1. Introduction
Head and neck squamous cell carcinoma (HNSCC) is an aggressive cancer that can develop in the mucosal epithelium of oral sites, including the oral cavity, pharynx, and larynx
[1],
[2], with a five-year overall survival (OS) of approximately 50%
[3]. Cervical lymph node (LN) metastasis can be clinically detectable in many HSNCC patients at initial diagnosis and is the leading cause of cancer-related death
[4]. Despite advances in therapeutic approaches over the years, a large proportion of HNSCC patients do not experience significant benefits because of the high incidence of metastasis and recurrence
[5],
[6],
[7]. The molecular mechanisms underlying LN metastasis in HNSCC remain incompletely understood, and effective therapies for patients with metastatic HNSCC are still lacking. Therefore, identifying new driving factors involved in HNSCC lymphatic metastasis may provide novel therapeutic strategies for metastatic HNSCC.
The tumor metastatic process and metabolic reprogramming are highly intertwined
[8], with the targeting of metabolic vulnerabilities emerging as an effective anticancer therapy
[9]. Alterations in cellular metabolism, often triggered by the upregulation of pyruvate kinase M2 (PKM2) and lactate dehydrogenase A (LDHA), isocitrate dehydrogenase (IDH) mutations, and succinate dehydrogenase (SDH) and fumarate hydratase (FH) deficiency, are hallmarks of cancer and frequently occur in aggressive and metastatic cancer types
[10]. Metabolic rewiring-mediated tumor metastasis is associated with various mechanisms, including epigenetic regulation, such as the acetylation and lactylation of histones and nonhistone proteins
[11]. These effects can be induced by oncometabolites, such as succinate, fumarate, 2-hydroxyglutarate, and lactate, which modify the tumor microenvironment (TME) and sustain invasive phenotypes in cancer cells
[12]. Within the TME, cancer cells can secrete soluble cytokines and metabolites to shape immune cells into immunosuppressive and tumor-promoting phenotypes, promoting faster tumor growth and metastasis
[13]. However, the contribution of metabolic crosstalk within the TME to HNSCC LN metastasis remains inadequately characterized. Understanding the molecular mechanisms through which cancer cells alter the immune cell phenotype to facilitate metastasis could reveal new therapeutic options for combating HNSCC metastasis.
Enolase 2 (ENO2) serves as a crucial enzyme in the glycolytic pathway, catalyzing the conversion of 2-phosphoglycerate (2-PG) to phosphoenolpyruvate (PEP)
[14]. Elevated ENO2 expression is often associated with a poor prognosis across various cancer types, including pancreatic ductal adenocarcinoma
[15], papillary renal cell carcinoma
[16], and colorectal cancer
[17]. ENO2 has been implicated in promoting tumor growth, drug resistance, and metastasis, either by enhancing glycolysis
[15] or activating the β-catenin and Yes-associated protein (YAP) signaling pathways
[17],
[18]. A prior study has demonstrated that ENO2 interacts with PKM2, promoting its stabilization and nuclear translocation, thereby increasing glycolytic flux and accelerating growth in HNSCC
[19]. Additionally, our previous study revealed that ENO2 conferred antiangiogenic therapy resistance in colorectal cancer, which was associated with PEP-mediated inhibition of histone deacetylases 1 (HDAC1) activity
[20]. Additionally, PEP has been shown to inhibit Th17 cell differentiation and autoimmunity
[21] and to modulate T-cell antitumor effector functions
[22]. Nevertheless, the precise role and mechanism of ENO2 in modulating HNSCC metastasis remain largely elusive. Furthermore, whether ENO2 and its metabolite PEP influence the interaction between HNSCC cells and immune cells, thereby affecting metastasis, remains to be determined.
Here, through the integration of tumor databases, public datasets, and clinical relevance analysis, we identified ENO2 as a key driver of HNSCC LN metastasis. Our findings demonstrated that ENO2, along with its oncometabolite PEP, not only promoted HNSCC invasiveness but also induced M2 macrophage polarization. Importantly, both effects were effectively abrogated by ENO2 inhibition. Furthermore, we revealed that PEP enhanced H3 lysine 18 lactylation (H3K18la) and upregulated the expression of genes essential for M2 macrophage polarization. Overall, our study provides novel mechanistic insights into the ENO2-mediated crosstalk between tumor cells and macrophages in HNSCC lymphatic metastasis, thereby offering promising therapeutic potential.
