Normal human gastric mucosal GES-1 and GC cell lines (HGC-27, MKN-74, SNU-216, MKN-45, AGS and SNU-719) were obtained from American Type Culture Collection and cultured in RPMI-1640 medium enriched with 10% FBS (both purchased from Gibco (Thermo Fisher Scientific). All cells were cultivated in a 37°C incubator with 5% CO₂.

The microarray or RNA sequencing datasets GSE70880 (ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE70880), GSE99416 (ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE99416) and GSE109476 (ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE109476) were downloaded from the Gene Expression Omnibus (GEO) database and analysis was conducted to obtain the differentially expressed genes (DEGs) between GC and healthy tissue samples, with the screening criterion set as a fold change >1.2. DEGs were integrated with RNA-binding protein (RBP) prediction data downloaded online from the RBPmap website (rbpmap.technion.ac.il/) (24). Bioinformatics analysis, including the construction of Venn diagrams, was performed using R software (v4.0.2; r-project.org/). The present study used expression levels of DAZAP1 from the FireBrowse (firebrowse.org/) and UALCAN (ualcan.path.uab.edu/) databases to further validate the results obtained from bioinformatics analysis. Gene searches were performed using the PubMed database (pubmed.ncbi.nlm.nih.gov/). To obtain EMT-related genes, the gene set from the EMT database (emtome.org) (25). The Cancer Genome Atlas (cancer.gov/tcga) database was used for bioinformatics analysis, focusing on the STAD (stomach adenocarcinoma) cohort. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis for target genes was conducted with the WebGestalt database (webgestalt.org/). Analysis of potential transcription factors of DAZAP1 was performed using the AnimalTFDB v4.0 database (guolab.wchscu.cn/AnimalTFDB4/#/).

Gastric adenocarcinoma tissue microarray (TMA; cat. no. HStm-Ade180Sur-04) was acquired from Shanghai Outdo Biotech Co Ltd. The TMA analysis included 90 GC samples and 90 matched adjacent healthy tissue samples (70 male; 20 female; median age, 65 years; age range, 34-83 years; distance, 5 cm). The use of tissue chips was approved by the Ethics Committee of Shanghai Outdo Biotech Co. Ltd. (approval no. SHYJS-CP-1804015).

A total of 10 formalin-fixed and paraffin-embedded GC specimens and adjacent healthy tissue specimens were collected from patients (seven male; three female; median age, 63 years; age range, 42-81 years) with GC at Nanfang Hospital (Guangzhou, China) between December 2023 and January 2024. Normal tissues were collected 5 cm from the tumor edge and immediately stored in liquid nitrogen at −196°C for subsequent experimentation. The diagnosis of primary GC was validated by two pathologists. Patients who had received chemotherapy, radiotherapy or molecular-targeted therapy before surgery were excluded. These tissues were used for western blotting, reverse transcription-quantitative (RT-q) PCR and RNA in situ hybridization (ISH). The study received ethical approval from Nanfang Hospital, Southern Medical University (approval no. NFEC-202105-K18; Nanfang, China). All patients provided written informed consent.

A total of 10 paired GC tumor and adjacent non-cancerous gastric mucosa samples were surgically obtained in 2020 from the Department of Surgery, Nanfang Hospital, Southern Medical University, for analysis by IHC and western blotting. Table SI presents the antibodies used. All experimental protocols received approval from the Ethics Committee of Southern Medical University, China.

Following 24 h fixation with 4% paraformaldehyde at room temperature, tumor specimens were embedded in paraffin wax. Sections of 4-mm thickness were cut from paraffin-embedded tissue blocks and placed on glass slides. The slides were deparaffinized using xylene, rehydrated using ethanol. For antigen retrieval, sections were microwaved in 0.01 mol/l sodium citrate buffer (pH=6.0) at 95°C for 15 min. After immersing the sections in 3% hydrogen peroxide for 10 min at room temperature to inhibit endogenous peroxidase activity. Blocking was performed at room temperature for 20 min using 5% BSA; cat. no. A8850, Solarbio Science & Technology Co., Ltd.) to eliminate non-specific binding. The sections were then incubated overnight at 4°C with the primary antibody against DAZAP1(Santa Cruz Biotechnology Co., Ltd.). Subsequently, they were incubated with the biotin-linked anti-Mouse IgG (cat. no. ab6823, dilution 1:1,000; Abcam) at room temperature for 30 min. Following another three PBS washes, chromogen detection was performed using DAB. The staining results were observed under a light microscope and analyzed using Image J 1.8 software (National Institutes of Health). Normal rabbit or mouse IgG (Sigma-Aldrich; Merck KGaA) was used as the isotype controls. The TMA was stained for DAZAP1 (Santa Cruz Biotechnology Co., Ltd.) according to the same protocol.

