Quantitative proteomics analysis of lactylation modification in gastric adenocarcinoma and adjacent tissues

Protein lactylation modifications were compared between five gastric adenocarcinoma tissues and five adjacent normal tissues (Fig. 1A). Detailed clinicopathological parameters for these patients are provided in Table 1. The cohort included both male and female patients, with ages ranging from 60 to 83 years. Tumor differentiation grades included both well-differentiated (G1) and moderately differentiated (G2) adenocarcinomas, and all cases were classified as stage T2N1M0. This selection reflects a typical spectrum of gastric adenocarcinoma seen in clinical practice, supporting the representativeness of the samples for proteomic analysis. Protein samples were extracted from both tissue types, followed by enzymatic digestion and fractionation. Lactylated proteins were then enriched using immunoaffinity methods. The enriched samples were subjected to quantitative analysis by LC–MS/MS to profile lactylation modifications. The results showed significant differences in lactylation modification sites and protein levels between tumor and normal tissues (Fig. 1B). Specifically, 121 lactylation sites were upregulated and 53 were downregulated in tumor tissues compared to normal tissues. At the protein level, 75 proteins exhibited upregulated lactylation, while 37 proteins showed downregulated lactylation in tumors (Fig. 1B-C, Supplementary Table 1).

Fig. 1

Protein lactylation profiling in gastric adenocarcinoma and prognostic significance of LDHB. (A) Protein lactylation modification analysis workflow for gastric adenocarcinoma (n = 5) and normal tissues (n = 5). After protein extraction from gastric adenocarcinoma and normal tissues, digestion, fractionation, and enrichment were performed. Methods such as immunoaffinity, phosphopeptide, and glycopeptide binding were used for modification identification, followed by LC–MS/MS analysis. (B) Differential analysis of lactylation modifications. Comparison of modification sites and proteins between tumor and normal tissues, with 121 upregulated and 53 downregulated modification sites, and 75 upregulated and 37 downregulated proteins. (C) List of proteins with differential lactylation modifications. (D) Heatmap of gene expression for genes corresponding to differentially lactylated proteins in normal (N = 35) and tumor (N = 415) samples from the TCGA-STAD database (red: high expression, blue: low expression). (E) Genes associated with progression-free interval and disease-specific survival. From 79 DEGs, genes significantly correlated with PFS and DSS (p < 0.1) were selected. Among them, LDHB gene expression was significantly associated with both PFS and DSS. (F-G) Kaplan–Meier (K-M) curve analysis of the difference in progression-free interval (PFI) and disease-specific survival (DSS) between patients with higher and lower LDHB gene expression in TCGA-STAD. (H-I) Kaplan–Meier (K-M) curve analysis of the difference in PFI and DSS between patients with higher and lower LDHB gene expression in Kaplan–Meier plotter. (J-L) Comparison of LDHB mRNA (J) and protein (K-L) expression between tumor and adjacent normal tissues (n = 5 pairs) used in panel A

Table 1 The clinicopathological parameters of patients included in this studyBioinformatics analysis to identify the survival-related candidate genes

Next, we performed differential expression and survival analyses on the 111 protein-coding genes corresponding to proteins with differential lactylation modifications. Gene expression data from 35 normal and 415 tumor samples in the TCGA-STAD database were analyzed. Heatmap and group analyses revealed that 79 out of the 111 genes exhibited differential expression (DEGs) between normal and tumor samples (Fig. 1D-E). We then analyzed the association between the 79 DEGs and survival outcome, including progression-free interval (PFI) and disease-specific survival (DSS) in patients with primary STAD (Fig. 1E). By setting p < 0.1 as the cutoff, we identified the following genes significantly associated with PFI: PDLIM5, LDHB, TF, C3, HP, FGG, FGA, and CRIP2, and with DSS: TF, C3, HP, FGG, FGA, and LDHB. Among them, only the expression of LDHB was significantly correlated with both PFI and DSS (Fig. 1E). We further performed Kaplan–Meier (K-M) survival analysis to evaluate the relationship between LDHB gene expression and survival time in gastric adenocarcinoma patients. Higher expression of LDHB was significantly associated with poorer PFI (Fig. 1F) and DSS (Fig. 1G) in the TCGA-STAD dataset. Similarly, analysis using the Kaplan–Meier Plotter database showed that higher LDHB expression was associated with shorter PFI (Fig. 1H), but this trend was not significant for OS (Fig. 1I).

