RBM10 Interacts with CTBNNBIP1 and Represses Lung Adenocarcinoma Progression Through the Wnt/β- catenin Pathway

Background: RNA-binding motif protein 10 (RBM10), one of the RNA-binding protein (RBP) family, has a tumor suppressor role in various tumors. However, the functional role of RBM10 in lung adenocarcinoma (LUAD) and the molecular mechanism remain unclear. The aim of this study was to explore the effect of RBM10 on LUAD growth and metastasis and its molecular mechanism. Methods: Bioinformatics analysis was used to predict RBM10 expression and its associations with clinicopathological features and prognosis in LUAD. Gain- and loss- of function experiments were conducted to investigate the biological functions of RBM10 both in vitro and in vivo. RNA-seq, bioinformatics programs, western blot, qRT-PCR, TOP/FOP ash reporter, co-immunoprecipitation (co-IP), nuclear and cytoplasmic protein extraction and rescue experiments were used to reveal the underlying mechanisms. Results: Bioinformatics analysis showed that RBM10 was signicantly downregulated and closely correlated with poor prognosis in LUAD patients. RBM10 silencing signicantly promoted the LUAD proliferation, migration, invasion ability, while RBM10 overexpression had the opposite effects. Furthermore, upregulation of RBM10 inhibited growth and metastasis of LUAD in vivo. Additionally, RBM10 suppressed tumor progression through inhibiting epithelial to mesenchymal transition (EMT) in LUAD cells. Mechanistically, RBM10 interacts with β-catenin interacting protein 1 (CTNNBIP1) and positively regulates its expression, thus inactivating the Wnt/β-catenin pathway. Conclusions: This is the rst study that reported how RBM10 suppresses cell proliferation and metastasis of LUAD by negatively regulating the Wnt/β-catenin pathway through interaction with CTNNBIP1. These data suggest that RBM10 may be a promising new target or clinical biomarker for LUAD therapy. and H1299 Cell proliferation was examined by CCK8; (c EdU; (g colony formation assays. The results were as mean ± SD, *P < A wound-healing assay was used to test the migration capacity of RBM10 in LUAD cells. The cells migrating into the wounded areas were photographed at 0h and 48h. The results were represented as mean SD, Scale is The migration and invasion capacity of RBM10 in LUAD were also examined by Transwell assays. The results mean inhibits the Wnt/β-catenin pathway by blocking the β-catenin-TCF/LEF interaction. (a, b) Studies from StarBase datasets showed that RBM10 expression was negatively corrected with the expression of LEF1 (a) and TCF4 (b) in LUAD samples. (c, d) qRT-PCR results of TCF3, TCF4, LEF1 mRNA levels under RBM10 overexpression, GAPDH was used as loading controls. (e, f) Western blot data of TCF3, TCF4, LEF1 protein levels under RBM10 overexpression, β-actin was used as loading controls. (g, h) Co-IP was used to test β-catenin interaction analyses using a β-catenin antibody. Western blot showed that upregulation of RBM10 decreased β-catenin-TCF/LEF interaction, whereas promoting CTNNBIP1/β-catenin interaction. (i) A schematic diagram of the functional roles of RBM10 in LUAD cells. The results were represented as mean ± SD. All *P < 0.05 All experiments were repeated three times.

Cells were counted and seeded in 96-well plates (5 × 10 3 cells/well). After incubation for 24h, 10µl of Cell Counting Kit-8 (CCK-8, Dojindo, Kumamoto, Japan) was added to the culture medium and incubated for 1.5h at 37°C. Then, the optical density (OD) value at 450nm was measured. Three independent experiments were performed.

Clone formation assay
Transfected cells (700 cells/well) were counted and plated in 6-well plates. After 14 days of culture, cells were xed with 0.4% paraformaldehyde for 15min and were then stained with 0.5% crystal violet for 30min. Colonies containing more than 50 cells were counted. Three independent experiments were performed.

5-Ethynyl-2′-deoxyuridine (EdU) incorporation assay
After transfection, the LUAD cells were inoculated into 24-well plates. EdU kit (RiboBio, Guangzhou, China) was used for labeling cells following the manufacturer's instructions. Photographs were taken using an inverted uorescent microscope (Leica Microsystems Inc., USA), and the experiment was repeated three times.