2. Material and methods
2.1. Cell culture
The HSC3 HNSCC cells and 293T human embryonic kidney cells were cultured in culture dishes with a diameter of 100 mm using culture medium (DMEM culture medium, 10% (v/v) fetal bovine serum (FBS), and 1% (v/v) penicillin–streptomycin). The SCC9 HNSCC cells were cultured in culture medium (F-12, 10% (v/v) FBS, and 1% (v/v) penicillin–streptomycin) in 100 mm culture dishes. All cell lines were cultured at 37 °C under 5% CO2 conditions. The human monocyte cell line THP-1 was cultured in RPMI-1640 complete medium (containing 10% FBS, 1% penicillin–streptomycin), and passaged every 2–3 d. Logarithmic growth phase THP-1 cells were induced with phorbol 12-myristate 13-acetate (PMA) (100 ng∙mL−1) for 4 h to differentiate into macrophages (MØ macrophages).
2.2. Experimental animals
Male BALB/c-nu mice (four weeks old, weighing 16–20 g) were purchased from Guangdong Medical Laboratory Animal Center (license No. SCXK (Su) 2022-0008). Animals were housed in the Laboratory Animal Center of Jinan University School of Medicine (license No. SYXK (Yue) 2022-0174) under the following conditions: five mice per cage, specific pathogen free (SPF) level environment, temperature of 18–22 °C, humidity of 50%–60%, noise level below 60 decibels, lighting for 10–14 h daily, bedding and feed sterilized by high-temperature and high-pressure, and drinking water filtered and sterilized. All animal studies were conducted in accordance with the approval of the Laboratory Animal Ethics Committee of Jinan University (approval No. IACUC-20230716-14).
2.3. Clinical tumor specimens
Clinical samples from patients diagnosed with locally or metastatic oral squamous carcinoma were randomly collected from The First Affiliated Hospital of Jinan University since February 2019. All patients provided informed consent for the use of tissue specimens for scientific research. The study complied with the principles of the Helsinki Declaration and was approved by the Ethics Committee of The First Affiliated Hospital of Jinan University (approval No. JNUKY-2024-0058). The 57 cases of HNSCC tissue array included 58 cases of tongue squamous cell carcinoma, nine cases of gingival squamous cell carcinoma, one case of lower lip squamous cell carcinoma, three cases of floor of mouth squamous cell carcinoma, seven cases of buccal mucosa squamous cell carcinoma, two cases of cheek mucosa tissue, nine cases of adjacent tongue tissue, five cases of normal tongue tissue, and three cases of pharyngeal mucosa tissue.
2.4. Cell migration and invasion assay
Cell migration ability was assessed using a Transwell migration assay. Logarithmic growth phase HNSCC cells and ENO2-overexpressing HNSCC cells were digested and suspended in serum-free medium. Then, 4 × 104 cells per well were seeded in the upper chamber of 8.0 μm Transwell inserts, with 100 μL of cell suspension added per well. The lower chamber was filled with 500 μL of medium containing serum, with three replicate wells per group. After incubation in a cell culture incubator for 24 h, the medium in the upper and lower chambers was discarded, and the chambers were washed with phosphate buffer saline (PBS) three times. The cells were fixed with 4% paraformaldehyde solution at room temperature for 20 min, followed by staining with 0.1% crystal violet solution for 3 min. After staining, residual staining solution was washed off with PBS, and non-migrated cells in the upper chamber were gently wiped off. The migrated cells in the lower chamber were observed and photographed under an inverted microscope. Five random fields were selected for photography per insert, and cell migration was quantified using ImageJ software.
The invasion assay was performed using the same Transwell chambers. First, 30 µL of diluted 4% Matrigel was added to the Transwell inserts and allowed to solidify in a 37 °C cell culture incubator. The subsequent steps were the same as the migration assay. After incubation, invaded cells in the lower chamber were observed and photographed under an inverted microscope. Five random fields were selected for imaging per insert, and cell invasion was quantified using ImageJ software.
