IGF2BP1 drives GC cell proliferation and metastasis primarily by stabilizing target molecules [such as nuclear factor erythroid 2-related factor 2 (NRF2) and c-MYC] via METTL3-mediated m⁶A modification (36,37). At the noncoding RNA level, it also forms complexes with pro-oncogenic long noncoding RNAs (lncRNAs), such as ABL, GLCC1 and GHET1 to promote GC cell migration and chemotherapy resistance, these complexes act by inhibiting cytochrome c/APAF1 interactions (blocking caspase-9/3 activation), stabilizing c-MYC mRNA or activating the PI3K/AKT pathway (37-41). The core innovative mechanism of IGF2BP1 in GC lies in enhancing NRF2 mRNA stability via the methyltransferase-like 5 (METTL5)/m⁶A/NRF2 axis. This simultaneously inhibits ferroptosis whilst upregulating the mitochondrial complex protein NDUFA4 to activate the glycolysis-oxidation metabolic coupling. This energy metabolism regulatory pattern remains unreported in colorectal cancer (CRC) and breast cancer (BC), representing a unique strategy for GC cells to sustain their malignant phenotype (42-45).

However, functional controversy persists: A previous small-sample study reported that IGF2BP1 exhibits low expression and tumor-suppressive effects (by downregulating MYC) in EBV-positive GC (46), which is in contrast to its pro-oncogenic role in genomically stable subtypes. Furthermore, IGF2BP1 exerts tumor suppression in hepatocellular carcinoma (HCC) by degrading the lncRNA HULC (47), and the molecular mechanism underlying this tissue-specific functional divergence (such as dependence on distinct RBD targets) remains unclear. Clinically, the specific inhibitor AVJ16 can efficiently reduce the RNA-binding activity of IGF2BP1and the viability of GC cell lines with high IGF2BP1 expression, showing potential for targeted therapy (48).

IGF2BP3 also relies on METTL3-mediated m⁶A modification to stabilize downstream targets [Drp1, lactate dehydrogenase A (LDHA), CCND1 and CD44] and promote GC cell proliferation, migration and metabolic reprogramming (49). At the noncoding RNA level, it binds to circular (circ)-ARID1A and circNFATC3 to enhance the stability of SLC7A5, CCND1 and CD44, while circTNPO3 can competitively bind to IGF2BP3 and block its activation of the MYC-SNAIL axis to inhibit GC progression (50-53). Notably, IGF2BP3 is involved in the metabolic-immune microenvironmental regulatory network, where hypoxia-inducible factor-1α (HIF-1α) disrupts mitochondrial dynamics via upregulation of the METTL3/IGF2BP3/Drp1 axis (54). In the tumor microenvironment (TME), IGF2BP3 participates in metabolic-immune crosstalk: It coordinates with HDGF to activate GLUT4 and ENO2/PKM2, induces the Warburg effect via the OIP5-AS1/heterogeneous nuclear ribonucleoprotein (hnRNP) A1 axis (55-57) and forms a complex with HuR to upregulate glutaminase (GLS) for increased nitrogen and energy supply (58); meanwhile, the IGF2BP3/LDHA axis promotes lactate accumulation, which directly inhibits CD8⁺ T-cell function via the NFAT1-IRF1 pathway to weaken antitumor immunity (59,60). To summarize, IGF2BP1/3 drive GC progression via m6A-dependent mRNA stabilization and GC-specific immune-metabolic regulation, differing from their roles in other tumors.

YTHDF1 drives malignant progression and therapeutic resistance in GC through multidimensional mechanisms. Within tumor-autonomous regulation, YTHDF1 promotes c-MYC translation to activate the PI3K/AKT/mammalian target of rapamycin (mTOR) pathway, upregulates FZD7/TCF7 to maintain tumor stemness, induces SNAIL-mediated epithelial-mesenchymal transition (EMT), while simultaneously stabilizing PARP1 mRNA to preserve CD133⁺ stem cell properties and enhances SPHK2/USP14 translation to mediate oxaliplatin resistance (63-68). By contrast, within the tumor immune microenvironment, YTHDF1 exerts a dual regulatory role within the immune microenvironment: Suppressing T-cell activity via m⁶A-dependent programmed death-ligand 1 (PD-L1) translation, whilst simultaneously recruiting mature dendritic cells upon its depletion, upregulates MHC II/IL-12 and activates the JAK/STAT1 pathway to enhance CD4⁺/CD8⁺ T cell infiltration and IFN-γ secretion (22,69-71). This regulatory pattern exhibits a more multifaceted phenotype compared to the unidimensional action solely dependent on the circMAP2K4/miR-139-5p axis observed in hepatocellular carcinoma (72).

