Gastric cancer (GC) is a common malignancy of the digestive system. According to global 2018 cancer data, GC has the fifth-highest incidence and the third-highest fatality rate among malignant tumors. More than 60% of GC are linked to infection with Helicobacter pylori (H. pylori), a gram-negative, active, microaerophilic, and helical bacterium. This parasite induces GC by producing toxic factors, such as cytotoxin-related gene A, vacuolar cytotoxin A, and outer membrane proteins. Ferroptosis, or iron-dependent programmed cell death, has been linked to GC, although there has been little research on the link between H. pylori infection-related GC and ferroptosis.

To identify coregulated differentially expressed genes among ferroptosis-related genes (FRGs) in GC patients and develop a ferroptosis-related prognostic model with discrimination ability.

Gene expression profiles of GC patients and those with H. pylori-associated GC were obtained from The Cancer Genome Atlas and Gene Expression Omnibus (GEO) databases. The FRGs were acquired from the FerrDb database. A ferroptosis-related gene prognostic index (FRGPI) was created using least absolute shrinkage and selection operator–Cox regression. The predictive ability of the FRGPI was validated in the GEO cohort. Finally, we verified the expression of the hub genes and the activity of the ferroptosis inducer FIN56 in GC cell lines and tissues.

Four hub genes were identified (NOX4, MTCH1, GABARAPL2, and SLC2A3) and shown to accurately predict GC and H. pylori-associated GC. The FRGPI based on the hub genes could independently predict GC patient survival; GC patients in the high-risk group had considerably worse overall survival than did those in the low-risk group. The FRGPI was a significant predictor of GC prognosis and was strongly correlated with disease progression. Moreover, the gene expression levels of common immune checkpoint proteins dramatically increased in the high-risk subgroup of the FRGPI cohort. The hub genes were also confirmed to be highly overexpressed in GC cell lines and tissues and were found to be primarily localized at the cell membrane. The ferroptosis inducer FIN56 inhibited GC cell proliferation in a dose-dependent manner.

In this study, we developed a predictive model based on four FRGs that can accurately predict the prognosis of GC patients and the efficacy of immunotherapy in this population.

Gastric cancer (GC) is a common malignancy of the digestive system. According to global 2018 cancer data, GC has the fifth highest incidence and the third highest fatality rate among malignant tumors[1]. More than 60% of GCs are linked to infection with Helicobacter pylori (H. pylori), a gram-negative, active, microaerophilic, and helical bacterium[2]. This parasite induces GC by producing toxic factors, such as cytotoxin-related gene A, vacuolar cytotoxin A, and outer membrane proteins[3]. The World Health Organization classifies H. pylori as a Group I carcinogen[4]. Ferroptosis, or iron-dependent programmed cell death[5], has been linked to GC, although there has been little research on the link between H. pylori infection-related GC and ferroptosis. Therefore, this study investigated the coregulated differentially expressed genes (co-DEGs) among ferroptosis-related genes (FRGs) in H. pylori infection-related GCs and investigated the relationship between the expression of these co-DEGs and clinical prognosis.

Infection with H. pylori is a significant risk factor for distal GC development[1-3]. Ferroptosis is a programmed cell death mechanism associated with fatal lipid peroxidation[7]. Iron metabolism abnormalities are related to both ferroptosis and elevated GC risk[8]. Many mechanisms may be linked to ferroptosis, H. pylori infection, and GC. Nevertheless, few studies have evaluated the relationship between ferroptosis and the prognosis of H. pylori-related GC patients. Therefore, this study aimed to identify Co-DEGs, construct an FRGPI, and validate the predictive significance of the FRGPI in H. pylori-associated GC. The discovered molecular processes and signaling pathways may help explain the relationship between Co-DEGs and H. pylori-associated GC.

This research identified a total of 244 co-DEGs. GO and KEGG analyses revealed that the co-DEGs were involved primarily in ferroptosis and in the FoxO, NOD-like receptor, mTOR, HIF-1, IL-17, TLR, and MAPK signaling pathways, among others; all of these pathways are strongly associated with GC occurrence and development. The FoxO subfamily of the forkhead transcription factor family is involved in cell fate determination and is thought to function as a tumor suppressor in various malignancies[9]. FoxO is implicated in both mitochondria-dependent and mitochondria-independent apoptotic processes by activating death receptor ligands such as the Fas ligand, TNF apoptotic ligand, and Bcl-2 family members[10]. The PI3K/AKT pathway is the most critical pathway through which FoxO1 interacts with many cancers[11]. In addition, numerous other vital pathways, including the Ras–MEK–ERK, IKK, and AMPK pathways, are involved in the mechanism by which FoxOs influence carcinogenesis[10,11]. The inflammasome, which is composed of NOD-like receptors, is also a critical component of innate immunity-induced host defense and inflammation in cancer[12]. Increased inflammasome activity has been linked to various malignancies, including breast, gastric, brain, and malignant prostate cancer, and is protective against colitis-associated cancers[13]. The tumor stage and engagement of different caspase isoforms with the inflammasome pathway impact inflammasome function in cancer. Autophagy inhibition via mTOR-dependent mechanisms such as the AMPK/mTOR and PI3K/Akt/mTOR pathways contributes to the malignant progression of GC cells[14]. Autophagy, which is essential for homeostasis, has dual roles in GC, acting as a tumor suppressor and promoter[15]. Long-term H. pylori infection inhibits autophagy, potentially promoting GC. HIF-1 signaling is vital for metabolism, inflammation, vascular homeostasis, and cancer. HIF-1 activation occurs in several malignancies, including GC[16]. Tumor-associated neutrophils release IL-17a, stimulating GC cell migration and invasion via JAK2/STAT3 activation[17]. TLRs, pattern recognition receptors crucial for H. pylori lipopolysaccharide recognition, are pivotal in GC development, occurrence, and treatment[18,19]. TLR activators hold potential as therapeutic and adjuvant agents in combination with other immunotherapies, suggesting that TLRs are immunotherapy targets for GC[20]. The MAPK regulatory network, which includes many components that initiate a phosphorylation cascade upon activation, elicits precise cellular responses[21]. This network significantly influences GC progression[22].