2.5. Hematoxylin and eosin (H&E) staining experiment
Tissue sections were dewaxed in xylene for 30 min, followed by immersion in gradient ethanol for 5 min each and washing with distilled water three times for 3 min each. Subsequently, the sections were stained with hematoxylin for approximately 2–3 min until the cell nuclei turned blue, followed by rinsing with running water and observation under a microscope. Then, the sections were stained with eosin for 2 min, followed by rinsing with running water. After dehydration, the sections were mounted with neutral resin and observed under a microscope.
2.6. Immunohistochemistry experiment
The tumor xenografts of HNSCC were fixed using 4% paraformaldehyde, embedded in paraffin, and frozen at −20 °C before sectioning. The paraffin-embedded tumor tissue blocks were sectioned and sliced into 4 μm thick sections using a paraffin microtome. The sections were flattened in a 45 °C water bath and transferred to slides. After drying, the slides were baked in a 65 °C oven until completely dry, then stored at room temperature. After deparaffinization and hydration, antigen retrieval was performed followed by permeabilization with 0.1% Triton X-100. The slides were then blocked in bovine serum albumin (BSA) for 1 h. The blocked sections were incubated with primary antibodies overnight at 4 °C. After washing with PBS, the sections were incubated with corresponding horseradish peroxidase (HRP) secondary antibodies at room temperature for 1 h. After 24 h of solidification in the fume hood, excess xylene odor was removed, and the sections were observed under a microscope. Finally, ImageJ software was used for quantification of immunohistochemistry (IHC) images.
2.7. Live imaging of mouse xenografts
Pentobarbital powder and D-fluorescein sodium are respectively weighed and dissolved in saline to form solutions with concentrations of 1% and 150 mg∙mL−1. The mice are then anesthetized through intraperitoneal injection of the pentobarbital solution at a dose of 2 mg∙kg−1, with the injection volume adjusted based on individual animal weights. Following pentobarbital administration, D-fluorescein sodium is intraperitoneally injected at a volume of 100 µL per animal. Subsequently, the anesthetized mice are transferred to a small animal live imaging luminometer (in vivo imaging system (IVIS)) for the in-situ detection of primary and metastatic tumors. Live imaging is performed utilizing appropriate fluorescence detection settings to capture images of the mice and any fluorescent signals emitted by the tumors. The resulting imaging data are then analyzed to visualize and quantify the presence and distribution of primary and metastatic tumors within the mice. All procedures involving animals are conducted in strict accordance with institutional guidelines and regulations governing the ethical treatment of animals.
2.8. Liquid chromatography–mass spectrometry (LC–MS)
Supernatants are collected from transfected blank controls, ENO2 overexpression, and ENO2 enzymatic mutant-treated HNSCC cells, each totaling approximately 1 mL. The collected supernatants are then subjected to centrifugation at 12 000g at 4 °C for 15 min. Subsequently, the supernatant is passed through a 0.22 µm filter. Following filtration, the supernatant is centrifuged again at 12 000g for 15 min at 4 °C. After a 30 min incubation period at 4 °C, the supernatant is meticulously transferred to a sample vial for further analysis.
To detect PEP uptake in cells, the culture medium was removed, and cells were washed twice with PBS. Subsequently, cells underwent five freeze–thaw cycles using liquid nitrogen, with each cycle lasting 5 min. Thawed samples were then centrifuged at 12 000g for 15 min at 4 °C. The resulting supernatant was transferred to a new 1.5 mL Eppendorf (EP) tube and filtered through a 0.22 µm filter. Gradient concentration standards of PEP ranging from 10 to 0.078125 µmol∙L−1 were prepared. Mass spectrometry was initially used to determine the m/z sizes of PEP’s primary (Q1) and secondary (Q2) ions, and the collision energy (CE) and declustering potential (DP) values were optimized using multiple reaction monitoring (MRM) mode to establish conditions for PEP mass spectrometry detection. Liquid separation was performed using the ACQUITY UPLC BEH C18 column (Waters Corporation, USA) with reverse-phase ultra-high performance liquid chromatography. PEP in the supernatant and cell samples was detected using an AB SCIEX Triple Quad 4500 LC mass spectrometer (SCIEX, USA) in negative electrospray ionization mode. The Analyst 1.6.3 Software (SCIEX) was used to quantify PEP levels by fitting the different concentrations of standards to their respective liquid chromatographic peak areas. Samples were adjusted based on their total protein concentrations.
