LRRC4 Suppresses E-Cadherin-Dependent Collective Cell Invasion and Metastasis in Epithelial Ovarian Cancer

Epithelial ovarian cancer (EOC) is the most malignant gynecological carcinoma and is of a high incidence of death due to detection at late stages when metastasis already occurs. However, the mechanism underlying metastasis of EOC remains unclear. Analysis of the open database and experiments with immunochemistry showed that LRRC4 is lowly expressed in high-grade serous ovarian cancer (HGSC) cells and during EOC metastasis. The 3D cell culture system and the orthotopic ovarian xenograft model infected with LRRC4-containing adeno-associated virus serotype 9 (AAV9) were used to confirm collective invasion and metastasis of cells in vitro and in vivo. Phos-tag SDS-PAGE was used to detect the phosphorylation of LRRC4 and PIK3R1. A number of experiments with methods such as co-immunoprecipitation and immunoblotting were performed to explore the mechanism for the actions of LRRC4 and PIK3R1 in EOC metastasis. An inverse correlation between LRRC4 and E-cadherin expression was detected in the regions of invasion in primary EOC tissues and metastatic ascites. LRRC4 binds to the cSH2 domain of PIK3R1 and inhibits the activity of PIK3R1, without disrupting the physical interactions between PIK3R1 and PIK3CA. LRRC4 inhibits EOC metastasis by targeting E-cadherin-dependent collective cell invasion and does so by inhibiting the PIK3R1-mediated AKT/GSK3β/β-catenin signaling pathway. LRRC4 functions as a tumor suppressor gene to inhibit EOC collective invasion and metastasis in vitro and in vivo and does so by directly binding to the cSH2 domain of PIK3R1 to exert its regulatory function. Our findings provide a potential novel approach for metastasis prognosis and a new strategy for the treatment of EOC.

Epithelial ovarian cancer (EOC) is the most malignant gynecological carcinoma and is of a high incidence of death due to detection at late stages when metastasis already occurs. However, the mechanism underlying metastasis of EOC remains unclear. Analysis of the open database and experiments with immunochemistry showed that LRRC4 is lowly expressed in high-grade serous ovarian cancer (HGSC) cells and during EOC metastasis. The 3D cell culture system and the orthotopic ovarian xenograft model infected with LRRC4-containing adeno-associated virus serotype 9 (AAV9) were used to confirm collective invasion and metastasis of cells in vitro and in vivo. Phos-tag SDS-PAGE was used to detect the phosphorylation of LRRC4 and PIK3R1. A number of experiments with methods such as co-immunoprecipitation and immunoblotting were performed to explore the mechanism for the actions of LRRC4 and PIK3R1 in EOC metastasis. An inverse correlation between LRRC4 and E-cadherin expression was detected in the regions of invasion in primary EOC tissues and metastatic ascites. LRRC4 binds to the cSH2 domain of PIK3R1 and inhibits the activity of PIK3R1, without disrupting the physical interactions between PIK3R1 and PIK3CA. LRRC4 inhibits EOC metastasis by targeting E-cadherin-dependent collective cell invasion and does so by inhibiting the PIK3R1-mediated AKT/GSK3β/β-catenin signaling pathway. LRRC4 functions as a tumor suppressor gene to inhibit EOC collective invasion and metastasis in vitro and in vivo and does so by directly binding to the cSH2 domain of PIK3R1 to exert its regulatory function. Our findings provide a potential novel approach for metastasis prognosis and a new strategy for the treatment of EOC.
Keywords: collective invasion and metastasis, epithelial ovarian cancer, LRRC4, PIK3R1, E-cadherin INTRODUCTION Epithelial ovarian cancer (EOC) is the most malignant gynecological carcinoma with a high mortality rate and poor prognosis (1). An increasing trend in ovarian cancer mortality has been observed in females (1). High-grade serous ovarian cancer (HGSC) accounts for ∼75% of all ovarian cancer deaths due to advanced stage and cancer cell metastasis when patients are diagnosed (2). In the past 30 years, little improvement in overall survival has be achieved, while the standard treatment has not advanced beyond platinum-base combination chemotherapy (3). Despite considerable efforts to improve early detection and treatment, tumor metastasis remains a major challenge for clinical treatment (4). Metastasis is a complex and multistep process that requires cancer cells to detach from primary tumors and migrate to distant organs and is responsible for more than 90% of all cancer-associated deaths. However, EOC mostly metastasizes into the peritoneal cavity as individual cells and clusters by shedding from ovarian cancer instead of spread in haematogenously (5,6). Recent studies have revealed that cell-collective metastasis is the primary mechanism for metastasis in ovarian cancer and that cell-cell junction plays a key role in collective cell movement (7,8).
More recently, genomic analyses have revealed that the alteration of the PI3K/AKT pathway in HGSC is a common event and is associated with poor clinical outcomes (9). The regulatory subunit, PIK3R1 (p85α), of PI3K tightly interacts with PIK3CA, the catalytic subunit of PI3K-p110α (as heterogeneity in nSH2 or cSH2 of PIK3R1), resulting in the inhibition of PI3K-p110α activity (10). Moreover, inhibition of PI3K-p110α is released when PI3K-p110α is recruited to the plasma membrane by receptor tyrosine kinases (RTKs), tyrosine-phosphorylated adaptor proteins, and the members of the Ras superfamily of small G proteins (11), leading to the activation of downstream effectors, such as AKT (12). PIK3R1 is currently considered as an oncogene in ovarian cancer (13).
In this study, we examined the role of LRRC4 in EOC metastasis. We detected low LRRC4 expression in HGSCs and metastatic tissues. In ovarian cancer orthotopic xenograft, LRRC4 overexpression inhibits metastasis of ovarian cancer cells. The inhibition of metastasis is dependent upon E-cadherin, but independent of EMT. At the mechanistic level, we showed that LRRC4 binds to the c-SH2 domain of PIK3R1 to inhibit the activity of PIK3R1 and subsequently disrupts E-cadherindependent collective invasion of ovarian cancer cells mediated by the PIK3R1 signaling pathway.

