Abstract

Objectives

   Nasopharyngeal carcinoma (NPC) is an aggressive malignancy with high
   rates of morbidity and mortality, largely because of its late diagnosis
   and metastatic potential. Lactate metabolism and protein lactylation
   are thought to play roles in NPC pathogenesis by modulating the tumor
   microenvironment and immune evasion. However, research specifically
   linking lactate-related mechanisms to NPC remains limited. This study
   aimed to identify lactate-associated biomarkers in NPC and explore
   their underlying mechanisms, with a particular focus on immune
   modulation and tumor progression.

Methods

   To achieve these objectives, we utilized a bioinformatics approach in
   which publicly available gene expression datasets related to NPC were
   analysed. Differential expression analysis revealed differentially
   expressed genes (DEGs) between NPC and normal tissues. We performed
   weighted gene coexpression network analysis (WGCNA) to identify module
   genes significantly associated with NPC. Overlaps among DEGs, key
   module genes and lactate-related genes (LRGs) were analysed to derive
   lactate-related differentially expressed genes (LR-DEGs). Machine
   learning algorithms can be used to predict potential biomarkers, and
   immune infiltration analysis can be used to examine the relationships
   between identified biomarkers and immune cell types, particularly M0
   macrophages and B cells.

Results

   A total of 1,058 DEGs were identified between the NPC and normal tissue
   groups. From this set, 372 key module genes associated with NPC were
   isolated. By intersecting the DEGs, key module genes and
   lactate-related genes (LRGs), 17 lactate-related DEGs (LR-DEGs) were
   identified. Using three machine learning algorithms, this list was
   further refined, resulting in three primary lactate-related biomarkers:
   TPPP3, MUC4 and CLIC6. These biomarkers were significantly enriched in
   pathways related to “immune cell activation” and the “extracellular
   matrix environment”. Additionally, M0 and B macrophages were found to
   be closely associated with these biomarkers, suggesting their
   involvement in shaping the NPC immune microenvironment.

Conclusion

   In summary, this study identified TPPP3, MUC4 and CLIC6 as
   lactate-associated clinical modelling indicators linked to NPC,
   providing a foundation for advancing diagnostic and therapeutic
   strategies for this malignancy.

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s12967-024-05935-9.

   Keywords: Nasopharyngeal carcinoma, Lactylation-related genes,
   Infiltrating immunocytes, Machine learning algorithms, GEO

Introduction

   Nasopharyngeal carcinoma (NPC) is an aggressive malignancy arising from
   the epithelial cells of the nasopharynx, with a particularly high
   incidence in East Asia. NPC is characterized by complex interactions
   involving genetic predispositions, Epstein‒Barr virus (EBV) infection,
   and environmental factors [[46]1]. Despite advancements in NPC
   treatment, survival rates remain low, largely due to the high
   metastatic potential and limited biomarkers for early detection of this
   disease [[47]2, [48]3].

   Metabolic reprogramming is a hallmark of many cancers, including NPC,
   where it manifests as a pronounced shift toward glycolysis even in
   oxygen-rich conditions‒a phenomenon known as the Warburg effect
   [[49]4]. This shift leads to an accumulation of lactate, which
   contributes to an acidic tumor microenvironment (TME), thereby
   promoting tumor growth, immune escape and therapy resistance. Recent
   studies have shown that lactate, in addition to being a byproduct of
   glycolysis, serves as a key signalling molecule within tumors and
   inflamed tissues [[50]5, [51]6].

   Of particular interest is the role of lactate in a novel
   posttranslational modification (PTM) known as protein lactylation,
   where lactate groups bind to lysine residues on proteins, impacting
   gene expression and cellular function. Protein lactylation has been
   shown to modulate immune responses, tumor cell proliferation and
   metabolic adaptation [[52]7]. In NPC, lactylation may contribute to
   immune evasion and tumor aggressiveness by reprogramming immune cells
   within the TME, including M0 macrophages and regulatory B cells, to
   support a more immunosuppressive environment [[53]8].

