A viral–host redox axis: EBNA1–FOSL2–ALDH3A1 defines a targetable vulnerability in EBV-positive carcinomas

Abstract

Epstein–Barr virus (EBV)–associated carcinomas exhibit reprogrammed redox metabolism, yet the underlying regulatory network and potential metabolic vulnerabilities remain incompletely defined. Here we identify a viral–host transcriptional axis in which EBV EBNA1 induces the transcription factor FOSL2 to repress ALDH3A1. Restoration of ALDH3A1 in EBV-positive models disrupts NAD(P)H/NAD(P)⁺ homeostasis, inducing reductive stress. This reductive milieu upregulates GSNOR and TrxR1, potentiating the denitrosylation of GSK3β, leading to its stabilization and suppression of the Wnt/β-catenin pathway. We establish that S-nitrosylation at GSK3β Cys199 controls its stability, providing a mechanistic bridge from redox regulation to Wnt inhibition. Critically, ALDH3A1 elevation selectively curbs EBV-positive tumor growth, exploiting an infection-specific vulnerability in redox signaling. Thus, our findings integrate EBV-driven redox remodeling with Wnt/β-catenin signaling activation and propose ALDH3A1 induction as a promising therapeutic strategy for EBV-associated carcinomas.

Graphical abstract

Highlights

• •ALDH3A1 specifically suppresses the proliferation of EBV-positive carcinomas.
• •EBNA1 transcriptionally represses ALDH3A1 via upregulating FOSL2 in EBV-positive carcinomas.
• •In EBV-positive carcinomas, ALDH3A1 induces reductive stress, remodelling cellular denitrosylation.
• •ALDH3A1 enhances GSK3β stability by preventing its nitrosylation at the C199 site, thereby promoting β-catenin degradation.

1. Introduction

The Epstein–Barr virus (EBV) is strongly linked to the vast majority of nasopharyngeal carcinoma (NPC) cases and to approximately 10 % of gastric carcinoma (GC) cases [1]. A major pathogenic mechanism of EBV is thought to involve the reprogramming of redox homeostasis in host cells [[1], [2], [3]]. Redox metabolism, a fundamental component of cellular metabolism, not only provides energy and biosynthetic precursors but also regulates signaling networks through post-translational modifications, such as the oxidation of cysteine residues [4]. Such oxidative modifications, counteracted by antioxidant defense systems, serve as critical molecular signals that govern cellular functions and diverse biological processes [5,6].

While oxidative stress has been widely studied, the role and molecular mechanisms of “reductive stress” — resulting from the accumulation of reducing equivalents like NADH, NADPH, and GSH — remain poorly understood. Reductive stress arises when these reducing equivalents accumulate excessively, creating a highly reduced cellular environment that disrupts mitochondrial function, impairs disulfide bond formation in proteins, and disturbs key metabolic pathways, ultimately leading to suppressed tumor cell growth [[7], [8], [9], [10], [11], [12]]. However, the potential role of reductive stress as a functional state in virus-associated tumors and its interaction with classical oncogenic pathways are still unclear.

Aldehyde dehydrogenase (ALDH) family member ALDH3A1 plays a critical role in maintaining cellular homeostasis by detoxifying toxic aldehydes, such as 4-HNE and acrolein, produced during lipid peroxidation [13]. It has been linked to both antioxidant defense and metabolic changes that promote cancer [[14], [15], [16], [17], [18]]. However, our research suggests that ALDH3A1 behaves differently in EBV-positive tumors. In our models of NPC and GC, EBV infection reduces ALDH3A1 expression. Restoring ALDH3A1 in EBV-positive cells can significantly inhibit the proliferation, whereas ALDH3A1 promotes proliferation in EBV-negative AGS cells. This suggests a virus-dependent shift in its function, raising important questions about how EBV regulates ALDH3A1 and how it affects tumor cell behavior.

In this study, we demonstrate that the EBNA1-FOSL2 axis directly represses ALDH3A1 transcription, establishing a viral-host transcriptional regulatory node. Functionally, high ALDH3A1 expression drives cells toward redox stress by disrupting the NADH/NAD⁺ balance and promotes the de-nitrosylation of GSK3β, enhancing its stability. This suppresses the Wnt/β-catenin signaling pathway. Our findings uncover a specific vulnerability in EBV-positive tumors, revealing how metabolism and signaling pathways interact in cancer.

2. Methods

2.1. Patient cohorts

RNA-seq data of NPC from the GEO database were obtained from the National Center for Biotechnology Information Gene Expression Omnibus (GEO) database (GSE12452, GSE53819, GSE64634, GSE102349) [[19], [20], [21], [22]]. The limma package in R software (version 4.3.1) was used to determine the differential expression of mRNAs. Adjusted P < 0.0001 and log (fold change) > 1 were defined as the thresholds for screening differential expression of mRNAs. A false-discovery rate of<0.05 was set as the cut-off value. Survival analyses of the ALDH3A1 in HNSCC tumor patients were performed using the GEPIA2 database (http://gepia2. cancer-pku. cn) [23]. The cox proportional risk ratios and Kaplan - Meier curves were then obtained from this website.

Single-cell RNA (scRNA) data were obtained from the Genome Sequence Archive of the BIG Data Center at the Beijing Institute of Genomics, Chinese Academy of Science, under accession number HRA000087 (accessible at http://bigd.big.ac.cn/gsa-human). The samples used in this study included primary tumors or lymph node metastases from 7 EBV + NPC patients, and 7 non-cancerous nasopharyngeal samples from normal subjects. Downloaded FASTQ files were subsequently analyzed using the Seurat package (5.0.1) after obtaining downstream analysis files through the CellRanger (2.1.0) process. Cell screening criteria were as follows: nCount_RNA ≥ 500 & nFeature_RNA <5000 & nFeature_RNA ≥ 700 & percent.mt < 0.20 & log10GenesPerUMI >0.8, and 124780 cells were finally retained.

