Bioinformatic analysis

The GC-related miRNA microarray datasets GSE106817, GSE112264, GSE113486, and GSE113740 were acquired based on Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/), with differential miRNAs being identified through R software’s limma package. The miRNA and mRNA expression profiles related to GC were obtained in The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov/), and differential miRNA and mRNA expression profiles were utilized by R software edgeR package. miRDB (http://mirdb.org/), DIANA (https://diana.e-ce.uth.gr/), TargetScan (https://www.targetscan.org/vert_72/), and miRWalk (http://mirwalk.umm.uni-heidelberg.de/) databases were utilized for identifying miR-3662-regulated target genes. miR-3662 expression level in normal tissues and stomach adenocarcinoma (STAD) tissues was investigated by ENCORI database (https://rnasysu.com/encori/index.php). CAB39L gene expression in STAD and normal samples was investigated with Gene Expression Profiling Interactive Analysis (GEPIA) database (http://gepia.cancer-pku.cn/index.html). Thereafter, CAB39L protein expression level in STAD and normal stomach glandular samples was acquired based on Human Protein Atlas (HPA) database (https://www.proteinatlas.org/).

Cell culture

Human healthy gastric mucosal epithelial cells (GES-1) were acquired in Cellverse Co., Ltd. (Shanghai, China), while GC cells (MKN-45, AGS, HGC-27, SNU-1 and NCI-N87) were acquired in Cell Resource Center at the Institute of Basic Medical Sciences of Chinese Academy of Medical Sciences (Beijing, China). These cells were cultivated with Roswell Park Memorial Institute (RPMI)-1640 medium (Gibco, Carlsbad, CA, USA) that contained 10% fetal bovine serum (FBS; Gibco, Carlsbad, CA, USA) as well as 1% penicillin/streptomycin under 37 °C and 5% CO2 conditions.

Cell transfection

The AGS and HGC-27 cells were inoculated into 6-well plates and cultured until exceeding 70% cell confluence. Thereafter, miR-3662 mimics, mimic-NC, CAB39L overexpression plasmids (Oe-CAB39L), and empty plasmids (Vector) were transfected in AGS cells with Lipofectamine™ 3000 (Invitrogen, Carlsbad, CA, USA), while the miR-3662 inhibitor, inhibitor-NC, CAB39L interference plasmid (si-CAB39L: 5′-CCTACTGTGGAGTATATTA-3′), and negative control plasmid (si-NC) were transfected in HGC-27 cells. The miR-3662 inhibitor and mimics were procured in Guangzhou RiboBio Co., Ltd (Guangzhou, China). Meanwhile, CAB39L overexpression and interference plasmids were constructed in GenePharma Co., Ltd (Shanghai, China).

qRT-PCR assay

RNA extraction from GC cells was completed using RNAiso Plus (Takara, Dalian, China). Subsequently, 2 µg and 100 ng of total RNA were used as templates following the manufacturer's instructions for the Mir-X miRNA qRT-PCR TB Green kit and One Step TB Green PrimeScript RT-PCR kit instructions (Takara, Dalian, China), respectively. The synthesis of first-strand cDNA and subsequent qRT-PCR were carried out to detect miRNA and mRNA expression levels. Endogenous control genes U6 and GAPDH were utilized for normalization. The relative quantitative analysis of miR-3662 and CAB39L mRNA expression was conducted by 2−ΔΔCt approach. The amplification efficiency of the primers employed in this study ranged between 90 and 105%. The specific primer sequences are listed as follows: miR-3662 F, 5′-CGCTCACAGTTACACTTCTT-3′; R, 5′-GTGCTTCATCAGTCAC TACTCATC-3′; CAB39L F, 5′-TCTGTGTGGAACGAACGACAA-3′; R, 5′-TGAC ACTCATAGCTGACCTGCA-3′; U6 F, 5′-CTCGCTTCGGCAGCACA-3′; R, 5′-AA CGCTTCAGGAATTTG CGT-3′; GAPDH F, 5′-ACTGCCACCCAGAAGACT-3′; R, 5′-GCTCAGTGTA GCCCAGGAT-3′.

Cell counting kit-8 (CCK-8) assay

Those collected GC cells were subjected to trypsin treatment for obtaining the single-cell suspension, which was subsequently plated at 3 × 103/well into the 96-well cell culture plates. Later, CCK-8 regent (10 µl, Beyotime, Shanghai, China) was introduced into every well at 12, 24, 48, and 72 h before 2 h of incubation. Cellular proliferation capacity was measured through recording optical density values at 450 nm wavelength with the microplate reader (Thermo Scientific, Wilmington, DE, USA).

