Novel KDM2B/SAV1 Signaling Pathway Promotes the Progression of Gastric Cancer

Salvador homologue 1 (SAV1), which is reported to act as a tumor suppressor in different types of cancer, is one of the key components of the Hippo pathway. However, the expression and mechanisms of SAV1 in the development and progression of gastric cancer (GC) remain to be elucidated. Immunohistochemistry (IHC) was performed in the present study to assess the expression levels of SAV1 and lysine-specific demethylase 2B (KDM2B) in GC tissues. The biological effects of SAV1 on GC cell proliferation, migration, and invasion were studied in vitro. KDM2B transcriptionally regulates SAV1 expression in several GC cell lines, and molecular experiments were performed to investigate underlying mechanisms. The expression level of SAV1 was significantly decreased in GC tissues and cell lines, negatively associated with tumor invasion depth, lymph node metastasis, and TNM stage, and positively associated with the overall survival of patients with GC. SAV1 overexpression inhibited the proliferation, migration, and invasion of GC cells. Further mechanistic studies revealed that KDM2B transcriptionally regulated SAV1 expression and further regulated the Hippo signaling pathway. To conclude, the present study demonstrated that KDM2B transcriptionally regulated SAV1 expression and promoted GC progression.


Introduction
Gastric cancer (GC) remains a prevalent disease worldwide. GC resulted in ∼783,000 deaths in 2018, making it the third leading cause of cancer-related deaths worldwide. Te incidence rates of GC are signifcantly increased in Eastern Asia, especially in Japan, South Korea, and China [1]. In China, GC ranks third in annual morbidity causes and mortality causes [2]. Most patients are already at the late stages at the time of diagnosis, and the 5-year survival rate of advanced GC is ∼5.2%; hence, early detection of GC is crucial [3]. Although medical technology has improved, the overall diagnostic rate remains low. Terefore, it is important to study the mechanisms that promote the development and progression of GC and identify novel targets to improve therapeutic efects and prognosis.
Salvador adaptor protein (SAV), which contains two protein-protein interaction modules known as WW domains, is considered to function as a scafolding protein for the mammalian Hippo pathway [4]. Salvador homologue 1 (SAV1), also known as WW45, is the human homolog of Salvador that couples mammalian Ste20-like 1 and 2 kinases (MST1/2) to large tumor suppressor kinases 1 and 2 (LATS1/ 2) to form the Hippo signaling pathway [5]. As an adaptor protein, SAV1 acts as a coactivator of MST1/2 kinases and can directly bind to MST1/2 and induce the kinase cascade via promoting the phosphorylation of MST 1/2, LATS 1/2, yes-associated protein (YAP), and/or transcriptional coactivation with PDZ-binding motif (TAZ). Te phosphorylation of YAP and TAZ can lead to their cytoplasmic translocation, ubiquitination, and degradation. Te cytoplasmic translocation of YAP/TAZ inhibits the transcription of their downstream target oncogenes and leads to tumor suppression [6,7]. SAV1 acts as a tumor suppressor in diferent types of cancer, including pancreatic [8], colon [9], and lung cancer [10]. SAV1 downregulation induces tumorigenesis and metastasis and is closely associated with a poor prognosis of lung and pancreatic cancer [8,10]. However, the roles and mechanisms of SAV1 expression and function in GC remain to be clarifed.
Histone demethylase lysine-specifc demethylase 2B (KDM2B) plays a role in numerous cellular processes, including cell diferentiation, senescence, and the selfrenewal of stem cells. It was recently revealed that KDM2B expression was increased in diferent cancer types and acts as an epigenetic regulator in cancer development and progression. KDM2B (also known as JHDM1B, Ndy1, and FBXL10) regulates the demethylation of H3K36me2 and the expression of a series of genes at the transcriptional level. In pancreatic ductal adenocarcinoma (PDAC), KDM2B interacts with KRAS-G12D to promote tumorigenesis in mouse models [11]. Our previous studies demonstrated that KDM2B expression was increased in PDAC and promoted PDAC progression via the Hippo pathway by transcriptionally regulating MOB kinase activator 1A expression [12]. Another study showed that KDM2B regulated cell adhesion and migration of prostate cancer cells [13]. However, the roles and mechanisms of KDM2B in promoting GC progression remain to be further studied.
Te present study investigated the roles, expression, and regulatory mechanisms of SAV1 in GC. Te present study revealed that increased SAV1 expression decreased the growth and metastasis of GC. Mechanistic studies showed that KDM2B could directly bind to the promoter region of the SAV1 gene, which resulted in the methylation of H3K27 and the decreased transcription and expression of SAV1, further resulting in GC progression via the Hippo signaling pathway.