2.2. Gene set enrichment analysis

KEGG pathway enrichment analysis [24], and Gene Set Enrichment Analysis (GSEA) were performed with the R package “clusterProfiler” [25] based on gene sets of MsigDB. GSEA is an algorithm that can be applied for the analysis of functional differences by focusing on gene sets [26]. In the present work, NPC samples were divided into two groups according to the expression level of each gene, namely, the high-expression group and the low-expression group. The number of gene set permutations was set to 1000. The enriched pathways for each phenotype were ranked according to the nominal P value and the normalized enrichment score (NES).

2.3. Cell lines

HONE1, HONE1 EBV, CNE2, CNE2 EBV cells were grown in RPMI-1640 (Gibco BRL) media supplemented with 10 % v/v heat-inactivated fetal bovine serum (Invitrogen, Carlsbad, CA, USA), 1 % w/v glutamine and 1 % w/v antibiotics. AGS and AGS EBV cells were grown in DMEM (Gibco BRL) media supplemented with 10 % v/v heat-inactivated fetal bovine serum (Invitrogen, Carlsbad, CA, USA), 1 % w/v glutamine and 1 % w/v antibiotics. All cancer cell lines were obtained from the Cancer Research Institute of Central South University. All the cell lines involved were cultured at 37 °C in a humidified incubator containing 5 % CO2.

2.4. Detection of protein S-nitrosylation

The S-nitrosylation (SNO) of cellular proteins was detected using the S-Nitrosylated Protein Detection Kit (Biotin Switch, Cayman Chemical, Cat. No. 10006518) according to the manufacturer's instructions [27]. All procedures were meticulously carried out under indirect light conditions to prevent unintended photolysis of S–NO bonds.

Briefly, cells were washed with ice-cold kit Wash Buffer and harvested by centrifugation. The cell pellet was resuspended in five volumes of Blocking Buffer (Buffer A containing the supplied Blocking Reagent, freshly reconstituted and diluted 1:10) and incubated with gentle agitation for 30 min at 4 °C to block free thiols. After blocking, proteins were acetone-precipitated to remove the blocking reagent. The protein pellet was then resuspended in Labeling Buffer (Buffer B containing freshly prepared sodium ascorbate and the maleimide-biotin reagent) and incubated for 1 h at room temperature to selectively reduce S–NO bonds and biotinylate the resulting nascent thiols. Following a second acetone precipitation to remove excess biotinylation reagent, the labeled proteins were solubilized in Wash Buffer, resolved by SDS–PAGE, and transferred to a PVDF membrane. Biotinylated proteins were detected using the kit's HRP-Streptavidin reagent and a chemiluminescent substrate. For all immunoblots, membranes were blocked with 2 % (w/v) bovine serum albumin (BSA) in TBST to avoid background from endogenous biotin present in milk.

2.5. GSK3β-specific S-nitrosylation

To assess GSK3β S-nitrosylation, we performed a biotin-switch assay on total cell lysates followed by streptavidin-based enrichment and immunoblot detection. In brief, after completion of thiol blocking, ascorbate-dependent reduction, and biotinylation (biotin-switch), the resulting biotin-labeled protein samples were incubated with high-capacity streptavidin magnetic beads (Beyotime, Cat. No. P2151) overnight at 4 °C with gentle rotation. Beads were then washed extensively to remove non-specifically bound proteins, and bound proteins were eluted by boiling in 1 × SDS–PAGE loading buffer. Eluates were resolved by SDS–PAGE and immunoblotted with an anti-GSK3β antibody (Proteintech, Cat. No. 82061-1-RR) to detect the biotin-enriched (i.e., S-nitrosylated) fraction of GSK3β.

2.6. Control experiments of protein S-nitrosylation

To validate the specificity of S-nitrosothiol (SNO) detection and to control for potential redox- or handling-dependent artifacts, the following controls were performed in parallel.

Photolysis control: To promote cleavage of S–NO bonds prior to labeling, aliquots of control lysates were subjected to UV photolysis immediately before the thiol-blocking step of the biotin-switch assay. Lysates were placed on ice and exposed to 254 nm UV light for 30 min in a thin layer using UV-compatible vessels at a fixed distance from the UV source. UV-treated lysates were processed identically to non-UV lysates in all subsequent steps.

Light-exposure control: To assess light sensitivity of SNO signals during processing, the entire biotin-switch procedure (including reduction and labeling steps) was intentionally performed under ambient bench-top laboratory lighting without foil/light protection. All reagents, incubation times, and temperatures were otherwise identical to the standard light-protected workflow.

Chemical modulation controls: Unless otherwise specified, cells were incubated with either reduced glutathione (GSH) or S-nitrosoglutathione (GSNO; used here as a transnitrosylating agent) at a final concentration of 200 μM for 30 min. Stock solutions of GSH and GSNO were freshly prepared in sterile ultrapure water and added directly to culture medium (vehicle: sterile ultrapure water at a matched volume). For standard experiments, GSNO-containing solutions were protected from ambient light during preparation, handling, and incubation.

UV-photolyzed GSNO: To generate UV-photolyzed (decomposed) GSNO as a reagent control, GSNO stock solution was pre-irradiated with 254 nm UV light for 30 min to disrupt the S–NO moiety. The UV-photolyzed solution was then applied to cells at the same final concentration (200 μM) for 30 min under light-protected conditions.

2.7. Statistical analysis

Statistical analysis were analyzed using the R software (V.4.3.1, R Core Team, Foundation for Statistical Computing, Vienna, Austria) or GraphPad Prism 9 software (GraphPad Software Inc.). Student's t-test was used for variables that met the requirements for normal distribution. The Wilcoxon signed-rank test was used for continuous variables that did not meet the requirement of normal distribution. Survival curves were estimated with the Kaplan-Meier method and subsequently compared using log-rank tests. P-value was set at p < 0.05 indicates significance. For all analysis, statistically significant differences between groups were determined using two-tailed tests. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001.

3. Result 1. expression of ALDH3A1 in EBV-positive nasopharyngeal carcinoma and its clinical significance

To investigate the molecular mechanisms underlying EBV-mediated tumorigenesis in NPC, we analyzed transcriptomic data from three EBV-positive NPC datasets obtained from the GEO database (GSE12452, GSE53819, and GSE64634). Differential expression analysis of NPC tissue compared to non-cancerous nasopharyngeal tissue identified 198 genes with significant dysregulation across all datasets (Fig. 1A). Among these, ALDH3A1 expression was significantly downregulated in EBV-positive NPC tissues compared to normal controls (Fig. 1B–D).