   Lactate-driven metabolic shifts in NPC are facilitated by glycolytic
   enzymes such as hexokinase 2 (HK2) and pyruvate kinase M2 (PKM2), which
   drive glycolytic flux toward lactate production [[54]9–[55]11]. The
   rerouting of glycolysis intermediates to the pentose phosphate pathway
   (PPP) further supports tumor proliferation by providing
   ribose-5-phosphate for nucleotide synthesis and NADPH to counter
   oxidative stress [[56]12]. This metabolic reprogramming is crucial for
   sustaining NPC growth and progression.

   Given the significance of lactate metabolism and lactylation in NPC,
   identifying lactate-related hub genes (LRGs) could offer new insights
   into NPC pathogenesis and potential diagnostic biomarkers. In this
   study, we employed bioinformatics and machine learning techniques to
   analyse transcriptomic data from NPC samples, aiming to identify LRGs
   closely associated with glycolysis and immune modulation. By
   integrating immune infiltration analysis and constructing
   transcriptional networks, we aimed to elucidate the underlying
   mechanisms linking lactate metabolism to NPC progression, thus paving
   the way for innovative approaches for NPC diagnosis and treatment.

Materials and methods

Patients, tissue Collection, and tissue microarray

   From 2019 to 2021, a total of 85 tissue specimens, included 45 healthy
   nasal mucosa tissues and 40 NPC tissues, were collected from Shenzhen
   Hospital of Southern Medical University. A tissue microarray was
   constructed using these 85 tissue samples, which included 45 healthy
   nasal mucosa tissues and 40 NPC tissues.

   These NPC patients ranged in age from 33 to 78 years, comprising 28
   male and 12 female patients.

   This study was approved by the Ethics Committees of Shenzhen Hospital
   of Southern Medical University. Written informed consent was obtained
   from all participants or their legal guardians.

Data acquisition

   The research workflow is illustrated in Fig. [57]1. Gene expression
   profiles and clinical data from three east Asian population cohorts of
   NPC and normal samples ([58]GSE12452, [59]GSE64634 and [60]GSE102349)
   were obtained from the Gene Expression Omnibus (GEO) database
   ([61]http://www.ncbi.nlm.nih.gov/geo/). The [62]GSE12452 dataset
   comprises 10 normal and 31 NPC samples, whereas the [63]GSE64634
   dataset includes 4 normal and 12 NPC samples. These two datasets were
   combined and processed with batch correction via the ‘sva’ package to
   create a unified training set. For validation, the [64]GSE102349
   dataset, containing 113 NPC samples, was utilized. A total of 335
   lactate-related genes (LRGs) were identified from the Molecular
   Signatures Database ([65]http://www.gsea-msigdb.org/gsea/index.jsp) and
   supplemented by relevant literature.

Fig. 1.

   [66]Fig. 1
   [67]Open in a new tab

   Flowchart of the study design

Analysis of differentially expressed genes

   Principal component analysis (PCA) was performed via the ‘FactoMineR’
   and ‘factoextra’ packages to evaluate the integrity of the training
   set. DEGs between the normal and NPC groups were identified via the
   ‘limma’ package with a threshold of |log2-fold change (FC)| > 0.5 and
   an adjusted p value < 0.05. Volcano plots were generated via the
   ‘ggplot2’ package to visualize the DEGs, and a heatmap of the top 50
   DEGs was created via the ‘Pheatmap’ package. To assess the activation
   or inhibition of biological pathways associated with these DEGs,
   pathways were classified as activated with a Z score > 2 and inhibited
   with a Z score < -2. The GSEA provided insights into functionally
   relevant pathways, highlighting potential mechanisms underlying NPC
   progression and aiding in the identification of key targets for further
   study [[68]13].

RNA extraction and quantitative real-time PCR (qPCR)

   Total RNA was extracted using the RNA Trizol reagent (Invitrogen, Cat.
   No. 15596-026CN) following the manufacturer`s protocol. cDNA synthesis
   was performed using the All-in-One First-Strand cDNA Synthesis Kit
   (TransGen, Cat. No. AE341-02) according to the manufacturer`s
   instructions.The qPCR reactions were prepared using the PerfectStart
   Green qPCR SuperMix Kit (with Dye II, TransGen, Cat. No. AQ602-01),
   following the manufacturer`s guidelines, and run on an ABI Prism 7500
   system (ABI). The expression levels of target mRNAs were normalized to
   GAPDH as the internal control, and the relative expression differences
   were calculated using the 2^^−ΔΔCt method, expressed as fold changes
   relative to the control group.