Apoptosis analysis

The GC cells (1.5 × 104/well) were inoculated into 6-well plates before 24 h of incubation. Subsequently, the cell pellet was collected through centrifugation, followed by PBS resuspension and counting. A total of 1 × 105 cells were then suspended into Annexin V-FITC solution (195 µl), and later Annexin V-FITC (5 µl) as well as propidium iodide (10 µl) (Beyotime, Shanghai, China) were poured. After thorough mixing, the resultant mixed sample was subjected to 10 min of incubation away from light under ambient temperature. Following this, the cellular solution was introduced into a flow cytometer (Invitrogen, Carlsbad, CA, USA) for quantifying apoptosis rate.

Transwell assay

Diluted Matrigel was poured to cover the top chamber, later, serum-free RPMI-1640 medium was added. GC cells (5 × 104/ml) were seeded into the top chamber. The lower chamber received serum-containing medium, followed by the placement of Transwell chamber to incubate for 24 h. Subsequently, chambers were removed and both Matrigel and non-invaded cells were eliminated from the membranes. Cellular fixation employed 4% paraformaldehyde solution, followed by crystal violet (Beyotime, Shanghai, China) staining for a duration of 30 min. Following the drying process, cell invasion was observed utilizing a microscope (Olympus, Tokyo, Japan), and invading cells were counted by selecting five non-overlapping fields at random.

Wound-healing assay

GC cells (5 × 104/well) were plated in 6-well plates. When cellular confluence attained 70%, a standardized linear wound was generated across each well's surface with the 200 µl sterile pipette tip, and photos were taken. The culture was then continued for 24 h and photographed again. Wound-healing was quantitatively analyzed with Image J software, comparing the initial and final wound areas to calculate cellular migration capacity.

Western blotting analysis

RIPA buffer (Beyotime, Shanghai, China) was employed for lysing GC cells, followed by subsequent centrifugation to collect supernatants. Protein concentrations were analyzed by bicinchoninic acid (BCA) analysis. Following electrophoretic separation, protein electrotransfer on polyvinylidene fluoride (PVDF) membranes (Millipore, Billerica, MA, USA) was completed. After 2 h of blocking under ambient temperature, PVDF membranes underwent overnight incubation using primary antibodies against B-cell lymphoma-2 (Bcl-2) (ab182858, 1:2000), Bcl-2 associated X (Bax, ab32503, 1:1000), Cleaved caspase-3 (C-caspase-3, ab32042, 1:500), N-cadherin (ab76011, 1:5000), Vimentin (ab92547, 1:2000), GAPDH (ab9485, 1:2500) (Abcam, Cambridge, UK); E-cadherin (#3195, 1:1000), AMP-activated protein kinase α (AMPKα, #2532, 1:1000), phosphor-AMPKα (Thr172) (p-AMPKα, #2535, 1:1000), platelet-type phosphofructokinase (PFKP, #8164, 1:1000), lactate dehydrogenase A (LDHA, #3582, 1:1000), β-actin (#4970, 1:1000) (Cell Signaling Technology, Danvers, MA, USA); CAB39L (NBP1-74079, 1:1000; Bio-techne, Minneapolis, MN, USA). Subsequently, they underwent 1 h of secondary antibody (ab205718, 1:10000; Abcam, Cambridge, UK) incubation under ambient temperature. Enhanced chemiluminescence reagent (Solarbio, Beijing, China) enabled band visualization, with Image J software quantifying band intensities.

Dual-luciferase assay

Based on those candidate interaction sites of miR-3662 with CAB39L identified through TargetScan database analysis, we generated luciferase reporter plasmids that contained either wild-type or mutated CAB39L sequences. AGS cells subsequently underwent co-transfection using the above vectors with mimic-NC or miR-3662 mimics. After incubation for 48 h, the relative firefly luciferase activity was measured using Renilla luciferase as an internal control, following the protocol provided by the dual-luciferase reporter gene assay kit (Solarbio, Beijing, China).

Statistical analysis

SPSS 23.0 was adopted in statistical analysis. All in vitro experiments were conducted using three biological replicates (n = 3). Results were represented by mean ± standard deviation (SD). Between-group comparisons utilized the independent samples t-test, whereas among-group comparisons were completed with one-way ANOVA. p < 0.05 indicated significant differences.