Fig. 1.

Expression of ALDH3A1 in EBV-positive nasopharyngeal carcinoma and its clinical significance. A. Venn diagram showing the intersection of 2-fold differential expression analysis in NPC samples; B-D. Expression profile of ALDH3A1 in three GEO database NPC datasets; E-F. Single-cell analysis of ALDH3A1 expression levels across various cell types in the tumor microenvironment of NPC tissue and normal control tissue; G. Proportion of cells expressing ALDH3A1 in different cell types within the NPC tumor microenvironment; H. Expression profile of ALDH3A1 in normal epithelial cells versus NPC cancer cells; I-J. Cox regression analysis of ALDH3A1 expression levels in relation to overall survival (I) and disease-free survival (J) in TCGA-HNSC patients.

To explore the expression patterns of ALDH3A1 within the tumor microenvironment (TME), we analyzed scRNA-seq data from EBV-positive NPC samples (Fig. 1E). Single-cell analysis confirmed that ALDH3A1 was primarily expressed in normal epithelial cells, with markedly reduced expression in NPC tumor cells (Fig. 1E–H). Notably, ALDH3A1 expression was highly specific to epithelial cells within the TME, with no detectable expression in immune cells, fibroblasts, or other stromal cell types (Fig. 1F–G).

Given that NPC is a prominent subtype of head and neck squamous cell carcinomas (HNSC), we examined the association between ALDH3A1 expression levels and prognosis in HNSC tissues using data from The Cancer Genome Atlas (TCGA). Survival analysis revealed that patients with high ALDH3A1 expression in HNSC tumors had significantly improved overall survival (OS) and disease-free survival (DFS) compared to those with low ALDH3A1 expression (Fig. 1I and J).

4. Result 2. EBV-induced downregulation of ALDH3A1 expression in cancer

To examine the relationship between EBV infection and ALDH3A1 expression, we compared five pairs of isogenic cell lines with differing EBV status: three NPC (HK1, HONE1, CNE2), one GC (AGS), and one lymphoma (Akata). Quantitative PCR (qPCR) analysis confirmed a corresponding decrease in ALDH3A1 mRNA expression in EBV-positive cells (Fig. 2A). Western blotting showed that ALDH3A1 was undetectable in Akata cells but expressed in the other four lines. Importantly, the protein levels of ALDH3A1 were significantly lower in EBV-positive cell lines than in EBV-negative ones (Fig. 2B, Supplementary Fig. 1A).

Fig. 2.

EBV-induced downregulation of ALDH3A1 expression in cancer. A. qPCR analysis of ALDH3A1 mRNA expression levels in 5 pairs of EBV-negative and EBV-positive cell lines; B. Western blot analysis of ALDH3A1 protein expression levels in 5 pairs of EBV-negative and EBV-positive cell lines; C-D. IHC detection of ALDH3A1 expression in nasopharyngitis and nasopharyngeal carcinomas tissue from clinical patients; E-F. IHC detection of ALDH3A1 expression in cancer and adjacent tissues from EBV-positive gastric cancer patients. ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001; ∗∗∗∗P < 0.0001.

To validate these findings in clinical samples, we analyzed EBV-positive NPC tumors and nasopharyngitis controls using immunohistochemistry (IHC). ALDH3A1 expression was significantly lower in NPC tumor tissues than in non-cancerous nasopharyngeal tissues (Fig. 2C and D), consistent with our in vitro observations. We further analyzed paired tumor and adjacent normal tissues from 10 EBV-positive and 10 EBV-negative gastric cancer patients. IHC revealed distinct patterns depending on EBV status. In EBV-positive cases, ALDH3A1 expression was significantly higher in adjacent normal tissues than in tumors (Fig. 2E and F). In contrast, EBV-negative tumors exhibited the opposite pattern, with higher ALDH3A1 expression in tumor tissues compared to adjacent normal tissues (Supplementary Fig. 1B–C).

5. Result 3. Overexpression of ALDH3A1 inhibits proliferation of EBV-positive tumor cells

To explore the functional role of ALDH3A1 in EBV-positive tumors, we analyzed transcriptomic data from 113 EBV-positive NPC patient samples (GSE102349). Patients were stratified into high and low ALDH3A1 expression groups. Differential expression analysis revealed broad gene expression changes, with KEGG enrichment pointing to strong associations with cell cycle pathways (Supplementary Fig. 2A–B). Gene set enrichment analysis (GSEA) further demonstrated that cell cycle–related genes, particularly those governing the G2/M phase, were enriched in the low ALDH3A1 group and suppressed in the high ALDH3A1 group (Supplementary Fig. 2C–D).

To experimentally validate these findings, we established stable ALDH3A1-overexpressing (ALDH3A1 OE) cell lines in three EBV-positive models (HONE1-EBV, CNE2-EBV, AGS-EBV) (Supplementary Fig. 2E). HK1-EBV cells, which express high endogenous ALDH3A1, were excluded from overexpression studies. For comparison, EBV-negative cell lines (HONE1, AGS) were also overexpressed ALDH3A1 (Supplementary Fig. 2F). No significant increase in cell death was observed relative to empty-vector controls (Supplementary Fig. 2G). Flow cytometry revealed that ALDH3A1 OE induced a G1/S phase arrest in EBV-positive cells, but had minimal impact on EBV-negative HONE1 cells. In EBV-negative AGS cells, overexpression even led to a significant increase in the G2 phase population (Supplementary Fig. 2H).

To evaluate whether ALDH3A1 influences cell proliferation, we conducted CCK-8 viability assays, EdU incorporation assays, and colony formation assays. Overexpression of ALDH3A1 significantly suppressed proliferation in EBV-positive tumor cell lines (HONE1-EBV, CNE2-EBV, and AGS-EBV) (Fig. 3A–C). Consistent with these findings, subcutaneous xenografts in nude mice showed that ALDH3A1 overexpression markedly reduced tumor growth, as evidenced by decreased tumor volume and weight compared with controls (Fig. 3D–G). Immunohistochemical analysis of xenograft tissues further confirmed diminished Ki67 expression, reflecting impaired proliferative activity (Fig. 3H).

