Oncogenic SOX9-TRAIP signaling drives gastric cancer progression by mediating the degradation of the CPEB3-mTORC1 tumor suppressor axis.

Introduction
Gastric cancer (GC) represents a formidable clinical challenge. According to recent global cancer statistics, GC remains one of the leading causes of cancer-related mortality worldwide [1–3]. Its high mortality rate is largely attributed to the fact that most patients are diagnosed at an advanced stage. This contrasts with regions like Japan, where extensive screening programs facilitate early diagnosis and significantly improve survival outcomes, underscoring the urgent need to develop novel therapies for late-stage disease [4]. GC is characterized by profound molecular heterogeneity, late-stage diagnosis, and a dismal prognosis, underscoring the urgent need for a deeper understanding of its fundamental pathogenic mechanisms [5]. The dysregulation of protein homeostasis, orchestrated by the ubiquitin-proteasome system (UPS), is a hallmark of malignancy [6, 7]. However, the specific E3 ubiquitin ligases and their substrates that function as key oncogenic drivers in GC are not yet fully understood. The specificity of protein degradation is conferred by over 600 E3 ligases, which function as molecular matchmakers, targeting specific proteins for ubiquitylation and destruction [8, 9]. The oncogenic potential of E3 ligases is exemplified by paradigms such as MDM2-mediated degradation of p53 and SKP2-driven destruction of p27, which dismantle tumor-suppressive barriers [10]. Conversely, the loss of tumor-suppressive E3 ligases like VHL or FBXW7 unleashes potent oncoproteins such as HIF-1α and c-Myc [11]. While some E3 ligases have been implicated in GC [12], a systematic elucidation of novel, clinically relevant players is paramount for developing next-generation targeted therapies.
TRAF-interacting protein (TRAIP), a RING-domain E3 ligase, has been identified as an essential regulator of cell division and genome stability, primarily through its role in resolving mitotic stress and facilitating DNA damage repair [13–15]. Its fundamental importance is highlighted by the embryonic lethality of Traip-deficient mice. However, beyond these core homeostatic functions, its potential role as a context-dependent oncogenic driver, particularly in gastrointestinal malignancies, has remained largely unexplored.
The transcriptional programs that drive oncogenesis are frequently orchestrated by reactivated developmental factors. SOX9, a master transcriptional regulator of organogenesis, is a paradigmatic example, having been repurposed as a potent oncogene in numerous solid tumors, including GC, where it governs cancer stem cell properties and therapeutic resistance [16–18]. Concurrently, signal transduction pathways that integrate metabolic and mitogenic cues are almost universally co-opted in cancer. The mTORC1 signaling cascade, a central controller of cell growth and anabolism, is frequently hyperactivated in GC and represents a key vulnerability [19–21]. While established pathways such as PI3K/AKT/mTOR are known drivers of GC, a central challenge is to understand how these distinct layers of regulation—transcriptional, post-translational, and signal transduction—are functionally integrated to create a robust oncogenic network. Our study addresses this gap by delineating a complete, linear axis that functionally connects a master transcription factor (SOX9) with the post-translational machinery (TRAIP) and a core signaling output (mTORC1), thereby providing a more integrated view of GC pathogenesis.
In this study, we addressed these fundamental questions by performing an unbiased screen that identified TRAIP as a previously uncharacterized, prognostically significant oncogene in GC. We hypothesized that TRAIP functions as a critical node linking upstream transcriptional drivers to downstream signaling outputs. Here, we delineate a complete and linear signaling axis, demonstrating that the oncogenic transcription factor SOX9 directly drives the pathological overexpression of TRAIP. We further reveal that the central oncogenic function of TRAIP is to act as the specific E3 ligase for the RNA-binding protein and tumor suppressor, CPEB3. We establish that the TRAIP-mediated degradation of CPEB3 is the pivotal mechanism that unleashes mTORC1 signaling, thereby promoting the aggressive phenotype of gastric cancer. This work not only uncovers a novel oncogenic cascade but also provides a compelling mechanistic rationale for targeting this axis for therapeutic intervention.

Materials and methods

Bioinformatics analysis
Based on the raw RNA sequencing data from the GEO datasets (GSE51575, GSE208099, GSE237002, and GSE49051), differential expression analysis was performed using the DESeq2 package in the R environment (v4.2.3), with significance thresholds set at adjusted p-value < 0.05 and |log₂FoldChange| > 1.0. The expression levels of TRAIP and CPEB3 in STAD were evaluated using the GEPIA2 platform (http://gepia2.cancer-pku.cn/#index) and UALCAN (https://ualcan.path.uab.edu/index.html). For survival analysis, patients were stratified into high- and low-expression groups according to the median expression values, and Kaplan–Meier survival curves were generated using the Kaplan–Meier Plotter platform (https://kmplot.com/analysis/), with statistical significance determined by the log-rank test. Gene set enrichment analysis (GSEA) was conducted using the GSEA software (v4.1.0) developed by the Broad Institute, in combination with the Hallmark gene sets (h.all.v7.4.symbols.gmt) from the MSigDB database. Prediction of transcription factor binding sites was performed using the JASPAR database (https://jaspar.elixir.no/) with a relative score threshold > 0.9. Correlation analysis was conducted based on TCGA-STAD and GTEx-Stomach data.