Soft agar colony formation assay
Soft agar colony formation assay (GENMED SCIENTIFICS INC, USA) was performed according to manufacturer's instructions. Brie y, 1.5mL GENMED Cloning Solution (Reagent A) and 1.5mL GENMED Hypertrophic Solution (Reagent B) were mixed and added into the 12-well plate, after which the substrate was solidi ed. Next, 1ml GENMED aqueous reagent (Reagent C) with 500ul GENMED clonal reagent (Reagent A) and 200µl cell suspension (containing 2500 cells) were mixed and immediately added into the 12-well plate. The colloid was set at room temperature for 2h and incubated overnight at 37℃ and 5% CO 2 . The next day, 1mL GENMED Reagent D was added into the 12-well plate and cultured at 37℃ and 5% CO 2 for 4 weeks. Photographs were taken under an inverted microscope (Leica Microsystems Inc., USA). Three independent experiments were performed.

Wound healing assay
After transfection, the A549 and H1299 cells were seeded into 6-well plates. When the cell density reached over 80%, a 200µl pipette tip was used to scratch three separate wounds through the cells, moving perpendicular to the line. The cells were then gently rinsed twice with PBS to remove oating cells and cultured in the medium containing 0.5% FBS serum for 48 hours. Images of the scratches were taken using an inverted microscope (Olympus, Tokyo, Japan) at ×10 magni cation at 0 and 48 h of incubation. The experiments were run in triplicate.

Transwell assay
In brief, 3 ~ 5×10^5 cells were resuspended in 300ul serum-free medium and then seeded in the upper chamber (BD Biosciences, New Jersey, USA) pre-coated with or without 40µl diluted Matrigel, while 700µl medium supplemented with 10% FBS was added in the lower chamber. After 24h or 48h, cells on the top surface of the microporous membrane were wiped off with a cotton swab. The remaining cells were xed with 4% paraformaldehyde, stained with 0.1% crystal violet, and counted under a microscope (Leica Microsystems Inc., USA). The experiments were repeated three times.
FITC-phalloidine cytoskeleton staining and cell immuno uorescence staining Cells were slightly washed by preheated (37°C) PBS 3 times and xed in 4% paraformaldehyde for 20 min. Then, the cells were permeated with 0.5% Triton X-100 for 5 min and blocked in 5% BSA for 1h at room temperature. For FITC-phalloidine cytoskeleton staining, F-actin was stained with TRITC (SolarBio, Beijing, China) containing 1% BSA for 40 min at room temperature. For cell IF staining, the cells were incubated with the rabbit polyclonal anti-E-cadherin antibody (diluted 1:100) and the rabbit polyclonal anti-Vimentin antibody (diluted 1:100) primary antibodies at 4°C overnight. The next day, the relevant secondary antibodies were added to the above cells for 1h at room temperature. The nuclei were stained with DAPI for 5 ~ 8min. The cells were imaged using an inverted uorescence microscope (Leica Microsystems Inc., USA). TOP/FOP ash reporter assay A549 and H1299 with stable RBM10 overexpression were cultured in 24-well plates (2 × 10 4 cells per well). After 24h, cells were transfected with the TOP-Flash or FOP-Flash reporter plasmids together with pRL-TK using Lipofectamine 2000 (Invitrogen). After 48h of culture, the luciferase activity was analyzed using a dual-luciferase reporter kit (Promega). Data are presented as the ratio of relative light units of TOP ash to FOP ash from triplicate experiments.
Nuclear and cytoplasmic protein extraction Cytosolic and nuclear protein extraction was performed using a Minute™ Cytoplasmic and Nuclear Extraction Kit for Cells (Invent, SC-003) according to the manufacturer's instructions. In brief, the cells were washed twice with cold PBS, after which the buffer was completely aspirated. Cells were then mixed with an appropriate amount of cytoplasmic extraction buffer and placed on ice for 5min, centrifuged for 5min at 14,000×g at 4°C, after which the supernatant was collected (cytosol fraction). Next, samples were mixed with an appropriate amount of nuclear extraction buffer to pellet, vigorously vortexing for 60 seconds, and then incubated on ice for 15min; this procedure was repeated 4 times, after which samples were centrifuged for 2min at 14,000×g. Each fraction was tested for the presence of the cytosolic marker β-actin and the nuclear marker laminB1 by Western blotting as appropriate. Each experiment was performed three times.