Fig. 3.

Overexpression of ALDH3A1 inhibits proliferation of EBV-positive tumor cells. A. CCK-8 assay to assess cell viability in EBV-positive ALDH3A1 overexpression versus control groups; B. EdU assay to evaluate cell proliferation in EBV-positive ALDH3A1 overexpression versus control groups; C. Colony formation assay to assess colony-forming ability in EBV-positive ALDH3A1 overexpression versus control groups; D-G. Subcutaneous tumor formation ability in nude mice injected with HONE1-EBV Vector cells (Control group, transfected with empty vector) or HONE1-EBV ALDH3A1 overexpression cells; H. Immunohistochemistry for ALDH3A1 and Ki67 expression in tumors from nude mice; I. CCK-8 assay to assess cell viability in EBV-negative ALDH3A1 overexpression versus control groups; J. EdU assay to evaluate cell proliferation in EBV-negative ALDH3A1 overexpression versus control groups; K. Colony formation assay to assess colony-forming ability in EBV-negative ALDH3A1 overexpression versus control groups. —, 50 μm; ns, no significance; ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001; ∗∗∗∗P < 0.0001.

In contrast, the role of ALDH3A1 in EBV-negative tumors was more heterogeneous. In HONE1 cells, overexpression modestly inhibited proliferation, although the effect was weaker than in EBV-positive counterparts. Strikingly, in AGS cells, ALDH3A1 overexpression instead enhanced proliferation (Fig. 3I–K). Supporting this, knockdown of ALDH3A1 in EBV-negative AGS cells significantly reduced proliferation (Supplementary Fig. 3A–C). These results are consistent with our IHC observations in EBV-negative gastric cancer patients, indicating that ALDH3A1 functions as a tumor promoter in EBV-negative gastric cancer.

Collectively, these results highlight the dual roles of ALDH3A1 in tumor biology. While previous studies linked ALDH3A1 to tumor promotion through oxidative stress modulation and metabolic reprogramming [28], we observed a paradoxical tumor-suppressive effect in EBV-positive NPC and gastric cancer. This EBV-dependent functional reversal underscores a unique viral-host interaction and prompted us to further dissect the molecular mechanisms underlying ALDH3A1-mediated tumor suppression.

6. Result 4. High expression of ALDH3A1 induces NADH/NAD + imbalance and triggers reductive stress

ALDH3A1 is a metabolic enzyme that catalyzes the conversion of acetaldehyde and NAD⁺ to acetate and NADH [8,18]. Given that NADH and NAD⁺ are critical redox cofactors in cellular metabolism [7], we examined whether ALDH3A1 affects cell proliferation by disrupting NADH/NAD⁺ balance.

Baseline measurements revealed that EBV-negative cell lines generally exhibited higher NADH/NAD⁺ ratios than EBV-positive lines, with significant differences observed in HONE1 and CNE2 cells (Fig. 4A). In contrast, NADPH/NADP⁺ ratios differed significantly only in HONE1 cells (Fig. 4B). Upon ALDH3A1 overexpression in EBV-positive cell lines (HONE1-EBV, CNE2-EBV, AGS-EBV), both NADH/NAD⁺ and NADPH/NADP⁺ ratios were significantly elevated (Fig. 4C and D). Notably, these changes were absent in EBV-negative HONE1 cells overexpressing ALDH3A1 (Fig. 4C and D).

Fig. 4.

High expression of ALDH3A1 induces NADH/NAD + imbalance and triggers reductive stress. A-B. Measurement of NADH/NAD+ (A) and NADPH/NADP+ (B) ratios in EBV-negative and EBV-positive cancer cells; C-D. Measurement of NADH/NAD+ (C) and NADPH/NADP+ (D) ratios in ALDH3A1 overexpressing and control cell lines; E. CCK-8 assay showing changes in cell viability upon addition of NMN (1 mM) for 72 h in ALDH3A1 overexpression versus control cell lines; F-G. EdU assay showing changes in cell proliferation upon addition of NMN (1 mM) for 72 h in ALDH3A1 overexpression versus control cell lines; H. Immunohistochemistry analysis of NRF2 and 8-OHdG levels in tumor tissues from nude mice. —, 50 μm; ns, no significance; ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001; ∗∗∗∗P < 0.0001.

To determine whether restoring NAD⁺ could reverse the effects of ALDH3A1, we treated cells with β-nicotinamide mononucleotide (NMN), a NAD⁺ precursor. NMN effectively normalized the elevated NADH/NAD⁺ ratio in ALDH3A1-overexpressing cells (Fig. 4C). More importantly, NMN treatment significantly alleviated the proliferation inhibition caused by ALDH3A1 overexpression (Fig. 4E–G). These results indicate that ALDH3A1 impairs cell proliferation by inducing NADH/NAD⁺ imbalance, with EBV-positive tumor cells being particularly sensitive to this perturbation.

To assess whether ALDH3A1 overexpression induces reductive stress in vivo, we examined xenograft tumors using immunohistochemistry (IHC) for NRF2, a key regulator of reductive stress, and 8-OHdG, a marker of oxidative stress [[29], [30], [31]]. Compared with controls, ALDH3A1-overexpressing tumors showed significantly increased NRF2 expression and reduced 8-OHdG levels (Fig. 4H). These findings demonstrate that ALDH3A1 overexpression shifts the redox balance toward a reductive state within tumor cells. Collectively, these results show that ALDH3A1 overexpression induces an NADH/NAD⁺ imbalance, leading to reductive stress that suppresses tumor cell proliferation. EBV-positive tumor cells are more vulnerable to this imbalance, underscoring the unique metabolic context imposed by viral infection.