DAZAP1 staining intensity in carcinoma cells was scored as follows: 0 (negative), 1 (weak), 2 (moderate), and 3 (intense). Cells with scores of 0-1 were classified as low expression, while scores of 2-3 indicated high expression. The final IHC score for each sample was the average of evaluations by two independent observers.

Total protein from tissue or cells was extracted via RIPA lysis buffer (Beijing Solarbio Science & Technology Co., Ltd.) containing a protease inhibitor (MedChemExpress). The protein concentration of lysates was measured using the BCA assay. A total of 20 μg total protein were separated by 10% SDS-PAGE and were transferred to PVDF membranes. The membranes were treated with 5% non-fat dry milk at room temperature for 30 min. The membranes were blotted with primary antibodies (Table SI) overnight at 4°C, followed by incubation with HRP-conjugated secondary anti-rabbit or anti-mouse antibodies (Abcam; cat. nos. ab7090 and ab97046; 1:6,000) for 1 h at room temperature. We confirm that β-tubulin (cat. no. 10094-1-AP, 1:5,000; Proteintech) was used as the reference protein. Enhanced chemiluminescence was utilized for visualization with the Pierce ECL Western Blotting Substrate (Thermo Fisher Scientific, Waltham, MA, USA), and densitometric analysis was performed using Image J 1.8 (National Institutes of Health, Bethesda, MD, USA).

DAZAP1-pEnter, DAZAP1-flag, DAZAP1-RNA recognition motif 1 (RRM1)-flag, DAZAP1-RRM2-flag, DAZAP1-C-terminal domain (CTD)-flag, NOTCH1-pEnter and JAG1-pEnter plasmids were constructed by Kidan Biosciences. Human DAZAP1 small interfering RNA (siRNA), including siRNA1, siRNA2 and siRNA pool (siRNAp; a 1:1 mixture of siRNA1 and siRNA2 with the total siRNA amount constant in the transfection system). NOTCH1-siRNAp, JAG1-siRNAp and scrambled siRNA non-specific control (Scr-siRNA) were synthesized by Shanghai GenePharma Co., Ltd. DAZAP1, NOTCH1 and Scramble shRNA were all purchased from GenePharma Co., Ltd. Both short hairpin (sh)RNAs and siRNAs were used. The siRNAs were applied for short-term functional assays, while shRNAs were used to establish stable cell lines and xenograft assays. Briefly, cells were plated in 6-well culture plates at a density of 3×10⁵ cells per well. At 30-50% confluency, the cells were transfected with 50 nM target siRNA or Scr-siRNA via Lipo8000 (Beyotime Institute of Biotechnology) transfection reagent, at 37°C for 6 h. Subsequent experiments were conducted on the cells at 48 h post-transfection, and the transfection efficiency was validated by RT-qPCR) and western blot assays.

To establish stable AGS cell lines, the cells were transduced with lentiviruses for target gene manipulation (2nd-generation lentiviral system) purchased from GeneChem, Inc. The lentiviral vectors were engineered to express green fluorescent protein (GFP) as a reporter gene; with overexpression and knockdown constructed via GV348 and GV493 vectors (1 μg/μl, GenePharma, Inc). Lentiviral particles were produced by co-transfecting 293T cells (ATCC, USA) with 8 μg lentiviral target plasmid, 4 μg packaging and 2 μg envelope plasmid at 37°C for 6 h; viral supernatants were collected at 48-72 h post-transfection. AGS cells were seeded at 5×10⁵ cells/well in 6-well plates and cultured to 80% confluency, then transduced with the lentivirus at MOI=100 with polybrene and transfection reagent A (GenePharma, Inc.) at 37°C for 24-48 h following the manufacturer's protocol. Puromycin (10 μg/ml) was added at 72 h post-transduction for stable cell selection over 7 days; stably infected cells were maintained in 10 μg/ml puromycin and subjected to subsequent experiments 2-3 days after the completion of selection. Table SII lists the siRNA and shRNA target sequences.

Total RNA was isolated from AGS and MKN-45 cells, as well as 10 pairs of human GC and their adjacent normal tissues. The extraction was performed using TRIGene Plus reagent (cat. no. ZP108-101S, Genestar) in accordance with the manufacturer's instructions. RT was performed in strict accordance with the manufacturer's instructions for the was conducted via Hieff^® Fast Cell SYBR Green RT-qPCR Kit (cat. no. 11172ES40, Shanghai Yeasen Biotechnology Co., Ltd.). RT-qPCR was conducted via SYBR Green Master Mix (Shanghai Yeasen Biotechnology Co., Ltd.) according to the manufacturer's guidelines. The thermocycling conditions were as follows: Initial denaturation at 95°C for 2 min, followed by 40 cycles of denaturation at 95°C for 15 sec and combined annealing/extension at 60°C for 30 sec (two-step PCR). Quantification of relative levels of genes was conducted via the 2^−ΔΔCq method (26), and then normalization to the internal reference gene GAPDH was conducted. Table SIII illustrates the primer sequences.