Co-immunoprecipitation (Co-IP)
Co-immunoprecipitation was conducted according to manufacturer's operations (Absin Bioscience Inc, china). Brie y, the cells were washed three times with ice-cold PBS. The cell lysate was then collected at 4℃ using immunoprecipitation lysis buffer supplemented with protease inhibitor (Roche, Basel, Switzerland). The 500µl of cell lysates (containing total protein 200-1000ug) were precleared with 5µl of protein A and protein G agarose beads at 4℃ for 2h. Then, the cell lysates (500µL) were incubated with 5µg of the antibody and 1ug of the normal IgG antibody at 4°C overnight. The next day, samples were mixed with an immunoprecipitation mixture (5µl of protein A and protein G beads) for 3h. The immunecomplex was collected, washed 6 times with cold IP buffer by a 2min centrifugation at 12,000×g. Samples were analyzed by Western blotting. Each experiment was performed three times. Chemicals XAV-939 (a speci c inhibitor of Wnt/β-catenin signaling) and CHIR-99021 (a speci c activator of Wnt/βcatenin signaling) were purchased from Selleckchem. All agents were used according to the manufacturers' instructions.

Animal experiments
Female nude mice (BALB/c, 4 weeks) were purchased from Beijing Vital Li Hua Experimental Animal Technology Company (Beijing, China). Animals were raised in pathogen-free conditions with a temperature of 22 ± 1 ºC, relative humidity of 50 ± 1%, and a light/dark cycle of 12/12 hr. All animal studies (including the mice euthanasia procedure) were done in compliance with the regulations and guidelines of Harbin Medical University institutional animal care and conducted according to the AAALAC and the IACUC guidelines.
For xenograft model construction, 2.5 × 10 7 /150µl A549 cells with or without stable RBM10 overexpression were subcutaneously injected into 4 weeks BALB/c nude mice (n = 5 mice per group). The length and width of tumors were measured every 4 days with a caliper, and the tumor volume (mm 3 ) was calculated with the formula: tumor volume (mm 3 ) = 0.5 × (length × width) 2 . The progression of xenograft growth was analyzed on day 32 using in vivo imaging system, after which the mice were sacri ced, the tumor dissected, weighed, and xed in formalin.

Statistical analysis
Statistical analysis was performed with GraphPad Prism 6.0 software (San Diego, California, USA). All data were shown as the mean ± SD, unless declared. Data were analyzed using Student's t-test for two groups or one-way analysis of variance (ANOVA) for three or more groups. A P value < 0.05 was considered to be statistically signi cant.

Results
Low RBM10 expression is associated with a poor prognosis in LUAD Based on Bhattacharjee lung statistics from Oncomine database (http://www.oncomine.org), the mRNA expression of RBM10 was signi cantly downregulated in LUAD tissues compared with corresponding normal lung tissues (p < 0.05, Fig. 1a). The RBM10 expression gradually decreased with clinical stage progression from TCGA and GTEx database in GEPIA website (http://gepia.cancer-pku.cn/) ( Fig. 1b ~ c). Moreover, Western blot analysis showed that RBM10 protein expression was signi cantly downregulated in LUAD fresh tissues (Fig. 1d). Consistently, the results of qRT-PCR analysis also showed that RBM10 mRNA expression was lower in LUAD cell lines (H1299, H1915, H1650, A549, H1975, H661, H827, PC-9) compared with the normal lung epithelial cell line HBE (Fig. 1e). In addition, the Kaplan-Meier plotter database (http://kmplot.com) analysis showed that patients with low RBM10 expression levels had poor overall survival (OS) compared to patients with high RBM10 expression levels (HR = 0.72, Log-rank P = 0.0068, Fig. 1f). Taken together, these results indicated that expression of RBM10 is low and is positively associated with poor prognosis in LUAD patients.