Univariate analysis indicated that high LDHB expression was significantly associated with shorter PFI (HR = 0.659, p = 0.017) and DSS (HR = 0.630, p = 0.028). Importantly, multivariate analysis confirmed that LDHB expression remained an independent prognostic factor for PFI (HR = 0.609, p = 0.010) after adjusting for confounding factors (Table 2). For DSS, while the trend remained consistent, statistical significance was marginal in the multivariate model (p = 0.080) (Table 3).

Table 2 Univariate and multivariate analysis of PFI in TCGA-STADTable 3 Univariate and multivariate analysis of DSS in TCGA-STAD

Although TF, C3, HP, FGG, and FGA also demonstrated prognostic value in both datasets, we prioritized LDHB for subsequent validation based on its functional relevance to the modification being studied. Unlike the other candidates, which are predominantly secretory proteins involved in systemic immune or coagulation processes, LDHB is the key metabolic enzyme directly controlling the interconversion of pyruvate and lactate conversion (Zdralevic et al. 2018) and some other anti-oxidants metabolism related to cancer behaviors (Deng et al. 2025; Ge et al. 2024). We also analyzed LDHB expression at both the mRNA and protein levels in the primary matched tumor and normal tissues (n = 5) but found no significant difference (Fig. 1J-L).

LDHB is modified by lactylation at K58

Given that LDHB expression levels were similar between tumor and normal tissues, we investigated whether high lactate levels influence LDHB expression. L-lactate treatment did not alter LDHB expression in AGS and MKN-45 cells (Supplementary Fig. 1A-B). To validate whether LDHB is modified by lactylation, cells were transfected with Flag-tagged LDHB constructs and stimulated with 0, 2, or 10 mM L-lactate for 24 h. Co-immunoprecipitation followed by western blot analysis was performed to detect lactylated LDHB (Fig. 2A, B). Flag-LDHB exhibited robust lactylation upon L-lactate stimulation, as indicated by the presence of a Kla (lactylation) signal in the immunoprecipitants (Fig. 2A, B). Quantification of LDHB lactylation levels confirmed a dose-dependent increase in lactylation in both cell lines (Fig. 2C).

Fig. 2

LDHB is modified by lactylation at K58. (A, B) AGS (A) and MKN-45 (B) cells were transfected with Flag-tagged LDHB and treated with 0, 2, or 10 mM L-lactate for 24 h. Whole-cell lysates were immunoprecipitated with anti-Flag antibody, and immunoprecipitates were analyzed by western blotting for lactylation (Kla) and Flag. (C) Quantification of relative LDHB lactylation levels in AGS and MKN-45 cells treated with different concentrations of L-lactate, as indicated. (D, E) AGS (D) and MKN-45 (E) cells were transfected with wild-type Flag-LDHB, Flag-LDHB-K58R, or Flag-LDHB-K156R constructs and treated with 10 mM L-lactate for 24 h. Immunoprecipitation and western blotting were performed as above. (F) Quantification of relative LDHB lactylation levels in AGS and MKN-45 cells expressing wild-type or mutant LDHB constructs

In our lactylation proteomic analysis, only two lysine residues (K58 and K156) were identified as lactylation sites on LDHB (Supplementary Table 1). Both sites showed significant downregulation of lactylation in tumor tissues compared to adjacent normal tissues. To investigate the effects of LDHB lactylation on gastric adenocarcinoma cell behaviors, we generated AGS and MKN-45 cells with stable knockdown of LDHB (shLDHB). These cells were subsequently rescued by lentiviral transduction with shLDHB-resistant LDHB-WT, LDHB-K58R, or LDHB-K156R constructs. Both AGS and MKN-45 cells overexpressing LDHB-WT, LDHB-K58R, or LDHB-K156R showed restoration in LDHB expression comparable to wild-type LDHB, confirming successful rescue (Supplementary Fig. 1C-F).