MATERIALS AND METHODS
Cell Culture and Tissue Samples A2780 human ovarian cancer cells were obtained from BOSTER Biological Technology. SKOV3, OVCAR3, and OVCAR5 cells were obtained from the Cell Culture Center of PUME. HEK293, A2780, and HO8910 cell lines were cultured in high-glucose DMEM (Invitrogen, CA, USA) containing 10% fetal bovine serum at 37 • C in 5% CO 2 . SKOV3, OVCAR3, and OVCAR5 cell lines were grown in RPMI medium (Invitrogen, CA, USA) containing 10% fetal bovine serum at 37 • C in 5% CO 2 . Paraffinembedded specimens of serous cystoma, HGSC and omental metastasis tissues were obtained from the Second Xiangya Hospital of Central South University and with appropriate patient consent.

Vector, shRNAs, and siRNAs
The recombinant adeno-associated virus serotype 9 (AAV9) vector of LRRC4 (AAV9/GFP/LRRC4) was obtained from Weizhen biotechnology limited company. shRNA constructs against PIK3R1 and a non-targeting shRNA control were obtained from the BROAD Institute. Transient knockdown of LRRC4 in SKOV3 cells was performed using siRNAs purchased from GenePharma. FOP/TOP plasmids were obtained from Majian group of Cancer Research Institute in Central South University.

Stable Cell Lines
The pcDNA3.1-LRRC4 plasmid was transient transfected into SKOV3 cells, and cells were thereafter treated with G418 for one weak in order to select the SKOV3 cells with pcDNA3.1-LRRC4 stable expression. In addition, lentivirus, which was generated by HANBIO and carried GFP or GFP/LRRC4, was used to infect the SKOV3 cells. Cells with stable expression of GFP or GFP/LRRC4 SKOV3 were selected by the use of flow cytometry.

Immunohistochemistry Staining
All tissues were fixed in 4% paraformaldehyde and embedded in paraffin. For samples from human HGSC, metastasis and ascitic tissues, the anti-LRRC4 (Santa Cruz) and anti-E-cadherin (Cell Signaling Technology) primary antibodies were used at dilutions of 1:300 and 1:500, respectively. For mouse tissues, the anti-ki67 (Santa Cruz) primary antibody was used at 1:200. LRRC4 (Abcam), E-cadherin (Cell Signaling Technology), pAKT (ser473) (Cell Signaling Technology), AKT (Proteintech), GSK3β (Cell Signaling Technology), pGSKβ (ser9) (Cell Signaling Technology), Vimentin (Boster Biological Technology), and βcatenin (Sangon Biotech) primary antibodies were used 1:500. A biotinylated secondary antibody was used to recognized primary antibodies, and signals were visualized using the Ultra Sensitive TM SP system and peroxidase substrate DAB kit. Immunohistochemistry data were quantitated using staining intensity and proportion of tumor cells. The intensity was scored using a system from 0 to 3 as follows: negative as 0; weak as 1; moderate as 2; and strong as 3. The frequency of positive cells was defined using a system from score 1-4 as follows: negative or <5% as 0; 5-25% as 1; 26-50% as 2; 51-75% as 3; and >75% as 4 score. Each component was scored independently, and the results were summed up.