RBM10 inhibits cell proliferation, migration, and invasion of LUAD cells in vitro
The above data showed that A549 and H1299 cells have a moderate RBM10 expression level (Fig. 1e). Next, in one group, we silenced RBM10 in those cells with RBM10-siRNA to knock down RBM10 expression; in the other cell group, RBM10 was stably overexpressed using an RBM10 lentivirus. Both the overexpression and knockdown e ciencies of RBM10 were con rmed by qRT-PCR and Western blot assays (Additional le 1: Figure S1). We then performed a variety of in vitro experiments to evaluate the effect of RBM10 expression on cell proliferation, migration, and invasion of LUAD cells. The CCK-8 and EdU assays showed that RBM10 knockdown signi cantly increased cell viability and enhanced the DNA synthesis ability of both A549 and H1299 cells (Fig. 2a, c, e), while forced RBM10 expression caused an opposite effect (Fig. 2b, d, e). Moreover, the inhibition of RBM10 led to the generation of more and larger cell colonies compared with the control groups, while overexpressing RBM10 reduced both colony size and number in the A549 and H1299 cells ( Fig. 2f ~ h). These ndings were further con rmed in soft agar colony formation assays (Additional le 1: Figure S2). We also found that RBM10 silencing promoted the invasion and migration ability of A529 and H1299 cells, whereas RBM10-overexpressing LUAD cells reduced the cell invasion and migration capacity ( Fig. 2i ~ t). Thus, this data suggests that RBM10 may inhibit tumor cell proliferation, migration, and invasion of LUAD cells in vitro.

RBM10 suppresses LUAD cell tumorigenesis and metastasis in vivo
Next, we explored that RBM10 suppressed LUAD tumorigenesis and metastasis in vivo. A xenograft tumor mouse model was established by subcutaneously injecting 2.5 × 10 7 /150µl A549-Vector and A549-RBM10 cells into the left armpit of 4 weeks BALB/c nude mice (Additional le 2: Figure S1a). Western blot assay showed that RBM10 was stably overexpressed in A549 cell lines (Additional le 2: Figure S1b). As shown in Fig. 3a and 3b, tumor volumes signi cantly decreased in the A549-RBM10 group compared with the A549-vector group. At the end of the experiments, the mice were sacri ced, the subcutaneous tumors were isolated, and their volume and weights were measured. The results showed that tumor volume was markedly decreased in the A549-RBM10 group when compared with the A549-Vector group (Fig. 3c, d, P < 0.05). Furthermore, IHC analyses showed that the xenograft tumors from the A549-RBM10 group displayed a lower level of Ki67 relative to control (A549-Vector) (Fig. 3e).
We also used the tail vein injection mouse model to investigate the in uence of RBM10 in LUAD metastasis in vivo (Additional le 2: Figure S2). On day 49 after inoculation, the mice were sacri ced, the lungs were collected, and metastatic nodules were counted. The results indicated that the number and the size of lung metastasis lesions were signi cantly decreased in mice injected with A549-RBM10 cells relative to control (A549-Vector) cells (Fig. 4a ~ c). Moreover, fewer mice in the A549-RBM10 group (1/5, 20%) showed lung metastasis, while in the control group (A549-Vector), all mice developed lung metastasis (4/5, 80%, Fig. 4d). In addition, we also found that the A549-RBM10 group had smaller and fewer lung metastatic foci than those in the control group (A549-vector) (Fig. 4e). Collectively, these data indicated that RBM10 overexpression inhibited LUAD lung metastasis in vivo.