7. Result 5. EBNA1 mediates EBV-driven repression of ALDH3A1 by upregulating FOSL2

Type II latent EBV in NPC and GC encodes three latent proteins, EBNA1, LMP1, and LMP2A [1]. To determine which viral factor represses ALDH3A1, we overexpressed each protein in AGS, HONE1, and CNE2 cells (Fig. 5A). EBNA1, but not LMP1 or LMP2A, significantly reduced ALDH3A1 mRNA in both NPC and GC cell lines (Fig. 5B). Consistently, EBNA1 overexpression decreased ALDH3A1 protein levels (Fig. 5C–F), while siRNA-mediated depletion of EBNA1 restored ALDH3A1 expression (Fig. 5G).

Fig. 5.

EBNA1 mediates EBV-driven repression of ALDH3A1 by upregulating FOSL2. A. qPCR analysis of EBV-encoded protein mRNA overexpression in HONE1 EBV; B. qPCR analysis of changes in ALDH3A1 mRNA expression upon overexpression of EBV-encoded proteins EBNA1, LMP1, and LMP2A in NPC and GC cell lines; C-E. Western blot analysis of changes in ALDH3A1 protein expression following overexpression of EBV-encoded proteins EBNA1, LMP1, and LMP2A in HONE1 (C), CNE2 (D) and AGS (E); F. Western blot analysis of changes in ALDH3A1 protein expression following EBNA1 overexpression; G. Western blot analysis of changes in ALDH3A1 expression following siRNA-mediated knockdown of EBNA1; H. Prediction of upstream regulatory transcription factors for ALDH3A1 from FIMO_JASPAR, ENCODE, GTRD, and CHIP_Atlas databases; I-J. qPCR analysis of changes in transcription factor mRNA levels following overexpression (I) and knockdown (J) of EBNA1; K-M. Western blot analysis of changes in FOSL2 protein expression following EBNA1 overexpression (K–L) and knockdown (M). ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001; ∗∗∗∗P < 0.0001.

Given that EBNA1 is a nuclear protein acting through transcriptional regulation (42), we hypothesized that it represses ALDH3A1 indirectly by activating an inhibitory transcription factor. Using four resources (FIMO_JASPAR, ENCODE, GTRD, ChIP-Atlas), we predicted regulators of the ALDH3A1 promoter, yielding eight candidates (Fig. 5H). qPCR screening identified three EBNA1-induced factors—ATF3, FOSL2, and TEAD4 (Fig. 5I and J). Public data (GSE141385) showed that ATF3 knockdown had no effect on ALDH3A1 expression (Supplementary Fig. 4A). Correlative analysis of GSE102349 singled out FOSL2 as the only factor inversely associated with ALDH3A1 (Supplementary Fig. 4B–F). At the protein level, EBNA1 overexpression increased FOSL2 abundance, while EBNA1 knockdown decreased it in NPC and GC cells (Fig. 5K–M).

FOSL2 overexpression downregulated ALDH3A1 mRNA and protein in both NPC and GC lines (Fig. 6A–C). Conversely, siRNA-mediated knockdown of FOSL2 or treatment with the c-Fos inhibitor T5224 significantly upregulated ALDH3A1 expression (Fig. 6D–F). Dual-luciferase reporter assays confirmed that overexpression of EBNA1 or FOSL2 suppressed ALDH3A1 promoter activity, whereas knockdown of either factor enhanced activity (Fig. 6H and I). Notably, FOSL2 knockdown exerted a stronger effect than EBNA1 knockdown, suggesting that EBNA1 primarily regulates ALDH3A1 via FOSL2. JASPAR analysis predicted three FOSL2 binding motifs in the ALDH3A1 promoter (Fig. 6J). Mutagenesis and reporter assays revealed that mutation of the MT1 site abolished FOSL2-mediated repression of ALDH3A1 promoter activity (Fig. 6K).

Fig. 6.

FOSL2 as an upstream inhibitory transcription factor of ALDH3A1. A-C. qPCR and Western blot analysis of changes in ALDH3A1 mRNA and protein levels following FOSL2 overexpression; D-E. qPCR and Western blot analysis of changes in ALDH3A1 mRNA and protein levels following FOSL2 knockdown; F. Detection of changes in FOSL2 and ALDH3A1 protein levels following treatment with the c-Fos inhibitor T5224; H–I. Dual-luciferase reporter assay to assess the impact of EBNA1 and FOSL2 expression changes on ALDH3A1 promoter activity; J. Mutation of predicted binding sites in the ALDH3A1 promoter region; K. Dual-luciferase reporter assay of ALDH3A1 promoter activity following mutation of predicted binding sites.

8. Result 6. ALDH3A1 inhibits Wnt/β-catenin signaling pathway by upregulating GSK3β

KEGG pathway analysis revealed significant enrichment of Wnt signaling–related genes among patients with low ALDH3A1 expression. Gene set enrichment analysis (GSEA) confirmed that Wnt pathway genes were preferentially enriched in the low ALDH3A1 group (Fig. 7A). These findings suggested that high ALDH3A1 expression may suppress Wnt signaling activity.

Fig. 7.

ALDH3A1 inhibits Wnt/β-catenin signaling pathway by upregulating GSK3β. A. Gene set enrichment analysis (GSEA) showing differences in the enrichment of ALDH3A1-related differentially expressed genes in the WNT signaling pathway; B. Western blot analysis of β-catenin protein expression in three pairs of EBV-negative and EBV-positive cell lines (HONE, CNE2, AGS); C. Western blot analysis of β-catenin protein expression in ALDH3A1 overexpressing versus control cell lines; D. Western blot analysis of β-catenin phosphorylation levels and expression of GSK3β and CK1α in ALDH3A1 overexpressing versus control cell lines; E. Western blot analysis of β-catenin and GSK3β expression levels in the nuclear and cytoplasmic fractions of ALDH3A1 overexpressing versus control cell lines; F. Immunohistochemistry detection of β-catenin and GSK3β expression levels in tumor tissues from nude mice; G-I. Detection of β-catenin and GSK3β stability changes after CHX (100 μg/ml) treatment in ALDH3A1 overexpressing cells; J. Detection of β-catenin and GSK3β expression levels after treatment with the proteasome inhibitor MG132 (10 μM, 12h) in ALDH3A1 overexpressing versus control cell lines; K. Detection of β-catenin and GSK3β expression levels after treatment with the GSK3β inhibitor LiCl (20 mM, 24h) in ALDH3A1 overexpressing versus control cell lines. —, 50 μm; ∗∗∗∗P < 0.0001.

