FAM83H and SCRIB stabilize β-catenin and stimulate progression of gastric carcinoma

FAM83H primarily is known for its function in tooth development. Recently, a role for FAM83H in tumorigenesis, conjunction with MYC and β-catenin, has been suggested. Analysis of public data indicates that FAM83H expression is closely associated with SCRIB expression in human gastric cancers. Therefore, this study investigated the roles of FAM83H and SCRIB in 200 human gastric cancers and gastric cancer cells. In human gastric carcinomas, both the individual and combined expression patterns of the nuclear FAM83H and SCRIB were independent indicators of shorter survival of gastric carcinoma patients. In MKN-45 and NCI-N87 gastric cancer cells, the expression of FAM83H and SCRIB were associated with proliferation and invasiveness of cells. FAM83H-mediated in vivo tumor growth was attenuated with knock-down of SCRIB. Moreover, immunoprecipitation indicates that FAM83H, SCRIB, and β-catenin, form a complex, and knock-down of either FAM83H or SCRIB accelerated proteasomal degradation of β-catenin. In conclusion, this study has found that the individual and combined expression patterns of nuclear FAM83H and SCRIB are prognostic indicators of gastric carcinomas and further suggests that FAM83H and SCRIB are involved in the progression of gastric carcinomas by stabilizing β-catenin.


INTRODUCTION
FAM83H plays a pivotal role in dental enamel formation and in amelogenesis imperfecta [1,2].
However, the effects of knocking out FAM83H in mice are not restricted to teeth development [3]. In particular, knock-out of FAM83H retarded overall growth of mice and induced malformation of digits, AGING and abnormal teeth development was not part of the phenotype of all the knock-out mice [3]. These aberrant phenotypes in FAM83H knock-out mice suggest that FAM83H can have a pleiotropic role in cell proliferation and the molecules interacting with FAM83H might be involved in various roles of FAM83H [3]. In this aspect, the oncogene MYC is presented as a transcriptional regulator of FAM83H in hepatocellular carcinomas [4]. In kidney cancers, FAM83H stimulates cancer progression in cooperation with PANX2, and PANX2 has been suggested to be part of the down-stream signaling of FAM83H [5]. In addition, the cBioPortal database (http://www. cbioportal.org) provided evidence that FAM83H expression is closely associated with the expression of SCRIB in gastric cancers [6,7]. Moreover, FAM83H is involved in various aspects of cell biology, including tumorigenesis. Higher expression of FAM83H increases the proliferation and invasiveness of cancer cells [4,[8][9][10]. The expression of FAM83H is higher in human cancer tissue compared with normal tissue [10][11][12] and higher expression of FAM83H is associated with poor prognosis of cancers of the uterus [10,12], liver [4], kidney [5], and bone [8]. However, FAM83H is a favorable prognostic factor of glioma and head and neck cancers [12]. Therefore, tumorigenic roles of FAM83H might differ according to cancer types, or the molecule(s) interacting with FAM83H during tumorigenesis of specific cancer types. SCRIB (scribble) is a polarity protein expressed on the basolateral side of epithelial cells and is important in maintaining tight junctions [13]. The loss of SCRIB in malignant tumors suggest that it has potential to be a tumor suppressor [14]. Alteration of polarity through deletion, downregulation, overexpression, and mislocalization can induce structural and functional alteration of cells that might be related to tumorigenesis [15,16]. Mislocalization of SCRIB is involved in epithelial-to-mesenchymal transition (EMT) by inducing loss of E-cadherin [15,17,18]. Deregulated SCRIB expression induces tumorigenesis in breast [19] and liver [20].
Gastric cancer is ranked fifth in incidence [21] and third in cancer-associated death worldwide [21,22]. The survival rate of gastric cancer is primarily associated with cancer stage and histological phenotype [22]. Especially, the survival rate of gastric cancer with high tumor stage and poorly cohesive carcinoma with fibrous stroma remains low [22]. Therefore, identification of histologic and molecular profiles in the progression of gastric cancer is important for the future treatment of gastric cancers. In this respect, FAM83H and SCRIB might be molecular therapeutic targets of gastric carcinomas. Based on the characteristic of FAM83H, SCRIB, and β-catenin as components of cell junctions [15], alteration of FAM83H, SCRIB, and β-catenin might stimulate the progression of gastric carcinomas. Moreover, FAM83H is important in the stabilization of β-catenin in osteosarcoma in our previous study [8], and a public database indicates a significant correlation between the expression of FAM83H and SCRIB in gastric cancer [6,7]. The direct interaction of SCRIB with β-catenin has also been observed in synaptic vesicles [23]. Therefore, this study investtigated the roles and associations between FAM83H, SCRIB, and β-catenin in gastric carcinomas using human gastric carcinoma tissues and gastric cancer cells.