A total of 1,000 transfected cells were plated into 12-well plates and cultivated in a 37°C, 5% CO₂ incubator for 10-12 days. All liquids were removed, 4% paraformaldehyde was used to fix them for 30 min at room temperature, then crystal violet was used for staining at room temperature for 15 min. Colonies were defined as cell populations large enough to be observed with the naked eye and were counted manually directly under a Zeiss microscope.

MKN-45, AGS and MKN-74 cells were cultured in 10 μM EdU-supplemented medium (Guangzhou RiboBio Co., Ltd.; cat. no. #C10310) for 4 h at 37°C. Following removal of the EdU-supplemented medium, 4% paraformaldehyde was used to fix the cells at room temperature for 60 min. After washing with PBS solution and subsequent incubation with Apollo and Hoechst 33258 dyes at 37°C, images were obtained via a fluorescence microscope. Proportion of EdU-positive staining was calculated as follows: EdU-stained cells/Hoechst-stained cells) ×100%.

Serum starved MKN-45, AGS and MKN-74 cells were seeded into a 6-well culture plate at a density of 5×10⁵ cells/well and grown to 100% confluence. After cells were rinsed with PBS, they were scratched. Images were captured under an inverted light microscope. For MKN-45 cells, imaging was performed at 0, 24 and 48 h. For AGS and MKN-74 cells, imaging was conducted from 0 to 24 h at 12 h intervals. Migration index was calculated as follows: (Original wound width-wound width after healing)/original wound width ×100%.

Migration assays were performed via Transwell chambers in 24-well plates (BD Biosciences; cat. no. #353092). MKN-45 (8×10⁴), AGS (2×10⁴ cells) or MKN-74 cells (2×10⁴ cells) were suspended in 100 μl serum-free RPMI 1640 medium (Gibco, Thermo Fisher Scientific) and plated in the top chambers, while the bottom chambers contained 700 μl RPMI-1640 medium with 20% FBS. After incubating for 48 h at 37°C, non-migrated cells on the top membrane surface were removed with a cotton swab. Then, 4% paraformaldehyde was utilized to fix the migrated cells on the bottom surface for 60 min at room temperature, 0.1% crystal violet was used to stain them for 60 min at 37°C and counting was conducted in four randomly selected high-power fields of view/sample under a light microscope (Zeiss microscope). For invasion assays, the top chambers were coated with Matrigel (cat. no. #358248, Corning, Inc.) at 37°C for 30 min to allow for solidification; all other procedures were identical to the migration assay.

MKN-45 (1×10⁴ cells) were cultured in a 24-well plate overnight at 37°C. Cells were fixed with 4% paraformaldehyde at room temperature for 15 min, permeabilized with 0.1% Triton X-100 at room temperature for 10 min, and blocked with 5% FBS at room temperature for 30 min. F-actin staining was performed with rhodamine-conjugated phallotoxin (5 U/ml, Molecular Probes; Thermo Fisher Scientific, Inc.) according to the manufacturer's guidelines. Nuclei were stained with Hoechst33258 at room temperature for 10 min in the dark. F-actin cytoskeleton staining was assessed using a fluorescence microscope.

Following washing in PBS, 4% paraformaldehyde was used to fix the MKN-45 cells for 15 min at room temperature. Following permeabilization with PBS containing 0.2% Triton X-100 for 5 min, cells were blocked with 10% goat serum (cat. no. ZLI-9056, Beijing Zhongshan Jinqiao Biotechnology Co., Ltd.) at 37°C for 30 min. Cells were incubated with primary antibodies against E-cadherin (1:200, Proteintech Group, Inc.) and Vimentin (1:200, Cell Signaling Technology, Inc.) at 4°C overnight. After three rinses with PBS, the cells were incubated with fluorochrome-conjugated secondary antibodies (1:100, cat. no. ZF-0512/ZF-0513, Beijing Zhongshan Jinqiao Biotechnology Co., Ltd., Alexa Fluor^® 488/Alexa Fluor^® 594 Conjugate) at 37°C for 1 h in the dark. Nuclear counterstaining was performed with Hoechst 33258 dye at room temperature for 10 min in the dark, followed by two washes with PBS to remove excess dye. The stained cells were observed under a fluorescence microscope and all fluorescent images were acquired and analyzed using Image J 1.8 (National Institutes of Health, Bethesda, MD, USA).