2.9. Flow cytometry analysis
THP-1 human monocytes were treated with PMA for 4 h, allowing them to adhere and differentiate into MØ macrophages. The cells were then subjected to treatment with tumor supernatant or exogenously added PEP for 24 h. Following treatment, the cells were digested, and the collected cells were filtered through a 30 µm filter. The filtered cells were subsequently centrifuged at 500g for 5 min, and the supernatant was discarded to obtain a single-cell suspension. The cells were then blocked on ice for 15 min using flow cytometry blocking solution, followed by incubation with corresponding antibodies on ice for 1 h. For M1 macrophage identification, anti-CD80-PE and anti-iNOS-FITC antibodies were used, while anti-ARG1-APC and CD206-PE antibodies were used for M2 macrophage identification. After the incubation period, the cell suspension was transferred to a flow cytometry tube, and Flow Cytometry staining buffer was added to a volume of 300 µL. Analysis was performed using the CytoFLEX flow cytometer, and experimental results were processed using FlowJo VX software.
2.10. Western blot
Cells were cultured in a six-well plate according to the experimental design. The cells were washed twice with PBS, followed by the addition of an appropriate amount of radioimmunoprecipitation assay buffer (RIPA) lysis buffer based on cell quantity. The cells were then scraped to collect cell lysate, which was transferred to a 1.5 mL centrifuge tube. The concentration of each sample was measured using the bicinchoninic acid (BCA) assay. Subsequently, 5× sample buffer was added to the lysate, mixed well, and heated at 100 °C for 10 min. Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) was carefully run at 120 V for 1.5 h, and the proteins were transferred to a polyvinylidene fluoride (PVDF) membrane. The membrane was rinsed with tris-buffered saline with Tween 20 (TBST) and blocked with 5% BSA blocking solution at room temperature for 1 h. After blocking, the membrane was rinsed again with TBST. The membrane was then incubated with the corresponding primary antibody on a shaker at 4 °C overnight. Following primary antibody incubation, the membrane was washed with TBST three times for 5 min each and then incubated with the secondary antibody at room temperature for 1 h followed by imaging.
2.11. Real-time fluorescent quantitative PCR (RT-qPCR)
Cells were cultured in a six-well plate, and after treatment, the culture medium was discarded. The cells were washed twice with PBS, and then collected by enzymatic digestion and centrifugation. Total RNA was extracted from the collected cells following the instructions of the Total RNA Kit I (Qiagen, Germany) for small quantity total RNA. The RNA concentration was measured using a microspectrophotometer. Reverse transcription was performed using the reverse transcription premix kit (Bio-Rad, USA) to acquire complementary DNA (cDNA). The cDNA was then pre-mixed with 2× RealStar Fast SYBR (Bio-Rad) and nuclease-free water. Subsequently, 9 µL of the mixture was added to each well of a 96-well qPCR plate, along with 1 µL of primers (including upstream and downstream primers). The plate was centrifuged to mix the contents, and RT-PCR was performed using a real-time PCR instrument. Cycle threshold (Ct) values were exported from each well for subsequent analysis.
2.12. Chromatin immunoprecipitation sequencing (ChIP-seq) and ChIP-qPCR
25 µg of MØ macrophage chromatin were incubated with 4 µg of anti-H3K18la antibody for 4 h at 4 °C. Following incubation, protein agarose was added and incubated for an additional 2 h. The chromatin DNA was then eluted with elution buffer containing 1% sodium dodecyl sulfate (SDS) and 0.1 mol∙L−1 sodium bicarbonate. To remove RNA, ribonuclease (RNase) A was used. DNA recovery was performed using the QIAquick PCR purification kit (Qiagen) according to the manufacturer’s instructions. Subsequently, the ChIP-seq library was constructed using the Accel-NGS 2S Plus DNA Library Prep Kit (Swift Biosciences, USA) following the manufacturer’s protocol. The fragment size of the library was amplified and assessed using the Bioanalyzer (Agilent, USA), and quantified using the Qubit dsDNA HS assay kit (Thermo Fisher Scientific, USA). Indexed libraries were pooled and applied to a Hiseq4000 sequencer (Illumina, USA) using a 50-nucleotide single-read configuration. Upon obtaining sequencing results, all reads were mapped to the University of California, Santa Cruz (UCSC) reference genome using Bowtie v2.5, retaining only unique reads. The files were then converted to bam format using SAMtools v0.1.19, sorted, and PCR duplicates were removed. Peaks were called using MACS v2.2.1 at q (adjusted p-value that controls the false discovery rate (FDR)) = 0.01. The quantification and direct comparison of H3K18la in different samples (control and PEP groups) involved counting features of unique mapping H3K18la reads around promoter regions (± 2 kb around transcription start sites) using featureCounts v.1.5.0-p130, followed by normalization of ChIP reads under respective conditions (control and PEP groups). M2 macrophage-related genes in the ChIP-seq data were analyzed using Database for Annotation, Visualization, and Integrated Discovery (DAVID) Bioinformatics Resources 6.8 for Gene Ontology (GO) analysis.