Molecular docking
The predicted three-dimensional protein structures of human TRAIP (UniProt: Q6PGW2) and CPEB3 (UniProt: Q8NE35) were retrieved from the AlphaFold Protein Structure Database. To model their interaction, the structures were submitted to the HDOCK server, a hybrid docking algorithm that combines template-based modeling with ab initio free docking. The server returned the top 10 predicted complex structures, ranked by docking score. The top-ranked model, with a docking score of -292.22, was selected for detailed analysis. The protein-protein interface was visualized using PyMOL (The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC.). Key intermolecular hydrogen bonds stabilizing the complex were identified, including those between LEU-6, PHE-14, ILE-3, GLU-69, ARG-56, LYS-55 of TRAIP and GLN-640, THR-343, PHE-344, ASN-345, ILE-393, HIS-397 of CPEB3, with bond lengths ranging from 2.4 Å to 3.5 Å.

Cell culture and reagents
Human gastric cancer cell lines (AGS, MKN-45, SGC-7901) and the 293T cell line were procured from the American Type Culture Collection (ATCC). All cell lines were maintained in either RPMI-1640 (Gibco, Cat#11875-093) or DMEM (Gibco, Cat#11965-092), supplemented with 10% fetal bovine serum (FBS, Gibco, Cat#10099-141) and 1% penicillin-streptomycin solution (Gibco, Cat#15140-122), and cultured at 37 °C in a humidified 5% CO₂ incubator. Cell line identity was periodically authenticated via STR profiling, and cultures were routinely screened for mycoplasma contamination using a PCR-based assay. Cycloheximide (CHX, Cat#C7698) and the proteasome inhibitor MG132 (Cat#M7449) were purchased from Sigma-Aldrich and dissolved in DMSO.

Plasmids, siRNAs, and transfection
The full-length human SOX9 open reading frame was PCR-amplified from a commercial cDNA library and subcloned into the pcDNA3.1(+) expression vector. Lentiviral shRNA constructs targeting human TRAIP (TRCN0000233484, TRCN0000233485), CPEB3 (TRCN0000140263), and SOX9 (TRCN0000013899), along with a non-targeting pLKO.1-shRNA control (shControl), were obtained from the Sigma-Aldrich Mission shRNA library. For luciferase reporter assays, a ~ 1.1 kb fragment of the human TRAIP promoter (spanning − 1000 to + 100 relative to the TSS and containing the SOX9 binding site at -340 to -332) was amplified from human genomic DNA (Promega) and inserted into the pGL3-Basic vector. The Q5 Site-Directed Mutagenesis Kit (NEB, Cat#E0554S) was employed to introduce a specific point mutation (CCATTGTTC to CCAAAATTC) into the SOX9 binding motif. All constructs were verified by Sanger sequencing. siRNAs targeting TRAIP and a non-silencing control were synthesized by GenePharma (Shanghai, China). Transfections were performed using Lipofectamine 3000 (Thermo Fisher Scientific, Cat#L3000001) following the manufacturer’s protocol for optimal efficiency.

Lentivirus production and stable cell line generation
For lentivirus production, 293T cells were co-transfected with the respective pLKO.1-shRNA construct, the psPAX2 packaging plasmid, and the pMD2.G envelope plasmid using Lipofectamine 3000 (Thermo Fisher Scientific, Cat#L3000001). Viral supernatants were collected at 48 and 72 h post-transfection, cleared of cellular debris by filtration through a 0.45 μm filter, and concentrated by ultracentrifugation. Target GC cells were transduced with viral particles in the presence of 8 µg/ml polybrene (Sigma-Aldrich, Cat#TR-1003-G). Following a 48-hour incubation, stable transductants were selected and maintained in growth medium supplemented with 2 µg/ml puromycin (Sigma-Aldrich, Cat#P8833).

Quantitative real-time PCR (qRT-PCR)
Total cellular RNA was isolated using TRIzol Reagent (Thermo Fisher Scientific, Cat#15596026). RNA integrity and concentration were assessed using a NanoDrop 2000 spectrophotometer. One microgram of total RNA was reverse-transcribed into cDNA using the FastKing First-Strand cDNA Synthesis Kit with gDNA Removal (Tiangen Biotech, Cat#KR116) according to the manufacturer’s instructions. qRT-PCR reactions were performed in triplicate on a CFX96 Real-Time PCR Detection System with SYBR Premix Ex Taq II (Takara, Cat#RR820A). Gene expression was normalized to the housekeeping gene GAPDH, and relative quantification was calculated using the 2⁻ΔΔCt method. The primer sequences used for qRT-PCR are available upon request from the corresponding author.