Magna RIP RBPs Immunoprecipitation kit (MilliporeSigma) was used according to the manufacturer's instructions. In brief, cells were subjected to centrifugation at 300 × g for 5 min at 4°C, the cells were suspended in 200 μl RIP lysis buffer (MilliporeSigma) provided in the kit that contained 1 μl protease inhibitor cocktail and 0.5 μl RNase inhibitor. A total of 200 μl lysate was used per IP reaction. A total of 5 μg anti-mouse DAZAP1 and mouse IgG isotype control antibody (MilliporeSigma) were added to the magnetic bead suspension, and the mixtures were gently shaken at room temperature for 1 h to form bead-antibody complexes. The bead-antibody complexes were allowed to sit at 4°C overnight. Magnetic bead-RNA-protein complexes were washed three times with the RIP wash buffer 1 and 2 provided in the kit using a magnetic stand for magnetic separation of the complexes. After the RNA-protein complexes coupled to the beads were isolated, total RNA was extracted from the isolated complexes and subjected to sequencing analysis for target gene detection.

Anti-DAZAP1 antibody was utilized to immunoprecipitate DAZAP1-RNA complexes, with input as the control (IgG). DAZAP1-bound RNAs were extracted via TRIzol, and a cDNA library was created via the TruSeq RNA library preparation kit (Illumina, Inc. cat. no. RS-122-2201), the quality and integrity of the processed RNA samples were verified using a Agilent 2100 Bioanalyzer (Agilent Technologies) prior to library construction. Sequencing was performed on the Illumina HiSeq X Ten platform by ABLife Inc. with 150 bp paired-end sequencing; the sequencing reaction was conducted using the HiSeq X Ten Reagent kit v2.5 (300 cycles; cat. no. FC-510-1003; Illumina, Inc.). The final library was loaded at a concentration of 10 pM, and the molar concentration was measured via Qubit 4 Fluorometer with Qubit dsDNA HS Assay kit (Thermo Fisher Scientific, Inc.; cat. no. Q32854). Raw sequencing data were analyzed using HISAT2 (v2.2.1; daehwankimlab.github.io/hisat2/), StringTie (v2.2.1; ccb.jhu.edu/software/stringtie/) and DESeq2 (v1.38.3; bioconductor.org/packages/release/bioc/html/DESeq2.html) for read alignment, transcript assembly and differential expression analysis, respectively.

Scr-siRNA, DAZAP1-siRNAp, DAZAP1-siRNA1 and DAZAP1-siRNA2 were transfected into cells. After 24 h at 37°C, the medium was substituted with medium (Gibco; Thermo Fisher Scientific; cat. no. 11875093). enriched with actinomycin D (10 μg/ml, MedChemExpress) or DMSO. Total RNA was collected at 0, 2, 4, 6 and 8 h post-actinomycin D treatment using TRIGene Plus reagent (cat. no. ZP108-101S, Genestar) and mRNA half-life was estimated by determining NOTCH1 and JAG1 mRNA levels at each time point via RT-qPCR (normalized to GAPDH RNA) and curve fitting with the one-phase decay model in GraphPad Prism 9 (Dotmatics). NOTCH1 or JAG1 mRNA relative levels were assessed by RT-qPCR as aforementioned. mRNA levels were normalized to those of GAPDH RNA.

ISH was conducted to observe NOTCH1 or JAG1 expression in GC specimens with locked nucleic acid-based ISH via digoxigenin-labeled miRCURY miRNA probes (Exiqon; Qiagen GmbH). DNA probes designed for the spatial expression analysis of mRNA were derived from the cDNA plasmid pGEM-T Easy (Promega, cat. no. A1360). Both probes had a length of 36 nucleotides, and the specific sequences of the probes used for ISH are illustrated in Table SIV. GC tissue was fixed with 4% paraformaldehyde at 4°C for 24 h, embedded in paraffin and cut into 4 μm-thick serial sections on poly-L-lysine-coated glass slides (Thermo Fisher Scientific; cat. no. 12-550-15). The slides were deparaffinized, rehydrated and incubated with Proteinase K (Roche Diagnostics; cat. no. 3115887001) at 37°C for 15 min, slides were then subjected to post-fixation with 4% paraformaldehyde at room temperature for 10 min and followed by washing with PBS. Pre-hybridization was performed with 5X saline-sodium citrate buffer (SSC; Sigma-Aldrich; cat. no. S6639) hybridization buffer (without probe) at room temperature for 15 min, followed by probe denaturation at 95°C for 1 min. NOTCH1 mRNA and JAG1 mRNA probes (final concentration, 50 nM) were applied for hybridization in 5X SSC hybridization buffer with probe at 50°C for 1 h. The slides were rinsed with 2.0 and 0.1X SSC (at 50°C for 15 min each (30 min total) to remove excess probes, then blocked with 10% goat serum (Thermo Fisher Scientific; cat. no. 16210064) at RT for 30 min. An overnight incubation was performed with an anti-digoxigenin alkaline phosphatase-conjugated primary antibody [1:1,000, Roche Diagnostics (Shanghai) Co,. Ltd.; cat. no. 11093274910] at 4°C.BCIP/NBT Chromogenic Detection System (Roche Diagnostics; cat. no. 11681451001) was used to visualize the hybridization signals. Nuclear fast red (Sigma-Aldrich; cat. no. N3020) was used to counterstain the slides at room temperature for 1 min, and mounting was performed via an aqueous solution. For each patient, four slides were hybridized with the NOTCH1 mRNA or JAG1 mRNA probe. Hybridization signals were observed under a light microscope and ImageJ v1.8 (National Institutes of Health) was used for signal intensity analysis.