For chromatin immunoprecipitation followed by quantitative PCR (ChIP-qPCR), M2 macrophage chromatin was incubated with anti-H3K18la antibody, followed by protein agarose addition and DNA was purified after elution. PCR was performed to detect H3K18la binding to specific promoter regions, early growth response 1 (EGR1), vascular endothelial growth factor A (VEGFA), and interleukin-10 (IL–10), and so forth.
2.13. Laser Doppler imaging scanning
The nude mice were fasted for 8 h before intraperitoneal injection of 1% pentobarbital sodium (7.5 mL∙g−1 body weight) for deep anesthesia. Subsequently, the tumor area of the nude mice was disinfected, the skin was incised, and the surrounding tissue of the tumor was bluntly dissected. Then, the tumor local microvascular blood flow of the nude mice was scanned using a laser Doppler imaging instrument (Moor Instruments, UK). Finally, the blood flow rate was calculated in arbitrary perfusion units using data acquisition software (Moor FLPI measurement software, V3.0).
2.14. Animal ultrasound imaging
The BALB/c-nu mice were fasted for 8 h before intraperitoneal injection of 1% pentobarbital sodium (7.5 mL∙g−1 body weight) for deep anesthesia. Subsequently, the nude mice were placed in a supine posture, and the entire tumor area of the nude mice was scanned using an animal ultrasound imaging diagnostic instrument (DaWei, China), and the maximum cross-section of the tumor was taken to measure the maximum longitudinal and transverse diameters.
2.15. Bioinformatics analysis
Public transcriptomic data for 546 HNSCC samples (including 502 HNSCC cases and 44 normal tissues or adjacent cancers) were downloaded from The Cancer Genome Atlas (TCGA) database. Metabolic pathways and gene sets were downloaded from the MSigDB database. Data sets GSE30784, GSE37991, GSE70604, and GSE65858 were obtained from the Gene Expression Omnibus (GEO) database. Use the DEseq2 software package to analyze differential genes in HNSCC normal samples and tumor samples from the TCGA database. Conduct gene set enrichment analysis (GSEA) pathway analysis and metabolic gene joint analysis of the obtained differential genes through the clusterProfile package; validate the results from the TCGA database with data sets GSE37991 and GSE30784. Analyze patient prognosis using the Survival software package for TCGA and GSE65858 data sets. All analyses and visualizations were conducted in R (v4.3.2) and RStudio.
2.16. Statistical analysis
Statistical analysis was conducted using GraphPad and Prism 8.0 software six. Data are presented as mean ± standard error of the mean (SEM). Assess the significance of differences between two groups using student t-tests, and assess the significance of differences between two groups using one-way analysis of variance (ANOVA) and Tukey’s multiple comparison test. Statistical significance was considered at p < 0.05. All in vitro experiments were independently performed at least three times.