Furthermore, a clinical study indicated that elevated YTHDF1 expression is significantly associated with oxaliplatin resistance and poor prognosis in patients with GC. Its inhibitor SAC could reverse drug resistance in MGC-803 cells by inhibiting SPHK2/USP14 translation according to preclinical data (73). Patients with high YTHDF1 expression exhibited higher PD-L1 positivity rates and reduced CD8⁺ T-cell infiltration, suggesting its potential as a predictive biomarker for immunotherapy response. The precision treatment strategy combining SAC with PD-1 antibodies is currently being preclinically validated in GC organoids (69). This suggests that combining SAC with PD-1 antibodies may constitute a precision treatment strategy for YTHDF1-overexpressing GC.

As a core m⁶A effector protein, YTHDF2 exerts diverse functions in GC. It regulates tumor stemness and oxaliplatin resistance via m⁶A-dependent stabilization of one-cut homeobox 2 mRNA, activation of tissue factor pathway inhibitor transcription and degradation of lncRNA AC026691.1, which disrupts Ras-related dexamethasone-induced 1 mRNA stability to drive GC proliferation, migration and M2 macrophage polarization (74,75). For metabolic reprogramming, YTHDF2 synergizes with long intergenic non-coding RNA 00659/alkB homolog 5 (LINC00659/ALKBH5) to stabilize JAK1 mRNA. This activates the JAK/STAT pathway and upregulates the glycolytic enzyme phosphoglycerate kinase; additionally, YTHDF2 cooperates with the fat mass and obesity-associated protein, while IGF2BP1 promotes proliferation in BC via the estrogen receptor signaling pathway by stabilizing PRKAA1 (protein kinase AMP-activated catalytic subunit α1) mRNA, thereby enhancing glycolysis and oxidative stress adaptation (76,77). It also impairs CD8⁺ T-cell activity by JAK1-dependent inhibition of IFN-γ signaling and PD-L1 induction (70), while exerting tumor-suppressive effects via upregulating phosphatase 2 catalytic subunit a (PPP2CA) (blocking cell cycle) and regulating forkhead box C2 (FOXC2) pathway (66,78,79).

However, the bidirectional regulatory function of YTHDF2 remains a subject of core controversy. Within the same GC cell line, the spatiotemporal coexistence of progression-promoting mechanisms (stabilizing JAK1/PRKAA1) and tumor growth-inhibiting mechanisms (upregulating PPP2CA, regulating FOXC2) remains unclarified. The molecular switches governing this functional switch (such as m⁶A modification levels or differential interactions with long non-coding RNAs) remain unidentified. Furthermore, YTHDF2 exhibits tissue-specific functionality; whilst it induces PD-L1 expression via JAK1-dependent pathways in GC, YTHDF2 primarily regulates ARHGEF2 translation in CRC (80). The tissue-specific variations in its immunomodulatory functions remain under-explored, potentially limiting the universality of targeted therapies.

Among hnRNPs, hnRNPA1, which serves a central role in hnRNPA1, promotes GC metastasis by stabilizing WNT1 inducible signaling pathway protein 2 mRNA to activate Wnt/β-catenin signaling, upregulate SNAIL/Vimentin expression and induce EMT (84). Furthermore, hnRNPA1 regulates ferroptosis inhibition and chemoresistance in GC. hnRNPA1 binds downstream arachidonate 15-lipoxygenase and inhibits lipid peroxidation; levels of reactive oxygen species (ROS) increase in metastatic GC cells compared with that in primary tumor cells, ultimately contributing to ferroptosis resistance and chemoresistance (85); it also upregulates stearoyl coenzyme A desaturase 1, which maintains the oncogenicity of GC stem cells (86) and enhances solute carrier family 7 member 11 (SLC7A11) and glutathione peroxidase 4 (GPX4, an antioxidant enzyme) expression, thus inhibiting lipid peroxidation and promoting GC progression (87). Additionally, in the regulation of metabolic reprogramming in GC, the lncRNA OIP5-AS1 drives glycolysis-dependent proliferation and metastasis by inhibiting Trim21-mediated ubiquitination degradation of hnRNPA1, which promotes the expression of LDHA and PKM2 (57).

Notably, cancer-associated fibroblasts (CAFs) serve as a core stromal component within the TME. Through multiple pathways, CAFs interact with hnRNPA1, IGF2BP3, YTHDF2 and other RBPs to form an interactive network that reinforces the malignant phenotype of GC epithelial cells: i) miR-522 from CAF-derived exosomes targets the ubiquitination site of hnRNPA1, inhibiting its degradation to enhance the stabilization of SLC7A11 and GPX4 (87); ii) paracrine TGF-β activates epithelial SMAD2/3 signaling to directly upregulate IGF2BP3 transcription (88,89); and iii) secreted hepatocyte growth factor activates epithelial c-MET signaling, inducing YTHDF2 expression. This protein stabilizes JAK1 mRNA via m⁶A modification, activates the JAK/STAT pathway and upregulates PD-L1. It also forms a positive feedback loop with CAF-secreted IL-6 to suppress CD8⁺ T-cell infiltration (70). Such regulatory circuits involving CAFs-TGF-β-IGF2BP3-epithelial cells not only expand the functional dimensions of RBPs within the microenvironment but also demonstrate their pivotal role as epithelial transcriptome coordinators linking stromal signals to epithelial malignant transformation.