Western blotting and antibodies
Cells were harvested and lysed on ice in RIPA buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) freshly supplemented with a protease and phosphatase inhibitor cocktail (Roche, Basel, Switzerland). Protein concentration was quantified via the BCA Protein Assay (Thermo Fisher Scientific). Thirty micrograms of total protein per lane were resolved by SDS-PAGE and transferred to PVDF membranes (Immobilon-P, Millipore, Billerica, MA, USA). Membranes were blocked in 5% non-fat milk in TBST and incubated with primary antibodies overnight at 4 °C. The following primary antibodies were used: anti-TRAIP (1:1000, Abcam, #ab229007), anti-CPEB3 (1:1000, Proteintech, #13459-1-AP), anti-SOX9 (1:1000, Cell Signaling Technology, #82630), anti-pS6K1(T389) (1:1000, CST, #9234), anti-S6K1 (1:1000, CST, #2708), and anti-β-actin (1:5000, Sigma-Aldrich, #A5441). After incubation with appropriate HRP-conjugated secondary antibodies (Jackson ImmunoResearch), chemiluminescent signals were developed using the SuperSignal West Pico PLUS Substrate (Thermo Fisher Scientific) and visualized with a ChemiDoc Imaging System (Bio-Rad).

Chromatin immunoprecipitation (ChIP)-qPCR
ChIP assays were performed using the SimpleChIP Plus Enzymatic Chromatin IP Kit (Cell Signaling Technology, Cat#9005) as per the manufacturer’s protocol. Briefly, approximately 4 × 10⁷ AGS cells were cross-linked with 1% formaldehyde for 10 min at room temperature. Chromatin was digested with micrococcal nuclease to an average fragment size of 150–900 bp. Ten micrograms of sheared chromatin were incubated overnight at 4 °C with 5 µg of anti-SOX9 antibody (Cell Signaling Technology, Cat#82630) or normal rabbit IgG (Cell Signaling Technology, Cat#2729) as a negative control. After elution and reversal of cross-links, the co-precipitated DNA was purified and quantified by qRT-PCR using primers specifically designed to amplify the region containing the predicted SOX9 binding site in the TRAIP promoter. Results were expressed as a percentage of input.

Co-immunoprecipitation (Co-IP) and in vivo ubiquitylation assay
For endogenous Co-immunoprecipitation (Co-IP), AGS cell lysates were pre-cleared and incubated with anti-CPEB3 antibody or IgG overnight at 4 °C. Protein A/G PLUS-Agarose beads (Santa Cruz Biotechnology, Cat#sc-2003) were then added for 4 h. For the ubiquitylation assay, cells were pre-treated with 10 µM MG132 for 6 h. Cell lysates were incubated with Tandem Ubiquitin Binding Entities 2 (TUBE2) agarose resin (LifeSensors, Cat#UM401) for 4 h at 4 °C to enrich for poly-ubiquitylated proteins. Beads were washed extensively, and bound proteins were eluted in SDS sample buffer and analyzed by Western blotting.

GST pulldown assay
The coding sequences for human TRAIP and CPEB3 were cloned into pGEX-4T-1 and pET-28a vectors, respectively. Recombinant GST-TRAIP and His-CPEB3 proteins were expressed in E. coli BL21(DE3) and purified using Glutathione-Sepharose 4B (Cytiva, Cat#17-0756-01) and Ni-NTA agarose (Qiagen, Cat#30210), respectively. For the pulldown assay, 5 µg of GST-TRAIP or GST control protein was immobilized on glutathione beads and incubated with 5 µg of purified His-CPEB3 in binding buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.1% NP-40) for 4 h at 4℃. After extensive washing, bound proteins were analyzed by SDS-PAGE and Western blotting.

Cell-based functional assays
Cell viability was measured using the MTS assay (Promega, Cat#G3582). DNA synthesis was quantified using the BrdU Cell Proliferation ELISA Kit (Cell Signaling Technology, Cat#6813S). For clonogenic assays, 500 cells per well were seeded in 6-well plates and cultured undisturbed for 14 days, after which Colonies were fixed with 4% paraformaldehyde and stained with 0.1% crystal violet. For invasion assays, 5 × 10⁴ cells in serum-free medium were placed in the upper chamber of a 24-well Transwell insert (8 μm pore size, Corning, Cat#3422) coated with a thin layer of Matrigel (BD Biosciences, Cat#356234). The lower chamber was filled with medium containing 10% FBS as a chemoattractant. After 24 h, non-invading cells were scraped from the upper surface, and invaded cells on the lower surface were fixed, stained, and counted.

Subcutaneous xenograft model
All animal procedures were conducted in strict accordance with a protocol approved by the Institutional Animal Care and Use Committee (IACUC) of the First Affiliated Hospital of Gannan Medical University. Four-week-old female BALB/c nude mice were acclimated for one week. Mice were housed in a specific pathogen-free (SPF) facility under controlled conditions (12-hour light/dark cycle, temperature 22 ± 2 °C, humidity 55 ± 5%) with ad libitum access to standard chow and water. Mice were randomly assigned to three groups (n = 6 per group). 5 × 10⁶ AGS cells stably expressing either shControl, shTRAIP-#1 or shTRAIP-#2 were resuspended in 100 µL of a 1:1 mixture of PBS and Matrigel and injected subcutaneously into the right dorsal flank. Tumor dimensions were measured with a digital caliper every 4 days, and tumor volume was calculated using the formula V = (length × width²)/2. After 28 days, mice were euthanized by CO₂ asphyxiation, and tumors were excised, photographed, and weighed.