3. Results
3.1. High ENO2 expression correlates with LN metastasis in HNSCC
To identify critical factors implicated in the pathogenesis of HNSCC, we performed differential gene expression analysis via the TCGA database, which includes data from 546 HNSCC tumors and 44 normal tissues. Employing the DESeq2 package, we identified 4471 genes that were differentially expressed in HNSCC samples (Fig. S1(a) in Appendix A). GSEA revealed significant upregulation of metabolic pathways in HNSCC tumors (Fig. S1(b) in Appendix A). Among these altered metabolism-related genes, ENO2 was highly expressed in HNSCC tumor samples (
Figs. 1(a) and
(b)), ranking the highest among the glycolysis-related genes in the gene cluster (Fig. S1(c) in Appendix A). Additional datasets from the GEO (GSE30784 and GSE37991) verified the dramatic increase in
ENO2 messenger RNA (mRNA) levels in HNSCC tumor samples compared with those in normal tissues (Figs. S1(d) and (e) in Appendix A). Notably, an integration of data from the TCGA and GEO (GSE70604) datasets revealed that
ENO2 mRNA levels were significantly greater in tumors with LN metastasis than in their non-metastatic counterparts (
Fig. 1(c), and Fig. S1(f) in Appendix A). Interestingly, the expression level of another enolase, ENO1, was not associated with metastasis status (Figs. S1(g)–(i)) in Appendix A).
For validation, IHC was employed to quantify ENO2 protein expression in HNSCC patient-derived tissue samples (
Fig. 1(d)), along with positron emission tomography (PET)-Computed Tomography imaging (
Fig. 1(e)), and a significant correlation was found between ENO2 expression and extensive LN metastasis. Furthermore, analysis of a human tissue microarray consisting of 58 samples (39 tumors and 19 normal per adjacent nontumor tissues) from HNSCC patients revealed a significant increase in ENO2 expression in tumor tissues, particularly in patients with LN metastasis (
Figs. 1(f)–(h)). Clinical correlation analysis further demonstrated that elevated ENO2 levels were significantly associated with shorter OS and disease-free survival (DFS) rates in HNSCC patients (
Figs. 1(i) and
(j)). Together, these results suggest that high ENO2 expression may positively correlate with LN metastasis in HNSCC.
3.2. ENO2 promotes HNSCC lymphatic metastasis
We then investigated the role of ENO2 in modulating the migration and invasion of HNSCC cells via Transwell and wound healing assays. Compared with vector controls, ENO2 overexpression notably increased the migration and invasion of both HSC3 and SCC9 cells (Figs. S2(a)–(d) in Appendix A). Subsequent mechanistic studies indicated that the enhanced invasion and migration effects were mediated by epithelial–mesenchymal transition (EMT), as evaluated by elevated protein expression levels of stem cell markers (CD44, CD133, and EpCAM) and mesenchymal markers (neuronal cadherin (N-cadherin) and vimentin) and decreased expression of the epithelial marker E-cadherin (Fig. S2(e) in Appendix A).
To confirm the role of ENO2 in promoting HNSCC LN metastasis, we examined the impact of ENO2 overexpression in a murine model. BALB/c-nu mice were sublingually inoculated with HSC3 cells transfected with either vector (HSC3
vector) or ENO2-overexpressing (HSC3
ENO2) constructs. Tumor progression and metastasis occurrence were monitored via an IVIS, which measures luciferase bioluminescence (
Fig. 2(a)). Our results revealed that the overexpression of ENO2 significantly increased the ability of HSC3 cells to metastasize to the neck area (
Fig. 2(b)), an observation further confirmed by subsequent dissection (
Fig. 2(c)). Additionally, high-frequency ultrasound (HFUS) was performed to further validate the presence of cervical LN metastases in tumors derived from HSC3
ENO2 as opposed to those derived from HSC3
vector injections. The HSC3
ENO2 group exhibited obvious LN enlargement and a hypoechoic ultrasound signal indicative of potential LN metastasis (
Figs. 2(d) and
(e)). Furthermore, H&E-stained histological images confirmed a drastic increase in lymphatic metastasis in the HSC3
ENO2 mice compared with the HSC3
vector mice (
Figs. 2(f) and
(g)). Additionally, mice bearing HSC3
ENO2 xenograft tumors presented significantly shorter progression-free survival (PFS) and earlier onset of metastasis (
Fig. 2(h)). Furthermore, the levels of mesenchymal markers (N-cadherin and vimentin) were markedly elevated and the expression of the epithelial marker E-cadherin was decreased in HSC3 tumors overexpressing ENO2 (
Figs. 2(i) and
(j)). Overall, ENO2 promotes malignant behavior and facilitates LN metastasis in HNSCC.