Human Tissue Microarrays (TMAs) and IHC.
Te protein expression of SAV1 and KDM2B was tested with human tissue microarrays (TMAs) which were bought from Shanghai Outdo Biotech Company (China). Totally, the TMA contains 100 primary GC tissues and 80 adjacent normal gastric tissues. Clinical and demographic information, including gender, age, TNM stages, diferentiation, and overall survival from the time of diagnosis, was available. Immunohistochemical analysis was conducted with anti-SAV1 (HPA0018085, Sigma-Aldrich, diluted 1 : 100) and anti-KDM2B (SAB2702002, Sigma-Aldrich, diluted 1 : 300). Ten, the immunostaining signals of SAV1 and KDM2B were evaluated by at least two pathologists who were blinded to the clinical information. Te percentage of SAV1-or KDM2B-positive cells was divided into four groups: 1 was <25%, 2 was from 25% to 50%, 3 was from 50% to 75%, and 4 was >75%. Te staining intensity of SAV1-or KDM2B-positive cells was scored into four categories: 0 was absent, 1 was weak infltration, 2 was moderate infltration, and 3 was strong infltration. Te fnal score was the result of multiplying the intensity and the percentage. In statistical analyses, the expression of SAV1 and KDM2B was further divided into low SAV1 or KDM2B expression which was from 0 to 4 or high SAV1 or KDM2B expression which was from 6 to 12. . Te human GC cell lines, namely, HTB103,  SNU-1, AGS, and NCI-N87, were obtained from the  American Type Culture Collection. Te TMK-1 cell line was  obtained from Masashi Kanai (Kyoto University, Kyoto,  Japan), and the SK-GT5 cell line was obtained from Gary K. Schwartz (Memorial Sloan-Kettering Center, New York, NY).

Dual-Luciferase Reporter
Assay. GC cells were cotransfected with indicated vectors or a control vector, pGL3-SAV1, and the β-actin/Renilla luciferase reporter. After transfecting for 24 hours in each group, we used the dual luciferase assay system (Promega) to test the luciferase activity in each group cells.

Chromatin Immunoprecipitation Assay.
Te chromatin immunoprecipitation (ChIP) assay kit (Millipore Technology, Billerica, MA) was used to perform the ChIP assay. According to the manufacturer's protocol, 2 × 10 6 tumor cells were prepared, and the anti-KDM2B antibody was purchased from Millipore (#09-864). Te resulting DNA samples were tested using quantitative real-time PCR. Te primers for qPCR were as follows: 5′-ATC TGC GTC GAG CTT CCC AGA ATT-3′ (forward) and 5′-ATT CCT TCT TCA CGT ACT TCC CCT-3′ (reverse). In qPCR, an 80 bp region of the SAV1 promoter was amplifed and analyzed.
2.6. Gene Transfection. In transient transfection, GC cells were plated into six-well or twenty-four-well plates. Lipofectamine LTX (Invitrogen) and Lipofectamine 2000 CD (Invitrogen) were used to transfect the plasmids or shRNAs. After 48 hours, the functional assays were performed. For retroviral transduction, the GC cells were plated into six-well plates for 24 hours and the confuency was about 50%. A mixture of retroviruses and hexadimethrine bromide (Polybrene; 5 μg/mL) were used to infect the GC cells, and 2 Genetics Research puromycin with a concentration of 2 μg/mL was used to select stable populations.

Colony Formation and Spheroid Colony Formation Assay.
24-well plates were seeded with two hundred cells from each group as indicated. Te cells were allowed to grow for two weeks in the medium which was changed twice a week. After two weeks, 4% paraformaldehyde was used to fx the cells and 0.1% crystal violet solution was used to stain the cells for 10 minutes. A microscope at 40x magnifcation was used to count the colonies (>20 cells). Te percentage of the control was the result of the test. Te GC cells were infected with pBABE-SAV1 or pBABEpuro and were maintained in the DMEM/F12 medium supplemented with bFGF (20 ng/ml) and EGF (20 ng/ ml) and B27 supplements (Invitrogen), and the spheroids were counted. All experiments were performed in triplicate and repeated twice.