Upon L-lactate treatment (10 mM), both wild-type Flag-LDHB and the K156R mutant showed strong lactylation signals in AGS and MKN-45 cells (Fig. 2D, E). In contrast, the K58R mutant displayed markedly reduced lactylation, indicating that K58 is a major site for LDHB lactylation (Fig. 2D, E). Quantification of lactylation levels further confirmed that mutation of K58, but not K156, significantly decreased LDHB lactylation in both cell lines (Fig. 2F). These results demonstrate that LDHB is primarily lactylated at lysine 58 in response to L-lactate.

To further explore the regulatory enzymes responsible for LDHB K58 lactylation, we treated AGS and MKN-45 cells expressing Flag-LDHB or Flag-LDHB-K58R with the p300 activator CTB (10 μM) or the HDAC inhibitor SAHA (10 μM) during L-lactate stimulation. Activation of p300 by CTB or inhibition of HDACs by SAHA increased LDHB lactylation levels (Supplementary Fig. 1G-H). These results suggest that p300 and HDACs are involved in regulating the lactylation status of LDHB at K58 in gastric cancer cells.

LDHB-K58 delactylation promotes the invasion of gastric cancer cells

Although total LDHB expression is comparable between tumor and normal tissues, we hypothesized that while its basal expression is essential for cell growth, its specific oncogenic activity is critically modulated by lactylation. In both AGS and MKN-45 cells (Fig. 3A, B), LDHB knockdown significantly impaired cell proliferation rates compared to shNC controls. Restoration of LDHB-WT, LDHB-K58R, or LDHB-K156R all rescued the proliferation rates of the two cell lines (Fig. 3A, B). Notably, the LDHB-K58R group exhibited significantly higher proliferation rates than the LDHB-WT and LDHB-K156R groups (Fig. 3A, B), suggesting that preventing K58 lactylation (delactylation) may more potently drive cell proliferation. In colony formation assays, the shLDHB group showed impaired colony formation compared to shNC cells. Overexpression of LDHB-WT, LDHB-K58R, or LDHB-K156R restored colony formation rates to levels higher than those of shNC (Fig. 3C, D). Among these, LDHB-K58R exerted the strongest effect in promoting colony formation compared to LDHB-WT and LDHB-K156R (Fig. 3C, D).

Fig. 3

LDHB-K58 delactylation promotes the invasion of gastric cancer cells. (A, B) Cell proliferation rates in AGS (A) and MKN-45 (B) cells with stable LDHB knockdown (shLDHB) and rescue by vector (Vec.), wild-type LDHB (WT), LDHB-K58R, or LDHB-K156R were assessed using CCK-8 assays at 0, 24, 48, and 72 h. (C) Representative images of colony formation in AGS and MKN-45 cells with the indicated treatments. (D) Quantification of relative colony formation rates in AGS and MKN-45 cells. (E, F) Representative images (E) and quantification (F) of cell migration in AGS and MKN-45 cells with the indicated treatments. (G, H) Representative images (G) and quantification (H) of cell invasion in AGS and MKN-45 cells with the indicated treatments. (I-L) Representative images (I, K) and quantification (J, L) of cell migration (I-J) and cell invasion (K-L) in AGS and MKN-45 cells with or without GSH-monoethyl ester (GSH-Mee, 2 mM) supplementation. *p < 0.05; **p < 0.01; ***p < 0.001, compared to respective control groups

Transwell assays were conducted to evaluate the effect of LDHB mutants on cell migration and invasion. Both cell lines with LDHB knockdown had fewer migrated and invaded cells compared to the shNC controls (Fig. 3E-H). Cells overexpressing LDHB-K58R exhibited significantly greater migration and invasion abilities compared to shNC cells and cells expressing LDHB-WT or LDHB-K156R (Fig. 3E-H).