The expression of FAM83H, SCRIB, and β-catenin in gastric carcinomas
In human tissue, both the nuclear and cytoplasmic expression were identified in FAM83H, SRCIB, and βcatenin immunostaining ( Figure 1A). Immunohistochemical staining scores for the expression of FAM83H in nuclei (FAM83H-N) and cytoplasm (FAM83H-H), expression of SCRIB in nuclei (SCRIB-N) and cytoplasm (SCRIB-C), and the expression of βcatenin in nuclei (β-catenin-N) were significantly higher in gastric cancers compared with normal gastric mucosa and gastric dysplasia ( Figure 1B). Consistently, mRNA levels of FAM83H and SCRIB were higher in gastric cancers compared with normal gastric tissue in the GEPIA database (http://gepia.cancer-pku.cn. Accessed 10 January 2020) ( Figure 1C). The cut-off points for FAM83H-N, FAM83H-C, SCRIB-N, SCRIB-C, and βcatenin-N expression were five, five, six, three, and four, respectively ( Figure 1D). With these cut-off values, FAM83H-N was significantly associated with serum level of CA 19-9, tumor stage, tumor invasion, lymph node metastasis, venous invasion, histologic grade, and the expression of FAM83H-C, SCRIB-N, SCRIB-C, and β-catenin-N (Table 1). FAM83H-C was significantly associated with Lauren classification, and the expression of SCRIB-N, SCRIB-C, and β-catenin-N (Table 1). SCRIB-N was significantly associated with tumor stage, tumor invasion, lymph node metastasis, WHO classification of gastric cancer, and the expression of SCRIB-C and β-catenin-N (Table 1). SCRIB-C was significantly associated with age, WHO classification of gastric cancer, histologic grade, Lauren classification, and the expression of β-catenin-N (Table 1). Positive β-catenin-N expression was significantly associated with age, sex, and tumor stage (Table 1).

Individual and co-expression patterns of FAM83H-N and SCRIB-N predict shorter survival of gastric carcinoma patients with multivariate analysis
Next, we performed multivariate analysis with the factors significantly associated with OS and RFS. Preoperative serum levels of CEA and CA19-9 were not included in multivariate analysis since the data were missing for 30 patients. In multivariate analysis, tumor stage, FAM83H-N positivity, and SCRIB-N positivity were independent indicators of poor prognosis for both OS and RFS of gastric carcinoma patients (Table 3, Multivariate analysis model 1). FAM83H-N positivity predicted a 3.551-fold greater risk of death and a 3.155fold greater risk of relapse or death of patients. SCRIB-N positivity predicted a 2.128-fold greater risk of death and a 2.062-fold greater risk of relapse or death of patients. The co-expression pattern of FAM83H-N and SCRIB-N is also an independent indicator of poor prognosis of survival of gastric carcinoma patients ( Table 3, multivariate analysis model 2). The FAM83H-N + /SCRIB-N + poor prognostic subgroup had a 6.266fold greater risk of death and a 5.442-fold greater risk of relapse or death of patients compared with the favorable prognostic subgroups (FAM83H-N -/SCRIB-Nor FAM83H-N -/SCRIB-N + ) ( Table 3).