Cell Migration and Invasion.
Cell migration and invasion assays were conducted using modifed 24-well Boyden chambers which were purchased from BD Biosciences and were used to test migration and invasion abilities of GC cells. Briefy, cells from diferent groups were treated as indicated for 24 hours. Ten, the cells were suspended in DMEM at a concentration of 8 × 10 4 /ml. Te cells were prepared in 500 μl of DMEM and were loaded in the upper wells. Te medium, containing 20% FBS, was used as a chemoattractant stimulus in the lower wells. Te cells which migrated on the bottom surface of the flter were fxed and stained with H&E. Te migrated cells, in three randomly selected felds, were counted under a microscope at a magnifcation of 200x.

Statistical Analysis.
Te correlation of the expression of SAV1 and KDM2B in the TMA was analyzed using the Spearman rank correlation coefcients. Te signifcance of diferences among the covariates in TMA was determined with a two-tailedX 2 test or Fisher exact test. Te Kaplan-Meier method was used to estimate OS, and the logrank test was used to compare the diferences. Multivariate analysis was used to analyze the signifcant variables for independent prognosis. All the in vitro experiments were performed at least twice, and one representative result of the two or three experiments with similar results was presented. Te signifcance of the results of the in vitro experiments was analyzed with Student's t-test (two-tailed) or one-way analysis of variance. A P value less than 0.05 was considered to be statistically signifcant. Te statistical analysis was performed via the SPSS software program (version 13.0; IBM Corporation).

SAV1 Expression Is Directly Associated with Pathological Features of GC.
To determine the roles of SAV1 in GC pathogenesis, the present study frst analyzed SAV1 expression in GC tissue arrays by IHC. Te clinicopathological characteristics of the TMA are shown in Table S1. SAV1positive staining was mainly observed in the cytoplasm of adjacent normal gastric tissue and several GC cancer tissues. SAV1 expression in cancer tissues was much lower than that in tumor-adjacent normal gastric tissues (Figures 1(a) and 1(b)). Decreased SAV1 expression was positively associated with tumor invasion depth (T stage; Table S1), lymph node metastasis (N stage; Table S1), and TNM stages (Table S1). Te prognostic value of SAV1 and classical clinicopathological characteristics on patient survival was determined by the Kaplan-Meier analysis and log-rank test. Univariate analysis showed that SAV1 expression was associated with the OS of patients with GC (P < 0.001; Figure 1(j) and Table S3). Univariate analysis also indicated that age (P < 0.001; Figure 1(e) and Table S3), T stages (P � 0.011; Figure 1(f ) and Table S3), N stages (P � 0.032; Figure 1(g) and Table S3), TNM stages (P < 0.001; Figure 1(h) and Table S3), and tumor diferentiation (P � 0.002; Figure 1(i) and Table S3) were correlated with patient OS. Furthermore, multivariate analysis showed that age (P � 0.021; Table S3), tumor diferentiation (P � 0.024; Table S3), and TNM stages (P � 0.008; Table S3) were independent prognostic factors for patients with GC.
Te present study further assessed SAV1 expression in GC cell lines via western blotting. SAV1 levels were signifcantly lower in most cancer cell lines (Figure 1(c)). Te present study then analyzed the mRNA levels of SAV1 from 12 paired GC and adjacent normal gastric tissues using qPCR. Te results showed that the mRNA levels of SAV1 were signifcantly decreased in GC tissues compared with adjacent normal tissues (P < 0.05; Figure 1 that SAV1 downregulation may be involved in GC pathogenesis.