Orthotopic Xenograft
Briefly, mice were completely anesthetized by the use of pentobarbital, and a dorsolateral incision (1-2 cm long) was made using surgical scissors on the top right of the spleen. The fat pad was exposed to open access to the mouse ovary. The ovary was firmly grasped with forceps, and 10 6 tumor cells were injected into the ovary. After forceps were released, the wound was closed using sutures. After mice were sacrificed, the tumors were dissociated from the mouse ovary and omentum, while the size of the tumors was measured. For the intraperitoneal xenograft, 2 × 10 6 tumor cells were intraperitoneally implanted into 5-6 week-old mice, which were euthanatized after 8 weeks. The metastasis nodules were quantified, while the number of mice used in each experiment is depicted in figure legends.

Cell Proliferation and Invasion Assays
Proliferation was measured using EdU (EdU Apollo R 488 in vitro Imaging Kit). Briefly, 5,000 cells/well were seeded in a 96well plate in 100 µl of growth medium 24 h post-transfection. After 36 h, cell proliferation was measured according to the manufacturer's instructions. For the invasion assay, 10 5 cells were placed in the top compartments medium containing 1% FBS while medium containing 10% FBS was placed in the bottom compartments of a Boyden chamber (Corning). After 36 h, cells that invaded through the membrane were stained with crystal violet and counted using Image J.

Quantitative Real-Time PCR
Total RNA was extracted using the RNeasy Mini Kit (Qiagen). Total RNA (1 µg) was processed to generate cDNA using the Revert Aid First Strand cDNA Synthesis Kit (Thermo Fisher). Quantitative real-time PCR (RT-PCR) was conducted using a Bio-Rad CFX96 system with SYBR green (Takara) to determine the mRNA levels of genes of interest.

Immunoblotting
For total cell lysates, cells were lysed in lysis buffer, and proteins were quantified using a BCA assay. For nuclear and membrane proteins, proteins were separated using a kit (Beyotime Biotechnology and Thermo). Total protein (30 µg) was separated on 8-10% SDS/PAGE gels and then transferred onto polyvinylidene fluoride PVDF membranes (Millipore, Billerica, MA). PVDF membranes were blocked with 3-5% BSA for 1 h and then incubated with specific primary antibodies at 4 • C overnight. The primary antibodies used were the same as those used for immunohistochemistry. Membranes were incubated with HRP-conjugated secondary antibodies (Proteintech) and visualized using the ECL system (Millipore). Immunoblots were developed using a ChemicalDocTM XRS+ (Bio-Rad, Berkeley, CA, USA).

Co-immunoprecipitation
Cells were lysed with immunoprecipitation buffer containing protease inhibitors. Lysates were incubated with 6 µg/ml antibodies or normal IgG at 4 • C overnight in a rotary agitator. Protein A magnetic beads were added to lysates and incubated for 4-6 h at 4 • C. Magnetic beads were collected and washed three times with immunoprecipitation buffer. Total lysates and immunoprecipitants were separated by SDS/PAGE gel and analyzed using western blotting.

Immunofluorescence
Cells were fixed in 4% paraformaldehyde for 20 min at room temperature, permeabilized with 0.25% Triton-X, and blocked in 5% bovine serum albumin (BSA) for 30 min. Cells were incubated with primary antibodies at 4 • C overnight or 6-8 h at room temperature. Secondary antibodies were coupled with Alexa Fluor 488 and 647, and cells were incubated with secondary antibodies for 2 h at room temperature. Cells were also incubated with phalloidin at 1:1,000 to stain F-actin-containing cells.