This study utilized public databases to identify crucial ferroptosis-related DEGs in H. pylori-associated GC. In GC patients, FRGs were differentially expressed between normal and tumor tissues, indicating the substantial involvement of ferroptosis in this disease. Univariate Cox analysis revealed 18 genes linked to OS, highlighting the feasibility and utility of constructing a prognostic signature using the FRGPI. LASSO-Cox analysis identified a novel prognostic signature comprising four FRGs: NOX4, MTCH1, GABARAPL2, and SLC2A3. The four hub genes accurately predicted GC using ROC curves. Functional enrichment and analyses of the tumor microenvironment and immunotherapy response revealed that the hub genes might accurately predict GC patient prognosis and clinical state. To better understand the involvement of these genes in GC, we describe their primary molecular activity.

NOX4 is an NADPH oxidase that promotes the cellular ROS response[23]. When the intracellular iron ion concentration exceeds normal levels, NOX4 is activated, producing a large amount of oxygen free radicals, which combine with iron ions to form highly reactive hydrogen and oxygen free radicals. These radicals then destroy proteins, DNA, and lipid molecular structures in cells. Hydroxyl radical activity causes the release of intracellular iron ions, intensifying the ROS response[23]. Previous studies have shown that it interferes with cancer cell proliferation, apoptosis, and cell cycle progression in conditions such as brain glioma, non-small cell lung cancer, and melanoma[24-26]. Therefore, NOX4 stimulates tumor development through various channels and processes and is thus considered a potential therapeutic target[27]. NOX4 overexpression accelerates cancer development and is correlated with poor prognosis in patients with colorectal cancer[28]. Additionally, the NOX4 gene was deleted in liver cancer patients, demonstrating that it may exert tumor-suppressor effects by facilitating TGF-β1-induced apoptosis[29]. Although the mechanism through which NOX4 promotes GC development and invasion is unknown, studies have revealed that NOX4 may stimulate GC cell proliferation and invasion through epithelial-mesenchymal transition[30,31]. Moreover, Li et al[32] showed that the new biotin derivative XN4 could induce ferroptosis in GC cells by upregulating NOX4 expression[32].

MTCH1 encodes a protein with two proapoptotic domains that localize to the inner mitochondrial membrane, causing apoptosis independently of Bax and Bak[33]. MTCH1 regulates cellular iron levels, maintaining iron within normal limits. When cellular iron levels exceed normal levels, MTCH1 is activated, causing mitochondrial outer membrane permeabilization (MOMP) and the release of mitochondrial molecules into the cytoplasm[34]. MOMP releases iron ions from mitochondria into the cytoplasm, increasing cellular iron. The increased iron can interact with hydroxide ions to produce highly reactive hydroxide radicals. These free radicals can cause cell death by damaging proteins, DNA, and lipids[35]. Although bioinformatics analysis and functional validation suggest that increased MTCH1 expression is correlated with metastasis, cancer stage, and poor survival in liver hepatocellular carcinoma patients[36], no study has validated MTCH1 expression in GC.

SLC2A3, also known as GLUT3, is a transmembrane transporter with high affinity for glucose. It can regulate intracellular glucose metabolism and promote ferroptosis, but little to no expression has been detected in normal tissues[37]. SLC2A3 is activated when the intracellular glucose concentration increases, promoting NOX4 activation, increasing oxygen free radical production, and inducing ferroptosis[38]. Tumor cells enhance glucose consumption through aerobic glycolysis for proliferation. Hence, SLC2A3 is strongly overexpressed in various tumor types[39,40]. SLC2A1 overexpression may increase colorectal cancer cell proliferation and invasion by boosting nucleotide synthesis[41]. Glycolysis control in GC cells promotes macrophage infiltration and tumor development[42,43]. Therefore, SLC2A3 may be a predictive biomarker in these cancers.