FAM83H and SCRIB stimulate the proliferation and invasiveness of gastric cancer cells
Based on the prognostic significance of the expression of FAM83H and SCRIB in gastric carcinoma patients, we performed proliferation and migration/invasion assays after inducing knock-down or overexpression of FAM83H and SCRIB in gastric cancer cells. In MKN-45 and NCI-N87 gastric cancer cells, the proliferation of cells was inhibited with knock-down of FAM83H and increased with overexpression of FAM83H ( Figure  4A). In addition, FAM83H stimulated the migration and invasion activity of MKN-45 and NCI-N87 cells ( Figure 4B  Abbreviations: FAM83H-N, nuclear expression of FAM83H; FAM83H-C, cytoplasmic expression of FAM83H; SCRIB-N, nuclear expression of SCRIB; SCRIB-C, cytoplasmic expression of SCRIB; β-catenin-N, nuclear expression of β-catenin; CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9; LN, lymph node; WD, well-differentiated; MD, moderately differentiated; PD, poorly differentiated; PCC, poorly cohesive carcinoma; EGC, early gastric cancer; AGC, advanced gastric cancer; *; Preoperative serum level of CEA or CA19-9 were not measured in 30 patients. **; Histologic grading was carried in tubular and papillary type carcinomas according to the grading system of the WHO histological classification of gastric tumours. and increased the expression of E-cadherin in both MKN-45 and NCI-N87 cells ( Figure 5A, 5B). Overexpression of FAM83H increased protein and mRNA expression of SCRIB, cyclin D1, N-cadherin, TGF-β1, snail, vimentin, MMP2, and MMP9, and decreased the expression of Ecadherin in both MKN-45 and NCI-N87 cells ( Figure 5A, 5B). Also, the knock-down of FAM83H decreased the protein levels of β-catenin and active β-catenin and overexpression of FAM83H increased the protein levels of β-catenin and active β-catenin. However, knock-down or overexpression of FAM83H did not affect mRNA expression of β-catenin ( Figure 5A, 5B). The protein and mRNA expression of GSK3β were not altered with knock-down or overexpression of FAM83H ( Figure 5A, 5B). However, knock-down of FAM83H increased the protein level of phosphorylated GSK3β, and overexpression of FAM83H decreased the protein level of phosphorylated GSK3β ( Figure 5A). Furthermore, TOPflash reporter activity, but not FOPflash activity, was significantly decreased with knock-down of FAM83H and was significantly increased with overexpression of FAM83H ( Figure 5C).   Figure 7A, 7B). However, SCRIB did not affect the expression of FAM83H protein and mRNA ( Figure 7A, 7B). Furthermore, the loss of SCRIB was associated with the decrease of the protein expression of β-catenin and active β-catenin. However, the mRNA expression of β-catenin was not decreased with the loss of SCRIB ( Figure 7A, 7B). The protein and mRNA expression of GSK3β were not altered with knock-down or overexpression of SCRIB ( Figure 7A, 7B). However, the phosphorylated GSK3β protein was increased with knock-down of SCRIB and decreased with overexpression of SCRIB ( Figure 7A). Furthermore, TOPflash reporter activity, but not FOPflash activity, was significantly decreased with loss of SCRIB and was significantly increased with overexpression of SCRIB ( Figure 7C).

SCRIB mediates FAM83H-associated proliferation of gastric cancer cells
FAM83H and SCRIB stimulate the proliferation and invasiveness of gastric cancer cells, and the expression of SCRIB was dependent on FAM83H expression, but FAM83H was not affected by SCRIB expression. Therefore, we evaluated the effects of knock-down of SCRIB on gastric cancer cells on inducing the overexpression of FAM83H. Consistent with the results in Figure 3, overexpression of FAM83H increased proliferation, which was attenuated with knock-down of SCRIB in MKN-45 and NCI-N87 cells ( Figure 8A, 8B). Consistent with these findings, overexpression of FAM83H stimulated in vivo growth of NCI-N87 cells and loss of SCRIB suppressed in vivo tumor growth AGING compared with controls transfected with empty vectors ( Figure 8C, 8D). The in vivo tumor growth of the cells that were induced to overexpress FAM83H and have a knock-down of SCRIB was intermediate to those of the control and the FAM83H-overexpressing groups ( Figure 8C, 8D). Moreover, pulmonary metastasis of gastric cancer cells was seen in four of five mice implanted NCI-N87 cells overexpressing FAM83H ( Figure 8E). There was no pulmonary metastasis in mice implanted with NCI-N87 cells transfected with empty vectors, hSCRIB CRISPR/Cas9 KO vectors, or both the FAM83H overexpression vector and the hSCRIB CRISPR/Cas9 KO vectors ( Figure 8E).