Recent studies reported that LDHB might modulate the invasion and migration of some adenocarcinomas via a glutathione (GSH)-dependent manner (Ge et al. 2024; Zhao et al. 2025). Therefore, we further examined whether this represents a mechanism contributing to gastric cancer cell migration and invasion. As expected, supplementation with GSH-monoethyl ester (GSH-Mee, 2 mM) rescued the migration and invasion defects caused by LDHB depletion (Fig. 3I-L). These findings suggest that LDHB-K58 delactylation may promote the migration and invasion of gastric cancer cells by regulating intracellular GSH levels.

LDHB-K58 delactylation does not obviously affect the EMT process of gastric adenocarcinoma cells

To further investigate whether LDHB lactylation influences the epithelial-mesenchymal transition (EMT) in gastric adenocarcinoma cells, we examined the expression of key EMT markers in AGS and MKN-45 cells with stable LDHB knockdown (shLDHB) and subsequent overexpression of LDHB-K58R or LDHB-WT. Immunofluorescence staining revealed that the expression patterns of E-cadherin (an epithelial marker) and N-cadherin (a mesenchymal marker) were not significantly altered in either AGS or MKN-45 cells upon overexpression of LDHB-K58R or LDHB-WT, compared to the shLDHB or shNC controls (Supplementary Fig. 2A-B). Consistently, western blot analysis showed that the protein levels of E-cadherin, N-cadherin, Vimentin, Snail1, and Slug remained unchanged in both AGS and MKN-45 cells expressing LDHB-K58R or LDHB-WT mutants (Supplementary Fig. 2C-D). These results indicate that neither the K58R nor the LDHB-WT has a significant impact on the EMT process in gastric adenocarcinoma cells.

LDHB-K58 delactylation enhances the GSH generation in a SLC7A11-dependent manner

To further assess the impact of LDHB-K58 delactylation on GSH homeostasis, we measured total GSH, total GSSG, and the GSH/GSSG ratio in AGS and MKN-45 cells. LDHB knockdown significantly decreased total GSH levels and the GSH/GSSG ratio. Overexpression of LDHB-K58R, LDHB-K156R, or LDHB-WT markedly elevated both total GSH content and the GSH/GSSG ratio to levels significantly higher than those of controls (Fig. 4A-C). Among these restoration groups, LDHB-K58R exerted stronger effects than the other groups.

Fig. 4

LDHB-K58R promotes migration and invasion of gastric cancer cells via GSH-dependent manner. (A-C) Measurement of total GSH (A), total GSSG (B), and the GSH/GSSG ratio (C) in AGS and MKN-45 cells with stable LDHB knockdown (shLDHB) and overexpression of LDHB-WT, LDHB-K58R, or LDHB-K156R mutants. (D) Schematic diagram showing two possible routes by which LDHB may affect intracellular GSH levels, based on recent publications. (E) Lactate levels in the culture medium of AGS and MKN-45 cells with the indicated treatments. (F) Relative cystine uptake in AGS and MKN-45 cells with the indicated treatments. (G-J) Western blot analysis (G, I) and quantification (H, J) of SLC7A11, STAT1, and GPX4 protein levels in AGS and MKN-45 cells with stable LDHB knockdown (shLDHB) and overexpression of LDHB-WT, LDHB-K58R, or LDHB-K156R mutants. (K-M) Measurement of total GSH (K), total GSSG (L), and the GSH/GSSG ratio (M) in AGS and MKN-45 cells with or without stable LDHB knockdown (shLDHB) and overexpression of LDHB-K58R, after treatment with SLC7A11-IN-2 (20 μM) or erastin (10 μM) for 24 h. **p < 0.01, ***p < 0.001, compared to respective control groups