3D Cell Culture
Cells were digested into single-cell suspensions at a density of 3,000 cells/ml in conditioned medium. Cells were then embedded in Matrigel (BD Biosciences, 354236) or Collagen I (BD Biosciences, 354236) with 100 ng/ml EGF, 20 ng/ml FGF, 2% B27, and 1% antibiotics (100 units/ml penicillin and 100 mg/ml streptomycin) in 200 µl of medium. Cells were then placed in a 37 • C heating block for 3 days, and the medium mixture was replaced every 2 days.

Phos-tag SDS-PAGE
Cells were lysed in phosphorylation buffer, and protein was quantified. The SDS-PAGE system used to detect the phosphorylation of LRRC4 and PIK3R1 consisted of a separating gel and a stacking gel. The separating gel (7.5 mL) consisted of 6%w/v polyacrylamide and 1.5 mM Bis-Tris-HCl buffer (pH 8.8) and was mixed with 30 µM Phos-tag and two equivalents of MnCl2. The stacking gel (2.5 mL) consisted of 4.5% w/v polyacrylamide and 1.4 mM Bis-Tris-HCl buffer (pH 6.8). Western blotting was performed, and primary antibodies were incubated with the membranes.

Statistical Analysis
Student's t-test was used to assess the significance of mean differences using SPSS Statistics software. Differences were considered significant at a two-tailed P-value of 0.05. All experiments were performed at least three times.

Low LRRC4 Expression Is Associated With HGSC and EOC Metastasis
We first examined the expression of LRRC4 in 5 normal ovarian tissues, 11 low-grade serous ovarian cancer (LGSC), 8 low malignant serous ovarian cancer (LMSC), and 17 HGSC samples using Gene Expression Omnibus (GEO) databases. Compared with normal ovarian surface epithelial cells, LRRC4 expression was reduced in advanced HGSC (p = 0.0093), but no significant difference in LRRC4 expression was found between the normal ovarian surface epithelial cells and LGSC or LMSC cells ( Figure 1A). LRRC4 was down-regulated in EOC tissues from ascitic cytology-positive patients compared with those from ascitic cytology-negative patients ( Figure 1B). LRRC4 expression was also reduced in serous ovarian cancer cells with ascitic cytology-positive patients (Figure 1C), suggesting that low expression of LRRC4 is associated with high-degree EOC malignant progression and metastasis.

LRRC4 Suppresses EOC Tumourigenesis and Metastasis
The correlation between LRRC4 low expression and high degree of EOC invasion and metastasis led us to ask whether LRRC4 has a function in these responses. Experiments with Western blotting and real-time quantitative PCR showed that LRRC4 expression was either absent or at low levels in five EOC cell lines (Figures S1A,C). We used the SKOV3 cell line to stably express LRRC4 was generated using the lentivirus-based system, which allowed LRRC4 to effectively express ( Figure S1B). EDU and transwell assays showed that LRRC4 ectopic expression inhibited cell proliferation and invasion (Figures S1D,E). We next used the ovarian orthotopic xenograft mouse model to assess the effect of LRRC4 ectopic expression on EOC tumourigenesis in vivo (22). We inserted LRRC4 cDNA into a recombinant AAV9 vector, which produces high and sustained serum levels of transgene when delivered as in a single injection (22). Mice were treated with AAV9/GFP/LRRC4 fusion vector (n = 5) or the control AAV9/GFP vector (n = 5) by the use of intraperitoneal injection (23), while the left ovary was orthotopically xenografted with 10 6 SKOV3 EOC cells in each experimental group (Figure 2A). We confirmed LRRC4 expression in the mouse ovary, fallopian tube muscle layer, fallopian tube mucosa, and xenografted tumors by using IHC, indicating that AAV9/GFP/LRRC4 was successfully expressed ( Figure S2A). As shown in Figure 2B, the SKOV3 cells implanted orthotopically in the ovary formed tumors in the AAV9/GFP group, while the AAV9/GFP/LRRC4 group exhibited significantly reduced tumourigenesis. Much weaker nuclear staining of ki67 was observed in cells of the AAV9/GFP/LRRC4 group compared with those of the AAV9/GFP group ( Figure 2C). Moreover, LRRC4 ectopic expression reduced cancer cell metastasis to the omentum from ovarian orthotopicallyxenografts ( Figure 2D). Finally, we used an intraperitoneal xenograft mouse model, which showed that LRRC4 ectopic expression significantly reduced abdominal metastases of SKOV3 cells when compared with those with empty vectors (Figure 2E).