Statistical analysis
All quantitative data were derived from at least three independent biological replicates and are presented as mean ± standard deviation (SD). Statistical analyses were performed using GraphPad Prism (v10.1.2). Comparisons between two groups were made using a two-tailed, unpaired Student’s t-test. Comparisons among multiple groups were performed using one-way analysis of variance (ANOVA) followed by Tukey’s multiple comparisons test. A p-value less than 0.05 was considered statistically significant. Significance levels are denoted as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Results

Systematic multi-cohort analysis identifies traip as a critically upregulated E3 ligase with prognostic value in gastric cancer
To systematically uncover novel E3 ubiquitin ligases integral to gastric cancer (GC) pathogenesis, we initiated an unbiased transcriptomic screen using two independent, publicly available GEO datasets [22]. Differential expression analysis consistently revealed extensive transcriptional dysregulation in GC tissues relative to matched normal counterparts. Specifically, 779 upregulated genes and 1,388 downregulated genes were identified in the GSE51575 cohort, while 1,131 upregulated genes and 768 downregulated genes were detected in the GSE208099 cohort, confirming robust and reproducible alterations in the GC transcriptome (Fig. 1A and B). To specifically isolate candidate E3 ligases driving this phenotype, we performed an intersectional analysis between the upregulated gene sets from these two cohorts and a curated list of known E3 ligases. This stringent filtering strategy yielded a core set of nine E3 ligases consistently overexpressed across both datasets: KLHL23, CCNF, TRAIP, BRCA1, SKP2, TRIM59, FBXO5, ENC1, and UHRF1 (Fig. 1C). Notably, while several of these candidates are known oncogenic drivers, the functional role of TRAF-interacting protein (TRAIP) in gastric cancer has remained entirely unexplored, positioning it as a high-priority candidate for in-depth investigation. To validate this finding and assess its clinical significance, we interrogated the larger, independent TCGA-STAD (Stomach Adenocarcinoma) cohort. Analysis of 408 tumor (T) and 211 normal (N) samples confirmed that TRAIP mRNA expression is significantly elevated in GC tissues (Fig. 1D). To determine the clinical relevance of TRAIP upregulation, we performed Kaplan-Meier survival analysis. This revealed a significant association between high TRAIP expression and markedly poorer patient outcomes. Patients in the high-TRAIP-expression group exhibited significantly shorter Overall Survival (OS) (Fig. 1E, HR = 1.34 [95% CI: 1.12–1.60], p = 0.0013) and Post-Progression Survival (PPS) (Fig. 1F, HR = 1.36 [95% CI: 1.07–1.74], p = 0.012). Taken together, this multi-pronged bioinformatic and clinical correlation analysis strongly suggests that TRAIP is a significantly upregulated E3 ligase in GC, whose elevated expression serves as a robust and independent predictor of adverse clinical prognosis.

The oncogenic transcription factor SOX9 directly binds and activates TRAIP promoter-driven transcription
To elucidate the upstream regulatory mechanisms responsible for TRAIP’s aberrant expression, we employed a convergent bioinformatic strategy. We intersected the set of 1,131 genes upregulated in the GSE208099 GC cohort with a list of 38 transcription factors predicted by the JASPAR database to bind the TRAIP promoter. This stringent analysis uniquely identified the oncogenic transcription factor SOX9 as the sole candidate at the intersection, implicating it as the primary driver of TRAIP expression in GC (Fig. 2A). To validate this prediction, we first assessed the correlation between SOX9 and TRAIP expression across multiple cohorts. A pan-cancer analysis using TCGA data revealed a significant positive correlation in numerous cancer types, with stomach adenocarcinoma (STAD) exhibiting an exceptionally strong association (Fig. 2B, left panel). A focused analysis of the TCGA-STAD cohort confirmed a highly significant and robust positive correlation between SOX9 and TRAIP mRNA levels (Fig. 2B, right panel, R = 0.427, p < 2.2e-16). This correlation, while still significant, was notably weaker in normal stomach tissue from the GTEx database (Fig. 2C, R = 0.186, p = 0.0139), suggesting a pathologically enhanced regulatory relationship in cancer. To experimentally establish a causal link, we performed gain- and loss-of-function studies. Ectopic expression of SOX9 in GC cells was sufficient to induce a significant, approximately 4-fold upregulation of endogenous TRAIP mRNA expression (Fig. 2D, ****p < 0.0001). Conversely, shRNA-mediated silencing of SOX9 using two independent constructs led to a concomitant and significant reduction in TRAIP transcript levels (Fig. 2E, ***p < 0.001, ****p < 0.0001).