Recent studies indicated that LDHB might regulate GSH levels in tumor cells by affecting the availability of pyruvate in the tricarboxylic acid (TCA) cycle (Ge et al. 2024) or by reversing STAT1-dependent SLC7A11 suppression and enhancing SLC7A11-dependent GSH metabolism (Zhao et al. 2025) (Fig. 4D). LDHB knockdown might also induce upregulation of GPX4, acting as a compensatory mechanism for the loss of SLC7A11 (Deng et al. 2025). Therefore, we decided to explore the underlying mechanism by which LDHB-K58 delactylation regulates intracellular GSH levels in gastric cancer cells. We measured lactate levels in the culture medium of AGS and MKN-45 cells (Fig. 4E). Neither LDHB knockdown nor re-expression of LDHB-WT, LDHB-K58R, or LDHB-K156R significantly affected lactate release (Fig. 4E), suggesting that the classical lactate-pyruvate interconversion activity of LDHB is largely preserved. This is in line with previous findings that only simultaneous disruption of both LDHA and LDHB is sufficient to ablate glycolytic flux in cancer cells (Zdralevic et al. 2018). Thus, the functional transition observed with LDHB-K58 delactylation appears to be independent of its canonical enzymatic activity. In contrast, LDHB knockdown impaired cystine uptake in tumor cells, while LDHB-K58R overexpression significantly enhanced cystine uptake compared to the other groups (Fig. 4F).

LDHB knockdown or rescue with LDHB-K58R or LDHB-WT did not significantly alter STAT1 mRNA levels in AGS and MKN-45 cells (Supplementary Fig. 3 A, C). However, LDHB-K58R and LDHB-WT overexpression significantly increased SLC7A11 mRNA levels, and this effect was abolished by STAT1 overexpression (Supplementary Fig. 3 A, C), indicating that LDHB regulates SLC7A11 transcription via STAT1. Western blot analysis confirmed robust STAT1 overexpression in both AGS and MKN-45 cells. LDHB-K58R and LDHB-WT overexpression decreased STAT1 protein levels and increased SLC7A11 protein levels, while STAT1 overexpression reversed the effect of LDHB on SLC7A11 (Supplementary Fig. 3B, D). Importantly, treatment with the proteasome inhibitor MG132 restored STAT1 protein levels in LDHB-K58R and LDHB-WT rescue cells (Supplementary Fig. 3E, F). No detectable direct interaction between LDHB (WT or K58R) and STAT1 was observed (Supplementary Fig. 4A-B). Ubiquitination assays demonstrated that both LDHB-WT and LDHB-K58R overexpression markedly increased STAT1 poly-ubiquitination, with the K58R mutant exerting a stronger effect (Supplementary Fig. 4C-D). These findings imply that LDHB might promote STAT1 proteasomal degradation rather than affecting its transcription.

Besides, LDHB knockdown resulted in increased STAT1 and GPX4 expression, and decreased SLC7A11 expression in both AGS and MKN-45 cells (Fig. 4G-J). Overexpression of LDHB-K58R restored SLC7A11 expression and significantly downregulated STAT1 compared to parental cell lines, while LDHB-K156R and LDHB-WT had weaker effects (Fig. 4G-J). LDHB-WT, LDHB-K58R, and LDHB-K156R restored GPX4 expression to an extent similar to the basal level (Fig. 4G-J).

To further validate that the regulatory effect of LDHB-K58R on enhancing intracellular GSH is SLC7A11-dependent, we treated cells with both SLC7A11-IN-2 and erastin, a widely used and well-characterized small-molecule inhibitor of SLC7A11 in ferroptosis research. Treatment with either SLC7A11-IN-2 (20 μM, 24 h) or erastin (10 μM, 24 h) significantly decreased total GSH levels and the GSH/GSSG ratio in both shNC and LDHB-K58R rescue cells compared to their respective controls (Fig. 4K-M). Collectively, these findings imply that the LDHB-K58R mutant enhances GSH synthesis in a SLC7A11-dependent manner, likely by suppressing STAT1 and upregulating SLC7A11.