LRRC4 Inhibits Collective Cells Invasion by Down-Regulating E-Cadherin Without EMT Induction
How does LRRC4 inhibit EOC metastasis? Epithelialmesenchymal transition (EMT), in which cells typically lose cell-cell cadherin and gain migration and invasion ability, plays a crucial role in tumor metastasis (24). The central role of EMT in tumor metastasis led us to assess whether it also mediates LRRC4 inhibition of EOC metastasis. Experiments with phalloidin staining demonstrated that SKOV3 EOC cells highly polarized with F-actin accumulation, while LRRC4 ectopic expression inhibited F-actin accumulation (indicated by arrowhead) compared with control cells (Figure 3A). At the molecular level, ectopic expression of LRRC4 simultaneously reduced the mRNA levels of E-cadherin and Vimentin but did not affect those of N-cadherin, slug, twist, ZEB1, and ZEB2 ( Figure 3B). Western blotting also showed that LRRC4 ectopic expression reduced E-cadherin and Vimentin protein levels in a dose-dependent manner (Figure 3C). The failure of LRRC4 expression to up-regulate the EMT markers suggested that LRRC4 inhibits EOC cell metastasis independently of EMT.
If EMT does not mediate LRRC4 inhibition of EOC metastasis, what may be the mechanism? We found that human normal ovary epithelial cells exhibited relatively low E-cadherin but high LRRC4 expression (n = 5, Figure 3D). In contrast, EOC cells exhibited high E-cadherin but low LRRC4 expression in the invasion regions in human HGSC (n = 17, Figure 3E) and were organized into multicellular units, indicating collective invasion, and metastasis (red box). Furthermore, E-cadherin expression was present in the ascites of ovarian cancer patients and was accompanied by low LRRC4 expression ( Figure 3F). Finally, we examined E-cadherin expression in EOC orthotopic xenograft tumor models we previously generated using AAV9/GFP. We detected high E-cadherin expression in tumor epithelial cells undergoing collective invasive ( Figure 3G; red circle), but low E-cadherin and Vimentin expression in EOC orthotopic xenograft tumor models with AAV9/GFP/LRRC4. Together, these results showed that LRRC4 significantly reduces Ecadherin and Vimentin expression and collective invasion of EOC ( Figure 3G) and that the expression of pan-cadherin and N-cadherin in the invasive cells ( Figure 3G, column 3, red circle) (Figure 3B) remained unaffected. Thus, LRRC4 inhibits collective cell invasion by targeting E-cadherin.

LRRC4 Inhibition of EOC Invasion Is Mediated by the PIK3R1/AKT/GSK-3β/β-Catenin Signaling Pathway
We next investigated the potential mechanism by which LRRC4 regulates E-cadherin-dependent collective invasion. The PI3K/AKT pathway is one of the most frequently altered pathways in EOC (25), leading us to assess the link between LRRC and the PI3K pathway. We first conducted genomic analyses of 489 human ovarian serous cystadenocarcinomas using the Cancer Genomics Atlas (TCGA), which revealed an inverse relationship between LRRC4 and PIK3R1 expression ( Figure 4A). The inverse relationship between LRRC4 and PIK3R1 led us to ask whether LRRC4 mediates other components of the PI3K/AKT signaling pathway. Indeed, the expression of LRRC4 also inversely correlates with that of AKT1, GSK3β, and CTNNB1 (β-catenin) (Figure 4A).
In the adhesion complexes, β-catenin tightly binds to cadherin and is critical for cellular structures and actin cytoskeleton organization (27). During carcinogenesis and EMT, β-catenin translocates to the nuclei, enabling β-catenin to function as a transcriptional co-regulator and cooperate with transcription factors of the T-cell factor (TCF) family to repress E-cadherin expression and support EMT progress (28). We next focused on β-catenin and found that LRRC4 overexpression reduced β-catenin expression ( Figure 4C) and prevented its nuclear translocation ( Figure 4D). Experiments with the Top/Fop assay also demonstrated that LRRC4 overexpression markedly inhibited β-catenin activities ( Figure 4E). However, when the β-catenin mutant (S33Y), which is refractory to phosphorylation and is therefore unable to be degraded by the proteasome, was expressed in SKOV3 cells, the down-regulation of E-cadherin caused by LRRC4 overexpression was rescued (Figure 4F), suggesting that LRRC4 controls E-cadherin expression through β-catenin. Consistent with the observations from in vitro experiments, the levels of pAKT, pGSK3β and β-catenin were lower in the orthotopic primary implantation sites of mice in the AAV9/GFP/LRC4 group than those in the AAV9/GFP group ( Figure 4G).