Analysis with the JASPAR database identified a classical SOX9 binding motif (CCATTGTTC) within the TRAIP promoter region (− 340 to − 332 bp). Accordingly, reporter gene constructs harboring either the wild-type (WT) or mutant (Mut) SOX9 binding site sequences were generated for subsequent functional validation (Fig. 2F and G). A dual-luciferase reporter assay demonstrated that SOX9 robustly activated a reporter driven by the wild-type (WT) TRAIP promoter by approximately 5-fold, and this activation was completely abrogated upon site-directed mutagenesis of the SOX9 binding motif (Mut) (Fig. 2H, ****p < 0.0001). As the ultimate confirmation of direct physical interaction in vivo, chromatin immunoprecipitation (ChIP)-qPCR assays demonstrated a significant, ~ 6-fold enrichment of SOX9 occupancy at the endogenous TRAIP promoter region compared to the non-specific IgG control (Fig. 2I, ***p < 0.001). Collectively, these data delineate a precise and hierarchical regulatory axis where the oncogenic transcription factor SOX9 directly binds and activates the TRAIP promoter, thereby driving its aberrant overexpression and contributing to the pathological landscape of gastric cancer.

TRAIP is a potent upstream activator of the mTORC1 signaling axis in gastric cancer
Having established the mechanism of its upregulation, we next sought to define the functional role of TRAIP by investigating the global transcriptomic consequences of its depletion (Dataset GSE237002). This intervention induced a substantial and reproducible transcriptional reprogramming, as visualized by the clear separation of siTRAIP and Control samples in the expression heatmap (Fig. 3A) and the widespread differential gene expression depicted in the volcano plot (Fig. 3B). To uncover the key signaling pathways governed by TRAIP, we performed an unbiased Gene Set Enrichment Analysis (GSEA) [23]. This analysis revealed a significant negative enrichment of the HALLMARK_MTORC1_SIGNALING gene set in TRAIP-depleted cells (Fig. 3C, NES = -1.49, p = 0.0059), indicating a robust transcriptional suppression of this central oncogenic pathway upon loss of TRAIP. To validate and extend this finding in a clinical context, we interrogated the TCGA-STAD patient cohort. Corroborating our in vitro data, GSEA demonstrated a strong positive enrichment of the mTORC1 signaling signature in patients with high TRAIP expression (Fig. 3D, NES = 2.4, p = 0.0013). This was further substantiated by a direct and highly significant positive correlation between TRAIP mRNA levels and the overall mTORC1 signaling pathway activity score (Fig. 3E, R = 0.543, p < 2.2e-16). A pan-cancer analysis confirmed this positive correlation to be a prevalent feature across numerous malignancies, with stomach adenocarcinoma ranking the most exceptionally significant (Fig. 3F), underscoring the generalizability of this regulatory axis. To dissect this relationship at the component level, we examined the expression of core mTORC1 complex constituents. TRAIP expression was found to be significantly and positively correlated with the expression of MTOR itself (Fig. 3G, R = 0.473, p < 2.2e-16), as well as its essential partners RPTOR (Fig. 3H, R = 0.234, p = 1.5e-6) and MLST8 (Fig. 3I, R = 0.259, p = 8.84e-8). Collectively, these multi-layered transcriptomic and correlational analyses provide strong evidence that TRAIP functions as a key upstream activator of the mTORC1 signaling axis in gastric cancer, thereby providing a key mechanistic framework for its pro-tumorigenic function.

TRAIP is functionally indispensable for the malignant progression and in vivo tumorigenicity of gastric cancer
To experimentally validate the oncogenic role of TRAIP predicted by our transcriptomic analyses, we engineered stable TRAIP-knockdown gastric cancer cell lines using lentiviral vectors encoding five distinct shRNAs. Two constructs, designated #1 and #2, demonstrated the most potent and consistent silencing, reducing TRAIP mRNA levels by approximately 75% and 80%, respectively, compared to the pLKO vector control (Fig. 4A, ****p < 0.0001). This robust knockdown was confirmed at the protein level via Western blot, which also revealed a concomitant and marked reduction in the phosphorylation of S6 Kinase 1 (pS6K1), a canonical downstream effector of mTORC1, thereby providing direct biochemical evidence that TRAIP depletion phenocopies mTORC1 inhibition (Fig. 4B). Having validated our models, we systematically assessed the impact of TRAIP loss on key cancer hallmarks. TRAIP depletion significantly impaired cellular proliferation, as quantified by a significant ~ 50% reduction in metabolic activity in an MTS assay (Fig. 4C, **p < 0.01) and a corresponding ~ 40–50% decrease in DNA synthesis, measured by BrdU incorporation (Fig. 4D, **p < 0.01). Furthermore, the long-term clonogenic potential of these cells was significantly attenuated, with TRAIP-knockdown cells forming significantly fewer and smaller colonies than control cells, resulting in a ~ 75% reduction in relative cell numbers (Fig. 4E, ****p < 0.0001). The invasive capacity of GC cells was also largely dependent on TRAIP, as its silencing led to a ~ 70% reduction in the number of cells migrating through a Matrigel-coated Transwell membrane (Fig. 4F, ***p < 0.001). To determine if this essential role in vitro translated to a requirement for tumor growth in vivo, we performed a subcutaneous xenograft study in nude mice. Consistent with our cellular assays, tumors derived from TRAIP-knockdown cells exhibited significantly retarded growth kinetics throughout the 4-week observation period (Fig. 4G). At the experimental endpoint, the tumors formed by TRAIP-depleted cells were visibly smaller (Fig. 4H) and possessed a significantly lower final tumor weight, reduced by over 50% compared to the pLKO control group (Fig. 4I, ****p < 0.0001). Collectively, these comprehensive in vitro and in vivo functional data provide strong evidence that TRAIP is a essential oncogene, essential for sustaining the proliferation, survival, invasion, and ultimate tumorigenicity of gastric cancer cells, acting at least in part through the potentiation of the mTORC1 signaling pathway.