A total of 24 nude female BALB/c mice (nu/nu; age, 4-6 weeks; weight, 20-22 g) were obtained from the Guangdong Medical Laboratory Animal Center (Guangzhou, China) and housed in specific-pathogen-free conditions were as follows: Temperature, 20-26°C; humidity, 50-70%; light/dark cycle, 12/12-h; free access to food/water. The Animal Ethics Committee of Southern Medical University and Nanfang Hospital approved the animal experiments (approval no. IACUC-LAC-20250111-002).

For the tail vein-lung metastasis model, two groups of mice were utilized to assess the metastatic capability of malignant cells. The first group comprised the shRNA and DAZAP1-shRNA groups, while the second group included Vector + Scr-shRNA, DAZAP1 + Scr-shRNA, Vector + NOTCH1-shRNA, DAZAP1 + NOTCH1-shRNA, DAZAP1 + NOTCH1-shRNA + Vector and DAZAP1 + NOTCH1-shRNA + JAG1. An injection of 5×10⁶ lentivirus (LV)-transfected cells was administered into the tail veins of nude mice, which were euthanized after 30 days via CO₂ asphyxiation using a gradual-fill approach, with the gas introduced at a rate of 30% of the chamber volume/min. Cervical dislocation was performed to confirm death. Death was confirmed by disappearance of pain response, no reaction when the toes were pressed with hands or tweezers and cessation of heartbeat and respiration. The humane endpoint was body weight loss >20%; no animals met this endpoint during the experimental period. Subsequently, the lung metastases were counted. Paraffin-embedded lung tissues were sectioned into 4 μm sections and stained with hematoxylin and eosin and vimentin antibody as aforementioned.

Statistical analyses were performed using SPSS version 25.0 (IBM Corp.) and GraphPad Prism 10.1.2 (Dotmatics). All experiments were conducted in triplicate. All data are presented as the mean ± SD. For normally distributed continuous variables, Student's t-test (paired) or one-way ANOVA with Dunnett's post hoc test was applied as appropriate. For non-parametric comparisons, the Mann-Whitney U test was used. The Pearson correlation coefficient was used to assess the correlation between variables. Kaplan-Meier survival analysis was performed to evaluate overall or progression-free survival and the log-rank test was used to compare survival differences between groups. A two-tailed P-value <0.05 was considered to indicate a statistically significant difference.

To identify novel protein coding genes involved in the onset and progression of GC, the present study assessed the microarray data of GSE70880, GSE99416 and GSE109476 downloaded from the GEO database and visualized the DEGs in GC and normal tissue samples (Fig. 1A). A total of 19 genes were up- and 17 were downregulated in the GSE109476 dataset, 81 genes were up- and 30 downregulated in the GSE70880 dataset and 48 genes were up- and 37 downregulated in the GSE99416 dataset. The intersection of the datasets demonstrated that only DAZAP1 was upregulated, and three genes (PPARGC1A, CIRBP, PABPC4) were downregulated. DAZAP1 was selected for further study due to its enhanced potential for use as an early diagnostic marker or intervention target (8-12). DAZAP1 was highly expressed in most tumors compared with adjacent healthy tissues based on FireBrowse and UALCAN databases (Fig. S1A and B). Expression of DAZAP1 was increased in GC compared with normal samples (Fig. 1B). Compared with normal gastric cell line GES-1, DAZAP1 protein expression was markedly upregulated in HGC-27, MKN-74, SNU-216, MKN-45, AGS and SNU-719 cells (Fig. 1C). IHC was performed to determine the intensity of the DAZAP1 staining (Fig. 1D). TMA showed upregulation of DAZAP1 in tumor tissue (Fig. 1Ea). DAZAP1 deposition was displayed in the stroma, with scattered malignant cells in the peripheral margin of the tumors (tumor-associated stromal cells), and the cells displayed EMT-like morphological alterations, including the loss of typical epithelial cobblestone morphology, acquisition of a fusiform mesenchymal shape and increased cell scattering and spindle-like polarization. Semiquantitative scoring illustrated that DAZAP1 protein was significantly upregulated in cancer compared with adjacent healthy gastric mucosa tissue (Fig. 1Eb).