FAM83H and SCRIB stabilize β-catenin by protecting it from proteasomal degradation
Activation of the FAM83H-SCRIB pathway stimulates the proliferation and invasion of gastric cancer cells, and both FAM83H and SCRIB are involved in the βcatenin-related pathway and the EMT pathway. Interestingly, the expression of FAM83H and SCRIB influenced the protein levels of β-catenin and active βcatenin, but not the mRNA level of β-catenin. Cotransfection to induce the overexpression of FAM83H and knock-down of SCRIB affected the protein expression of β-catenin and active β-catenin ( Figure 9A). and FAM83H-N + /SCRIB-N + subgroups. 5y-OS; overall survival rate at five years, 10y-OS; overall survival rate at ten years, 5y-RFS; relapse-free survival rate at five years, 10y-RFS; relapse-free survival rate at ten years. Higher expression levels of β-catenin and active βcatenin protein via overexpression of FAM83H were attenuated with knock-down of SCRIB ( Figure 9A). Moreover, western blotting of immunoprecipitates indicated that FAM83H, SCRIB, and β-catenin form a complex ( Figure 9B). Furthermore, the nuclear fraction of β-catenin decreased with loss of FAM83H or SCRIB and increased with overexpression of FAM83H or SCRIB in cellular fractionation analysis of MKN-45 cells ( Figure 9C). Therefore, to further explore the roles of FAM83H and SCRIB in the stabilization of β-catenin, we evaluated the proteasomal degradation of β-catenin. The β-catenin protein was rapidly degraded with the treatment of 30 μM cycloheximide in MKN-45 cells that had a knockdown of FAM83H or SCRIB ( Figure 9D). Moreover, the ubiquitination of β-catenin was increased with the loss of FAM83H or SCRIB ( Figure 9E). In the immunoprecipitated of β-catenin after knock-down of FAM83H or SCRIB, there was more polyubiquitinated β-catenin compared with controls ( Figure 9E). Furthermore, the interaction between βcatenin and β-TrCP protein increased with the loss of FAM83H and weakened with overexpression of FAM83H ( Figure 9F). The interaction between βcatenin and USP47 protein was weakened with the loss of FAM83H and increased with overexpression of FAM83H ( Figure 9F). Therefore, these findings suggest that both FAM83H and SCRIB stabilize βcatenin protein by inhibiting its proteasomal degradation via interaction with β-TrCP and USP47.