GABARAPL2, also known as GATE-16 (16 kDa enhancer of Golgi-associated ATPase), is a member of the autophagy-associated protein 8 subfamily and is involved in the degradation of autophagosome membranes. It promotes intracellular ferroptosis[44]. GABARAPL2 forms a complex with LC3 to operate at various autophagosome phases; the LC3 subfamily promotes phagocyte membrane fusion, while the GABARAP subfamily functions downstream[45]. GABARAPL2 can promote MOMP, causing mitochondrial molecules to escape into the cytoplasm. Under MOMP, iron ions in mitochondria are released into the cytoplasm, increasing cellular iron and inducing ferroptosis[46]. High GABARAPL2 mRNA expression is associated with poorer prognosis in distinct malignancies, such as worse OS in patients with esophageal cancer or GC[47,48] but better OS in patients with kidney cancer[49]. This discrepancy primarily results from the autophagy-dependent and autophagy-independent functions of GABARAPL2.

We utilized multivariate analysis to create a prognostic model and then assessed the accuracy of the individual patient predictions. The prognostic nomogram and calibration curve findings were consistent with the actual results based on the 1-year, 3-year, and 5-year survival curves. Similar curves indicate that the model can accurately predict OS. However, the novel prognostic nomogram has the potential to increase the prediction accuracy and assist in predicting individual patient survival risk, with significant therapeutic implications[50]. According to the FRGPI, which was developed using hub gene characteristics, GC patients can be classified into low-risk and high-risk groups. The low-risk group had a better prognosis and longer OS than did the high-risk group. To further investigate the underlying mechanisms of the hub genes, we determined that they were significantly related to infiltrating immune cells. Moreover, the FRGPI was associated with the tumor microenvironment and immunotherapy response. There were substantial differences in immune cell infiltration between the FRGPI low-risk and high-risk patients. In the high-risk group, DCs, eosinophils, iDCs, macrophages, and neutrophils exhibited considerable infiltration, while in the low-risk group, NK cells, Tem cells, and Th1 cells exhibited substantial infiltration. NK cells inhibit tumor development and promote CD4+ Th1 polarization. Greater increases in the NK cell percentage correlate with prolonged survival in GC patients, suggesting that NK cells may be an independent prognostic biomarker[51]. Central memory T cells activated by tumor antigens create antitumor Tem cells, which destroy cancer cells and bolster immunity[52]. An imbalance in Th1 and Th2 cells is considered significant in GC development[53]. A shift from Th1 to Th2 cytokines can create an immunosuppressive state, compromising antitumor immunity[54]. Additionally, we observed significant differences in patient characteristics and ICPs between the high-risk and low-risk groups. The expression of genes that encode most ICPs was significantly upregulated in the high-risk group, implying a better immunotherapy response in low-risk patients. Previous research links ferroptosis and tumor immunotherapy. Tumor cell ferroptosis sensitivity correlates with immunological function, explaining the effects of immunotherapy on tumor cells[55]. Cytokines, especially interferon-γ, generated by immunotherapy-activated CD8+ T lymphocytes downregulate SLC3A2 and SLC7A11 expression and inhibit tumor cell cystine uptake[43]. Inflammatory tumors have high immunogenicity, preserved antigen presentation, and intense CD8+ T-cell infiltration. Ferroptosis-inducing agents can dramatically reduce immune checkpoint inhibitor immunotherapy efficacy by killing invading CD8+ (and CD4+) T cells and impairing naive DC maturation and function[56]. The ferroptosis inducer FIN56 used here blocks glutathione peroxidase 4 (GPX4)[57]. GPX4, the first ferroptosis inhibitor identified, uses GSH to reverse lipid peroxidation and prevent ferroptosis. Moreover, GPX4 protects cells against lipid peroxidation-related death and oxidative stress[58].

To determine whether the identified hub genes are required for ferroptosis in GC, we utilized FIN56 to trigger ferroptosis in two GC cell lines, HGC-27 and MGC-803. CCK-8 assays revealed that FIN56 significantly inhibited the proliferation of HGC-27 and MGC-803 cells, and low concentrations of FIN56 were fatal to GC cell lines. Lipid peroxides, abundant ROS in living organisms, primarily cause ferroptosis[59]. Thus, we detected significantly elevated ROS levels in FIN56-treated GC cells by flow cytometry and fluorescence microscopy, suggesting that FIN56-induced cell death is due to ferroptosis. Furthermore, both the mRNA and protein levels of the hub genes decreased in a dose-dependent manner in FIN56-treated GC cells, demonstrating that FIN56 regulates hub gene expression during ferroptosis. These data suggest that FIN56 is a potential drug for inhibiting GC cell invasion.

In conclusion, this study identified co-DEGs between H. pylori-associated GC and ferroptosis that promote GC formation and progression through various signaling pathways. The FRGPI, which included NOX4, MTCH1, GABARAPL2, and SLC2A3, demonstrated high predictive accuracy.