   The primer sequences used for RT-qPCR (5′ to 3′) were as follows:

   TPPP3
     * Forward: AAGTCTGCTCGGGTCATCAAC.
     * Reverse: GAGCCCGTGTATCTGCTGG.

   MUC4
     * Forward: CAGGCCACCAACTTCATCG.
     * Reverse: ACACGGATTGCGTCGTGAG.

   CLIC6
     * Forward: GGGACCCAACATCCCGAATC.
     * Reverse: TCAGGCAGAGGGCTATTTAAGT.

   GAPDH
     * Forward: GTCAGCCGCATCTTCTTT.
     * Reverse: AGGCTGTTGTCATACTTCTC.

Immunohistochemical staining

   Tissue samples were fixed in 10% buffered formalin and subsequently
   processed into paraffin-embedded sections of 3–4 μm thickness. After
   antigen retrieval, the sections were incubated with specific primary
   antibodies followed by species-specific HRP-conjugated secondary
   antibodies. Signal development was performed using freshly prepared DAB
   solution, and nuclei were counterstained with hematoxylin [[69]14].

   The antibodies and their working concentrations used for IHC were as
   follows:

   CLIC6 Antibody (AFSBio, Cat. No. DF3934): 1:500.

   TPPP3 Rabbit pAb (ABclonal, Cat. No. A6775): 1:1000.

   MUC4 Rabbit mAb (ABclonal, Cat. No. A3438): 1:1000.

WCGNA

   WGCNA was performed on the top 25% most variable genes to identify gene
   coexpression modules linked to nasopharyngeal carcinoma (NPC). The
   soft-thresholding power was set to 12 to achieve a scale-free network
   with R2 ≥ 0.85 [[70]15, [71]16]. Modules were defined with a minimum
   size of 30 genes, each represented by a module eigengene (the primary
   expression component), which was subsequently correlated with
   NPC-related clinical traits.

Acquisition of LR-DEGs

   The overlap of differentially expressed genes (DEGs), key module genes,
   and lactate-related genes (LRGs) was identified and termed
   lactate-related differentially expressed genes (LR-DEGs), which were
   visualized through a Venn diagram. Chromosomal locations of the LR-DEGs
   were mapped via RCircos (version 0.4.15), which marks target gene
   positions on the genome following its initialization. To investigate
   LR-DEG interactions, a protein‒protein interaction (PPI) network was
   built in the STRING database, applying a confidence score threshold of
   > 0.75. Functional enrichment analysis, including Gene Ontology (GO)
   and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses,
   was conducted on the LR-DEGs via the ‘clusterProfiler’ package (version
   4.4.4), with statistical significance set at p.adjust < 0.05.

Machine-learning methods

   Support vector machine recursive feature elimination (SVM-RFE), friend,
   and random forest (RF) algorithms were employed to identify key genes
   in the training set [[72]17–[73]20]. The intersection of genes
   identified by these three methods was considered to represent critical
   genes. The diagnostic performance of these key genes was then evaluated
   by constructing a receiver operating characteristic (ROC) curve via the
   ‘pROC’ package [[74]20–[75]22]. Genes with an area under the curve
   (AUC) greater than 0.7, which also displayed consistent expression
   patterns between the training set and the external validation set
   ([76]GSE102349), were classified as key biomarkers with high diagnostic
   potential for NPC.

Clinical nomogram model

   Nomograms are extensively applied in clinical research to estimate the
   likelihood of individual clinical events. In this study, a nomogram
   incorporating selected biomarkers was constructed via the ‘rms’ package
   to predict NPC risk. The model’s clinical utility was assessed through
   decision curve analysis (DCA) [[77]23, [78]24]. To evaluate the
   nomogram’s predictive accuracy, calibration curves and receiver
   operating characteristic (ROC) curves were generated.