TRAIP functions as a direct E3 ligase for CPEB3, mediating its ubiquitylation and proteasomal degradation
To dissect the molecular conduit through which TRAIP exerts its oncogenic effects, we investigated its interaction with the RNA-binding protein CPEB3. In silico molecular docking predicted a stable, high-affinity binding interface between TRAIP (cyan) and CPEB3 (purple), characterized by multiple hydrogen bonds and hydrophobic interactions within two primary contact regions (Fig. 5A). Key predicted interactions included hydrogen bonds between GLN-640 of CPEB3 and ILE-3 of TRAIP (Region 1), and between ARG-56 of TRAIP and ILE-393 of CPEB3 (Region 2), suggesting a structurally plausible and specific interaction. To validate this predicted interaction under physiological conditions, we performed co-immunoprecipitation (Co-IP) experiments using endogenous proteins from gastric cancer cells. Immunoprecipitation of endogenous CPEB3 specifically co-precipitated endogenous TRAIP, while the IgG control did not, confirming a robust in vivo association (Fig. 5B). To prove that this interaction is direct and not mediated by other cellular components, we performed an in vitro GST-pulldown assay with purified recombinant proteins. As shown, purified GST-tagged TRAIP, but not GST alone, was able to directly capture His-tagged CPEB3 from solution, providing definitive evidence of a direct physical interaction (Fig. 5C). Having established this direct binding, we next investigated its functional consequence. Notably, stable shRNA-mediated depletion of TRAIP (#1) resulted in a substantial, approximately 3-fold accumulation of CPEB3 protein, as determined by Western blot (Fig. 5D and E left panel, **p < 0.01). In stark contrast, CPEB3 mRNA levels remained unchanged upon TRAIP knockdown (Fig. 5E right panel, ns = not significant), strongly indicating that TRAIP regulates CPEB3 at the post-translational level. To test the hypothesis that this regulation occurs via protein degradation, we performed a cycloheximide (CHX) chase assay. In control cells (pLKO), CPEB3 exhibited a rapid turnover with a half-life of approximately 4 h. However, in TRAIP-depleted cells (#1), CPEB3 was profoundly stabilized, with minimal degradation observed even after 8 h of CHX treatment (Fig. 5F and G). Given that TRAIP is an E3 ligase, these data strongly suggested it mediates CPEB3’s ubiquitylation. To directly test this, we performed an in vivo ubiquitylation assay using TUBE2-based pulldown to enrich for poly-ubiquitylated proteins. This experiment revealed a high molecular weight smear of ubiquitylated CPEB3 in control cells, which was markedly diminished upon TRAIP knockdown (Fig. 5H). Taken together, these multilayered biochemical and cellular experiments provide compelling evidence: TRAIP functions as a direct E3 ubiquitin ligase for CPEB3, binding to it, mediating its poly-ubiquitylation, and thereby targeting it for rapid proteasomal degradation. This identifies the TRAIP-CPEB3 axis as a critical post-translational regulatory node in gastric cancer.