hnRNPK exhibits functional dynamics in GC and its role is regulated by microenvironmental signals and genetic background, such as p53 status. First, hnRNPK is involved in the maintenance of cancer stem cell properties. hnRNPK increases β-catenin stability and synergistically activates Wnt signaling to drive GCSC self-renewal and chemoresistance (90), promotes the expression of the prometastatic isoform CD44v6 and enhances invasiveness and stemness through the modulation of the selective splicing of CD44 precursor mRNAs (91). Binding to the lncRNA ELF3-AS1 activates the signal transducer and STAT3 pathway and induces the release of thrombopoietin, promoting tumor-associated thrombocytosis and the formation of a metastatic microenvironment (92). hnRNPK upregulates key enzymes involved in glycolysis, hexokinase 2 (HK2) and LDHA, which drive abnormal energy metabolism (93). Furthermore, hnRNPK promotes the TPT1-OCT1/CDX2 transcriptional axis and induces gastrointestinal epithelial chemotaxis, a precancerous stage of GC (94). However, the functional controversy surrounding hnRNPK poses challenges for clinical targeted therapies; in p53-mutant gastric carcinomas (GC), hnRNPK exhibits significant oncogenic effects (stabilizing β-catenin and driving glycolysis), whereas in p53-wild-type GC, hnRNPK exerts tumor-suppressing effects by stabilizing p53 (95). Currently, p53 mutations account for 40-60% of clinical gastric carcinoma cases (3). The primary controversy and technical challenge in current research lies in designing conditionally active inhibitors based on p53 status, such as those selectively blocking hnRNPK-β-catenin interactions only in mutant p53 contexts, to avoid compromising its tumor-suppressing function.

A mammalian target of rapamycin complex 1 (mTORC1) pathway-dependent immunosuppressive factor. PUM1 drives tumor progression in GC through multidimensional mechanisms: It activates the DEPTOR-mediated glycolytic pathway; and promotes GC cell development via circGMPS carried by exosomes through the miR-144-3p/PUM1 axis, while simultaneously regulating the NPM3/NPM1 axis to facilitate PD-L1-mediated immune evasion (122-124). Furthermore, PUM1 activates the mTORC1-HIF-1α axis by inhibiting the mRNA translation of the mTORC1 inhibitor DEPTOR. This not only upregulates glycolytic enzymes such as HK2 and LDHA but also directly binds to the PD-L1 promoter to promote its transcription, thereby regulating the GC immune microenvironment (122).

MSI1/2 regulates the GC immunosuppressive microenvironment by stabilizing Notch1 and Wnt3a mRNA: It binds to the 3'-UTR regions of Notch1 and Wnt3a mRNA, prolonging their half-lives (from 8.7 and 7.9 h to 14.2 and 13.5 h, respectively). The activated Notch1 pathway promotes Treg cell differentiation (increasing IL-10 secretion by 2.3-fold), while the Wnt3a pathway upregulates CCL2 to recruit M2 macrophages. Single-cell sequencing demonstrated that in MSI1/2-high GC tissues, the proportion of Treg cells (FOXP3⁺CD4⁺) and M2 macrophages (CD206⁺CD68⁺) was significantly higher compared with that in the low-expression group (Treg cells: 15.7% vs. 6.2%; M2 macrophages: Not specified vs. 8.3%, respectively; P<0.001) (125,126). Clinically, patients with high MSI1/2 expression demonstrated significantly shorter progression-free survival (4.2 months) following anti-PD-1 therapy compared with those with low expression (9.5 months; P=0.008) (127,128).

FMR1 inhibits GC ferroptosis by stabilizing suppressor of cytokine signaling 2 (SOCS2) mRNA. SOCS2 promotes the ubiquitin-mediated degradation of the cystine transporter SLC7A11, thereby reducing glutathione synthesis. FMR1 deficiency is reported to cause a 40% reduction in SLC7A11 expression, significantly increasing GC cell susceptibility to ferroptosis (129). A preclinical study demonstrated that the FMR1 inhibitor Sc1-VHLL, by degrading FMR1, increases CD8⁺ T cell infiltration in GC tumor tissue by 2.4-fold and reduces cell proportions of Tregs by 35%, thereby converting 'cold tumors' to 'hot tumors', which achieved a tumor growth inhibition rate of 58.7% (P<0.01) (130). In summary, oncogenic RBPs promote GC progression by regulating the epithelial-immune-metabolic axis, thereby providing novel targets for GC prevention and treatment (Fig. 1).