GSEA and immune infiltration analysis

   Single-sample gene set enrichment analysis (GSEA) was performed via the
   ‘clusterProfiler’ package to identify potential KEGG pathways
   associated with the biomarkers, with a significance threshold of
   p.adjust < 0.05. Additionally, the CIBERSORT algorithm was utilized to
   determine the relative abundances of 22 immune cell types present in
   the NPC microenvironment. Spearman correlation analysis was then
   conducted to explore the relationships between the biomarkers and the
   differentially abundant immune cells [[79]25, [80]26]. The ‘estimate’
   package was employed to calculate and compare the immune, stromal and
   ESTIMATE scores between the normal and NPC groups. Spearman correlation
   analysis between the biomarkers and these scores was performed via the
   ‘ggExtra’ package [[81]27–[82]29].

Construction of ‘TF-miRNA-gene’ networks

   The TRRUST database was utilized to predict transcription factors (TFs)
   associated with key biomarkers, whereas the miRWalk database was used
   to identify miRNAs linked to these biomarkers [[83]30]. The resulting
   TF‒miRNA‒gene interaction network was refined and visualized via
   Cytoscape software.

Statistical analysis

   All statistical analyses were conducted using R software (version
   4.1.1) and Perl. A p-value of < 0.05 was considered statistically
   significant for all tests.

Results

Analysis of differentially expressed genes and functional roles in NPC

   Box plot analysis indicated effective correction of the data samples,
   whereas principal component analysis (PCA) confirmed that batch effects
   were successfully removed in the merged dataset (training set)
   (Supplementary Figure [84]S1A, B). As shown in Fig. [85]2A, B and a
   total of 1,058 differentially expressed genes (DEGs) were identified
   between the normal and NPC groups, with 682 genes downregulated and 376
   genes upregulated. KEGG pathway analysis further revealed that these
   DEGs were involved in multiple pathways (Fig. [86]2C-F), including
   pathways related to the “cell cycle”, “mitosis” and “extracellular
   matrix and ECM receptor interaction”. Functional and disease pathway
   analyses revealed that the DEGs primarily participated in
   “intercellular signalling and interaction” as well as the “immune
   microenvironment”.

Fig. 2.

   [87]Fig. 2
   [88]Open in a new tab

   Identification and analysis of lactate-related DEGs in nasopharyngeal
   carcinoma. (A) Volcano plot showing the DEGs between the NPC and normal
   groups, including 682 downregulated and 376 upregulated genes. (B)
   Heatmap showing the top 50 DEGs. (C-F) KEGG pathway analysis further
   revealed that these DEGs were involved in multiple pathways

Identification of key module genes associated with NPC

   To identify NPC-associated modules, weighted gene coexpression network
   analysis (WGCNA) was applied. Sample clustering indicated no outliers
   (Supplementary Figure [89]S2). A soft threshold power of 12 was
   selected as optimal, achieving a signed R² of 0.85, which supported a
   scale-free network topology as the average connectivity approached zero
   (Fig. [90]3A, B). Using a dynamic tree-cutting algorithm with
   subsequent merging of similar modules, five distinct modules were
   identified (Fig. [91]3C-E). Among these, three modules—MEblue, MEbrown
   and MEyellow—were significantly correlated with NPC (|cor| > 0.6,
   P < 0.05) (Fig. [92]3C). From these genes, 372 key genes associated
   with NPC were extracted for further analysis (Fig. [93]3F-H).

Fig. 3.

   [94]Fig. 3
   [95]Open in a new tab

   (A-B) Analysis of the scale-free topology fit index and mean
   connectivity across various soft-thresholding powers. (C‒D)
   Hierarchical clustering produced a gene dendrogram, yielding 12 modules
   through the dynamic tree cut algorithm and subsequent merging of
   similar modules. (E) Heatmap indicating significant correlations
   between the MEblue, MEbrown, and MEyellow modules and NPC (|cor| > 0.3,
   P < 0.05). (F-H) Scatterplots of gene significance (GS) versus module
   membership (MM) reveal 372 key module genes associated with NPC

