The human gastric adenocarcinoma GES-1, BGC-823, MGC-801, SGC-7901, and HGC-27 cell lines were cultured in RPMI-1640 medium (Gibco, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS), except for AGS in F12 medium (HyClone, USA) supplemented with 10% FBS (Gibco).

Fourteen RNA samples were harvested from the GC patients undergoing surgical resection in Shandong Tumor Hospital. Atrophic gastritis (AG) specimens were from 63 patients with or without H. pylori infection of Jinan Central Hospital. The 13C urea breath test was used to detect the H. pylori. Histological typing of the tumors was used to confirm the diagnosis.

Three H. pylori strains were cultured in Brucella broth containing 5% FBS at 37°C for 48 h under microaerophilic conditions. The bacteria were added to cells with the multiplicity of infection (MOI) at 100. C57BL/6 mice (male) were separated into three groups. Among them, group 1 is the control group. Groups 2 and 3 were first given the N-methyl-N-nitrosourea (MNU, 30 ppm) in the drinking water; 10 weeks later, they were given distilled water for another 2 weeks. Thereafter, Group 3 was given the SS1 strain (1 × 10⁹ colony-forming units (CFU)/ml) every other day for three times. Group 1 was provided with distilled water without MNU or SS1. Finally, all mice were sacrificed after 50 weeks.

JARID1B siRNAs (Sigma-Aldrich, USA), miR-29c mimic, and inhibitor (Ribobio, Guangzhou, China) transfection were performed with the Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). Sequences for these siRNAs: 5′-CAUAUUCUCUCUUAGAGGUdTdT-3′ (JARID1B si1) and 5′ -CAGUGAAUGAGCUCCGGCAdTdT-3′ (JARID1B si2).

Cells (600 cells/well) were seeded into the plates after respective treatments, then they were incubated for about 8–14 days until colonies appeared. The colonies were then treated with methanol and Giemsa. The experiments were performed three times.

DNA synthesis was assessed by the Cell-Light EdU (5-ethynyl-2'-deoxyuridine) Apollo488 In Vitro Imaging Kit according to the manufacturer’s instructions (RiboBio).

Paraffin-embedded sections acquired from samples were deparaffinized and rehydrated. Then, the citric acid buffer was used for antigen retrieval. Next, after H₂O₂ treatment and nonspecific antigen blocking for 30 min, sections were incubated with the primary JARID1B antibody (1:100, ab181089, Abcam, Cambridge, UK). Next day, the colorimetric detection was performed using DAB (3,3'-diaminobenzidine) staining kit (Vector Laboratories, USA).

Total RNA was extracted with TRIzol reagent (Invitrogen, Carlsbad, CA, USA) as per the manufacturer’s instructions. RNAs were then reverse-transcribed by the PrimeScript RT reagent Kit with gDNA Eraser (Takara, Japan). Finally, SYBR-Green (TaKaRa, Japan) was then used for the real-time PCR analysis. The primers used in this real-time quantitative PCR (RT-qPCR) assay are as follows: 5′-TGTCACAGTGGAATATGGAGCTGAC-3′ and 5′-GCCACTATCAAGATACTCCTCTTCC-3′ for JARID1B; 5′-ATGGAACACCAGCTCCTGTG-3′ and 5′-ACCTCCAGCATCCAGGTGGC-3′ for CCND1; 5′-AGTTGCGTTACACCCTTTCTTG-3′ and 5′-CACCTTCACCGTTCCAGTTTT-3′ for β-actin; 5′-GAATTGCTATGTGTCTGGGT-3′ and 5′-CATCTTCAAACCTCCATGATG-3′ for β2-M.

Cells were lysed with the protein lysis buffer supplemented with protease inhibitors for 30 min on ice to extract the protein. Next, proteins were subjected through sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to the polyvinylidene fluoride (PVDF) membranes. Then, 5% nonfat milk was used to block the membranes. The specific primary antibodies and the respective secondary antibodies were also used for incubation. Immunoblots were visualized with ECL (Enhanced chemiluminescent) reagent (Millipore). The primary antibodies were specific for JARID1B (Abcam), β-actin (Sigma-Aldrich, St. Louis, MO, USA), CCND1 (Cell Signaling, Danvers, MA, USA), and CagA (Santa Cruz Biotechnology, Santa Cruz, CA, USA).

The human JARID1B 3′-untranslated region (UTR) and corresponding mutants were cloned into the pMIR-GLO basic luciferase reporter vector (Promega, Madison, WI, USA). GC cells were transfected with the internal control vector pRL-TK and the promoter or 3′-UTR reporters using Roche Transfection Reagent (Roche, Basel, Switzerland) 24 h after seeding. Luciferase reporter activity was measured by a Luciferase Assay System (Promega) as per the manufacturer’s instructions 48 h after transfection.

For chromatin immunoprecipitation (ChIP) assay, the SimpleChIP^® Enzymatic Chromatin IP Kit (Cell Signaling, Danvers, MA, USA) was used according to the manufacturer’s protocol. The precipitated DNA samples were detected with PCR method.