3.3. ENO2-mediated HNSCC metastasis is associated with M2 macrophage polarization
To understand the role of ENO2 in HNSCC progression, we examined its impact on the TME via the HNSCC TCGA database. By employing the ESTIMATE algorithm, we investigated the association between ENO2 expression and the presence of tumor-infiltrating immune cells. Our results indicated that elevated expression of ENO2 corresponded to an increased tumor purity and a reduced immune score, indicating a potential adverse impact on patient prognosis (Fig. S3(a) in Appendix A). Further analysis of immune cell infiltration via the CIBERSORT algorithm revealed a significant increase in the proportion of macrophages in the ENO2-high (ENO2
high) cohort (
Fig. 3(a)). A pairwise comparison of ten predominant immune cell phenotypes revealed a significant correlation between ENO2 levels and macrophage infiltration in HNSCC samples (Fig. S3(b) in Appendix A). Additional GEO datasets were assessed, which revealed a strong correlation between the M2 macrophage fraction and ENO2 expression (Fig. S3(c) in Appendix A). Additionally, we calculated the correlation of the ENO2 expression levels with tumor purity and immune cell infiltration via the TIMER 2.0 database, which revealed a positive correlation between ENO2 expression and macrophage infiltration (Fig. S3(d) in Appendix A). Elevated M2 macrophage infiltration was further confirmed by IHC staining of ARG1 (Fig. S3(e) in Appendix A), and its correlation with metastasis was confirmed via TCGA datasets (Figs. S3(f) and (g) in Appendix A). To address whether M2 macrophages drive HNSCC metastasis, we performed immunofluorescence staining. Our results revealed that ENO2 negligibly affected the infiltration of T cells, neutrophils, and myeloid-derived suppressor cells (MDSCs) (Fig. S3(h) in Appendix A).
Furthermore, we confirmed what we observed from public datasets via immunofluorescence staining. Tumor samples derived from patients with ENO2
high and LN metastatic HNSCC showed increased CD68
+ARG1
+ M2 macrophage infiltration at the tumor margin (
Fig. 3(b)). A similar infiltration pattern was observed in HSC3
ENO2 xenograft tumors (
Fig. 3(c)). Collectively, our findings suggest that ENO2 may contribute to HNSCC metastasis by modulating M2 macrophage infiltration. To comprehensively assess the influence of ENO2 on macrophages within the TME, we utilized PMA to induce the differentiation of human THP-1 monocytes into MØ macrophages, followed by exposure to conditioned medium (CM) collected from HSC3
ENO2 cells. After 24 h of indirect coculture, the macrophages were harvested for RT-PCR analysis. Compared with macrophages treated with HSC3
vector-CM, those exposed to HSC3
ENO2-CM presented significant downregulation of M1 macrophage markers (
NOS2,
CD80, and
TNFα) and upregulation of M2 macrophage markers (
CD206,
ARG1, and
CD163) (Fig. S4(a) in Appendix A). Furthermore, flow cytometry analysis revealed a significant increase in the proportion of ARG1
+CD206
+ M2 macrophages, accompanied by a significant reduction in iNOS
+CD80
+ M1 macrophages (Fig. S4(b) in Appendix A). These findings suggest that tumor cells might secrete soluble factors to modulate M2 macrophage polarization.
3.4. ENO2 induces M2 macrophage polarization through its metabolite PEP
To test the hypothesis that ENO2 regulates M2 macrophage polarization through its metabolite PEP, we engineered an enzymatically inactive variant of ENO2, ENO2
K394R, which is deficient in PEP production. MØ macrophages were incubated with CM from HSCs transfected with empty vector (Vector), ENO2
WT, or ENO2
K394R (
Fig. 3(d)). Flow cytometry analysis further revealed that the proportions of ARG1
+CD206
+ and iNOS
+CD80
+ macrophages were restored after exposure to CM generated from ENO2
K394R-transfected HSC3 cells (
Fig. 3(e)). The expression levels of M1 macrophage markers (
CD80 and
NOS2) and M2 macrophage markers (
CD206,
ARG1, and
CD163) were assessed via RT-qPCR (
Fig. 4(a)). These results indicated that CM from ENO2
WT-transfected cells significantly promoted the M2 polarization of macrophages, as indicated by elevated M2 and decreased M1 marker expression levels. In contrast, compared with that from ENO2
WT-transfected cells, CM from ENO2
K394R-transfected cells significantly restored marker expression (
Fig. 4(b)). Additionally, Western blot and flow cytometry analyses revealed that CM generated from ENO2
K394R-transfected HSC3 and SCC9 cells failed to facilitate M2 macrophage polarization (Figs. S4(c)–(e) in Appendix A). Collectively, these findings suggest that ENO2 may depend on its enzymatic activity to regulate M2 macrophage polarization.