Identification and Functional Enrichment Analysis of Lactate-related DEGs

   Seventeen lactate-related differentially expressed genes (LR-DEGs) were
   identified in NPC by intersecting the DEGs, key module genes, and
   lactate-related genes (LRGs) (Fig. [96]4A). Protein‒protein interaction
   (PPI) analysis revealed that SLPI, LCN2, BPIFB1 and PIGR extensively
   interact with other proteins (Fig. [97]4B). These genes are distributed
   across chromosomes except for chromosomes 1, 3, 7, 9, 12, 16, 17, 18,
   20 and 21 (Fig. [98]4C). To investigate the functional roles of the
   LR-DEGs in NPC, functional enrichment analysis was performed. KEGG
   pathway analysis revealed significant enrichment in pathways such as
   “metabolic pathways”, “cytokine signalling” and the “extracellular
   microenvironment” (Fig. [99]4D). Gene Ontology (GO) analysis indicated
   that the LR-DEGs were associated mainly with “immune response”, “immune
   receptor activation” and “extracellular matrix organization”
   (Fig. [100]4E‒G).

Fig. 4.

   [101]Fig. 4
   [102]Open in a new tab

   Acquisition and functional analysis of lactate-related differentially
   expressed genes (LR-DEGs). (A) Venn diagram illustrating the overlap of
   17 LR-DEGs among DEGs, key module genes and lactate-related genes
   (LRGs) in NPC. (B) Protein‒protein interaction (PPI) network of
   LR-DEGs. (C) Chromosomal distribution of the LR-DEGs visualized in a
   circular format. (D) KEGG pathway enrichment analysis of LR-DEGs. (E-G)
   Gene Ontology (GO) enrichment analysis of LR-DEGs

Development and validation of an LR-DEG-based Signature for NPC

   To identify key genes, support vector machine recursive feature
   elimination (SVM-RFE) analysis was conducted on the 17 lactate-related
   DEGs (LR-DEGs), resulting in nine prominent genes: TPPP3, DNALI1,
   CHST9, MUC4, CLIC6, SLPI, KIF18B, FAM3D and PIGR (Fig. [103]5A, B).
   Moreover, the random forest (RF) algorithm identified the top 10
   characteristic genes, including CLIC6, MUC4, CHST9, TPPP3, DNALI1,
   RACGAP1, FAM3D, PARPBP, INHBA and SLPI (Fig. [104]5C, D). Additionally,
   the Friend algorithm selected 10 critical genes: INHBA, LTF, LCN2,
   SLPI, FAM3D, RACGAP1, BPIFB1, MUC4, KIF18B and CD1C (Fig. [105]5E). The
   cross-referencing results from these three algorithms highlighted three
   core genes: TPPP3, MUC4 and CLIC6 (Fig. [106]5F). In external
   validation TPPP3, MUC4 and CLIC6 showed strong diagnostic and
   predictive values for NPC, with AUC values over 0.7, and their
   expression trends were consistent with those of the training set,
   establishing them as lactate-related biomarkers for NPC (Fig. [107]5G,
   H). Analysis of these genes in both the training and validation
   datasets revealed their downregulation in NPC samples compared with
   normal controls, with further decreased expression observed in the PD
   (progressive disease) group relative to the SD (stable disease) group
   (Fig. [108]5I, J). We also validated the expression of these three
   genes in clinical specimens. Both qPCR (including 10 paired NPC and
   normal tissues) and tissue array (including 40 NPC and 45 normal
   tissues) experiments indicated that the expression levels of TPPP3,
   MUC4 and CLIC6 were significantly downregulated in tumour tissues
   compared to normal nasopharyngeal tissues (Fig. [109]5K, L).

Fig. 5.

   [110]Fig. 5
   [111]Open in a new tab

   Screening of lactate-related biomarkers for NPC. (A, B) Core gene
   identification within LR-DEGs via the SVM algorithm. (C, D) Core gene
   screening of the LR-DEGs via the RFE algorithm. (E) Core gene screening
   with the Friends algorithm. (F) Venn diagram showing three key LR-DEGs
   identified at the intersection of all three machine learning
   algorithms. (G, H) ROC curves for the three key genes in both the
   training set and the external validation set. (I, J) Expression levels
   of TPPP3, MUC4 and CLIC6 across groups in the training and validation
   sets. (K) qPCR validation of TPPP3, MUC4 and CLIC6 mRNA expression
   levels in nasopharyngeal carcinoma (NPC) (n = 10) and normal
   nasopharyngeal tissues (NP)(n = 10). (L) Immunohistochemical validation
   of TPPP3, MUC4 and CLIC6 protein expression levels in nasopharyngeal
   carcinoma(n = 40) and normal nasopharyngeal tissues(n = 45). ns, not
   significant. *p < 0.05, **p < 0.01, ****P < 0.0001

