was reasonable to speculate that overexpression of RBM24 could down-regulate Twist1 in vivo, which was also re ected in vitro.

Discussion
RNA binding proteins (RBPs) play vital roles in regulating RNA metabolic pathways, including transcription, alternative splicing, modi cation, polyadenylation, stability, subcellular localization, translation and decay 11,28 . Due to their signi cant importance in gene regulation, it has been reported that aberrant expression of RBPs or mutations in RBPs as well as their binding sites with RNA targets are involved in diverse human diseases, including cancers 29 . Recently, a review has revealed RBPs participate in tumour metastasis by regulating mechanotransduction 30 .
RBM24, an RNA binding protein, belongs to the RRM family. RBM24 has been identi ed as a participant in the p53 pathway through the mutual interaction with p53 and regulating p21 mRNA stability in different kinds of human cancer cell lines 13 31 . It has been reported RBM24 also regulates p63 expression via mRNA stability 32 . Notably, RBM24 is found to inhibit cancer progression in several tumors 18-20 . However, there is still a lack of understanding of its role in tumor initiation and progression, especially in hypopharyngeal carcinoma. In the early work, our team has carried out RNA sequencing to explore genes with altered expression in HSCC patients with or without LN metastasis, and RBM24 was one of the down-regulation genes. In the present study, we rst veri ed RBM24 was down-regulated in HSCC patients with LN metastasis. Moreover, the lower expression of RBM24 was associated with several clinicopathological features and worse overall survival. It indicated that RBM24 might be a potential target for both early diagnosis and therapy in HSCC. To determine the biological function of RBM24 in vitro, we utilized siRNAs and lentiviral particles to regulate the expression of RBM24 in FaDu cells. And the results showed RBM24 could inhibit cell proliferation, migration, invasion abilities as well as EMT process in vitro. Based on these results, we suspected RBM24 might be a tumour suppressor to inhibit HSCC progression.
HSCC patients often have developed cervical lymph node metastasis at their initial diagnosis, leading to treatment failure and relatively poor overall survival 4,[33][34][35] . Thus, it is particularly urgent to study the mechanism of HSCC tumor progression as well as LN metastasis. EMT has been widely reported to play crucial roles in the cancer process and is characterized by the loss of polarity of the epithelial cells and the conversion of epithelial cells to a mesenchymal phenotype 36 , which endows the cancer cells with the ability to escape from the primary environment and develop distant metastasis. And EMT is de ned by the decreased expression of epithelial marker E-cadherin and increased expression of mesenchymal marker N-cadherin and Vimentin 37 . Recent studies have demonstrated the close associations between EMT and lymph node metastasis 38, 39 . As we know, EMT is mainly regulated by EMT-inducing transcription factors (EMT-TFs), and Twist1 is one of the core EMT-TFs 27 . Over the past years, Twist1 has been uncovered to be an oncogene, and its abnormal overexpression in various tumours facilitates cancer initiation, progression and metastasis. Twist1 is able to promote tumour process by enhancing the proliferation and suppressing apoptosis of cancer cells, increasing cancer cell resistance to chemotherapy, up-regulating cancer stem cell (CSC) population, and driving cancer cell invasion and metastasis 40,41 . Considering the signi cant importance of EMT in tumor metastasis and a previous study discovered RBM24 overexpression upregulated E-cadherin expression while downregulated the expression of Vimentin in human colorectal cancer HCT116 cells 19 , we turned to investigate the interactions between RBM24 and TGFβ-induced EMT. Our study had con rmed that RBM24 could regulate the expression of several typical markers of EMT and Twist1 in HSCC FaDu cells. This suggested RBM24 might reduce TGFβ-induced EMT and HSCC LN metastasis by targeting Twist1. To verify the hypothesis, we carried out rescue experiments by overexpressing Twist1 in FaDu cells. The results indicated cell motility, invasion ability as well as the EMT process could be reversed by Twist1 overexpression. In addition, we found RBM24 overexpression could suppress HSCC tumour growth and LN metastasis in animal experiments. Meanwhile, IHC staining showed Twist1 and several EMT markers changed correspondingly. Thus, it was feasible to speculate that overexpression of RBM24 might inhibit HSCC tumour growth and it might reduce LN metastasis by restraining Twist1 expression. However, it is still necessary to conduct further studies to verify whether RBM24 can combine with Twist1 and elucidate the speci c mechanism. As well, the detailed mechanisms of how they regulate EMT and LN metastasis after RBM24 combing with Twist1 need further research.
In summary, RBM24 serves as a novel tumour suppressor gene and suppresses the LN metastasis and EMT process by downregulating the level of Twist1 in HSCC. Therefore, RBM24 and Twist1 may be potential biomarkers for diagnosis and therapeutic targets for HSCC patients with LN metastasis. Figure 1 <p>RBM24 was down-regulated in patients with LN metastasis and correlated with poor prognosis in HSCC. a-b Expression of RBM24 in HSCC was con rmed by qRT-PCR and western blot analyses. c Representative IHC staining of RBM24 in HSCC tissues from patients with or without LN metastasis. PBS instead of the primary antibody was used as negative control (left panel, magni cation 100x; right panel, magni cation 400x). d Kaplan-Meier analysis demonstrated down-regulation of RBM24 was negatively related to overall survival in patients with HSCC (n=57). The upper line represents the survival rate of patients with high expression of RBM24, while the nether line represents the survival rate of patients with low expression of RBM24.<strong> </strong>*p &lt; 0.05, **p &lt; 0.01</p><p>Abbreviations: N, normal tissues; NL, tumour tissues from patients without lymph node metastasis; L, tumour tissues from patients with lymph node metastasis; NC, negative control&nbsp;</p>