SAV1 Inhibits GC Cell Proliferation, Migration, and Invasion In Vitro.
To investigate the biological roles of SAV1 in GC, AGS and NCI-N87 cells (which express low levels of endogenous SAV1) were transfected or infected with SAV1 expression vectors (AGS/NCI-N87-pSAV1 and AGS/NCI-N87-pBABE-SAV1). Empty expression vectors were used as the control (AGS/NCI-N87-Control and AGS/NCI-N87-pBABEpuro). Infected cells were selected using puromycin, and it was found that pooled drug-resistant cells had signifcantly elevated SAV1 expression (Figure 2(a)). HTB103 cells (which express higher levels of endogenous SAV1) were transfected with shSAV1-1, shSAV1-2, and control vectors. Te protein levels of SAV1 in these cells were measured using western blotting (Figure 2(a)). Western blotting revealed that SAV1 levels were signifcantly overexpressed in AGS and NCI-N87 cells. shSAV1-2 resulted in signifcantly lower expression of SAV1 compared with shSAV1-1. Hence, shSAV1-2 was chosen as shSAV1 for subsequent experiments.
To investigate the roles of restored SAV1 expression in GC cell proliferation, AGS and NCI-N87 cells were used for colony formation assays. As shown in Figures 2(b) and 2(c), restoration of SAV1 expression signifcantly suppressed colony formation in AGS cells, but SAV1 knockdown promoted colony formation in HTB103 cells. Furthermore, the role of SAV1 in GC cell spheroid formation was assessed. Restored SAV1 expression signifcantly reduced spheroid numbers and sizes in the frst and second generations of AGS and NCI-N87 cells (Figures 2(d) and 2(e)). Tese results revealed the suppressive roles of SAV1 in GC cell proliferation.
To further assess the efects of restored SAV1 expression on the migration and invasion of GC cells, AGS and HTB103 cells were transfected with pSAV1 or shSAV1, respectively. Similarly, restored SAV1 expression suppressed the migration and invasion of AGS cells, but SAV1 knockdown promoted the migration and invasion of HTB103 cells (Figures 3(a) and 3(b)). Collectively, these data demonstrated that SAV1 served as a tumor suppressor by inhibiting the proliferation, migration, and invasion of GC cells.

KDM2B Transcriptionally Inhibits SAV1 Expression.
SAV1 expression is modulated by hypermethylation in PDAC [8]. However, the mechanism of decreased SAV1 expression in GC has not been demonstrated. Tzatsos et al. reported that KDM2B binds to TSS, resulting in decreased H3K27me2 and suppressing the expression of a series of genes involved in development [11]. Te present study then assessed whether KDM2B regulated SAV1 expression. KDM2B was knocked down by shKDM2B-1 and shKDM2B-2 in AGS and NCI-N87 cells. Both shKDM2B-1 and shKDM2B-2 could efectively knock down KDM2B, with shKDM2B-2 showing more efective knockdown (Figure 4(a)). shKDM2B-2 was then used as shKDM2B in subsequent experiments. Western blotting revealed that KDM2B knockdown led to decreased H3K27me3 levels (Figure 4(a)). In addition, KDM2B knockdown increased both the mRNA and protein levels of SAV1 (Figures 4(a) and  4(b)). ChIP revealed that KDM2B directly bound to the TSS region of SAV1 in AGS cells (Figures 4(c)). Te SAV1 promoter reporter (pLuc-SAV1) was then generated, which contained the surrounding bases of TSS. Te luciferase assay results showed that KDM2B knockdown signifcantly elevated the transcriptional activity of the SAV1 promoter reporter (Figure 4(d)). Te aforementioned data demonstrated that SAV1 was a direct downstream target of KDM2B and that KDM2B transcriptionally regulated SAV1 expression.
SAV1 is the core component of the Hippo signaling pathway and suppresses the oncogenic transcriptional module, YAP, which functions as transcriptional coactivators and promotes GC progression [6,7]. KDM2B knockdown led to increased levels of SAV1 protein and YAP phosphorylation and decreased protein levels of YAP and its typical downstream target CTGF (Figure 4(e)). Tese results demonstrated that KDM2B regulated the Hippo pathway via transcriptionally regulating SAV1.

KDM2B Expression Is Correlated with Pathological Features of GC and Negatively Associated with SAV1.
Te present study provided evidence that KDM2B transcriptionally suppressed SAV1 expression. To further confrm the results, the protein levels of KDM2B in the serial GC tissue array of SAV1 were analyzed using IHC. KDM2B was mainly positively stained in the nuclei of cancer tissues and was more highly expressed in cancer tissues than in tumoradjacent normal gastric tissues (Figures 5(a) and 5(b)). Furthermore, KDM2B expression was positively associated with T stages (P � 0.001; Table S2), lymph node metastasis (P < 0.001; Table S2), and higher TNM stages (P < 0.001; Table S2). Te prognostic value of KDM2B expression on GC patient survival was tested using the Kaplan-Meier analysis and log-rank tests. Univariate analysis showed that KDM2B was negatively associated with the OS of patients with GC (P < 0.001; Figure 5(c) and Table S3). However, multivariate analysis showed that KDM2B was not an independent prognostic factor for GC (P � 0.446; Table S3). Te present study then further analyzed the correlation between SAV1 and KDM2B expression within the same cohort. As shown in Figures 5(d) and 5(e), a direct negative correlation between SAV1 and KDM2B expression was found in GC tissues (r � −0.535; P < 0.001). Tese data further confrmed that KDM2B is a negative regulator of SAV1.