The Wnt/β-catenin pathway is a conserved signaling cascade regulating proliferation, differentiation, apoptosis, migration, invasion, and tissue homeostasis [[32], [33], [34]]. Western blotting revealed elevated β-catenin protein levels in EBV-positive cell lines compared with EBV-negative counterparts, indicating EBV-driven Wnt pathway activation (Fig. 7B). Overexpression of ALDH3A1 significantly reduced β-catenin protein abundance, as confirmed by Western blotting and immunofluorescence (Fig. 7C, Supplementary Fig. 5A). However, qPCR analysis showed no significant changes in CTNNB1 mRNA (Supplementary Fig. 5B), indicating that ALDH3A1 regulates β-catenin post-transcriptionally.

Because β-catenin stability is controlled by the destruction complex [35], we examined its components. Western blotting revealed that ALDH3A1 overexpression increased GSK3β expression and β-catenin phosphorylation, whereas CK1α levels were unchanged (Fig. 7D). Nuclear-cytoplasmic fractionation showed that ALDH3A1 enhanced GSK3β accumulation in both compartments, while β-catenin reduction occurred primarily in the cytoplasm (Fig. 7E), consistent with cytoplasmic destruction complex activity. Immunohistochemistry in xenograft tumors confirmed higher GSK3β and lower β-catenin in ALDH3A1-overexpressing groups (Fig. 7F). Notably, the regulatory effect of ALDH3A1 on the Wnt/β-catenin signaling pathway varied between EBV-negative NPC and GC cells. In HONE1 cells, although GSK3β expression was upregulated, the pathway remained in a relatively low-activation state, thereby limiting the impact of ALDH3A1 on this signaling cascade. By contrast, in AGS cells, over expression of ALDH3A1 enhanced Wnt/β-catenin pathway activity, accompanied by a pronounced reduction in the phosphorylation of both GSK3β and β-catenin (Supplementary Fig. 5C).

Notably, GSK3β mRNA was not altered by ALDH3A1 overexpression (Supplementary Fig. 5D), suggesting regulation at the post-translational level. Cycloheximide (CHX) chase assays demonstrated that β-catenin degraded more rapidly in ALDH3A1-overexpressing cells, whereas GSK3β degradation was slower compared with controls (Fig. 7G–I). These results indicate that ALDH3A1 stabilizes GSK3β protein, thereby promoting β-catenin degradation.

To confirm the degradation pathway, we treated cells with the proteasome inhibitor MG132. MG132 treatment restored β-catenin levels suppressed by ALDH3A1 overexpression (Fig. 7J), demonstrating proteasome-dependent degradation. Furthermore, treatment with the GSK3β-specific inhibitor lithium chloride (LiCl) reversed ALDH3A1-induced β-catenin downregulation (Fig. 7K), confirming that ALDH3A1 suppresses Wnt/β-catenin signaling through GSK3β activity.

9. Result 7. High expression of ALDH3A1 enhances GSK3β stability by reducing its Cys199 nitrosylation

While GSK3β stability is classically regulated by N-terminal serine phosphorylation, notably at Ser9 [35], this modification was unaffected by ALDH3A1 expression (Fig. 8A). We therefore sought an alternative mechanism. Given that ALDH3A1 is cytoplasmic yet increases nuclear and cytoplasmic GSK3β, we reasoned that its effect is indirect and system-wide.

Fig. 8.

High expression of ALDH3A1 enhances GSK3β stability by reducing itS-nitrosylation. A. Western blot analysis of GSK3β phosphorylation levels in ALDH3A1 overexpressing versus control cell lines; B. ALDH3A1 overexpression inhibits overall nitrosylation modification in EBV-positive NPC and GC cells; C. Changes in GSK3β nitrosylation modification levels in EBV-negative, EBV-positive, and ALDH3A1 overexpressing cancer cells; D-F. Changes in GSK3β stability following GSNO (200 μM, 30min) induced nitrosylation modification in three EBV-positive ALDH3A1 overexpressing cell lines.

ALDH3A1 functions as a key antioxidant enzyme that potently scavenges intracellular ROS. ROS act as signaling molecules that regulate protein activity through oxidative modifications of cysteine thiols [4,28]. Importantly, S-nitrosylation—a cysteine thiol modification mediated by reactive nitrogen species (RNS)—has been shown to modulate GSK3β activity in certain contexts [36] and is a recognized regulator of protein function and stability [37,38]. Given the close biochemical and functional interplay between ROS and RNS [39,40], we hypothesized that ALDH3A1 promotes GSK3β stability by regulating its S-nitrosylation.

To test this hypothesis, we measured global protein S-nitrosylation levels (Supplementary Fig. 6A). The results showed that ALDH3A1 overexpression in EBV positive NPC and GC cells reduced global protein S-nitrosylation (Fig. 8B). Importantly, GSK3β was more heavily S-nitrosylation in EBV positive versus EBV negative cells, and this modification was suppressed by ALDH3A1 (Fig. 8C). To test whether S-nitrosylation directly regulates GSK3β stability, we treated cells with the transnitrosylating agent S-nitrosoglutathione (GSNO). At non-cytotoxic doses (Supplementary Fig. 6B), GSNO—but not glutathione (GSH) or its photolysis products—elevated global S-nitrosylation (Supplementary Fig. 6C–D) and reversed ALDH3A1-induced GSK3β denitrosylation (Supplementary Fig. 6E). Crucially, GSNO accelerated GSK3β degradation in cycloheximide chase assays (Fig. 8D–F), confirming that S-nitrosylation destabilizes GSK3β and that ALDH3A1 counteracts this process.

GSK3β contains nine cysteines (Cys14, Cys76, Cys107, Cys178, Cys199, Cys218, Cys245, Cys317, Cys335; UniProtKB P18266) (Supplementary Fig. 6F). Using site-directed mutagenesis (Cys→Ser), we found that C76S, C178S, and C199S markedly reduced GSK3β S-nitrosylation, identifying these residues as major S-nitrosylation sites (Fig. 9A and B). Cycloheximide (CHX) chase assays further showed that, relative to C76S and C178S, the C199S mutation substantially slowed GSK3β degradation (Fig. 9C), implicating Cys199 as a key stability control point.