LRRC4 Binds to PI3KR1 and Inhibits PI3KR1 Phosphorylation
How does LRRC4 regulate PI3K? We analyzed potential LRRC4 protein interactions using Scansite3 software, which predicted that PIK3R1 can putatively interact with LRRC4 and does so via the PIK3R1-SH2 domain (http://scansite3. mit.edu/). Experiments with immunofluorescence analysis showed that LRRC4 and PIK3R1 proteins colocalized in the plasma membranes and the cytoplasm in HEK293 cells, while both were only present in the cytoplasm of SKOV3 EOC cells (Figure 5A). We also found that LRRC4 co-immuoprecipitated with PIK3R1 in HEK293 cells ( Figure 5B). Furthermore, we preformed the GST pull-down assay, which revealed direct binding between LRRC4 and PIK3R1 through the c-SH2 domain of PIK3R1 ( Figure 5C). PIK3R1 serves as the regulatory subunit of PI3K and complexes with PIK3CA (p110α), leading to the inhibition of PIK3CA activity and subsequently PI3K/AKT signaling, while the inhibition is prevented when RKTs recruit PI3K to the plasma membranes (16). Although LRRC4 functioned to suppress PIK3R1 and PIK3CA expression in SKOV3 cells, as shown by Western blotting (Figure 4B) and immunofluorescence analyses (Figure 5D), LRRC4 overexpression did not disrupt the interaction between PIK3R1 and PIK3CA ( Figure 5E). It was previously shown that that PIK3R1 phosphorylation leads to increased membrane binding of PI3K and promotes PI3K activity when PIK3R1 complexes with the p110 subunit (29), leading us to assess the effect of LRRC4 onPIK3R1 phosphorylation. We detected phosphorylation of both LRRC4 and PIK3R1 using the Phos-tag SDS-PAGE analysis, which detects phosphorylated proteins based on differences in protein migration speed. PIK3R1 overexpression did not affect the level of LRRC4 phosphorylation, while LRRC4 overexpression reduced the level of PIK3R1 phosphorylation. The reduction in PIK3R1 phosphorylation was alleviated when PIK3R1 was ectopically expressed (Figure 5F, shown as red arrows).
The E-cadherin-Catenin adhesion Complex can be found in the plasma membranes and in the cytoplasm. To ask which pool of E-Cadherin and β-catenin was reduced after LRRC4 overexpression, we separated plasma membranes and cytoplasm and determined the levels of the two proteins. We found that LRRC4 overexpression caused the levels of membrane-associated E-cadherin and β-catenin to significantly reduce while exerting little on the cytoplasmic pool of the two proteins ( Figure 6D). As expected, the reduction in Ecadherin and β-catenin levels in the plasma membranes caused by LRRC4 overexpression was rescued after knockdown of PIK3R1 ( Figure 6E). Together, these results suggested that  LRRC4 inhibits collective invasion and metastasis of EOC by targeting membrane-bound E-cadherin via the PIK3R1/AKT signaling pathway.