DISCUSSION
In the analysis of expression of FAM83H and SCRIB in human gastric tissue samples, their expression levels were significantly higher in cancer tissue compared with both non-neoplastic and precancerous lesions. Furthermore, their expression levels were much higher in advanced gastric carcinomas, suggesting that they play a role in the progression of gastric carcinomas. Similarly, elevated expression of the FAM83H gene was present in cancers of breast, colon, liver, lung, pancreas, and stomach [12]. Moreover, in this study, the expression of FAM83H-N and SCRIB-N predicted shorter survival of gastric carcinoma patients. Especially, the combined expression patterns of FAM83H-N and SCRIB-N were strongly predictive of the survival of gastric carcinoma patients. Consistently, FAM83H was an indicator of poor prognosis of cancers of the liver [4], kidney [5], uterus [10,12], and bone [8].
The expression of SCRIB mRNA was also higher in hepatocellular carcinomas compared with non-tumorous liver tissue [20]. Also, the higher expression of SCRIB mRNA was significantly associated with poor prognosis of hepatocellular carcinoma [20] and breast cancer patients [24]. However, controversially, FAM83H expression was low in brain tumors, and increased expression of FAM83H was associated with favorable prognosis of patients [12]. Therefore, the prognostic impact of the expression of FAM83H and SCRIB might differ according to cancer type. However, despite some controversies, our data indicate that the survival of gastric carcinoma patients is independently associated AGING AGING with the expression of FAM83H-N and SCRIB-N. Moreover, the co-positivity of FAM83H-N and SCRIB-N was also an independent indicator of poor prognosis. Similarly, higher expression of SCRIB was associated with poor prognosis of estrogen receptor-negative and vimentin-positive breast cancer subtypes [25] and clear cell renal cell carcinoma patients [5]. Altogether, our results suggest that the expression of FAM83H-N and SCRIB-N might be novel prognostic indicators of gastric cancer patients.
In the cBioPortal public database, there was a strong correlation between the expression of FAM83H mRNA and SCRIB mRNA in gastric cancers (Pearson's correlation; 0.77, Spearman's correlation; 0.81, http://www.cbioportal.org) [6,7]. In agreement with AGING data from the public database, FAM83H expression showed a significant association with SCRIB expression in human gastric carcinoma tissue samples. Our indepth univariate and multivariate analyses further found that the FAM83H-N + /SCRIB-N + subgroup of gastric carcinoma had the shortest survival. Moreover, FAM83H and SCRIB stimulated the proliferation and invasion of gastric cancer cells, and FAM83H overexpression-stimulated proliferation of gastric cancer cells was attenuated with a knock-down of AGING SCRIB in vitro and in vivo. The expression of mRNA and protein of SCRIB was significantly altered according to FAM83H expression. However, the expression of FAM83H was not influenced by SCRIB alteration. These findings suggest that FAM83H and SCRIB are closely related to the progression of gastric carcinomas, and that there are tumorigenic roles of FAM83H in mediating SCRIB; furthermore, the expression of SCRIB was dependent on FAM83H. However, despite the close correlation between FAM83H and SCRIB, the knock-down of SCRIB could not completely attenuate the effect of FAM83H overexpression in Figure 8. Therefore, there might be another role of FAM83H that is unrelated to SCRIB in cancer progression. Therefore, further study is necessary to explore the exact mechanism of how AGING FAM83H is involved in cancer progression in conjunction with SCRIB or independent of SCRIB.
Concerning the role of FAM83H in conjunction with SCRIB in tumorigenesis, the oncogene MYC might be closely associated. In our previous study, we demonstrated that FAM83H was an intermediate of the oncogene MYC on the MYC-mediated proliferation and invasiveness of liver cancer cells [4]. The expression of FAM83H was transcriptionally controlled by MYC in liver cancer cells [4]. Cytoplasmic SCRIB was also involved in MYC-mediated hepatic tumorigenesis [20]. Although the mechanism by which SCRIB is involved in MYC-mediated tumorigenesis is puzzling, SCRIB is involved in MYC-mediated mammary tumorigenesis. In a mammary tumorigenic model, loss of SCRIB AGING promoted tumorigenesis by suppressing MYC-mediated apoptosis, which was associated with the deregulation of SCRIB that results in the loss of its cytoplasmic membrane expression [19]. In addition, although there are only two cases of neuroendocrine carcinoma, nuclear expression of FAM83H in neuroendocrine carcinoma might be related to the MYC-FAM83H relationship because it has been reported that MYC drives progression of neuroendocrine carcinoma [26]. Therefore, our findings that FAM83H is involved in the expression of SCRIB suggests that the MYC-FAM83H-SCRIB pathway is important in tumorigenesis in both liver and stomach.
In this study, the effect of FAM83H and SCRIB on the proliferation of gastric cancer cells was associated with the canonical Wnt/β-catenin pathway. The reduced expression of both FAM83H and SCRIB decreased the protein level of β-catenin and the expression of cyclin D1. When considering the important role of the Wnt/βcatenin pathway on cellular proliferation, FAM83H and SCRIB had an effect on cellular proliferation through the regulation of the Wnt/β-catenin pathway by stabilizing the β-catenin protein. In our results, FAM83H and SCRIB were shown to bind to β-catenin and prevent its proteasomal degradation. Consistently, the possibility of FAM83H-mediated stabilization of βcatenin has been suggested in colorectal cancers [27] and osteosarcomas [8]. Furthermore, in this study, during the search for the roles and relationship between FAM83H and SCRIB in gastric carcinomas, we found that both FAM83H and SCRIB form