Clinical relevance and functional enrichment analysis of NPC biomarkers

   To further investigate the link between the identified biomarkers and
   NPC, a nomogram incorporating these biomarkers was constructed
   (Fig. [112]6A). The model’s performance and accuracy were validated
   through calibration, receiver operating characteristic (ROC) and
   Kaplan‒Meier (KM) curves (Fig. [113]6B‒D). Additionally, decision curve
   analysis (DCA) was employed to assess the clinical utility of the
   1-year and 3-year prognostic models (Fig. [114]6E, F). To explore the
   roles of TPPP3, MUC4 and CLIC6 in NPC pathogenesis, single-gene gene
   set enrichment analysis (GSEA) was performed for each biomarker. KEGG
   pathway enrichment revealed associations with key pathways, including
   “antigen processing and presentation”, “extracellular matrix
   organization” and “immune microenvironment regulation”
   (Fig. [115]6G-I).

Fig. 6.

   [116]Fig. 6
   [117]Open in a new tab

   Clinical and functional enrichment analysis of NPC biomarkers. (A)
   Nomogram incorporating identified biomarkers for clinical prediction.
   (B-D) Calibration, receiver operating characteristic (ROC) and
   Kaplan‒Meier (KM) curves were used to validate the nomogram’s
   predictive accuracy. (E-F) Decision curve analysis (DCA) confirmed the
   clinical utility of the prognostic model. (G-I) Single-gene GSEA for
   TPPP3, MUC4 and CLIC6 revealed associations with pathways related to
   “antigen processing and presentation”, “extracellular matrix
   organization” and “immune microenvironment regulation.”

Influence of TPPP3, MUC4, and CLIC6 on the immune microenvironment in NPC

   Considering the link between NPC pathophysiology and the immune
   microenvironment, we further investigated immune cell distributions
   within NPC. Using three computational algorithms—Cibersort, ssGSEA and
   xCell—we analysed immune cell composition (Fig. [118]7A-C). The
   expression levels of immune cells in the training set were assessed
   with CIBERSORT and ssGSEA (Fig. [119]7D, E). Notably, 14 immune cell
   types, including naïve and memory B cells, CD8 + T cells, naïve and
   resting CD4 + T cells, activated memory CD4 + T cells, follicular
   helper T cells, M1 macrophages, resting dendritic cells, aDCs, Th1
   cells, NK CD56 dim cells, Th17 cells, Th2 cells, NK CD56 bright cells
   and mast cells, were significantly different in abundance between the
   NPC samples (Fig. [120]7D, E).

Fig. 7.

   [121]Fig. 7
   [122]Open in a new tab

   Immune infiltration analysis of lactate-related biomarkers in NPC.
   (A-C) Relative proportions of infiltrating immune cells in NPC, as
   determined by the CIBERSORT, ssGSEA and xCell algorithms. (D-E)
   Significant differences in the abundances of 23 distinct immune cell
   types in NPC samples. (F-I) Correlation analysis between
   lactate-related biomarkers and 23 immune cell types, illustrating their
   associations. ns, not significant. *p < 0.05, **p < 0.01, ***p < 0.001,
   ****P < 0.0001

   Further analysis revealed strong correlations between key biomarkers
   (MUC4, CLIC6 and TPPP3) and specific immune cells, particularly B cells
   and M0 macrophages (Fig. [123]7F-I).

   Furthermore, we observed that the stroma, immune, estimated and
   microenvironment scores were significantly greater in the SD group than
   in the PD group (Fig. [124]8A). These scores were positively correlated
   with the expression levels of TPPP3, MUC4 and CLIC6, suggesting an
   association between these biomarkers and the tumor microenvironment
   (Fig. [125]8B).

Fig. 8.