DISCUSSION
The high mortality of EOC is mainly due to detection at late stages with vast intraperitoneal dissemination and to chemotherapy resistance (30). Ovarian cancer often metastasizes to the omentum, intestine, and liver through the peritoneal cavity with peritoneal fluid dissemination (31), in which cancer cells directly extend from primary tumors and metastasize to peritoneal tissues as a single cell or clusters (4). Nevertheless, haematogenous metastasis also exists with circulating tumor cells (CTCs) found in ovarian cancer patients (6,32). Epithelial ovarian cancer metastasis occurs prior to a predominantly intraperitoneal dissemination mechanism (4,33). Here we present evidence to show that decreased LRRC4 expression modulates EOC collective invasion and metastasis, especially to omentum, and ascites. Thus, LRRC4 functions as a tumor suppressor gene that inhibits EOC tumourigenesis and omentum metastasis.
Cancer cell metastasis is a complex process and involves various molecular mechanisms (34). The mechanism for cancer cell metastasis can be grouped into two categories including single cell migration and collective cell migration, in which cells move individually or as clusters, respectively (35). Previous studies have indicated that collective cell metastasis occurs through cell connections via cell-cell junctions, mostly cadherinmediated cell-cell interactions, to enable coordinated cell movement by following the leader cells (36). Several studies have suggested that E-cadherin or P-cadherin plays important roles in junctions during collective metastasis (37). In addition, in-depth single-cell phenotypic characterization of high-grade serious ovarian cancer has shown that cancer cells co-express E-cadherin and vimentin, which correlates significantly with tumor recurrence and metastasis (38). Although EMT and MET are methods of tumor metastasis in ovarian cancer patients (39), our current study shows that LRRC4 acts as a suppressor gene to inhibit EOC cell collective invasion in an E-cadherin-dependent manner but not through EMT. This is mainly because LRRC4 ectopic expression not only simultaneously inhibits the expression of E-cadherin and Vimentin in vitro and in vivo but also attenuates EOC collective invasion and omentum metastasis. In primary EOC tissues, cell collective invasion exists, while high-level Ecadherin expression and low-level LRRC4 expression are present in the region of collective invasion. In addition, high-level expression of E-cadherin is also accompanied by the loss of LRRC4 expression in cancer cell clusters in ascites of EOC patient.
PI3K transduces phosphatidylinositol-4,5-bisphosphate (PI-4,5-P2) to generate phosphatidylinositol-3,4,5-trisphosphate (PIP3) and then actives AKT kinase (40). The PI3K/AKT pathway is frequently altered in human cancer and is critical in tumor initiation and progression (41). As the regulatory subunit of PI3K, PIK3R1 maintains PI3K in a low activity state via interacting with the PIK3CA catalytic subunit (42). GSK3β allows crosstalk between PI3K/AKT and Wnt pathways and antagonizes Wnt signaling by forming protein complexes with Axin1 and CK1 to polyubiquitinate β-catenin, the central component of canonical Wnt signaling, to inhibit cancer cell proliferation and invasion (43,44). Our study shows that LRRC4 binds to the cSH2 domain of PIK3R1 and inhibits the expression of PIK3R1 and PIK3CA and that LRRC4 does not affect the interaction between PIK3R1 and PIK3CA. LRRC4 does not only repress the phosphorylation of PIK3R1 but also inhibits pAKT, pGSK3β, and β-catenin expression and subsequently collective cell invasion. Moreover, during EMT, β-catenin translocates from the cytosol to the nuclei, allowing it to function as a transcriptional co-regulator and cooperate with transcription factors of the TCF family to repress E-cadherin expression and support EMT (28). Although LRRC4 represses β-catenin expression and nuclear translocation, LRRC4 mainly attenuates E-cadherin and β-catenin levels in the plasma membranes. Because β-catenin binds tightly to the cytoplasmic domain of type I cadherins, such as E-cadherin, and plays an essential role in the actin cytoskeletal organization (27), the LRRC4/PIK3R1 axis may inhibit the intrinsic invasive ability of ovarian cancer cells to and reduce tumor metastasis behavior by disrupting cell-cell junctions via the AKT/GSK-3β/β-catenin/E-cadherin signaling mechanism.

CONCLUSIONS
In summary, we find that LRRC4 is significantly downregulated and is closely associated with metastasis in EOC patients. LRRC4 functions as a tumor suppressor gene to inhibit EOC collective invasion and metastasis in vitro and in vivo and does so by directly binding to the cSH2 domain of PIK3R1 to exert its regulatory function. Our findings provide a potential novel approach for metastasis prognosis and a new strategy for the treatment of EOC.

DATA AVAILABILITY STATEMENT
Publicly available datasets were analyzed in this study. This data can be found here: http://scansite3.mit.edu/.

ETHICS STATEMENT
This study was approved by the Joint Ethics Committee of the Central South University Health Authority, and was performed according to the ethical standards of the Declaration of Helsinki. All animal studies were approved and performed according to the guidelines of the Institutional Animal Care and Use Committee (IACUC) of Central South University.

AUTHOR CONTRIBUTIONS
MW, SX, and LL designed the manuscript. XS, YZ, CL, PL, SC, BS, YL and YS collected the samples and performed the experiments. JF and JL analyzed and interpreted the data. CZ and MW were the major contributors in writing the manuscript. All authors read and approved the final manuscript.