a complex with βcatenin and are involved in post-translational stabilization from proteasomal degradation. As we have shown in Figure 9C, the expression of FAM83H and SCRIB are involved in nuclear localization of βcatenin. Therefore, nuclear expression patterns of FAM83H, SCRIB, and β-catenin and their interactions might be important in the progression of gastric cancers, and their nuclear expression may be of prognostic significance. The regulation of cytoplasmic and nuclear localization of β-catenin by escaping ubiquitination-mediated degradation is a classical model that explains how cancer develops in situations where the β-catenin degradation complex does not function due to inactivation or mutation of the degradation complex components, such as in the APC gene. Alternatively, the molecule might prevent degradation of β-catenin protein and might be involved in β-catenin activation-mediated proliferation of cells.
In this context, our results present a model of β-catenin stabilization where β-catenin is protected from proteasomal degradation by forming a complex with FAM83H and SCRIB. Supportively, the interaction between β-catenin and β-TrCP protein was weakened by FAM83H and SCRIB, and the interaction between β-catenin and deubiquitinase USP47 protein was increased by FAM83H and SCRIB. Therefore, FAM83H and SCRIB stabilize β-catenin protein by protecting it from proteasomal degradation via interaction with β-TrCP and USP47. However, despite direct interaction between SCRIB and β-catenin, the localization of β-catenin was not affected by SCRIB [23]. Moreover, there is a controversial report that SCRIB is a negative regulator of Wnt/β-catenin signaling in a epithelial cell polarity model [28]. Therefore, further study is needed to explore the general or specific roles of FAM83H and SCRIB in the regulation of Wnt/β-catenin pathway during tumorigenesis.
In gastric cancer cells, FAM83H and SCRIB stimulated the invasiveness of cancer cells and FAM83H induced pulmonary metastasis in mice transplanted cancer cells with FAM83H overexpression. Acquiring invasive and metastatic potential is important in the progression of cancers, and EMT is an important phenotype of advanced carcinomas [18,29,30]. The hallmark of EMT is the deregulation of adhesion molecules with the loss of E-cadherin [14,29,31,32]. FAM83H and SCRIB are also main components of cell adhesion and have roles in maintaining cytoskeletal structures [15,27], and deregulated mislocalization of FAM83H and SCRIB are suggested to be involved in tumorigenesis [8,19,20,24,33]. Therefore, cytoplasmic and nuclear localization of FAM83H and SCRIB might be important in cancer progression via a regulation of EMT that accompanies disruption of cellular adhesion. Supportively, the cytoplasmic expression of SCRIB is inversely correlated with the membranous expression of E-cadherin [20]. In addition, as the degraded membranous components of SCRIB and E-cadherin, the levels of serum-soluble SCRIB and E-cadherin were higher in endometrial cancer patients compared with healthy volunteers [34]. Our results also showed that both FAM83H and SCRIB affected the expression of molecules closely associated with the EMT such as Ecadherin, N-cadherin, TGF-β1, snail, vimentin, MMP2, and MMP9. Moreover, our results indicate nuclear expression of FAM83H and SCRIB as independent indicators of poor prognosis of gastric carcinoma patients. In contrast to membranous SCRIB, cytoplasmic SCRIB stimulated the development and progression of tumors. Induced expression of cytoplasmic SCRIB via transfecting with mutant SCRIB promoted hepatic and mammary tumorigenesis by activating PI3K-Akt signaling [20,24]. In the liver, cytoplasmic localization of SCRIB was higher in hepatocellular carcinoma compared with non-tumorous liver tissue, and cytoplasmic SCRIB induced expression of EMT genes [20]. In addition, EMT is another consequence of the activation of the Wnt/β-catenin AGING hSCRIB CRISPR/Cas9 KO plasmid and treated with 30 μM cycloheximide or 30 μM MG132 for 0.5 to 4.0 hours. Thereafter, the total protein was immunoblotted for β-catenin and actin. (E) The MKN-45 cells were transfected with empty vector, shRNA for FAM83H, or hSCRIB CRISPR/Cas9 KO plasmid, and were treated with 30 μM MG132 for two hours. The protein lysate was immunoprecipitated with anti-βcatenin antibodies and immunoblotted with anti-ubiquitin antibodies. The immunoblot was performed on total protein lysate. (F) The MKN-45 cells were transfected with empty vectors, shRNA for FAM83H, or an overexpression vector for FAM83H. The protein lysate was immunoprecipitated with anti-β-catenin antibodies and immunoblotted with anti-β-TrCP, anti-USP47, or anti-GAPDH antibodies. The immunoblots were performed on total protein lysate.
pathway [29,31,35]. When considering the role of FAM83H and SCRIB in the stabilization of β-catenin, it might be suggested that cytoplasmic FAM83H and SCRIB form a complex with β-catenin in the cytoplasm and facilitate translocation to nuclei and activate the downstream signaling of the β-catenin pathway. In this manner, FAM83H-and SCRIBmediated stabilization of β-catenin might activate cancer progression by regulating EMT. Therefore, when considering the importance of the loss of Ecadherin and activation of Wnt/β-catenin pathway in EMT, our results suggest that FAM83H-SCRIB-βcatenin-related alteration of the EMT pathway might be significantly involved in the progression of gastric cancer.
In conclusion, we demonstrate that FAM83H and SCRIB cooperatively activate the progression of gastric carcinoma by stabilizing β-catenin. Moreover, individual and co-expression patterns of the nuclear FAM83H and SCRIB might be used as novel prognostic markers of gastric cancers. Therefore, the FAM83H-SCRIB-β-catenin pathway might be a novel therapeutic target for the poor prognostic subgroups of gastric cancer, which express high levels of FAM83H and SCRIB.