RBM10 inhibits the EMT program of LUAD
Current evidence suggests that epithelial-mesenchymal transition (EMT) is a key step in the progression of cancer metastasis [22]. Through mRNA expression correlation analysis on the TCGA database, we found that the expression of RBM10 is positively correlated with CHD1 (also called E-cadherin) but negatively correlated with Vimentin (VIM), ZEB1, and ZEB2 ( Fig. 5a and Additional le 3: Figure S1), thus indicating that RBM10 might participate in EMT process of LUAD. The cytoskeleton can trigger micro lament structural changes and increase the number of pseudopodia (lamellipodia and lopodia), which is responsible for cancer cells' invasive and migratory properties [37]. FITC-phalloidine cytoskeleton staining was performed to evaluate morphological alterations in LUAD cells. The results showed that RBM10-silencing cells formed a large number of visible actin laments and pseudopodia compared with control cells, while RBM10-overexpressing cells had clear and round cell shapes bearing scarcely actin remodeling (Fig. 5b, c).
Next, we evaluated whether EMT markers were altered. Using immuno uorescence (IF) assays, we observed that the uorescence intensity of E-cadherin decreased and Vimentin increased in the RBM10 silencing cells, whereas RBM10 overexpression upregulated E-cadherin uorescence but downregulated Vimentin (Fig. 5d ~ g). A similar result was also revealed by IHC in tumor tissues from xenograft tumors (Additional le 3: Figure S2). Western blot analysis demonstrated that RBM10 silencing increased Vimentin, N-cadherin, slug, and twist protein expression levels, whereas E-cadherin and EPCAM protein expression levels were decreased in A549 and H1299 cells (Fig. 5h, i). Conversely, overexpression of RBM10 showed the opposite effect (Fig. 5j, k). Additionally, we also evaluated the mRNA expression of Ecadherin, Vimentin, slug, and twist in LUAD cells, and changes of mRNA expression were consistent with that observed at the protein level (Additional le 3: Figure S3). Furthermore, qRT-PCR assays showed that downregulation of RBM10 clearly increased the mRNA levels of ZEB1, ZEB2, MMP3, MMP7, and MMP10, while upregulation of RBM10 markedly reduced their mRNA expression (Additional le 3: Figure S4). Thus, these results strongly suggested that RBM10 inhibited EMT.

RBM10 negatively regulates the Wnt/β-catenin signaling pathway
To elucidate the underlying molecular mechanisms through which RBM10 regulates LUAD progression, the RNA sequencing (RNA-seq) was performed using H1299 cells that express either control si-NC or si-RBM10. Gene ontology (GO) enrichment analysis revealed that RBM10-dependent genes were involved in either biological processes such as biological adhesion, cell proliferation, and growth, or cellular component, including cell junction (Additional le 4: Figure S1a), supporting a role for RBM10 in cell proliferation and EMT. KEGG pathway analysis showed that these genes were signi cantly associated with cancer-related functions, including cellular motility, growth and death and etc. (Additional le 4: Figure S1b). More importantly, pathway enrichment analysis suggested that multiple signaling pathways might participate in the tumor-promoting mechanism of silencing RBM10, such as the Wnt/β-catenin signaling pathway, NF-KB signaling pathway, TGF-β signaling pathway (Additional le 4: Figure S2). Based on ChipBase and StarBase databases, we found that the expression of RBM10 was markedly negatively correlated with the four common Wnt/β-catenin signaling pathway target genes such as CTNNB1(also called β-catenin), Wnt5a, c-MYC, and CD44 in LUAD (Additional le 4: Figure S3). Hence, Wnt/β-catenin signaling was selected for further research.
We performed TOP/FOP ash luciferase reporter assays. The results showed that RBM10 overexpression signi cantly reduced the activity of the TOP/FOP-ash reporter genes in both A549 and H1299 cells compared to control cells (Fig. 6a), thus suggesting that RBM10 inhibits the WNT/β-catenin signaling activity in LUAD cells. Interestingly, qRT-PCR and Western blot indicated that RBM10 knockdown signi cantly increased the expression of endogenous β-catenin in LUAD cells (Fig. 6b, c, Additional le 4: Figure S4a, b), while overexpression of RBM10 markedly reduced the expression of β-catenin (Fig. 6d, e, Additional le 4: Figure S4c, d). Furthermore, we performed the nuclear and cytoplasmic cellular fractions. Western blot assays results showed that the level of β-catenin in the nucleus was increased, while that in the cytoplasm was decreased by silencing RBM10 (Fig. 6f, g). In contrast, overexpressing RBM10 made impaired nuclear β-catenin and induced cytoplasmic β-catenin (Fig. 6h, i).
Moreover, the subcellular localization of β-catenin in A549 cells detected by immuno uorescence (IF) analysis further supported our hypothesis. The results indicated that downregulation of RBM10 increased the concentration of β-catenin in nuclear and blocked the β-catenin in the cytoplasm in A549 cells (Fig. 6j), while upregulation of RBM10 decreased the expression of β-catenin in nuclear and enhanced the β-catenin in the cytoplasm (Additional le 4: Figure S5).
We also examined the expression of c-MYC, cyclin D1, and MMP7, which are important downstream target genes of the Wnt/β-catenin signaling pathway [38,39]. As shown in Fig. 6b ~ e and Additional le 4: Figure S4a ~ d, the results indicated that the mRNA and protein expression levels of c-MYC, MMP7, and cyclinD1 were up-or down-regulated when RBM10 was silenced or overexpressed in A549 and H1299 cells. In addition, IHC staining showed that c-MYC and cyclinD1 expression was lower in xenograft tumors with RBM10 overexpression (Additional le 4: Figure S6).
To further investigate the functions of Wnt/β-catenin signaling on the progression of LUAD, cells were treated with Wnt/β-catenin pathway activator CHIR 99021 [40] or inhibitor XAV93 [41]. As shown in Fig. 6k, l, when cells were treated with XAV-939 (10µl) for 24h, Transwell assays indicated that the migration and invasion ability of A549 and H1299 RBM10 silencing cells was signi cantly decreased.
However, CHIR 99021 promoted the migration and invasion ability of RBM10-overexpressing cells (Fig. 6m, n). Furthermore, Western blot showed that CHIR 99021 reverses the effect of RBM10 overexpressed on EMT markers (E-cadherin and twist) expression (Fig. 6o). Altogether, all these data indicated that depletion of RBM10 might promote LUAD cell proliferation and metastasis through promoting the activation of the Wnt/β-catenin pathway.