LDHB-K58 delactylation alleviates RSL3-induced ferroptosis of gastric adenocarcinoma cells

To determine whether LDHB-K58 delactylation modulates ferroptosis sensitivity in gastric adenocarcinoma cells, we evaluated the effects of LDHB-K58R and LDHB-WT mutants on RSL3-induced ferroptosis in AGS and MKN-45 cells. Cell viability assays revealed that LDHB knockdown significantly sensitized both AGS and MKN-45 cells to RSL3, as evidenced by a leftward shift in the dose–response curves and a marked reduction in RSL3 IC50 values (Fig. 5A-D). Notably, re-expression of LDHB-K58R, as well as LDHB-WT, substantially enhanced resistance to RSL3, resulting in IC50 values that exceeded those of control cells, but LDHB-K58R showed stronger effects than LDHB-WT (Fig. 5A-D). Colony formation assays further demonstrated that LDHB knockdown impaired clonogenic survival under RSL3 treatment, while LDHB-K58R and LDHB-WT overexpression significantly rescued colony formation in both cell lines. Similarly, LDHB-K58R presented stronger rescuing effects (Fig. 5E-H). Supplementation with GSH-monoethyl ester (GSH-Mee) partially restored colony formation in LDHB-deficient cells, supporting the role of GSH metabolism in ferroptosis resistance (Fig. 5E-H). To assess lipid peroxidation, a hallmark of ferroptosis, we performed C11-BODIPY staining and flow cytometry. LDHB knockdown led to a pronounced increase in lipid ROS accumulation upon RSL3 exposure, whereas LDHB-K58R and LDHB-WT expression markedly reduced lipid ROS levels in both AGS and MKN-45 cells, among which LDHB-K58R showed a greater reduction (Fig. 5I-L).

Fig. 5

LDHB-K58R confers enhanced resistance to RSL3-induced ferroptosis in gastric cancer cells compared to wild-type LDHB. (A, C) Dose–response curves showing cell viability of AGS (A) and MKN-45 (C) cells with stable LDHB knockdown (shLDHB) and overexpression of LDHB-K58R or LDHB-WT following treatment with increasing concentrations of RSL3. (B, D) Quantification of RSL3 IC50 values in AGS (B) and MKN-45 (D) cells for each group. (E, F) Representative images of colony formation assays in AGS (E) and MKN-45 (F) cells treated with RSL3 (0, 0.5, and 1 μM) with or without GSH-monoethyl ester (GSH-Mee) supplementation. (G, H) Quantification of relative colony formation rates in AGS (G) and MKN-45 (H) cells under the indicated conditions. (I, J) Flow cytometry analysis of lipid ROS levels using C11-BODIPY staining in AGS (I) and MKN-45 (J) cells treated with DMSO or RSL3 for 48 h (0.5 μM for AGS cells and 1 μM for MKN-45 cells). (K, L) Quantification of lipid ROS-positive cell percentages in AGS (K) and MKN-45 (L) cells. (M, O) Representative images of transwell migration (M) and invasion (O) assays in AGS and MKN-45 cells with the indicated treatments (with or without RSL3 treatment, 0.5 μM for AGS cells and 1 μM for MKN-45 cells). (N, P) Quantification of relative cell migration (N) and invasion (P) rates in AGS and MKN-45 cells under the indicated conditions. **p < 0.01, ***p < 0.001, compared to respective control groups

Finally, transwell migration and invasion assays revealed that LDHB knockdown significantly suppressed the migratory and invasive capacities of gastric cancer cells, particularly under RSL3 treatment. In contrast, LDHB-K58R and LDHB-WT overexpression preserved both migration and invasion abilities, even in the presence of RSL3, among which LDHB-K58R showed better preservation (Fig. 5M-P). Collectively, these results indicate that LDHB-K58 delactylation confers robust protection against RSL3-induced ferroptosis and maintains the malignant phenotypes of gastric adenocarcinoma cells, highlighting the functional importance of LDHB lactylation status in ferroptosis regulation and tumor progression.