CPEB3 functions as a clinically relevant tumor suppressor whose loss is linked to mTORC1 pathway activation in gastric cancer
Having identified CPEB3 as a direct substrate of the oncogenic E3 ligase TRAIP, we next sought to ascertain its intrinsic clinical and functional relevance as a putative tumor suppressor. We first examined CPEB3 expression across large patient cohorts. Analysis of the TCGA-STAD dataset (n = 415 tumors vs. 34 normal) revealed a significant downregulation of CPEB3 mRNA in primary gastric tumors compared to normal tissues (Fig. 6A). This finding was independently and robustly validated in the GSE49051 cohort, which confirmed a profound reduction in CPEB3 expression in GC tissues (Fig. 6B, ****p < 0.0001). Moreover, this downregulation holds significant prognostic power. Kaplan-Meier survival analysis demonstrated that low CPEB3 expression is a powerful predictor of adverse clinical outcomes. Patients with low CPEB3 levels exhibited markedly shorter Overall Survival (OS) (Fig. 6C, HR = 0.64 [95% CI: 0.53–0.76], logrank p = 4.4e-07) and Post-Progression Survival (PPS) (Fig. 6D, HR = 0.60 [95% CI: 0.48–0.74], logrank p = 4.1e-06). To mechanistically link CPEB3 loss to the oncogenic signaling pathways previously identified, we assessed its relationship with the mTORC1 cascade. In the TCGA-STAD cohort, CPEB3 expression exhibited a significant and robust inverse correlation with the HALLMARK_MTORC1_SIGNALING pathway activity score (Fig. 6E, right panel, R = -0.214, p = 1.13e-5). A broader pan-cancer analysis confirmed that this inverse correlation between CPEB3 expression and mTORC1 signaling is a conserved feature across numerous human malignancies, highlighting the fundamental nature of this regulatory axis (Fig. 6E, left panel). Consistent with its inverse correlation with mTORC1—a master negative regulator of autophagy—Gene Set Enrichment Analysis revealed that high CPEB3 expression is, as expected, associated with the activation of the “GOBP_POSITIVE_REGULATION_OF_AUTOPHAGY” gene set (Fig. 6F, NES = 1.23, p = 0.035). Collectively, these data establish CPEB3 as a clinically relevant tumor suppressor in gastric cancer. Its loss, which is characteristic of the disease and strongly predictive of poor prognosis, is mechanistically linked to the disinhibition of the oncogenic mTORC1 signaling pathway.

The pro-tumorigenic functions of traip are predominantly mediated through the targeted degradation of CPEB3
To definitively establish the functional hierarchy and epistatic relationship between TRAIP and its substrate CPEB3, we performed a series of genetic rescue experiments. We engineered four stable cell lines: control (shRNA-control), CPEB3-depleted (shRNA-CPEB3), TRAIP-depleted (shRNA-TRAIP), and double-depleted (shRNA-TRAIP + shRNA-CPEB3). Western blot analysis first confirmed the efficacy and specificity of each knockdown and validated the integrity of the regulatory axis at the protein level. As expected, TRAIP knockdown led to a robust accumulation of CPEB3 protein. This accumulation was completely reversed by the concurrent depletion of CPEB3 in the double-knockdown cells, which restored CPEB3 levels to an undetectable state, thereby perfectly setting the stage for functional interrogation (Fig. 7A).

We next assessed the impact of these genetic perturbations on core malignant phenotypes. In clonogenic assays, TRAIP depletion profoundly suppressed colony formation to less than 25% of control levels. Conversely, depletion of the tumor suppressor CPEB3 enhanced colony formation nearly 3-fold. Remarkably, the concurrent knockdown of CPEB3 in TRAIP-depleted cells led to a significant and substantial rescue of the clonogenic defect, restoring colony numbers to a level significantly greater than that of the TRAIP-knockdown group and not significantly different from the control group (Fig. 7B and C, *p < 0.05, **p < 0.01, ****p < 0.0001). This epistatic relationship was further validated by assessing cell proliferation via BrdU incorporation. TRAIP silencing reduced proliferation by approximately 40%, while CPEB3 silencing increased it by over 50%. The concurrent depletion of CPEB3 in TRAIP-knockdown cells completely rescued the anti-proliferative phenotype, restoring BrdU incorporation to control levels and significantly above that of the TRAIP-knockdown alone (Fig. 7D, **p < 0.01, ***p < 0.001, ****p < 0.0001). Finally, we examined cellular invasion using a Transwell assay. TRAIP knockdown virtually abolished invasive potential, reducing it to ~ 10% of control, whereas CPEB3 knockdown enhanced invasion nearly 4-fold. Once again, the co-depletion of CPEB3 in TRAIP-silenced cells resulted in a potent functional rescue, significantly restoring invasive capacity to a level far exceeding that of TRAIP knockdown alone (Fig. 7E and F, *p < 0.05, ****p < 0.0001). Collectively, these rigorous epistatic rescue experiments provide strong evidence that the pro-tumorigenic functions of TRAIP are predominantly mediated through its targeted degradation of the tumor suppressor CPEB3. This work definitively positions the TRAIP-CPEB3 axis as a critical signaling node that governs gastric cancer progression.

Discussion
The elucidation of precise molecular circuits that drive tumorigenesis is essential for advancing cancer therapy [24]. In this study, we have delineated a novel, hierarchical, and functionally integrated signaling axis that is central to the pathogenesis of gastric cancer. Our findings establish a linear cascade wherein the oncogenic transcription factor SOX9 drives the expression of the E3 ubiquitin ligase TRAIP, which in turn targets the tumor suppressor CPEB3 for proteasomal degradation, ultimately resulting in the hyperactivation of the mTORC1 signaling pathway (Fig. 8). This work provides a comprehensive mechanistic framework that connects a master transcriptional regulator to the post-translational machinery and a core metabolic signaling hub, thereby explaining a significant facet of GC’s malignant behavior.