RBM10 interacts with CTNNBIP1
TCGA analysis showed that mRNA expression of CTNNBIP1 was markedly lower in most solid cancer tissues, including in LUAD (Additional le 5: Figure S1a). UALCAN (http://ualcan.path.uab.edu/index.html) and StarBase analysis showed that RBM10 expression was signi cantly downregulated in LUAD compared to that of normal lung tissues (Additional le 5: Figure  S1b, c). GEPIA website indicated that the mRNA expression of CTNNBIP1 was decreased in lung adenocarcinoma tissues and negatively corrected with clinical stage of lung adenocarcinoma patients, which was similar to the expression pattern of RBM10 (Additional le 5: Figure S1d, e). Furthermore, qRT-PCR assays showed that the CTNNBIP1 mRNA expression was low in LUAD cell lines (Additional le 5: Figure S1f). Kaplan-Meier survival curves revealed that low CTNNBIP1 expression in LUAD correlated with poor survival (HR = 0.67, Log-rank P = 0.012, Additional le 5: Figure S1g). These results suggested that CTNNBIP1 has low expression in lung adenocarcinoma and that it has a tumor-suppressive role.
We further analyzed the expression pattern of RBM10 and CTNNBIP1 in Landi lung statistics from the Oncomine database. The results showed that CTNNBIP1 exhibited a similar expression pattern with RBM10, both of which had low expression in LUAD tissues (Fig. 7a, b). In addition, the correlation between RBM10 and CTNNBIP1 was analyzed using the LUAD-TCGA data collection from the StarBase database; the results showed that RBM10 expression was positively correlated with CTNNBIP1 (Fig. 7c). Based on these data, we inferred that RBM10 interacts with CTNNBIP1, which was then con rmed by the co-IP assay (Fig. 7d, e); co-IP assay of the nuclear and cytoplasmic cellular fractions showed that RBM10 mainly interacted with CTNNBIP1 in the nucleus (Fig. 7f).
For further exploration of the regulatory relationship between RBM10 and CTNNBIP1, we used Western blot and qRT-PCR assays to detect their expressions. As shown in Fig. 7g, h, RBM10 silencing decreased the CTNNBIP1 protein level in A549 and H1299 cells, while RBM10-overexpression caused an opposite effect. In addition, down-regulation or upregulation of RBM10 did not signi cantly alter the levels of CTNNBIP1 mRNA transcripts in A549 and H1299 cells (Fig. 7i). These ndings suggested that CTNNBIP1 might be downstream of RBM10. RBM10 positively regulates CTNNBIP1, and its expression is regulated by RBM10 at the protein levels.
In line with these results, we also conducted rescue experiments. In A549 and H1299 cells with stable overexpression of RBM10, si-CTNNBIP1 was transiently transfected to down-regulate the expression of CTNNBIP1. Using wound healing assay and Transwell assay, we observed that RBM10 signi cantly reduced cell invasion and migration in A549 and H1299 cells while silencing CTNNBIP1 reversed this process (Additional le 5: Figure S2). Collectively, these data demonstrate that RBM10 interacts with CTNNBIP1 and reduces the protein expression of CTNNBIP1 in LUAD.