Discussion
SAV1 is one of the key components of the Hippo signaling pathway and acts as a coactivator of MST1/2 kinases by directly binding to MST 1/2 and promoting the phosphorylation of MST 1/2, LATS1/2, YAP, and TAZ. Te Hippo signaling pathway was reported to play essential roles in the development and progression of cancer and is considered a potential therapeutic target [15]. Tus, the roles of SAV1 in cancer have gained increasing attention. It was reported that SAV1 repressed the growth of colorectal and pancreatic cancer [9]. In glioblastoma, repressed SAV1 expression promoted the stem cell phenotype of glioblastoma cells [16]. In another study, deletion of both PTEN and SAV1 in the liver promoted the development of liver cancer in mice [17]. In addition, SAV1 plays an important role in the chemosensitivity of ovarian cancer cells to cisplatin [18]. However, the roles of SAV1 in GC have not been identifed. Te present study provided four lines of evidence to demonstrate the tumor suppressor role of SAV in GC. First, the present results showed that SAV1 expression was decreased in GC cell lines and tissues. Second, SAV1 expression was negatively associated with tumor invasion depth, lymph node metastasis, TNM stages, and patient survival. Tird, restored expression of SAV1 suppressed the proliferation and cell spheroid formation of GC in vitro. Moreover, SAV1 knockdown promoted colony formation and cell spheroid formation of GC cells. Fourth, SAV1 overexpression inhibited GC cell migration and invasion in vitro, but SAV1 knockdown promoted the migration and invasion of GC cells. Collectively, these fndings demonstrated that SAV1 suppressed the proliferation, migration, and invasion of GC cells and also functioned as a tumor suppressor in GC. SAV1 expression was reported to be decreased in diferent types of cancer, including colon, lung, renal cell, liver, and pancreatic cancer [8,9,[19][20][21]. It was reported that microRNA (miRNA)-21, miRNA-181c, miR-149-5p, miRNA-130b, long noncoding HOTAIR, and hypermethylation of the DNA promoter region modulated the expression of SAV1 [8,9,16,18,22,23]. In another study, MST2 was found to be coexpressed with SAV1, phosphorylated SAV1 at Tr-26, Ser-27, Ser-36, and Ser-269, and promoted cell death [24]. However, the mechanism of suppressed SAV1 expression requires further study. A study on pancreatic cancer showed that KDM2B bound to TSS, decreased the levels of H3K27me3, and regulated the expression of a series of genes, and ChIP-sequencing results showed that SAV1 is one of the potential target genes of KDM2B [11].  [7].
Loss of SAV1 induced STAT3 activation and promoted tubulointerstitial fbrosis [25]. In lung cancer, SAV1 could bind to zinc fnger protein Gli1 and negatively regulate the Hedgehog signaling pathway [19]. YAP expression was increased in GC, and the Hippo signaling pathway plays a critical role in GC development and progression [6]. Hence, the present study further analyzed the efects of KDM2B on the Hippo signaling pathway in GC. Te results showed that KDM2B knockdown led to increased protein levels of SAV1 and phosphorylation of YAP and decreased protein levels of total YAP and its typical downstream target CTGF. Tese data demonstrated that KDM2B regulated the Hippo pathway via SAV1. However, whether KDM2B further regulates other signaling pathways via SAV1 may be further studied in the future.
In conclusion, the present study provided both clinical and mechanistic evidence, identifying that KDM2B regulated SAV1 expression at the transcriptional level and promoted GC progression. Te present study not only identifed that KDM2B regulated SAV1 expression but also identifed a promising molecular target for new therapeutic strategies for GC.

Data Availability
All data generated or analyzed during this study are included in this article.

Ethical Approval
Te human tissue experiments were approved by the Ethics Committee of Shanghai East Hospital.