A key advance of our study is the de-orphanization of TRAIP as a potent and clinically relevant oncogene in gastric cancer. While TRAIP’s fundamental roles in cell cycle and DNA repair are established [25–27], its context-specific functions in tumorigenesis have been nascent. Our systematic, multi-cohort analysis not only identified TRAIP as a robustly overexpressed gene but also linked it directly to poor patient prognosis, positioning it as a high-value biomarker and therapeutic target. The subsequent identification of SOX9 as its direct transcriptional activator provides a crucial etiological link. The SOX9-TRAIP regulatory axis exemplifies how developmental transcription factors, when aberrantly activated in cancer, can co-opt the cellular protein degradation machinery to dismantle tumor-suppressive checkpoints. This finding suggests that a subset of the pleiotropic oncogenic effects attributed to SOX9 may be channeled through its downstream control of the UPS via targets like TRAIP.
The central mechanistic insight of our work is the identification and validation of the TRAIP-CPEB3 regulatory module. We have rigorously established CPEB3 as a bona fide substrate of TRAIP, providing a clear mechanism for the observed downregulation of CPEB3 protein in GC. CPEB3 belongs to a family of RNA-binding proteins that control translation, and its role as a tumor suppressor is increasingly recognized in other malignancies, where it represses the translation of oncogenes such as EGFR, c-Myc, and HIF1α [28–30]. Our data firmly establish CPEB3 as a tumor suppressor in GC, linking its loss to adverse clinical outcomes and, for the first time, providing a definitive mechanism for its post-translational inactivation. The TRAIP-mediated destruction of CPEB3 represents a novel strategy employed by cancer cells to silence a key guardian of protein synthesis.
Furthermore, our study forges a critical link between this ubiquitylation event and the mTORC1 pathway, a central driver of cancer cell growth and metabolism [31]. The inverse correlation between CPEB3 levels and mTORC1 activity, coupled with the profound suppression of mTORC1 signaling upon TRAIP depletion, strongly supports a model where CPEB3 functions as an upstream suppressor of the mTORC1 cascade. The precise mRNA targets of CPEB3 that mediate this effect are a critical area for future investigation. Plausible candidates could include mTOR itself, essential mTORC1 components like RPTOR, or key upstream activators within the PI3K-AKT pathway, all of which are known to be subject to post-transcriptional control [32, 33]. By degrading CPEB3, TRAIP effectively removes a brake on the mTORC1 pathway, which is a master regulator of metabolic reprogramming. This leads to a sustained, pro-anabolic state that fuels tumor growth, providing a novel, non-canonical route to mTORC1 hyperactivation. Our conclusions are further strengthened by the epistatic rescue experiments, which provide strong genetic evidence of the functional hierarchy within this axis. The ability of CPEB3 co-depletion to almost completely reverse the potent anti-tumorigenic effects of TRAIP loss demonstrates that the degradation of CPEB3 is not merely one of many functions of TRAIP, but rather its principal oncogenic mandate in this context. This establishes a clear dependency and highlights the TRAIP-CPEB3 interface as a key vulnerability.

Limitations of the study
Despite the comprehensive nature of our study, several important questions remain that represent exciting avenues for future research. First, while our multi-cohort transcriptomic analysis provides strong evidence for TRAIP upregulation, validating TRAIP protein overexpression in an independent human GC tissue cohort via immunohistochemistry is a critical next step for clinical translation. Second, while we definitively established CPEB3 as a critical TRAIP substrate that suppresses mTORC1, the precise mRNA repertoire controlled by CPEB3 to mediate this effect remains to be fully elucidated. Future studies using techniques like RIP-Seq will be powerful for identifying these direct translational targets. Third, our in vivo work relied on subcutaneous xenograft models. Future studies using more physiologically relevant models, such as genetically engineered mouse models (GEMMs) or patient-derived organoids (PDOs), would provide deeper insights into the function of the SOX9-TRAIP-CPEB3 axis. Finally, a comparative prognostic analysis of this axis against other established molecular markers using multivariate models would be valuable to determine its independent predictive power.
From a therapeutic standpoint, our work provides a strong rationale for targeting this newly identified axis. The development of small molecule inhibitors that specifically disrupt the TRAIP-CPEB3 protein-protein interface represents an attractive, albeit challenging, therapeutic strategy. Such molecules could act as “stabilizers” of CPEB3, restoring its tumor-suppressive function. Alternatively, given that TRAIP is an E3 ligase, developing specific inhibitors of its catalytic activity is another viable approach. The emerging field of Proteolysis-Targeting Chimeras (PROTACs) could also be leveraged to develop molecules that induce the degradation of TRAIP or SOX9 itself [34, 35].
In conclusion, our study has identified and fully delineated a critical oncogenic signaling cascade in gastric cancer, linking a master transcription factor to a specific E3 ligase-substrate pair and a core metabolic pathway. We have established the SOX9-TRAIP-CPEB3-mTORC1 axis as a central driver of GC progression and a key determinant of patient outcome. This work not only deepens our fundamental understanding of gastric cancer biology but also uncovers a set of highly promising, mechanistically defined targets for future therapeutic intervention.