   [126]Fig. 8
   [127]Open in a new tab

   Analysis of biomarker roles within the NPC immune microenvironment. (A)
   Matrix, immune, ESTIMATE and microenvironment scores were significantly
   higher in NPC patients in the SD group than in those in the PD group.
   (B) Correlation analysis revealed that these four scores were
   positively correlated with the expression of TPPP3, CLIC6 and MUC4,
   suggesting that these biomarkers are not significantly associated with
   immune microenvironment characteristics. *p < 0.05, **p < 0.01,
   ***p < 0.001, ****P < 0.0001

Analysis of the regulatory network and drugs used in NPC

   To investigate the regulatory mechanisms of TPPP3, MUC4 and CLIC6, a
   “TF-miRNA-gene” network comprising 28 nodes and 34 edges was
   constructed (Fig. [128]9A). In this network, CREB1 was identified as a
   likely regulator of MUC4 and TPPP3, whereas hsa-miR-574-5p appeared to
   regulate CLIC6 expression. Additionally, drug candidates targeting
   TPPP3, MUC4 and CLIC6 were identified through the DrugBank and ChEMBL
   databases. The resulting biomarker‒drug interaction network, shown in
   Fig. [129]9B, includes 32 nodes and 71 edges. Notably, gefitinib and
   bortezomib were found to be common therapeutic targets for all three
   biomarkers, suggesting their potential significance in NPC treatment.

Fig. 9.

   [130]Fig. 9
   [131]Open in a new tab

   Analysis of the role of biomarkers in NPC. (A) ‘TF‒miRNA‒gene’ network
   presenting the regulatory mechanisms of TPPP3, MUC4 and CLIC6. (B)
   Relationships between biomarkers and drugs predicted from the DrugBank
   and ChEMBL databases

Discussion

   Nasopharyngeal carcinoma (NPC) is a malignant tumor originating from
   the epithelial cells of the nasopharynx; its distinct geographic
   distribution is strongly linked to Epstein–Barr virus (EBV) infection
   [[132]3, [133]12, [134]31]. Clinically, NPC poses significant
   challenges because of its typical late-stage diagnosis, aggressive
   behavior and high metastatic potential [[135]32]. Uncovering the
   molecular mechanisms underlying NPC is essential for advancing
   diagnostic and therapeutic strategies. Recent studies have highlighted
   the role of metabolic reprogramming—specifically lactate
   modification—and immune dysfunction in NPC pathogenesis.

   Lactate, which has long been considered a byproduct of anaerobic
   metabolism, has emerged as a key factor in the tumor microenvironment.
   The Warburg effect, which is commonly observed in many cancers,
   describes the preference of cancer cells for glycolysis over oxidative
   phosphorylation, even in oxygen-sufficient conditions
   [[136]33–[137]35]. This shift results in elevated lactate levels,
   leading to an acidic microenvironment that can modulate immune
   responses [[138]36]. Elevated lactate in tumors has been shown to
   promote immune evasion by altering the function of immune cells,
   including macrophages and T cells, underscoring the importance of
   understanding the influence of lactate on immune regulation within NPC.

   Our study focused on the lactate-related modifications of three key
   genes—TPPP3, CLIC6 and MUC4—each of which plays an important role in
   NPC progression. TPPP3, which is involved in microtubule dynamics, cell
   proliferation, and apoptosis, has been associated with various cancers,
   including NPC, where it may contribute to tumor growth and metastasis
   [[139]37–[140]41]. CLIC6, a chloride intracellular channel protein,
   plays a role in ion transport and cellular signalling, potentially
   influencing immune cell activation and inflammation [[141]42–[142]44].
   Its altered expression in cancers suggests that it may contribute to
   immune evasion mechanisms in NPC. MUC4, a mucin family member, supports
   epithelial cell protection and signalling [[143]45–[144]48]. In NPC,
   MUC4 is associated with tumor growth and metastasis through its roles
   in cell adhesion and immune evasion. The involvement of lactate in
   lactate metabolism could provide insight into how lactate influences
   immune suppression in NPC by affecting antigen presentation and immune
   recognition.

   Our research identified TPPP3, CLIC6 and MUC4 as core lactate-related
   genes with diagnostic potential in NPC [[145]24, [146]49, [147]50].
   These genes are significantly associated with pathways such as “antigen
   processing and presentation”, “extracellular matrix organization” and
   “immune microenvironment regulation”, which are crucial for initiating
   immune responses against tumors [[148]51–[149]53]. The influence of
   lactate on these pathways may enable NPC cells to evade immune
   detection, thus facilitating tumor progression.