The original files (.CEL files) and platform files of the GSE54129 dataset in the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo/) were downloaded. The raw data files were analyzed using the Robust Multi-array Average algorithm with the “affy” package and the “impute” package in R/Bioconductor software (version 3.5.3). For the upregulated differentially expressed genes (DEGs), we used the “limma” package and set P < 0.05. The inside-software plugin cytoHubba was used to identify hub genes among the genes of interest. This program provides a user-friendly interface to explore important nodes in biological networks and computes using 11 methods (24). Metascape is a free online resource with an automated meta-analysis tool for understanding the biological significance of a large number of genes and can identify enriched pathways and be used for the construction of protein–protein interaction networks from lists of gene and protein identifiers (25).

All data are presented as means ± standard deviations from three independent assays. The Student’s t-test or Mann–Whitney U test was used to determine the significance of differences in each two-group comparison. Statistical analysis was performed using SPSS version 23.0. Correlation analysis of the data was performed using linear regression. Differences were considered statistically significant when P < 0.05.

To identify potential roles of histone demethylases in human GC pathogenesis, we first undertook logistic regression analysis of 18 histone demethylase genes to establish hazard ratio (HR) values for their influence on GC prognosis using data from The Cancer Genome Atlas (TCGA) database. This approach showed that JARID1B and KDM4A (Lysine-specific demethylase 4A) were statistically significant ( Figure 1A ). Further analysis indicated that only JARID1B had a significant prognostic value in GC patients ( Figures 1B and S1 ). Additionally, we performed a bioinformatics analysis of JARID1B expression profiles from 876 GC patients using the Kaplan–Meier method and found that high expression of JARID1B was significantly correlated with poorer overall survival, first progression survival, and post-progression survival of GC patients ( Figures 1C–E ). Furthermore, we also carried out an analysis of the GEO databases (e.g., GSE27342 and GSE63089), which showed that JARID1B expression was upregulated in GC tissues ( Figures 1F, G ). Analysis of 14 GC biopsies and their corresponding adjacent non-tumor tissues confirmed this result ( Figure 1H ). Similarly, JARID1B was also differentially expressed in several gastric cell lines, including GES-1, AGS, BGC-823, HGC-27, MGC-803, and SGC-7901 ( Figure S2 ). Taken together, the data imply an oncogenic role for histone demethylase JARID1B in GC.

The tumor microenvironment consists of tumor cells, stromal cells, and infiltrating immune cells, and it has been proposed that these variables are valuable for diagnostic and prognostic assessment of tumors. We therefore utilized TIMER (Tumor IMmune Estimation Resource) to assess the role of JARID1B in GC by adopting the ESTIMATE algorithm (from https://bioinformatics.mdanderson.org/estimate/). This analysis showed that the level of JARID1B messenger RNA (mRNA) was significantly higher in the low immune score and low stromal score group ( Figures 2A, B ). Furthermore, high expression of JARID1B was statistically significantly associated with immune score (low, P < 0.0001) and stromal score (low, P = 0.017; Figures 2A, B ). The expression of JARID1B was negatively correlated with the ESTIMATE score (r = -0.26, P <0.001; Figure 2C ), suggesting that JARID1B was positively associated with tumor purity of GC. Moreover, analysis of tumor-infiltrating immune cells showed that high expression of JARID1B was negatively associated with the amount of CD8+ T cells, neutrophils, and dendritic cells in the tumor ( Figure 2D ).

Next, we investigated the mechanism responsible for JARID1B upregulation in GC. We searched for information on JARID1B gene mutations, as well as the inherent connection between gene copy number and mRNA expression in the “Stomach Adenocarcinoma (TCGA, Provisional)” dataset via the cBioportal program (http://www.cbioportal.org). Notably, gene amplification of JARID1B was frequently found in TCGA GC tissues, and the copy number significantly correlated with levels of JARID1B mRNA expression ( Figures 3A, B ). This suggests that copy number variation may underlie JARID1B amplification in GC.

H. pylori infection is a risk factor for GC. Hence, we investigated whether H. pylori infection influences the abnormal expression of JARID1B in GC. We first analyzed JARID1B expression in H. pylori-infected gastritis tissue by RT-qPCR and documented its significant upregulation relative to H. pylori-positive gastritis tissues ( Figure 4A ). To further investigate the regulation of JARID1B expression by H. pylori infection, we used different H. pylori strains (H.p11637 and H.p26695) to infect GC cells (AGS and BGC-823) at different time points. RT-qPCR showed that H. pylori infection can upregulate the expression of JARID1B mRNA at different time points ( Figures 4B–E ). Furthermore, Western blotting confirmed the upregulation of JARID1B in H. pylori-infected GC cells at the protein level ( Figures 4F, G and S3 ).

H. pylori gavage can promote the progression of precancerous lesions in murine gastric tissue (26). To further investigate the regulation of JARID1B expression by H. pylori infection in vivo, we constructed MNU-treated mice with H. pylori infection. Immunohistochemistry (IHC) showed that JARID1B was slightly upregulated in the low-dose MNU group compared with the normal gastric mucosa in the control group but was significantly upregulated when low-dose MNU was combined with H. pylori infection ( Figure 4H ). These findings indicate that H. pylori promotes the expression of JARID1B both in vivo and in vitro.

