Antibodies
The following antibodies used in this study were purchased from BD PharMingen (San Diego, CA): mouse anti-CD4 PE (H129.19), anti-CD8a PE (53–6.7), anti-CD11b PE (M1/70), anti-F4/80 PE (BM8), and anti-Ly6C FITC (AL-21) monoclonal antibodies. In addition, PE-conjugated anti-mouse CD44, PE rat IgG1, and FITC rat IgG2a isotype control antibodies were purchased from eBioscience (San Diego, CA). The remaining antibodies used were as follows: anti-MST3/STK24 (EP1468Y; Abcam, UK), mouse anti-CDH1 (BD Transduction Laboratories, San Jose, CA), rabbit anti-beta catenin (GeneTex, Inc., San Antonio, TX), rabbit anti-AKT1 and peroxidase-conjugated goat anti-rabbit IgG (Cell Signaling, Boston, MA), mouse anti-β-actin (GeneTex, Inc.), and peroxidase-conjugated sheep anti-mouse IgG (Chemica, San Diego, CA).
Cell culture
MKN45 cells were kindly provided by Professor Ming-Derg Lai (National Cheng Kung University, Tainan, Taiwan). The MKN45 cell lines were authenticated in 2013 by DNA short tandem repeat profiling at Bioresource Collection and Research Center. These cell lines were maintained in RPMI 1640 medium containing 10% fetal bovine serum (FBS) (Gibco, Life Technologies, Grand Island, NY) and 1% penicillin/streptomycin. M12 cells were maintained in Dulbecco’s modified Eagle’s medium/high glucose supplemented with 10% FBS and 1% penicillin/streptomycin.
Mice
For use in all animal experiments, 8-week-old C578BL/6 mice were purchased from the Laboratory Animal Center of National Cheng Kung University (Tainan, Taiwan) and maintained under pathogen-free conditions.
Metastatic model of gastric cancer
The metastatic abilities of M12 cells were evaluated in vivo using a hepatic metastasis model in immunocompetent C57BL/6 mice [11]. To establish this model, mice were first anesthetized by an intraperitoneal injection of Zoletil (50 mg/kg; Parnell Laboratories, Alexandria, Australia) and xylazine (10 mg/kg; Troy Laboratories, Glendenning, Australia). A small midline incision was then made in the abdomen. The spleen was exteriorized and 5 × 105 tumor cells in 0.05 mL of PBS were injected into the spleen using a 1-cm3 U-100 disposable insulin syringe (Becton-Dickinson, Franklin Lakes, NJ). Fourteen days after this injection, the mice were sacrificed. Hepatic and splenic masses were examined macroscopically and histologically. Formalin-fixed/paraffin-embedded sections of the stomach, liver, and spleen were subjected to hematoxylin and eosin staining. Each animal experiment was performed at least twice.
Construction of CRISPR knockout STK24 plasmid and generation of stable cancer cell clones
Single guide RNAs (sgRNAs) targeting mouse Stk24 and human STK24 were purchased from the National RNAi Core Facility (Academia Sinica, Taiwan; http://rnai.genmed.sinica.edu.tw). The DNA sequences for generating mouse sgRNA were prepared as previously described [2]. The DNA sequences of human sgSTK24-RNA1 and sgSTK24- RNA2 were as follows: sgSTK24-RNA1 forward: 5′-CAC CGC GCC AAA GTC CGC CAG CTT C-3′; sgSTK24-RNA1 reverse: 5′-AAA CGA AGC TGG CGG ACT TTG GCG C-3′; sgSTK24-RNA2 forward: 5′-CAC CGT AGT TTC CTT CCA ACG TCG G-3′; and sgSTK24-RNA2 reverse: 5′-AAA CCC GAC GTT GGA AGG AAA CTA C-3′. A Cas9/gRNA vector construct was introduced into the MKN45 and M12 cells by transfection with Lipofectamine 3000 (Invitrogen, Waltham, MA) for 48 h. To create stable clones, selection was performed with puromycin (1 μg/mL) (Sigma-Aldrich, St. Louis, MO) for 2 weeks. Single-cell clones of the transfectants were selected using the limiting dilution method. To monitor the efficacy of STK24 silencing, STK24 expression in the stable transfectants was analyzed by western blotting.