   Additionally, we found a significant correlation between these
   biomarkers and immune cell populations, particularly M0 macrophages and
   regulatory B cells [[150]54]. M0 macrophages, which exhibit phenotypic
   plasticity, can differentiate into subtypes on the basis of the
   microenvironment [[151]4, [152]53, [153]55, [154]56]. In NPC, elevated
   lactate may skew M0 macrophages toward an immunosuppressive phenotype,
   further promoting tumor growth. B cells, which are known for
   maintaining immune homeostasis, suppress effector T-cell responses and
   contribute to immune evasion in tumors. The associations of TPPP3,
   CLIC6 and MUC4 with these immune cell types highlight the role of
   lactate in modulating the immune microenvironment of NPC.

   This study provides key insights into the role of lactate in NPC and
   its effects on immune function. By identifying TPPP3, CLIC6 and MUC4 as
   lactate-related biomarkers [[155]37–[156]41, [157]57–[158]66], we
   highlight their importance in understanding NPC pathogenesis.
   Furthermore, the involvement of these genes in immune modulation
   presents new therapeutic opportunities. Targeting lactate metabolism or
   the pathways associated with these biomarkers could improve treatment
   efficacy and patient outcomes.

   Despite our best efforts, we acknowledge that future research will need
   to address three significant aspects [[159]1]. Given that the
   mechanisms of diseases related to lactate and lactylation are emerging
   hot topics in recent years, we should plan to perform lactate treatment
   experiments in cell lines to study the underlying molecular regulatory
   mechanisms, when we secure research funding support [[160]2].
   Biomarkers should be closely linked to clinical applications. To better
   verify the reliability and potential significance of our data, we have
   initially validated the expression levels of TPPP3, MUC4 and CLIC6 in
   NPC specimens and normal tissues using qPCR (Fig. [161]5K), potentially
   supporting the transcriptional regulation of these genes in NPC.
   Furthermore, we have thoroughly validated our results by using tissue
   microarrays (Fig. [162]5L). We believe that this should provide useful
   insights and suggestions for the future research of our team and other
   colleagues on the regulation mechanism of lactylation in NPC [[163]3].
   biomarkers should be closely linked to clinical applications. Among the
   three proteins mentioned, we found that MUC4 is a secreted protein and
   can potentially be detected in serum samples. The other two proteins
   are not secretory, but they may potentially have clinical utility in
   immunohistochemistry analyses by pathologists. They could likely be
   used for clinical biopsies or postoperative assessment of prognosis or
   treatment effectiveness in the future.

   In conclusion, this research identified three lactate-related core
   genes—TPPP3, CLIC6 and MUC4—with significant diagnostic potential in
   NPC. These genes may contribute to NPC pathogenesis through immune
   modulation mechanisms, particularly those involving M0 macrophages and
   B cells. By clarifying the relationships among lactate, immune
   function, and these biomarkers, this study lays the groundwork for the
   development of new diagnostic and therapeutic strategies for NPC.
   Further studies are needed to explore how lactate impacts the immune
   microenvironment in NPC and to validate the clinical relevance of these
   biomarkers for diagnosing and treating this challenging malignancy.

Conclusion

   In summary, this study identified three lactate-related hub
   genes—TPPP3, MUC4 and CLIC6—as promising biomarkers for diagnosing and
   predicting disease progression in nasopharyngeal carcinoma (NPC). These
   genes appear to play roles in NPC pathogenesis through pathways related
   to “antigen processing and presentation”, “extracellular matrix
   organization” and “immune microenvironment regulation”. Furthermore,
   associations between these biomarkers and immune cells, particularly M0
   macrophages and regulatory B cells, were observed. These findings
   underscore the critical influence of lactate on NPC progression,
   offering a foundation for the future development of diagnostic and
   therapeutic approaches targeting these biomarkers for NPC management.

Electronic supplementary material

   Below is the link to the electronic supplementary material.
   [164]Supplementary Material 1^ (6.6MB, tif)
   [165]Supplementary Material 2^ (5.8MB, tif)

Acknowledgements