LDHB-K58R promotes lung metastasis and confers resistance to RSL3-induced ferroptosis in vivo

To validate our in vitro findings in a physiologically relevant context, we established a lung metastasis model using tail vein injection of AGS cells in nude mice. Five-week-old male BALB/c nude mice were injected with 1 × 10⁶ AGS cells expressing either shNC, shLDHB, shLDHB + LDHB-K58R, or shLDHB + LDHB-WT constructs. The mice were subsequently treated with either RSL3 (10 mg/kg) or DMSO control via intraperitoneal injection twice weekly for four weeks (Fig. 6A). At the end-point of the animal study, histological examination of lung tissues revealed distinct differences in metastatic burden among the experimental groups. H&E staining demonstrated that mice injected with shLDHB cells exhibited significantly reduced lung metastases compared to the shNC control group, consistent with our in vitro observations of impaired invasive capacity upon LDHB knockdown (Fig. 6B). Re-expression of either LDHB-K58R or LDHB-WT in LDHB-depleted cells substantially restored metastatic potential, with lung tissues showing increased tumor foci formation compared to the shLDHB group (Fig. 6B). Quantitative analysis of metastatic tumor burden confirmed these observations, showing that LDHB knockdown significantly reduced lung metastasis, while overexpression of LDHB-K58R or LDHB-WT rescued this phenotype, with LDHB-K58R exhibiting a stronger effect (Fig. 6C). Importantly, RSL3 treatment effectively suppressed lung metastasis in both shNC and shLDHB groups, but had minimal impact on mice bearing shLDHB + LDHB-K58R or shLDHB + LDHB-WT cells, demonstrating enhanced resistance to ferroptosis-inducing therapy in vivo (Fig. 6C).

Fig. 6

LDHB-K58R promotes lung metastasis and confers resistance to RSL3-induced ferroptosis in vivo. (A) Schematic of the experimental design. Five-week-old male BALB/c nude mice were injected via the tail vein with 1 × 10⁶ AGS cells (shNC, shLDHB, shLDHB + LDHB-K58R, or shLDHB + LDHB-WT). Mice were randomly assigned to receive intraperitoneal injections of RSL3 (10 mg/kg) or DMSO (2%) twice weekly for four weeks. Lung metastasis was monitored by micro-computed tomography (micro-CT) at the indicated time points. (B) Representative hematoxylin and eosin (H&E) staining of lung tissue sections from each group, showing metastatic tumor foci (scale bars: left, 100 μm; right, 50 μm). (C) Quantification of metastatic tumor burden in the lungs of each group (n = 6 per group). (D) Quantification of mean staining intensity of LDHB within metastatic tumors in each group. (E) Representative immunohistochemical staining of lung sections for LDHB and 4-HNE in each group (scale bars: 50 μm). (F) Quantification of mean staining intensity of 4-HNE within metastatic tumors in each group. Data are presented as mean ± SD. **p < 0.01, ***p < 0.001

Immunohistochemical staining for LDHB confirmed successful knockdown or restoration of the respective constructs in the lung metastases, with strong LDHB signal observed in shNC, shLDHB + LDHB-K58R, and shLDHB + LDHB-WT groups, and weak staining in the shLDHB group (Fig. 6D-E). Notably, 4-hydroxynonenal (4-HNE) staining, a marker of lipid peroxidation and ferroptosis, revealed elevated levels in shLDHB tumors, particularly following RSL3 treatment, while shLDHB + LDHB-K58R and shLDHB + LDHB-WT tumors showed markedly reduced 4-HNE accumulation even under RSL3 exposure (Fig. 6E-F). Quantitative analysis further confirmed that LDHB-K58R and LDHB-WT expression effectively suppressed ferroptosis-associated lipid peroxidation in vivo, with LDHB-K58R showing a greater reduction in 4-HNE levels (Fig. 6F).