Western blot analysis
Total cell lysates were prepared and analyzed by SDS-PAGE as previously described [11]. For quantification, the bands were measured using the AlphaImager 2200 system (Alpha Innotech, San Leandro, CA) and normalized using the density of β-actin. The expression of STK24 was quantified and given as the STK24-to-β-actin ratio. These experiments were repeated three times using independent batches of cell clones or cell lysates. Quantitative data are presented as values relative to those in control cells.
Cell migration and wound healing assays
Cell migration was evaluated in modified Boyden chambers (NeuroProbe, Inc., Gaithersburg, MD) for 8 h, as previously described [11]. MKN45 cells (70 μL of 1 × 106 cells per mL) or M12 cells (70 μL of 2 × 105 cells per mL) were seeded in an ibidi culture insert (Applied BioPhysics, Inc., Martinsried, Germany) on top of a 24-well plate. After overnight incubation, the insert was carefully removed to form a cell-free gap in the attached cells. The time of incubation was dependent on the tumor cells used. The number of migrating cells was calculated and analyzed. Six fields were randomly selected for analysis.
Flow cytometry analysis
To characterize the immune cells from the spleens of tumor-bearing mice, individual spleens were isolated and subjected to flow cytometry as previously described [12].
Patients
For patient studies, fresh specimens were collected from 38 patients with gastric adenocarcinoma who underwent radical resection at the National Cheng Kung University Hospital between August 2003 and August 2008. In total, 38 pairs of cancerous tissues and matched adjacent normal gastric mucosa were collected and analyzed as previously described [13].
Bioinformatics
A search was conducted in the Oncomine database (http://www.oncomine.com) [14] to systematically assess the expression level of CDH1 genes in gastric cancer. For differential analyses, we compared normal tissues and cancer tissues, specifically via analysis of P values, fold changes, and cancer subtypes. The prognostic value of CDH1 genes in gastric cancer was also analyzed using the Kaplan-Meier Plotter (http://kmplot.com/analysis/) as previously described [15]. The overall survival (OS), progression-free survival (PFS), and postprogression survival (PPS) were recorded, and the cut-off points for gene expression were automatically selected using the default setting. The probe of CDH1 gene was “201131_s_at.” The hazard ratio (HR), 95% confidence intervals, and log rank P values were displayed. Data from the Oncomine database and Kaplan-Meier Plotter were extracted between July and August 2020. Finally, the association between CDH1 protein expression (CDH1-to-b-actin ratio) and the Lauren classification (intestinal, diffuse, and mixed) of patients with gastric adenocarcinoma was assessed in the fresh specimens. The statistical differences between each two groups were analyzed.
Epithelial mesenchymal transition (EMT)-related genes were defined according to a meta-analysis of 14 gene expression studies [16]. The gene lists were applied to the raw data of gastric cancer in The Cancer Genome Atlas (TCGA). Hierarchical clustering was performed in R to produce a heatmap. Gene expression data were also obtained from GSE112369, a dataset of gastric organoids for which the raw data was publicly available [17]. Expression levels of STK24 and CDH1 were extracted. Gastric organoids forming from CDH1-single-knockout and parental cells were selected for further comparison.
Statistical analysis
Data were expressed as means ± standard deviations (SDs). Statistical analyses were performed in Prism (Graphpad Software, San Diego, CA). Student’s t-test was used for two-group comparisons, whereas one-way ANOVA followed by Tukey’s test was used for multiple-group comparisons. P values <0.05 were considered statistically significant.