RBM10 inhibits the Wnt/β-catenin pathway by blocking the β-catenin-TCF/LEF interaction
The precise molecular mechanisms through which RBM10 suppresses the Wnt/β-catenin pathway activity in LUAD cells were further elucidated. CTNNBIP1, as an inhibitor of β-catenin, can directly bind with β-catenin and impair the interaction between β-catenin and TCF/LEF complex, and subsequently inhibits Wnt/β-catenin signaling pathway. StarBase database showed that the mRNA expression of RBM10 was positively correlated with the expression of CTNNBIP1 (Fig. 7c) but negatively corrected with the expression of TCF4 and LEF1 (Fig. 8a, b). Upregulation of RBM10 led to low expression levels of TCF3, TCF4, and LEF1 (Fig. 8c ~ f). Therefore, we performed co-IP assays to further con rm whether RBM10 regulated the Wnt/β-catenin pathway through CTNNBIP1. As shown in Fig. 8g and h, we observed that RBM10 overexpression markedly inhibited the association of β-catenin with TCF/LEF while enhancing the interaction between β-catenin and CTNNBIP1. In general, these results proved that RBM10 inactivated the Wnt/β-catenin pathway by increasing the inhibitory role of CTNNBIP1 and blocking the βcatenin-TCF/LEF interaction.