The present study analyzed correlations between DAZAP1 expression and clinicopathological characteristics and prognosis in 90 GC tissue samples. High DAZAP1 expression was significantly associated with local invasion (T1/2 vs. T3/4), American Joint Committee on Cancer (AJCC) stage (I/II vs. III/IV), lymph node metastasis (N0 vs. N1) and tumor size (<5 vs. ≥5 cm). However, DAZAP1 was not linked to additional clinical characteristics, such as age, sex and differentiation in GC (Table SV). Kaplan-Meier survival analysis demonstrated a relationship between DAZAP1 levels and overall survival time (Fig. 1Fa). This correlation detected in individuals with late-stage GC (AJCC stage III/IV; Fig. 1Fc) was more noticeable than that in early-stage GC (AJCC stage I/II; Fig. 1Fb). DAZAP1 expression demonstrated predictive value for overall survival in early-(AJCC I/II) and late-stage (AJCC III/IV) GC patient subgroups. These data illustrated that DAZAP1 is upregulated and has an oncogenic function in GC.

MKN-45 GC cell line with relatively low and AGS and MKN-74 cell lines with relatively elevated endogenous DAZAP1 levels were selected for transfection and functional experiments (Fig. 1C). Stable transfectants were constructed via DAZAP1-sense plasmids in MKN-45 cells (Fig. S2A) or DAZAP1 was suppressed using siRNAp in AGS and MKN-74 cells (Fig. S2B), as demonstrated via western blot analysis and qPCR. Subsequently, colony formation assays showed that overexpression of DAZAP1 led to significantly more colonies than the vector in MKN-45 cells (Fig. 2A). EdU assay showed that overexpression of DAZAP1 significantly induced cell proliferation compared with the control vector in MKN-45 cells (Fig. 2B). Proliferation of AGS and MKN-74 cells was inhibited by the knockdown of DAZAP1 (Fig. 2A and B).

The present study explored the effect of DAZAP1 on cell migration and invasion in GC. Wound healing assay demonstrated that the migration ability was promoted by DAZAP1 overexpression in MKN-45 cells, whereas DAZAP1 knockdown had the opposite effect on AGS and MKN-74 cells (Fig. 2C). Transwell experiments with or without Matrigel demonstrated a significant increase in migration and invasion following DAZAP1 overexpression and a significant decrease in migration and invasion in response to DAZAP1 knockdown (Fig. 2D). All these data illustrated that DAZAP1 mediated malignant phenotypes in GC cells in vitro.

A key step in tumor spreading is EMT, which occurs when epithelial cells associated with tumors transform into mesenchymal cells (27,28). The present study assessed cell morphology to see if DAZAP1 is essential for GC EMT. By contrast with the control cells, the MKN-45 cells that had DAZAP1 overexpression had an elongated spindle-like phenotype, which was distinguishable from normal mesenchymal cells (Fig. 3A).

Previous studies have shown that EMT improves motility and actin reorganization of cells (29,25). Therefore, F-actin was stained with phalloidin. In DAZAP1-overexpressing cells, F-actin filaments were more dispersed throughout the cytoplasm and localized at the protrusion rim compared with control cells (Fig. 3B). Immunofluorescence analysis revealed that DAZAP1 overexpression in MKN-45 cells downregulated E-cadherin and upregulated vimentin expression (Fig. 3C). Western blot analysis confirmed that DAZAP1 transfection altered the expression of EMT-associated markers. Overexpression of DAZAP1 significantly suppressed E-cadherin, whereas knockdown of DAZAP1 increased the expression of E-cadherin in GC cells (Fig. 3D). Conversely, overexpressing DAZAP1 promoted the expression of N-cadherin and vimentin, while knockdown of DAZAP1 suppressed their expression in GC cells (Fig. 3D).

Additionally, to explore the association between DAZAP1 and migration/invasion, the function of DAZAP1 in metastases was examined by injecting AGS/LV-pEGFP-DAZAP1-shRNA and AGS/LV-pEGFP-Scr-shRNA expressing GFP into nude mice (Fig. 3E). The knockdown efficiency of DAZAP1-shRNA was verified in AGS cells (Fig. S4C). Smaller lung metastatic nodules were detected in the LV-DAZAP1-shRNA compared with the LV-pEGFP-Scr-shRNA group (Fig. 3F). Histological examination confirmed the presence of GC cell metastases in the lung tissue (Fig. 3G). Moreover, IHC analysis demonstrated that vimentin protein levels were promoted in cancer tissue of the LV-pEGFP-Scr-shRNA compared with those in the LV-DAZAP1-shRNA group (Fig. 3H). These results indicated that DAZAP1 in GC cells may be a vital contributor to EMT progression, thereby mediating the ability of GC cells to migrate and invade.