We further explored the mechanisms by which H. pylori regulates the expression of JARID1B, focusing on miRNAs, which are important posttranscriptional regulators. Abnormal expression of multiple miRNAs is a known result of H. pylori infection (27, 28). To investigate the potential role of miRNAs in the regulation of JARID1B, we first screened candidates using TargetScan and found that miR-29c, miR-29b, and miR-29a can bind to the 3′-UTR of JARID1B ( Figure 5A ). Further analysis of the GEO database (GSE51306) revealed that the expression of miR-29c was downregulated in gastric epithelial cells infected with H. pylori, while the expression of miR-29a and miR-29b was not significantly affected ( Figures 5B–D ). Furthermore, RT-qPCR showed that miR-29c expression in H. pylori-infected gastritis tissues was significantly lower than that in uninfected tissues ( Figure 5E ). In addition, miR-29c expression was downregulated in H. pylori-infected AGS and BGC-823 cell lines ( Figure 5F ).

To determine whether miR-29c regulates JARID1B expression, we transfected these cell lines with miR-29c analogs (mimics) and inhibitors ( Figure 5G ). Western blotting confirmed that the former decreased the amount of JARID1B protein, while the latter increased it ( Figures 5H and S4 ). RT-qPCR showed that transfection of AGS and BGC-823 cells with miR-29c analogs significantly inhibited JARID1B mRNA expression ( Figure 5I ). To further investigate whether JARID1B is a direct target of miR-29c, we constructed a fluorescent reporter plasmid containing the JRID1B 3′-UTR of the miR-29c binding site [JARID1B-wild type (WT)-3′-UTR] and a miR-29c binding site-mutated reporter plasmid (JARID1B-Mut-3′-UTR) ( Figure 5J ). We found that the miR-29c analogs with JARID1B-WT-3′-UTR reduced the luciferase activity, which was not the case for JARID1B-Mut-3′-UTR ( Figure 5K ). In addition, H. pylori infection promoted the expression of JARID1B in AGS and BGC-823 cells, while JARID1B expression was downregulated when cells were transfected with miR-29c analogs ( Figures 5L, M ). These results indicate that H. pylori regulates the expression of JARID1B through miR-29c.

To gain a deeper understanding of the effects of JARID1B, we performed gene set enrichment analysis (GSEA) on the microarray data from TCGA database. This revealed that high JARID1B expression was positively associated with the G2M checkpoint and cell cycle control in GC ( Figures 6A, B ). As JARID1B can promote the proliferation of GC cells ( Figure S5 ), to further explore the role of JARID1B in the proliferation induced by H. pylori, we performed colony formation and EdU staining experiments. We showed that the colony-forming capacity of GC cells in which JARID1B expression was inhibited was decreased relative to controls and that H. pylori infection increased colony formation. Furthermore, H. pylori infection together with JARID1B interference decreased colony formation relative to cells only infected with H. pylori ( Figures 6C, D ). The results of the EdU staining experiment were consistent with the colony formation findings ( Figures 6E, F ). Our data thus indicate that JARID1B promotes the proliferation of GC cells and is a node molecule for GC cell proliferation induced by H. pylori.

Finally, we aimed to elucidate the mechanisms by which JARID1B promotes GC cell proliferation. In this study, we analyzed the GSE54129 dataset in the GEO database. A search for upregulated DEGs positively associated with JARID1B was performed with Cytohubba. This showed that a number of genes are related to JARID1B expression, with CCND1 being the strongest ( Figure 7A ). Biological function and pathway enrichment analyses with Metascape indicated effects on six biological processes, most significantly was cancer pathways ( Figure 7B ). The analysis of 45 pairs of GC and corresponding adjacent normal tissues in the GSE63089 also revealed a significant positive correlation between JARID1B and CCND1 ( Figure 7C ). Furthermore, the expression levels of JARID1B and several proliferation-associated genes including cyclins (CCND1, CCND2, CCND3, CCNB1, and CCNE1) and cell cycle protein-dependent kinases (CDK1, CDK2, CDK3, CDK4, and CDK7) were analyzed. We observed that silencing of JARID1B decreased the expression of CCND1 in AGS and BGC-823 cells, both by RT-qPCR and Western blotting ( Figures 7D–F and S6 ). Because JARID1B was reported to decrease H3K27me3 levels in osteosarcoma (29), we then tested the change of H3K27me3 in GC cells. The results showed that knockdown of JARID1B could enhance H3K27me3 levels in AGS and BGC-823 cells ( Figures 7G and S7, 8 ). In addition, as H3K27me3 is associated with transcriptional repression (30), we further tested the correlation between JARID1B expression and H3K27me3 levels at the promoter of CCND1. ChIP results indicated that knockdown of JARID1B could enhance H3K27me3 levels in AGS cells ( Figure 7H ). Furthermore, the results of colony formation assays and EdU staining also showed that CCND1 is a node molecule in the proliferation of GC cells regulated by JARID1B ( Figures 7I, J ). Therefore, our data indicate that JARID1B can regulate CCND1 expression and thereby mediate the proliferation of GC cells.