Human tissue samples and gastric carcinoma patients
To investigate the clinicopathological significance of the expression of FAM83H and SCRIB in human gastric carcinoma, this study evaluated 200 gastric carcinomas from patients operated on between January 1997 and December 2005 at Jeonbuk National University Hospital. In addition, tissue samples from 20 cases of non-neoplastic gastric mucosa unrelated with gastric cancer, and 40 cases of gastric dysplasia were included in this study. At first evaluation, the cases were staged according to the 6 th edition of the American Joint Committee Cancer Staging System [36], and 50 cases were selected from stage IV patients. Thereafter, 50 cases each were selected from stage I, II, and III gastric carcinomas and matched according gender, age (±2 y), and calendar year of operation (±2 y). The 200 cases of gastric carcinomas were classified according to WHO classification [22] and re-staged according to the 8th edition of the American Joint Committee Cancer Staging System [37]. Information regarding clinicopathological factors was obtained through a review of the medical records and histologic reports. This study obtained institutional review board approval from Chonbuk National University Hospital (IRB number, CUH 2016-07-023-001) and was performed according to the Declaration of Helsinki. Based on the retrospective and anonymous character of the study the approval contained a waiver for written informed consent.

Cell proliferation assay
Cell proliferation was evaluated by counting cells, a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) cell proliferation assay, and a colonyforming assay. Cell counts were performed by seeding 3x10 3 cells in a 6-well plate and counting with a hemocytometer after 24, 48, and 72 hours. The MTT assay was performed by seeding 3x10 3 cells per well in 96 well plates for 24, 48, and 72 hours. The colonyforming assay was performed by plating 1x10 3 cells per well in 6-well culture dishes in triplicates and allowing them to grow and form colonies for ten days. The culture plates were fixed with cold methanol and stained with 0.01% crystal violet for 1 hour; then, the number of colonies was counted with GeneTools analysis software (Syngene Imaging system, Frederick, MD).

In vitro trans-chamber invasion and migration assay
To evaluate migration-and invasion-activity of cells in vitro, a 24-transwell migration or an invasion chamber (BD Biosciences, San Jose, CA) was used. The migration assay was performed by seeding 1x10 5 cells in serum-free medium in the upper chamber with 8 μm-pore filters. The invasion assay was performed by seeding 2x10 5 cells in serum-free medium in the upper 8μm-pore Matrigel Invasion Chamber (BD Biosciences, San Jose, CA). The bottom chamber consisted of 10% FBS containing RPMI-1640 for both migration and invasion assays. The migration and invasion chambers were incubated for 48 hours at 37 °C. The un-migrated and un-invaded cells on the upper surface of the chamber were removed using a cotton swab and the migrated and invaded cells on the underside of the insert were stained with DIFF-Quik staining solutions (Sysmex, Kobe, Japan). The migrated or invaded cells were counted in five microscopic fields (magnification x200) per well.

Quantitative reverse-transcription polymerase chain reaction
Total RNA was obtained with an RNeasy Mini Kit (Qiagen Sciences, Valencia, CA) and reverse transcription was performed on 1.5 μg RNA with TaqMan Reverse Transcription Reagents (Applied Biosystems, Foster City, CA). A quantitative reversetranscription polymerase chain reaction was performed with Applied Biosystems Prism 7900HT sequence Detection System and SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA). All experiments were carried out in triplicate, and the results were normalized to the expression of glyceraldehyde-3phosphate dehydrogenase reference housekeeping gene. The primer sequences for the polymerase chain reaction are listed in Table 4.