As DAZAP1 is an RBP and regulates mRNAs by binding UAGKWWR motifs (9,11,12) (Fig. 4A), RIP-seq was performed in AGS cells to detect potential RNA targets. A total of 980 genes (1,376 transcripts) potentially bound to DAZAP1 (Table SVI). EMT-related genes were obtained from the EMT database (emtome.org) (Table SVII) (25,30).

The 980 DAZAP1-bound genes and the EMT-related genes from the database were used to construct a Venn diagram for identifying common genes (Fig. 4B). The intersection illustrated 88 genes (Table SVIII), among which 44 genes were associated with GC. These target genes were used for KEGG) pathway enrichment analyses. A total of 10 significantly enriched KEGG pathways, including 'pathways in cancer', were identified (Fig. 4C; Table SIX). NOTCH1 and JAG1 are essential in tumor progression and enriched in 'pathways in cancer' (16,18). Therefore, NOTCH1 and JAG1 were selected for subsequent study.

qPCR was performed to determine NOTCH1 and JAG1 mRNA expression in GC cells with DAZAP1 suppression; DAZAP1 knockdown group exhibited a significant decrease in NOTCH1 and JAG1 mRNA expression compared with the negative control group (Fig. 4D). RIP was performed to determine whether DAZAP1 binds to the NOTCH1 or JAG1 genes. Compared with IgG control, DAZAP1 antibody demonstrated notable enrichment of NOTCH1 and JAG1 RNA (Fig. 4E and F). The results demonstrated a physical association between the DAZAP1 protein and either NOTCH1 or JAG1 mRNA in vitro.

The present study explored the effect of DAZAP1 on NOTCH1 or JAG1 mRNA stability. Actinomycin D assay demonstrated that the half-life of NOTCH1 or JAG1 mRNA was significantly lower following DAZAP1 knockdown than in the control group in AGS cells (Fig. 4G and H). These data suggest that DAZAP1 can increase NOTCH1 or JAG1 expression by binding NOTCH1 or JAG1 mRNA and enhancing its stability.

DAZAP1 protein is composed of three primary domains: RRM1 (10-97 aa), RRM2 (113-190 aa) and CTD (250-407 aa). To investigate the domain required for DAZAP1-mediated NOTCH1 or JAG1 expression, DAZAP1 [full-length (FL)], including RRM-F1 (1-112 aa), RRM2-F2 (113-249 aa) and CTD-F3 (250-407 aa) plasmids with a FLAG-tag, were constructed (Fig. 4I). The in vitro binding assay showed that for NOTCH1 (Fig. 4J) and JAG1 (Fig. 4K) mRNA, the Flag group of full-length DAZAP1 (-FL) and its CTD domain (-F3) exhibited markedly high fold enrichment, while RRM1 (-F1) and RRM2 (-F2) only had weak enrichment. Thus, CTD is key for DAZAP1 to physically bind NOTCH1/JAG1 mRNA, thereby regulating their stability.

Ligands for NOTCH1 are transmembrane proteins expressed on adjacent cells, including JAG and Δ (14,15,23). The present study assessed whether DAZAP1/NOTCH1 or DAZAP1/JAG1 pathways contribute to GC cell progression. First, the present study constructed stable transfectants using NOTCH1- and JAG1-sense plasmids or knockdown of NOTCH1 and JAG1 with siRNAp in MKN-45 and AGS cells (Fig. S3A), as confirmed by western blot and RT-qPCR analysis. DAZAP1 knockdown attenuated the migratory and invasive abilities of GC cells, whereas overexpression of NOTCH1 or JAG1 partially reversed the phenotypes induced by DAZAP1 suppression compared with the control, as demonstrated by wound healing and Transwell assays (Figs. 5A-D and S4A-D). Rescue experiments showed that DAZAP1 overexpression markedly increased cell motility ability, which was abrogated by downregulation of NOTCH1 or JAG1 (Figs. 5A-D and S4A-D).