Fig. 9.

Nitrosylation of GSK3β Cys199 is the main determinant of its stability. A. Western blot analysis of SNO modification levels at cysteine sites in GSK3β mutants in HONE1 EBV cells; B. Western blot analysis of SNO modification levels in GSK3β-WT and single or double Cys site mutants (C76/C178/C199) of GSK3β; C. Changes in protein stability of GSK3β C76/C178/C199 single site mutants; D-F. Changes in GSK3β stability after GSNO (200 μM, 30min) treatment in three EBV-positive ALDH3A1 overexpressing cell lines with GSK3β-WT or GSK3β-C199S mutations.

To test whether S-nitrosylation at Cys199 directly governs GSK3β stability, we introduced the C199S mutation into ALDH3A1-overexpressing cell lines, then elevated cellular S-nitrosylation with GSNO and monitored protein turnover with CHX. Compared to the wild type (WT), C199S effectively resisted GSNO-induced GSK3β degradation (Fig. 9D–F). Collectively, these results indicate that S-nitrosylation at Cys199 is a dominant regulator of GSK3β stability; preventing modification at this site markedly stabilizes the protein and ultimately suppressing Wnt/β-catenin pathway activation.

10. Result 8: ALDH3A1 suppresses protein S-nitrosylation by upregulating GSNOR and TrxR1

To explore how ALDH3A1 modulates protein S-nitrosylation, we examined two key enzymatic systems that shape cellular S-nitrosothiol (SNO) tone: GSNOR and TrxR1 [[40], [41], [42]]. Immunoblotting revealed that GSNOR protein levels were markedly lower in EBV-positive than in EBV-negative cell lines, whereas TrxR1 levels were not significantly changed (Fig. 10A). In EBV-positive cells, ALDH3A1 overexpression increased GSNOR expression across all tested cell lines, and elevated TrxR1 in HONE1-EBV and AGS-EBV cells (Fig. 10B and C), supporting an association between ALDH3A1 and induction of denitrosylation machinery. Notably, NMN supplementation attenuated GSNOR/TrxR1 upregulation in ALDH3A1-overexpressing cells, while exerting cell-line–dependent effects in parental EBV-positive backgrounds (Fig. 10D).

Fig. 10.

ALDH3A1 suppresses protein S-nitrosylation by upregulating GSNOR and TrxR1. A. Western blot analysis of GSNOR and TrxR1 expression in three pairs of EBV-negative and EBV-positive cell lines (HONE, CNE2, AGS); B. Western blot analysis of GSNOR and TrxR1 expression in Vector and ALDH3A1 OE cell lines; C. Immunohistochemical staining of GSNOR and TrxR1 in xenograft tumor sections (scale bar, 50 μm); D. Protein levels of GSNOR, TrxR1, β-catenin, and GSK3β after NMN treatment (1 mM, 72 h); E. Cell viability after treatment with the GSNOR inhibitor N6022 or the TrxR1 inhibitor DVD-445 at indicated concentrations; F-G. Western blots showing β-catenin and GSK3β levels following DVD-445 mediated TrxR1 inhibition (F) or N6022 mediated GSNOR inhibition (G); H. Total cellular S-nitrosylation levels upon treatment with NMN (1 mM, 72 h), DVD-445 (1 μM, 24 h), or N6022 (25 nM, 24 h); I. S-nitrosylation of GSK3β under the same treatment conditions as in (G). ∗∗P < 0.01, ∗∗∗P < 0.001.

To evaluate whether GSNOR and TrxR1 contribute functionally to downstream signaling changes, we used the GSNOR inhibitor N6022 (IC₅₀ = 8 nM) [43] and the TrxR1 inhibitor DVD-445 (IC₅₀ = 0.60 μM) [44]. In viability assays, DVD-445 showed no overt cytotoxicity up to 2 μM, and N6022 was non-cytotoxic up to 125 nM (Fig. 10E). Importantly, pharmacological inhibition of either enzyme decreased GSK3β protein abundance and increased β-catenin levels (Fig. 10F and G), suggesting that these denitrosylation systems are required to maintain the ALDH3A1-dependent state of the GSK3β/β-catenin axis.

Finally, we directly measured S-nitrosylation levels. Treatment with NMN, N6022, or DVD-445 each increased global protein S-nitrosylation and enhanced S-nitrosylation of GSK3β (Fig. 10H and I). Taken together, these results support a model in which ALDH3A1 suppresses GSK3β S-nitrosylation by enhancing GSNOR- and TrxR1-dependent denitrosylation, thereby modulating Wnt/β-catenin signaling in EBV-positive cancer cell lines.

11. Discussion

Redox balance remodeling is a hallmark of cancer [45,46]. EBV infection imparts unique metabolic features to tumor cells, and its viral proteins like EBNA1 and LMP1 can establish a new redox balance by inducing high oxidative stress while enhancing antioxidant pathways [31,[47], [48], [49], [50]]. However, our study reveals a novel mechanism in which EBV suppresses, rather than enhances, antioxidant molecule ALDH3A1, contrasting with the prevailing view that EBV's oncogenic properties primarily counteract oxidative stress by upregulating antioxidant systems.

ALDH3A1 is a well-characterized antioxidant enzyme that is upregulated in a variety of cancers [[14], [15], [16], [17], [18]]. In line with this, we found that in EBV-negative GC, ALDH3A1 expression is elevated and functions as a pro-tumorigenic factor. In striking contrast, in EBV-positive cancer cells, EBV-mediated suppression of ALDH3A1 revealed a tumor-suppressive role, effectively reversing its typical pro-cancer function. Specifically, overexpression of ALDH3A1 created a hyper-reductive environment, marked by imbalances in the NADH/NAD⁺ and NADPH/NADP⁺ ratios, activation of NRF2, and reduced levels of 8-OHdG. This tumor-suppressive effect, observed exclusively in EBV-positive contexts, underscores both the complexity of ALDH3A1's role in cancer and the unique reprogramming of host redox metabolism by EBV.