Discussion
Our previous study indicated a high frameshift mutation of RBM10 in LUAD (2/19, approximately 10.5%), which was distributed throughout the coding region rather than at speci c sites; this data are similar to the classical tumor suppressor gene mutation spectrum (article to be published) [42]. Previous studies have con rmed that frameshift mutations usually decrease or lose the gene's function. In this research, bioinformatics analysis showed that RBM10 was down-regulated in LUAD tissues compared with normal lung tissues, and the expression of RBM10 gradually decreased with the progression of clinical stages.
Kaplan-Meier analysis results showed that low RBM10 expression had a signi cantly shorter survival time and poor prognosis, thus suggesting that the low RBM10 expression was a marker of poor prognosis in LUAD. Based on this, we speculated that RBM10 is involved in LUAD progression as a tumor suppressor gene.
To date, only a few studies have shown that RBM10 is involved in tumor metastasis [18,21]. Garrisi et al [21] reported that high RBM10 expression was correlated with increased disease aggression in melanoma cancer. Furthermore, Julie et al [43] found that RBM10 promotes transformation-associated processes in RBM5-null SCLC cells. However, so far, there are no studies on the involvement of RBM10 in metastasis and progression of lung adenocarcinoma. Our study revealed that RBM10 knockdown markedly promoted the proliferation, migration, and invasion ability of LUAD cells, while RBM10 overexpression had the opposite effects. We further con rmed the inhibiting effect of RBM10 on tumor growth and lung metastasis of LUAD in vivo. Together, our studies clearly indicate that RBM10 inhibited proliferation and metastasis of LUAD cells by regulating EMT.
Next, we explored the underlying mechanism of RBM10 inhibiting the process of LUAD. Firstly, using RNAseq and KEGG pathway enrichment analysis, we found that RBM10 is involved in a variety of signaling pathways, including the Wnt signaling pathway, NF-kB signaling pathway, and TGF-β signaling pathway.
Combined with TGCA databases, we found that RBM10 expression was negatively regulated with Wnt/βcatenin pathway target genes (CTNNB1, c-MYC, MMP7, CD44). The TOP/FOP, luciferase activity assay showed the Wnt/β-catenin signaling activity was attenuated by RBM10 overexpression. Additionally, the expression of β-catenin, cyclin D1, MMP7, and c-MYC, key molecules of the Wnt/β-catenin pathway, were decreased or increased by up-or down-regulation of RBM10. We also detected that LUAD cell invasion and migration induced by RBM10 overexpression or silencing were reversed by either CHIR-99021 or XAV-939. To the best of our knowledge, this is the rst study that reported how RBM10 inhibits the EMT process of lung adenocarcinoma by regulating the Wnt/β-catenin signaling pathway. Recently, the Wnt/βcatenin signaling pathway has been gradually recognized as a potentially important target for anticancer therapy [27,50]. Preclinical and clinical studies have shown that inhibitors targeting the Wnt/β-catenin pathway, such as Wnt974, LGX818, OMP-18R5 (Vantictumab), OMP-54F28 (ipafricept), and CWP232291, can successfully inhibit tumors progression. Hence, our results may help improve treatment strategies for the selection of LUAD patients who may particularly bene t from agents that selectively target blocking the Wnt/β-catenin pathway.
Our results revealed that RBM10 interacts with CTNNBIP1 and positively regulates its expression in LUAD. Previous studies reported that the CTNNBIP1 gene is an antagonist of Wnt signaling [51]. By interacting with β-catenin, CTNNBIP1 disrupts the binding of β-catenin with TCF/LEF complex and down-regulates the expression of downstream target genes of the Wnt signaling pathway (such as c-MYC, CyclinD1, MMP7, etc.), and prevents Wnt/β-catenin signaling pathway activation from inhibiting the progression of LUAD [52,53]. This nding may explain why TCF3, TCF4, and LEF1 protein levels were reduced by RBM10 in the present study. In our study, down-regulation of CTTNBIP1 expression reversed the cell migration and invasion ability inhibited by overexpression of RBM10 in LUAD cells. Furthermore, we also found that overexpression of RBM10 promoted the inhibitory role of CTNNBIP1, reduced the interaction between βcatenin and TCF/LEF complex while promoting the interaction between β-catenin and CTNNBIP1. Above all, these results indicated that RBM10 abolished the binding of β-catenin and TCF/LEF complex and nally inhibited Wnt/β-catenin signaling to prevent LUAD progression by interacting with CTNNBIP1 and regulating its expression. Whether RBM10 has a direct interaction with CTNNBIP1 and which domain or sequence participates in the above interaction needs to be further studied.

Conclusions
The present study revealed a rst working model for how RBM10 inhibits LUAD tumor growth and metastasis (Fig. 8i), indicating that RBM10 may be a potential biomarker and therapeutic target for lung adenocarcinoma. Speci cally, RBM10 interacts with CTNBBIP1 and downregulates CTNNBIP1 expression, thereby disrupting the interaction between β-catenin and TCF/LEF complex and inactivating the Wnt/β-catenin pathway. This nding may broaden the understanding of mechanisms involved in LUAD progression; moreover, as a prognostic predictor, RBM10 might be a potential target for LUAD therapy.       (d, e) Co-immunoprecipitation (Co-IP) was used to validate the interaction of RBM10 and CTNNBIP1. (d) RBM10 was pulled down by anti-RBM10, and Western blot was used to detect RBM10 and CTNNBIP1. (e) CTNNBIP1 was pulled down by anti-CTNNBIP1, and then RBM10 and CTNNBIP1 were detected by western blot. (f) co-IP assay of the nuclear and cytoplasmic cellular fractions revealed that RBM10 mainly interacted with CTNNBIP1 in the nucleus. (g, h) Western blot assay was utilized to detect RBM10 and CTNNBIP1 expressions in RBM10 knockdown (g) or overexpression (h) cells. β-actin was used as loading controls. (i) qTR-PCR result of RBM10 and CTNNBIP1 mRNA expressions under RBM10 knockdown or overexpression. Both down-regulation or upregulation of RBM10 did not alter CTNNBIP1 mRNA. GAPDH was used as loading controls. The results were represented as mean ± SD. *P < 0.05, All experiments were repeated three times. Figure 8