EMT-associated marker levels were detected by western blotting, indicating that DAZAP1 suppression elevated the expression of the epithelial marker (E-cadherin) and decreased expression of mesenchymal markers (vimentin and N-cadherin) in GC, Notably, this effect was reversed when NOTCH1 or JAG1 was introduced alongside DAZAP1 suppression (Figs. 5E and F and S4E-F). Conversely, DAZAP1 overexpression decreased E-cadherin expression and increased vimentin/N-cadherin levels in GC- and this trend was abolished when NOTCH1 or JAG1 expression was concurrently reduced. (Figs. 5E-F and S4E-F). These results indicated that upregulation of DAZAP1 facilitated NOTCH1 or JAG1 expression to promote motility of GC cells via EMT.

The present study examined whether NOTCH1 and JAG1 mediate the influence of DAZAP1 on GC cell migration and invasion. DAZAP1 overexpression vector increased motility capacity compared with the control, while DAZAP1/NOTCH1-siRNA suppressed migration/invasion capacity compared with DAZAP1 in GC cells (Figs. 6A-C and S5A and B). Reintroduction of JAG1 partially reversed the suppression compared with DAZAP1/NOTCH1-siRNA, as demonstrated by wound healing and Transwell assay (Figs. 6A-C and S5A and B).

Levels of epithelial marker E-cadherin were downregulated in DAZAP1 compared with control, while DAZAP1/NOTCH1-siRNAed result in upregulated E-cadherin compared with DAZAP1. However, E-cadherin upregulation was partly reversed following DAZAP1/NOTCH1-siRNA/JAG1 transfection. Mesenchymal markers vimentin and N-cadherin were upregulated in the DAZAP1 overexpression group relative to the vector, downregulated in the DAZAP1/NOTCH1-siRNA group compared with the DAZAP1 overexpression group and partially restored following DAZAP1/NOTCH1-siRNA/JAG1 transfection (Fig. 6D).

To investigate the effect of DAZAP1-NOTCH1-JAG1 on cell metastasis in vivo, the AGS cells expressing the LV-vector, LV-DAZAP1, DAZAP1/NOTCH1-shRNA or DAZAP1/NOTCH1-shRNA/JAG1 were implanted into the tail vein of nude mice, thereby causing lung metastases (Fig. 6E). The knockdown efficiency of NOTCH1-shRNA was verified in AGS cells (Fig. S3C). More metastatic tumors were determined in DAZAP1 overexpression compared with the vector group; DAZAP1/NOTCH1-shRNA caused a significant decrease in the number of visible tumors than in the control (Fig. 6F). However, the reintroduction of JAG1 partially reversed the suppression compared with DAZAP1/NOTCH1-shRNA, which was associated with an increased count of metastasis loci (Fig. 6F). The GC metastases in the lung of nude mice were confirmed through hematoxylin and eosin staining (Fig. 6G), and the difference in cell motility was detected by vimentin IHC staining. Vimentin is a well-characterized mesenchymal marker that is associated with cancer cell migratory and invasive potential as well as metastatic ability (31-32). Vimentin was upregulated in the DAZAP1 overexpression group relative to the vector, downregulated in the DAZAP1/NOTCH1-siRNA group compared with the DAZAP1 overexpression group and partially restored following DAZAP1/NOTCH1-siRNA/JAG1 transfection (Fig. 6H). Therefore, DAZAP1 modulated NOTCH1/JAG1 signaling in GC cells to promote invasion and metastasis.

The present study assessed the expression of DAZAP1, NOTCH1 and JAG1 in 10 freshly collected GC biopsy samples. Western blotting (Fig. 7A) and RT-qPCR (Fig. 7B and C) demonstrated that the expression of DAZAP1, NOTCH1 and JAG1 was significantly upregulated compared with adjacent normal tissue. DAZAP1 expression was positively associated with NOTCH1 or JAG1 expression (Fig. 7D and E).

IHC and ISH were demonstrated that the expression of DAZAP1, NOTCH1 and JAG1 was significantly higher, while that of E-cadherin was significantly lower, in GC compared with healthy tissue (Fig. 7F).

Aberrant NOTCH signaling activation is typically linked to its receptors and ligands (33-35). To determine whether DAZAP1 impacts the NOTCH1 signaling pathway, the present study examined the expression of key components in the NOTCH1 pathway using western blot analysis in GC cells. Knockdown of DAZAP1 decreased NOTCH1, JAG1, MYC and hairy and enhancer of split homolog-1 (HES-1) protein expression in AGS cells. Conversely, forced expression of DAZAP1 resulted in increased NOTCH1, JAG1, MYC and HES-1 levels in MKN-45 cells.

Collectively, these data demonstrated that DAZAP1 expression was positively associated with levels of NOTCH1 and JAG1 in GC samples and DAZAP1 facilitated the activation of the NOTCH1 signaling pathway.

For the potential transcription factors of DAZAP1, predictions were made using the AnimalTFDB v4.0 database, which identified a total of 796 putative transcription factors (Table SX).