As a central pathway regulating cellular fate, the Wnt/β-catenin signaling pathway plays a critical role in tumorigenesis by coordinating key biological processes such as cell proliferation, differentiation, and apoptosis [[32], [33], [34]]. Activation of the Wnt pathway has been observed in EBV-associated tumors, though the specific mechanisms driving this activation remain unclear [51]. Notably, the redox system interacts with the Wnt pathway at multiple levels [[51], [52], [53], [54]]. Some studies have reported a negative correlation between ROS and Wnt signaling [55,56], while others suggest that oxidative stress can enhance β-catenin protein stability through H₂O₂ generation mediated by NOXs [52,53]. Additionally, a study by Zhao et al. highlighted a nonlinear regulatory relationship, showing that when ROS levels surpass a critical threshold, Wnt signaling is rapidly inhibited via an unidentified mechanism [54]. These contradictory findings underscore the need for further exploration of the relationship between redox status and Wnt signaling.

In this study, we demonstrate that EBV infection markedly increased β-catenin expression, whereas ALDH3A1 overexpression stabilized GSK3β by preventing itS-nitrosylation, thereby promoting β-catenin degradation. Through site-specific analysis, we identified C76, C178, and C199 as major S-nitrosylation sites on GSK3β and showed that modification at C199 is particularly critical for protein stability. This finding reveals a precise redox-sensitive switch that governs GSK3β regulation, adding a new dimension to redox control in EBV-associated cancers.

Having established that ALDH3A1 regulates GSK3β S-nitrosylation, we next sought to define how it influences the cellular denitrosylation machinery. Cellular S-nitrosothiol (SNO) homeostasis is maintained by a dynamic balance of SNO generation and clearance [57]. Endogenous nitric oxide (NO) produced by nitric oxide synthases (NOS) provides the upstream nitrosative input [58], and under nitrosating/oxidative conditions, NO-derived nitrosating species can convert GSH to S-nitrosoglutathione (GSNO) [59]. GSNO can then mediate transnitrosylation to protein cysteine residues, supporting the propagation of NOS-derived nitrosative signals [57]. To maintain homeostasis, SNO signals are removed primarily by two enzymatic systems: GSNO reductase (GSNOR), which irreversibly reduces GSNO in an NADH-dependent manner to limit transnitrosylating donor availability [60], and the thioredoxin (Trx) system (Trx/TrxR/NADPH), which directly catalyzes protein denitrosylation [61].

Here, we observed that GSNOR protein levels are markedly reduced in EBV-positive cell lines compared with their EBV-negative counterparts, whereas TrxR1 expression remains unchanged. The suppression of GSNOR suggests that EBV-positive context may remold S-nitrosylation homeostasis, potentially leading to the accumulation of S-nitrosylated proteins. In contrast, ALDH3A1-driven reductive stress upregulates the expression of both GSNOR and TrxR1, an effect that is reversed by NMN supplementation, consistent with a shift in NAD(H) redox balance. These findings suggest that ALDH3A1-induced redox remodeling is an important mechanism influencing the expression of these denitrosylation enzymes. Notably, across EBV-positive carcinoma cell models, GSNOR responds to ALDH3A1 overexpression and NMN rescue more consistently than TrxR1, indicating that the GSNOR axis may constitute a comparatively stable and potentially more influential route through which ALDH3A1 reshapes SNO signaling.

Targeting redox homeostasis in tumors is challenging due to metabolic heterogeneity, compensatory pathways, and the potential for off-target toxicity [45,46]. However, EBV infection offers a unique opportunity by defining a specific tumor subtype—EBV-positive carcinomas—that consistently exhibit suppression of ALDH3A1. This shared redox signature represents a common vulnerability that enables more precise therapeutic targeting. In this context, inhibiting FOSL2 to restore ALDH3A1 expression emerges as a promising strategy to enhance treatment selectivity while minimizing toxicity to normal tissues. Nevertheless, a major limitation lies in the pleiotropic nature of FOSL2. As a transcriptional regulator, FOSL2 can both activate and repress diverse downstream genes [62,63], raising the risk of broad off-target effects if directly targeted. Thus, the development of approaches that either activate ALDH3A1 more specifically or modulate its regulatory network represents a critical step forward.

In conclusion, our study identifies a novel mechanism by which EBV reshapes host redox balance to sustain its oncogenic potential through the EBNA1–FOSL2–ALDH3A1 axis. This axis regulates NADH/NAD⁺ redox balance and S-nitrosylation homeostasis. The overexpression of ALDH3A1 in EBV-positive carcinomas causes a pronounced perturbation of this balance, producing potent anti-tumor effects. These findings not only expand the understanding of EBV-mediated redox regulation but also offer a promising avenue for the development of precision metabolic therapies that target the redox vulnerabilities of EBV-positive carcinomas.

Lead contact

Further information and requests for reagents should be directed to the LEAD CONTACT, Xiangjian Luo (luocsu@csu.edu.cn).

Materials availability

All unique/stable reagents generated in this study are available from the lead contact with a completed Materials Transfer Agreement.

Data and code availability

• •All data relevant to the results of this study can be found in the article and its supplementary materials.
• •This paper does not report original code.
• •Any additional information required to reanalyze data reported in this paper is available from the lead contact.

Funding

This work was supported by grants from Fundamental Research Funds for the Central Universities of Central South University (2025ZZTS0140), Hunan Province Graduate Research Innovation Project (CX20250336), Hunan Provincial Natural Science Foundation of China (2024JJ6289), Changsha City Technology Program (kq2403120), the Climb Plan of Hunan Cancer Hospital (QH2023006), and the High-Level Talent Support Program of Hunan Cancer Hospital (20250731-1050).

CRediT authorship contribution statement

Qian Liu: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing – original draft. Binliang Liu: Methodology, Validation, Writing – original draft. Zhenbao Liu: Methodology, Writing – review & editing. Lidan Gong: Writing – review & editing. Min Li: Funding acquisition, Supervision, Writing – review & editing. Xiangjian Luo: Conceptualization, Data curation, Funding acquisition, Project administration, Supervision, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no conflict of interest.

Appendix A. Supplementary data

The following are the Supplementary data to this article: