Abstract Background Stomach adenocarcinoma (STAD) poses a major public health challenge across various populations, necessitating the construction of robust models for prognostic prediction and effective clinical therapies. Dysregulation of lactylation, a key regulatory mechanism in cell metabolism and gene expression, can either impede or promote tumor growth and metastasis. Methods This study got into the bottom of TCGA-STAD-sourced transcriptome data to profile lactylation-related genes and construct a gene signature through LASSO regression. A nomogram was further created to assess the prognostic performance of this model. Our investigation primarily concentrated on the expression of Dehydrogenase/reductase 7 (DHRS7) in STAD, with the verification of its correlations with clinical characteristics, immune cell infiltration, and cellular signaling pathways. Results DHRS7 expressed lower in STAD tissues, and that modulating DHRS7 levels could either promote or inhibit malignant behaviors associated with STAD. In the later stages of tumor progression, DHRS7 appeared to facilitate tumor growth through mechanisms such as immune evasion and activation of PI3K/AKT/mTOR signaling pathways, ultimately contributing to an unfavorable prognosis. Conclusions DHRS7 has the potential to shift from acting as a tumor suppressor to functioning as an oncogene in modified TMEs, despite its lower expression levels in STAD tissues relative to normal tissues. This transformation accounts for the association between high DHRS7 expression in the later stages of STAD and a negative prognosis. Supplementary Information The online version contains supplementary material available at 10.1186/s12920-025-02244-y. Keywords: Lactylation, STAD, DHRS7, Immune infiltration, Prognosis Introduction Gastric cancer (GC) stands as a highly aggressive and life-threatening malignancy, posing a robust risk to global health and ranking as the fourth most common cause of cancer-related fatalities worldwide [[34]1]. Stomach adenocarcinoma (STAD) dominates within the spectrum of GC, making up roughly 95% of all GC cases [[35]2]. The development of STAD is driven by a combination of diverse factors, such as Helicobacter pylori infection, precancerous conditions, genetic predispositions, etc [[36]3]. STAD is commonly managed through surgical intervention followed by adjuvant therapies clinically [[37]4]. Despite these efforts, the outlook remains grim for patients grappling with recurrent or metastatic STAD, with failure of remarkably boosting the dismal five-year survival rates even employing cutting-edge treatments [[38]5]. Confronted with this dire medical challenge, there’s an urgent need to pioneer novel therapeutic strategies for enhancing patients’ quality of life and survival outcomes. On a brighter note, the pivotal role of genomic networks in the progression of cancer got magnified owing to the breakthroughs in molecular biology, sequencing technologies, and bioinformatics [[39]6]. So far, the vast majority of genomic elements fall short when it comes to meeting the stringent clinical benchmarks for specificity and sensitivity, although certain ones have emerged as potential prognostic markers [[40]7, [41]8]. Consequently, in the fight against this devastating disease, the priority is to hunt for precise and dependable diagnostic and prognostic biomarkers. Lactylation has recently been considered as a major participator in regulating cellular metabolism and gene expression as it has been confirmed to be a new post-translational modification. It involves the addition of lactate, a byproduct of glycolysis, to lysine residues on histone and non-histone proteins. The discovery of lactylation has opened up new possibilities for understanding cellular metabolism-epigenetic regulation interplay, particularly in the context of cancer biology. Lactylation-related genes (LRGs), which encode enzymes and proteins involved in this modification, play pivotal roles in tumorigenesis, tumor progression, and tumor microenvironment (TME) [[42]9, [43]10]. Studies have been published on the dysfunction of several LRGs and related enzymes [e.g., lactate dehydrogenase (LDH), monocarboxylate transporters (MCTs), and acyltransferases] in cancer to mediate tumor progression and metastasis [[44]11–[45]13]. Dehydrogenase/reductase 7 (DHRS7), a member of the short-chain dehydrogenase/reductase superfamily, may alter the landscape of cancer biology obviously. DHRS7 encodes an enzyme that relies on NADPH to facilitate the reduction of carbonyl compounds, contributing to the metabolism of steroids, lipids, and xenobiotics. Its function in maintaining cellular balance and metabolic processes has drawn considerable interest, particularly in the context of cancer [[46]14]. DHRS7 is often misregulated across a range of cancers, such as those of the breast, lung, colon, and liver, as either a tumor suppressor or an oncogene, depending on the specific cancer type and accompanying conditions. In breast cancer, for instance, a lower expression level of DHRS7 hinted a poorer prognosis as well as heightened tumor aggressiveness and metastasis [[47]15]. In lung cancer, DHRS7 exhibited both tumor suppressive and oncogenic roles, varying with the subtype of cancer [[48]16]. In colorectal cancer, DHRS7-deficient tumors presented with an immunosuppressive TME, characterized by increased regulatory T cells and diminished activity of cytotoxic T cells, resulting in accelerated tumor advancement [[49]17]. Moreover, reduced levels of DHRS7 lined to the development, spreading, and unfavorable outcomes of hepatocellular carcinoma, as DHRS7 deficiency might trigger hyperactivation of SREBP1 to worsen lipid accumulation and tumor progression [[50]18]. Nonetheless, we have poor interpretation of DHRS7 in STAD, not to mention its specific roles and mechanisms. On these basis, this study gathered and analyzed all known genes associated with lactate metabolism to examine their expression levels in STAD tissues and adjacent normal tissues. Our goal was to pinpoint specific genes that could serve as biomarkers for predicting patient outcomes. This study also constructed an LASSO-assisted prognostic model characterized by its high accuracy and clarity. Additionally, we conducted survival analysis, explored clinical correlations, assessed immune cell interactions, and performed pathway enrichment analysis on DHRS7, to demonstrate that elevated DHRS7 expression in STAD was associated with a poorer prognosis. Our results may offer important clinical perspectives for STAD patients considering the TME and cellular signaling pathways. Moreover, this study confirmed the distinct expression levels of DHRS7 in the collected cancerous tissues. Cell biology experiments were also utilized to evaluate the influence of DHRS7 on STAD cell proliferation and invasion, as well as its role in M2 macrophage polarization. Ultimately, the impact of DHRS7 on various cellular signaling pathways was validated by Western blot to delve deeper into the potential molecular mechanisms underlying the effect of DHRS7 on STAD progression and prognosis. Materials and methods Data acquisition TCGA was employed to download RNA transcriptome profiles of 375 STAD patients and corresponding clinical information. Profiles of gene expressions were normalised using R “limma” (v3.54.0) based on the scaling method. Based on existing data [[51]19–[52]21], in addition to the 327 identified lactation-associated proteins, the LRGs also include one lactase (EP300) and six histone deacetylases (HDAC1–3 and SIRT1–3). Thus, a total of 334 LRGs were included in this study. (Supplementary Table S1). Identification of differentially expressed genes (DEGs) and prognostic genes The TCGA-sourced mRNA data were extracted using the R “limma” to analyze the expression profiles of DEGs, and to identify statistically significant DEGs based on the criteria of FDR < 0.05 and |log2 FC| ≥ 1. After the merging of clinical information with expressions in R “bioconductor” (v3.16), univariateor multivariate Cox analyses were performed to screen for prognostic genes (p < 0.05). Prognostic LRG signature construction and validation After the generation of gene profiles by R “glmnet” (v4.1.6) and “survival”, univariate and multivariate Cox analyses were performed on the overall survival (OS) of lactylation-associated DEGs. The TCGA-STAD cohorts were employed as the training dataset. Furthermore, based on LASSO regression, this study intended to determine a suitable set of potential genes for prognostic features and to generate the optimal λ (penalty parameter) to determine the gene coefficients in the risk score equation. Corresponding model trained from the TCGA-STAD data was: riskscore = (0.1184) * DHRS7 + (−0.1134) * NOP2 + (0.03) * CRABP2 + (0.0298) * CALD1 + (0.0285) * ALB + (0.177) * RIMS1. Subsequently, High- and Low-risk groups were created from the studied TCGA-STAD cohorts on that basis, followed by the assessment of inter-group difference in OS time by plotting Kaplan-Meier (K-M) curve. For further evaluation of this model, we generated the receiver operating characteristic (ROC) curves in R “timeROC” (v0.4). The analysis methods described above were subsequently applied to the validation set, [53]GSE84437 cohort. Construction of a nomogram A nomogram was created via “pN_stage” and “pM_stage” packages after the identification of potential predictors that would predict OS independently through unifactorial and multifactorial COX regression. Afterwards, the clinical value of the model established above was validated by decision curve analysis (DCA). Identification of correlations between LRG model and immune cell infiltration To evaluate the association between different LRGs and immune cell infiltration in STAD, the proportion of diverse immune cells was computed by four immune infiltration algorithms (EPIC, QUANTISEQ, MCPCOUNTER, and TIMER). Subsequently, the findings were depicted as heatmaps using R “Pheatmap”, followed by the examination of the correlation between individual LRGs and immune infiltrating cells. However, the computational principles underlying these four algorithms are not entirely identical. Among them, the EPIC algorithm is based on RNA-seq data and utilizes reference gene expression profiles to estimate the relative proportions of immune and cancer cells within the tumor microenvironment. The QUANTISEQ algorithm employs a linear least squares regression method and uses specific marker gene sets to quantify the proportions of immune cells. The MCPCOUNTER algorithm estimates the abundance of eight immune cell types and two non-immune cell types based on the average expression levels of marker genes. Finally, TIMER uses a deconvolution method to estimate the abundance of six immune cell types. Collectively, these four algorithms cover a broad range of immune cell types, enabling a comprehensive characterization of the immune composition of the tumor microenvironment and providing validation of our findings from multiple perspectives. All four algorithms have been widely applied and validated in numerous high-quality studies, and their selection aligns with current research standards. Survival analysis and clinical relevance analysis for DHRS7 Sourced from TCGA, STAD and normal tissues were compared in terms of their differential expression of DHRS7. After that, the TCGA-STAD cohort was stratified into subgroups based on the single gene expression of DHRS7 (High and Low), after which the K-M method was adopted for plotting the survival curves of DHRS7-associated STAD patients. Subsequently, the specific correlations of DHRS7 expression with clinical features were investigated using R “ggpubr” (v0.4.0), with result visualization using a box plot. Additionally, we examined the associationof different clinical stages and lymph node metastases with STAD patients’ prognosis. Functional annotation analysis Pathway enrichment targeting DEGs was carried out in Metascape to assess these genes’ potential functions. With GO and KEGG analyses as benchmarks, enrichments were completed in R “clusterProfiler” (3.12.0). Correlation analysis of genes and pathways A valuable tool of R “GSVA” (1.32.0) was also applied to identify biologically relevant patterns and functional differences at the pathway level. In addition to the above assessment, GSVA enabled the conversion of gene-level expression data into scores at the pathway level. Rather than relying on traditional differential gene expression analysis, GSVA would provide a more holistic approach by considering the collective expression behaviors of genes within predefined gene sets or pathways. The association between gene expressions and pathway scores was assessed using R “pSpearman” (4.0.3). Collection of tissue samples from STAD patients From Nov. 2024 to Feb. 2025, 28 pairs of STAD tumorous and non-tumorous samples were harvested from Fujian Provincial Geriatric Hospital. This study was approved by the Ethics Committee of Fujian Provincial Geriatric Hospital (Approval No. 20241003). All participants provided written informed consent in accordance with the Declaration of Helsinki.” Cell culture and transfection The STAD cell lines, HGC-27 and AGS (ATCC, Virginia, USA), were cultured using RPMI 1640 medium (L220KJ, basalmedia, China) and DMEM/Nutrient Mixture F-12 medium (L370KJ, basalmedia, China), respectively. Both media were enriched with 10% FBS (Biological Industries; Cromwell, CT, USA). These cells were kept in a humidified incubator set at 37 °C with an atmosphere of 5%CO[2]. We synthesized DHRS7-specific siRNA (Qingke Bio; Beijing, China; sequence in Supplementary Table 2) to decipher its role. We transfected HGC-27 and AGS cells using Lipofectamine 2000 (Invitrogen; Waltham, MA, USA) as instructed, which were then collected for further experimentation following an additional incubation period of 24 to 48 h. RNA extraction and RT-PCR Total RNA of the aforementioned STAD tumorous and non-tumorous tissues was isolated using TRIzol (AC0101-B; SparkJade, China). PCR reactions were conducted utilizing the 2 × HQ SYBR qPCR Mix (ZF501; ZOMANBIO; Beijing, China) on an Applied Biosystems 7500 Fast Real-Time PCR System (Foster City, CA, USA) The primers used in this study were listed in Supplementary Table 2. IHC staining Tissue sections were prepared from human GC surgical specimens that had been deparaffinized. The samples underwent rehydration through a series of ethanol washes, followed by an hour of blocking. The sections were subsequently treated with 3% hydrogen peroxide at room temperature for 15 min and the next step was incubation throughout the night at 4 °C with the specific primary antibody (details of all antibodies listed in Supplementary Table 2). Detection of horseradish peroxidase-conjugated antibodies was achieved through IHC staining using 3,3′-diaminobenzidine. Analysis of IHC-stained sections was performed using the H-score system. The H-score is calculated as follows: H-score = (percentage of cells with weak intensity × 1) + (percentage of cells with moderate intensity × 2) + (percentage of cells with strong intensity × 3), resulting in a final score ranging from 0 to 300. Cell proliferation For cell viability, PTC cells (1,000 cells/well) were seeded into 96-well plates, and then left to react for 2.5 h with 10 µL of CCK-8 reagent (C6005M; US Everbright; Silicon Valley, CA, USA) at 37 °C in a 5% CO[2] environment. After that, they were processed successively for 0, 24, 48, 72, and 96 h. For the EdU (C6015M; US Everbright) assay, 5 × 104 cells were grown in complete culture medium after seeding into a 24-well plate. The cells were stained and photographed twenty - four hours later. Cell invasion and migration For detecting cell invasion, we employed 8-millimeter micropore inserts placed in 24-well plates for cell culture, and seeded 5 × 10⁴ cells into the upper wells. These upper wells were coated with 50 µl of diluted matrigel and had no FBS in them. For cell migration, we added 10% FBS to the lower wells. After that, we fixed the wells with 4% paraformaldehyde for half an hour and then stained them with 0.25% crystal violet for 20 min. Statistical analysis Using R (v4.0.2) software, this study completed all analyses with the statistical significance labeled by asterisks (ns: not statistically significant, * p < 0.05, ** p < 0.01, *** p < 0.001 and *** p < 0.0001). Results Lactylation-related DEGs in STAD We first obtained TCGA-sourced RNA expression data of 375 STAD tumour tissues and 32 normal gastric epithelial tissues as well as corresponding clinical information. Initially, we screened lactylation-related DEGs, and identified 12 DEGs related to the prognosis of STAD: KRT1, CBR1, ALDOB, AHNAK, CSRP1, DHRS7, FABP5, NEFL, CRABP2, CALD1, RIMS1, and ALB. Both volcano plots and heat maps were drawn to characterize varied expressions of the screened lactate metabolism-related genes (DECGs), Among these genes, ALB was upregulated in tumors, while the other 11 genes were downregulated (Fig. [54]1A, B). Fig. 1. [55]Fig. 1 [56]Open in a new tab Identification of LRGs in STAD. A, B Volcano plot and Heatmap presenting the expression levels of the lactylation-related DEGs TCGA cohort-based prognostic model and validation cohort [57]GSE84437 We determined the risk models for the 16 prognostic genes and obtained a six-gene model (Fig. [58]2A, B). The high-risk and low-risk groups were categorised after the calculation of the risk score for each patient. Furthermore, DHRS7, CRABP2, CALD1, ALB, and RIMS1 were upregulated in the high-risk group, whereas NOP2 was upregulated in the low-risk group (Fig. [59]2C). As expected, the high-risk group had a notably lower survival in relative to the low-risk group (p = 0.00328), as supported by K-M curve in Fig. [60]2D. The AUCs were 0.611 at 1-year, 0.608 at 3-year and 0.604 at 5-year, respectively (Fig. [61]2E). Subsequently, the robustness of the prognostic risk model was validated using the independent [62]GSE84437 validation cohort. Consistent with observations in the TCGA-STAD cohort, patients stratified into the high-risk group exhibited significantly reduced survival probabilities (Figure S2A). The differential expression patterns of the six signature genes in [63]GSE84437 mirrored those observed in TCGA-STAD (Figure [64]S2A), with significantly elevated expression of DHRS7, CRABP2, CALD1, ALB, and RIMS1 in the high-risk group, whereas NOP2 expression was markedly higher in the low-risk group. Furthermore, high-risk patients within the [65]GSE84437 cohort demonstrated significantly poorer prognosis, consistent with the TCGA-STAD findings (Figure S2B). Time-dependent ROC analysis further substantiated the model’s predictive capacity within this validation cohort, yielding AUC values of 0.612, 0.584, and 0.569 for 1-, 3-, and 5-year survival, respectively (Figure S2C). Fig. 2. [66]Fig. 2 [67]Open in a new tab Construction of TCGA cohort-based prognostic signature. A The gene coefficient profiles. B The partial likelihood deviance plotted with log (λ). C Distribution of risk score and survival status with increasing risk score in the TCGA cohort. D K-M survival curves of the STAD OS in the TCGA cohort. E ROC curves showing satisfactory predictive efficacy of the 6-gene prognostic signature in the TCGA cohort Prognostic value of the LRG signature In univariate prognostic analyses, the expression of CALD1, CRABP2, DHRS7 and RIMS1, as well as the pT stage, pN stage, pM stage and pTNM stage were all associated with the hazard ratio (HR) of STAD; moreover, patients with high expression of CALD1, CRABP2, DHRS7 and RIMS1 revealed worse prognostic outcomes (Fig. [68]3A). In contrast, from the multivariate perspective, only DHRS7 expression, pN staging and pM staging were significantly associated with the HR of STAD, showing worse prognosis observed in patients with higher DHRS7 expression, with a HR of 1.29417 (95% CI: 0.99894–1.67664, P value = 0. 005094) (Fig. [69]3B). On the basis of corresponding data of pN staging and pM staging, nomograms were generated to predict the survival of patients at 1, 2, 3, and 5 years (Fig. [70]3C). The TCGA cohort showed that the predictions were more in line with the actual probabilities (Fig. [71]3D). Fig. 3. [72]Fig. 3 [73]Open in a new tab Univariate and multivariate independent prognostic analyses based on the TCGA cohort. A Univariate independent prognostic analysis. B Multivariate independent prognostic analysis. C Nomogram. D Nomogram calibration plots for 1, 2, 3, and 5 years Lactylation-related DEGs and Immune-cell infiltration Next, the proposed differences in DECGs were analysed by applying four analyses related to immune cell infiltration. Data in EPIC, QUANTISEQ, MCPCOUNTER and TIMER documented significantly positive correlation of DHRS7 with macrophage infiltration, and positive with NK cell infiltration in QUANTISEQ and MCPCOUNTER. RIMS1 was positively correlated with CD4 + T cells and CD8 + T cells in both EPIC and TIMER analyses, but negatively correlated with M1 and positively with M2 phenotype in QUANTISEQ. NOP2 correlated with M1 negatively in EPIC. and negatively with CD4 + T cells in QUANTISEQ. CRABP2 correlated with T cells and CD8 + T cells negatively in QUANTISEQ, MCPCOUNTER and TIMER. In all the four types of analyses, positive correlation of CALD1 was observed with macrophage infiltration. In addition, ALB was positively correlated with CD8 + T cells. Consequently, different DECGs have different TME characteristics, partially explained by poorer prognosis in patients with high CALD1, CRABP2, DHRS7 and RIMS1 expressions from the point of view of immune cell infiltration (Fig. [74]4A-D). According to the results of our analysis, the EPIC algorithm (Fig. [75]4A) shows a negative correlation between RIMS1 and NK cells, whereas the MCPCOUNTER algorithm (Fig. [76]4C) indicates a positive correlation. This discrepancy arises because the EPIC algorithm estimates the proportion of NK cells present in a sample. If regions with high RIMS1 expression correspond to areas with low NK cell infiltration, EPIC will report a lower NK cell proportion, resulting in a negative correlation between RIMS1 expression and NK cell levels. Conversely, the MCPCOUNTER algorithm calculates the total abundance of NK cell marker gene expression. In samples with high RIMS1 expression, even if the actual NK cell count is low, other immune cell subpopulations may also express NK cell markers, leading to a positive correlation between RIMS1 expression and the estimated NK cell proportion. Fig. 4. [77]Fig. 4 [78]Open in a new tab Associations between LRG signature and immune-cell infiltration evaluated by four different approaches. Analyses of the correlation of RIMS1, NOP2, DHRS7, CRABP2, CALD1 and ALB with tumor immune infiltrating cell abundance in TCGA-STAD using EPIC (A), QUANTISEQ (B), MCPCOUNTER (C) and TIMER (D), respectively (* p < 0.05 and ** p < 0.01) Prognostic investigation of the Lactylation-related DEG DHRS7 On the basis of the aforementioned data, DHRS7 might be a key prognostic DECG. Specifically in multiple tumour types, DHRS7 expression was higher than that of normal tissues in GBM, BRCA, PRAD, HNSC, LIHC, KICH, and CHOL, but lower in COAD, STES, KIRP, KIPAN, STAD, UCEC, KIRC, LUSC, THCA and READ (Figure S1). Moreover, it was much lower in STAD (Fig. [79]5A), and had a negative correlation with STAD prognosis (Fig. [80]5B). Next, DHRS7 expression was noticed to vary obviously only between stages II and IV, with higher expression and thus poorer prognosis in patients with stage IV (Fig. [81]5C, E and G). In stage N, N1 patients showed higher DHRS7 expression and worse prognosis as well in relative to N0 patients (Fig. [82]5D, F and H). Collectively, despite a generally low DHRS7 expression in STAD, DHRS7 may gradually play a pro-oncogenic role in STAD as the tumour develops further. Fig. 5. [83]Fig. 5 [84]Open in a new tab Correlations between DHRS7 expression and clinical characteristics. A DHRS7 expression levels in tumor and normal tissues. B K-M survival curves between DHRS7-high and -low expression groups. C Association between DHRS7 expression and pathological staging in STAD. D Relationship between DHRS7 expression and lymph node metastasis in STAD. E Associations among different pathological stages, lymph node metastasis, and OS in STAD. F Associations among different pathological stages and DSS in STAD. G Associations among different lymph node metastasis and PFS in STAD (ns: not statistically significant, * p < 0.05, ** p < 0.01, *** p < 0.001 and **** p < 0.0001) Effect of DHRS7 on different signaling pathways in STAD Comparison of GO analysis revealed predominant location of DHRS7 subcellular localisation (CC) in nucleus, nucleoplasm and cytosol, while the molecular function and biological process were mainly manifested as DNA binding and an effect on the regulation of transcription by RNA polymerase II, respectively (Fig. [85]6A). In terms of KEGG pathways, DHRS7 was mainly enriched in EGF/EGFR signaling pathway, TGF-beta signaling pathway, IL-4 signaling pathway, TGF-beta signaling pathway and PI3K-AKT signaling pathway when it was highly expressed (Fig. [86]6B). DHRS7 expression was also significantly positively correlated with mucin type O glycan biosynthesis, biosynthesis of unsaturated fatty acids, fatty acid elongation, galactose metabolism, glycosaminoglycan degradation, amino sugar and nucleotide sugar metabolism, ferroptosis, glycosaminoglycan biosynthesis heparan sulfate heparin, fructose and mannose metabolism, EMT markers, degradation of ECM and PI3K AKT mTOR pathway (Fig. [87]7A-L). The analysis results above indicate that DHRS7 may play a significant role in tumor progression, primarily through the biological processes described. Fig. 6. [88]Fig. 6 [89]Open in a new tab Results of GO (A) and KEGG (B) enrichment analyses Fig. 7. [90]Fig. 7 [91]Open in a new tab Correlation analysis between DHRS7 and cell signaling pathways. Correlation analysis of DHRS7 expression with Mucin type O glycan biosynthesis (A), Biosynthesis of unsaturated fatty acids (B), Fatty acid elongation (C), Galactose metabolism (D), Glycosaminoglycan degradation (E), Amino sugar and nucleotide sugar metabolism (F), Ferroptosis (G), Glycosaminoglycan biosynthesis heparan sulfate heparin (H), Fructose and mannose metabolism(I), EMT markers (J), Degradation of ECM (K) and PI3K AKT mTOR pathway (L) Effects of knockdown and overexpression of DHRS7 on STAD cells To assess the expression of DHRS7 in STAD, RT-qPCR and IHC assays were continued on 28 matched pairs of STAD tissues. Our analytical data in Fig. [92]8A and B indicated a marked reduction in DHRS7 levels in tumor tissues when compared to normal counterparts. Furthermore, HGC-27 and AGS cell lines confirmed significant reductions and increases in knockdown and overexpression efficiencies, respectively (***p < 0.001; Fig. [93]8C). Moreover, knocking down DHRS7 significantly boosted the activity and growth capacity of HGC-27 and AGS cells. Conversely, overexpressing DHRS7 led to a notable decline in their activity and proliferation (***p < 0.001; Fig. [94]8D, E). In addition, DHRS7 knockdown and overexpression respectively facilitated and hindered cell migration potential (***p < 0.001; Fig. [95]8F). Fig. 8. [96]Fig. 8 [97]Open in a new tab Knockdown of DHRS7 promotes cell proliferation and metastasis. A Expression of DHRS7 in 28 pairs of STAD and adjacent normal tissues was measured via qRT-PCR. B DHRS7 expression levels were assessed by IHC and quantified using the H-score. C Post-transfection expression of DHRS7 by RT-qPCR. D Cell absorbance values at different time points after DHRS7 knockdown and overexpression via CCK-8 assay. E Cell proliferation after DHRS7 knockdown and overexpression via EdU assay. F Cell migration after DHRS7 knockdown and overexpression via Transwell assay (***p < 0.001) Discussion In this study, we identified six LRGs associated with the prognosis of STAD. Furthermore, we examined the involvement of DHRS7 in the advancement of malignancy and its correlation with unfavorable outcomes in STAD. Lactylation is a valuable and nonnegligible player in various facets of tumor biology, including immune evasion, blood vessel formation, and metastasis. Within the TME, the buildup of lactate resulting from increased glycolysis in cancer cells fosters an immunosuppressive atmosphere. The lactylation of histones in immune cells (e.g., macrophages) encourages their transformation into an M2 phenotype to advance tumor proliferation and immune escape [[98]22]. Furthermore, lactylation of non-histone proteins, including PKM2, boosts glycolytic activity and aids in tumor cell proliferation [[99]23]. It also relates to a hallmark in cancer metastasis, namely, epithelial-mesenchymal transition (EMT). The lactylation of Snail, Twist and other transcription factors has been demonstrated to enhance their stability and transcriptional effectiveness, thereby driving EMT and tumor cell invasion [[100]24]. We downloaded the TCGA-STAD transcriptome data from TCGA for differential expression analysis of LRGs. Twelve of these genes (KRT1, CBR1, ALDOB, AHNAK, CSRP1, DHRS7 FABP5, NEFL, CRABP2, CALD1, RIMS1, and ALB) had the most significant expression differences, and then a prognostic model was constructed focusing on 6 genes, with DHRS7 finally determined as the primary gene for subsequent study. Advances in mass spectrometry and proteomics have unexpectedly led to the identification of DHRS7 as an LRG, which opens up new breakthroughs for deciphering lactate-forming proteins in cancer cells. In-depth exploration revealed the pivotal role of DHRS7 in maintaining intracellular lactate levels by precisely regulating the activities of LDH and MCTs [[101]13]. The profound role of DHRS7 in metabolic regulation, epigenetic regulation, and key signaling pathway should be highlighted in view of its multidimensional impact on tumor cell biology. In terms of metabolic regulation, there would be a metabolic paradigm shift, presenting with remarkable increase in glycolysis process and lactate production in tumor cells, when DHRS7 expression was reduced, undoubtedly providing sufficient energy supply and biosynthesis material basis for the crazy proliferation and tenacious survival of cancer cells [[102]25]. Epigenetically, by regulating the emulsification process of histones and other proteins, DHRS7 indirectly up-regulates genes related to tumor growth and immune evasion, and this complex regulatory network would be regarded as a key contributor in tumor development [[103]26]. Of interest, DHRS7 has been documented to establish intricate molecular-level interaction with core pathways (e.g., PI3K/AKT and MAPK) involved in tumorigenesis, which in turn profoundly affected key biological processes, such as cell proliferation, survival, and metastasis [[104]16]. In our study, a higher DHRS7 expression was observed in normal tissues compared to STAD tissues. However, when analyzing different stages of STAD, its expression was much higher in stage IV STAD than in stages II and III. The OS and DSS of stage IV STAD patients were obviously worse than those of stage II and III patients. Moreover, patients with lymph node metastasis were also detected with higher DHRS7 expression, and similarly, poorer prognosis. These findings suggest that, despite relatively lower DHRS7 expression in STAD, when the intratumoral environment of STAD undergoes changes, more aggressive STAD tumors tend to exhibit