Luciferase reporter assay
For cell-based luciferase reporter assay, gastric cancer NCI-N87 and MKN-45 cells were plated in 6-well plates at ~50% confluency. The cells were cotransfected with 100 ng/well TOPflash or FOPflash plasmid DNA and 2 ng/well pRL-TK Renilla luciferase plasmid DNA (Promega, Madison, WI) in addition to empty vectors, shRNA for FAM83H, hSCRIB CRISPR/Cas9 KO plasmid, overexpression vector for FAM83H, or overexpression vector for SCRIB by using Lipofectamine ® 2000 (Thermo Fisher Scientific, Waltham, MA). Twelve hours after transfection, cells were lysed with 1x Passive Lysis buffer using Dual-Glo Assay (Promega, Madison, WI). The firefly luciferase signals were monitored from 10 μl of each protein lysate by using the Dual-Luciferase Reporter Assay System (Promega, Madison, WI), and normalized to the signals obtained from positive controls of co-transfected Renilla luciferase expression. The assay was performed in quadruplicate and repeated three times.

Tumorigenic assay
Eight-week-old male FoxnN.Cg/c nude mice (Orient Bio, Seongnam, Korea) were used to establish the xenografted gastric cancer model. The mice were randomly divided into four groups with five mice in each group. According to the experimental groups, NCI-N87 cells were transfected with empty vector, FAM83H-overexpressed vector, vectors to induce knockdown of SCRIB, or both of FAM83H-overexpression vector and vector for knockdown of SCRIB. After that, 2x10 6 NCI-N87 cells were injected subcutaneously on the back. The size of the implanted tumor was measured every week, and the tumor volume was calculated by the equation: tumor volume = length x width x height x 0.52. At five weeks after tumor cell inoculation, the mice were euthanized after anesthetizing with sodium pentobarbital and evaluated for the tumors. All animal experiments were performed with the approval of the institutional animal care and use committee of Jeonbuk National University (approval number: CBNU 2018-033).

Ubiquitin proteasomal degradation and ubiquitination analysis
MKN-45 gastric cancer cells were transfected with empty vector, shRNA for FAM83H, or hSCRIB CRISPR/Cas9 KO plasmid. After 24 hours posttransfection, the cells were treated with 30 μM cycloheximide (CHX; Sigma, St. Louis, MO) or 30 μM MG132 (Sigma, St. Louis, MO) for 30 minutes to two hours. Western blotting for β-catenin (BD Biosciences, San Jose, CA) and actin (Santa Cruz Biotechnology, Santa Cruz, CA) was performed on whole protein lysates. In addition, to evaluate the ubiquitination of βcatenin with knock-down of FAM83H or SCRIB, western blotting with anti-ubiquitin antibody after immunoprecipitation with anti-β-catenin antibody was performed. For this evaluation, MKN-45 cells were transfected with empty vector, shRNA for FAM83H, or hSCRIB CRISPR/Cas9 KO plasmid, and the cells were treated with 30 μM MG132 for two hours. The total cell lysates were immunoprecipitated with anti-β-catenin antibodies and blotted with anti-Ubiquitin antibody (Santa Cruz Biotechnology, Santa Cruz, CA).

Statistical analysis
The positivity for the immunohistochemical expression of FAM83H, SCRIB, and β-catenin were determined by receiver operating characteristic curve analysis [31,41,42]. The cut-off points were determined at the points with the largest area under the curve to predict OS. Survival analysis was performed for OS and RFS of gastric carcinoma patients. An event in OS analysis was the death of the patients from gastric carcinoma, and the patients who were alive at last contact or died from other causes were treated as censored through June 2013. The relapse or death of patients from gastric carcinoma was an event of RFS analysis. The patients who were alive without relapse or died from other causes at last contact were treated as censored. The prognostic significance was analyzed with Cox proportional hazards regression analysis and Kaplan-Meier survival analysis. The relationship between factors was analyzed with Pearson's chi-square test, Student's t-test, and one-way ANOVA analysis. All experiments were done in triplicate, and the representative data are presented. All statistical analysis was performed using SPSS software (IBM, version 20.0, Armonk, NY). P values less than 0.05 were considered statistically significant.