higher DHRS7 expression, thereby advancing to poor prognosis. Further clarification of the correlation between DHRS7 and immune-infiltrating cells revealed positive correlation with macrophages, particularly M2-type macrophage infiltration. Thus, DHRS7 may influence tumor immune escape by regulating the function of M2-type macrophages. In advanced STAD, high expression of DHRS7 may be related to a poor prognosis owing to its function of promoting the formation of an immunosuppressive microenvironment. From the perspective of signaling pathway regulation, DHRS7 may exert diverse effects by modulating key signaling pathways, such as EGF/EGFR, TGF-β, IL-4, PI3K/AKT, and MAPK. In early-stage tumors, its low expression may activate these pathways to stimulate tumorigenesis; whereas in advanced-stage tumors, its high expression may further exacerbate the abnormal activation of these pathways to drive tumor progression. Numerous genes and signaling pathways have been outlined to exhibit “dual functions” or “contradictory expression” within the TME. TGF-β, for example, acts as a “gatekeeper” in early-stage tumors by inhibiting cell growth; however, as the disease process evolves, it gradually goes rogue and contributes to cancer by inducing EMT and immunosuppression [[105]27, [106]28]. Intriguingly, HIF-1α, occupying an indispensable position in hypoxia adaptation, is the “guardian” of cell survival in normal physiology. Nevertheless, this “guardian” becomes an “accomplice” once it progresses to an advanced tumor stage, and its aberrant activation exacerbates tumor invasiveness, metastatic tendency and metabolic reprogramming [[107]29]. Moreover, Wnt/β-catenin and Notch pathways are supposed to be the “master regulators” of cell proliferation and differentiation, both of which may become the “pushers” of tumor invasion, proliferation and acquisition of stem cell properties once activated incorrectly in the late stage of cancer [[108]30–[109]33]. It is even more ironic that NF-κB, a key molecule that serves as a “double insurance” for inflammatory response and cell survival in healthy tissues, still “deviates” from its original purpose, i.e., curbing tumor progression, in the advanced stage of cancer, and promotes tumor invasion, metastasis and immune escape [[110]34]. Given the circumstances described above, DHRS7 may play a dual role in STAD depending on the tumor microenvironment. In early-stage tumors, DHRS7 may function as a tumor suppressor, with low expression potentially promoting tumorigenesis. Conversely, in advanced tumors, elevated DHRS7 expression may facilitate tumor progression through mechanisms such as modulating the immune microenvironment, driving metabolic reprogramming, or activating signaling pathways, ultimately leading to a poor prognosis. Nevertheless, this study has several limitations that should be acknowledged. Our study research solely exclusively the functional phenotypic effects of DHRS7 at the cellular level and did not explore investigate mechanistic basis or potential interactions within the tumor immune microenvironment. Nevertheless, However, intriguing observations, including such as correlation with macrophage infiltration, offer provide compelling rationale for further research. investigation. Future will aim to comprehensively elucidate the role and mechanistic underlying mechanisms DHRS7 throughout the progression of STAD. Conclusions In summary, this study innovatively revealed the differential expression of LRGs in STAD and established a prognostic gene signature. Furthermore, by employing multiple immune infiltration algorithms, potential correlations between these genes and various immune cell types were identified. Among them, DHRS7 emerged as the most prognostically significant candidate, warranting further investigation. Subsequent analysis demonstrated that although the overall expression level of DHRS7 was lower in STAD tissues compared to normal tissues, its expression increased with advancing tumor stage and was ultimately associated with poor prognosis. Based on these observations, we hypothesize that DHRS7 may play stage-specific roles during STAD progression, potentially mediated through the regulation of distinct signaling pathways and variable immune cell infiltration within the dynamically evolving tumor microenvironment. Supplementary Information [111]Supplementary Material 1.^ (383.2KB, docx) [112]Supplementary Material 2.^ (13KB, xlsx) [113]Supplementary Material 3.^ (13.3KB, docx) Acknowledgements