We further analyzed the effect of the ENO2 metabolite PEP on macrophage polarization. Using LC–MS, we quantified PEP levels in both tumor cells and PMA-induced MØ macrophages (
Fig. 4(c)). LC–MS confirmed elevated PEP levels in HSC3 and SCC9 cells transfected with ENO2
WT compared with those in the control group, which was not observed in the ENO2
K394R group. The concentrations of extracellular PEP in HNSCC cells were as follows: (390 ± 42.5) μmol∙L
−1 (HSC3
vector), (1015 ± 32.3) μmol∙L
−1 (HSC3
ENO2-WT), (347 ± 42.6) μmol∙L
−1 (HSC3
ENO2-K394R), (746 ± 43.9) μmol∙L
−1 (SCC9
vector), (2134 ± 29.3) μmol∙L
−1 (SCC9
ENO2-WT), and (1588 ± 3.2) μmol∙L
−1 (SCC9
ENO2-K394R) (
Fig. 4(d)). Additionally, the introduction of exogenous PEP resulted in its increased uptake by MØ macrophages (
Fig. 4(e)). Direct treatment of MØ macrophages with 1–2 mmol∙L
−1 PEP (
Fig. 4(f)) and subsequent validation via Western blot, RT-qPCR, and flow cytometry analyses revealed M2 polarization in the macrophages (
Figs. 4(g) and
(h), Figs. S5(a) and (b) in Appendix A).
3.5. PEP triggers M2 macrophage polarization via upregulation of H3K18la
Epigenetic modifications, particularly lactylation and acetylation, are pivotal in the process of macrophage polarization
[23]. Therefore, we investigated the mechanisms by which PEP influences M2 macrophage polarization. Upon the addition of 1 mmol∙L
−1 PEP to MØ macrophage cultures for 48 h, we observed significant increases in the pan-lactylation and pan-acetylation levels (
Fig. 5(a)). Considering that histone lactylation and acetylation undergo dynamic changes during the macrophage polarization process
[24], with different temporal dynamics, and that H3K18la
[23] and H3 lysine 27 acetylation (H3K27ac)
[25] have been reported in macrophage polarization, we measured H3K18la and H3K27ac levels in MØ macrophages at different time points following PEP treatment. In contrast to the downregulation of H3K27ac after 24 h, H3K18la, alongside M2 macrophage marker genes, progressively increased over time (
Fig. 5(b)). To confirm the contribution of lactylation to macrophage polarization, the E1A-binding protein p300/CREB-binding protein (P300/CBP) inhibitor C646 was applied, and the expression levels of H3K18la and markers of M1 and M2 macrophages were assessed. We found that treatment with C646 reduced H3K18la levels, which in turn decreased macrophage polarization (
Fig. 5(c)).
Next, ChIP-seq was further applied to assess the role of PEP in mediating M2 macrophage polarization through H3K18la, revealing preferential enrichment of H3K18la at promoter regions in PEP-treated macrophages (
Fig. 5(d)). GO analysis revealed that H3K18la-enriched genes were correlated mainly with wounding reponse, cell migration, and angiogenesis, which was consistent with the M2 macrophage transcriptome profile (
Fig. 5(e)). Additional analysis revealed a 61.88% enrichment of peaks at promoter regions in PEP-treated cells compared with controls (Fig. S5(c) in Appendix A). Using the integrative genomics viewer (IGV), we visualized the enrichment of H3K18la on target genes and identified significant peaks at M2 macrophage-associated gene loci, including
EGR1,
JAK3,
VEGFA, and
IL-
10 (
Fig. 5(f)). Given that EGR1 is a key transcription factor for M2 polarization
[26], our analysis of H3K18la binding motifs revealed that EGR1 and EGR2 were activated, whereas transcription factors associated with M1 macrophages, such as interferon regulatory factor 3 (IRF3) and MAF bZIP transcription factor F (MAFF), were repressed (
Fig. 5(g)). ChIP-qPCR validation revealed H3K18la enrichment at M2 marker gene promoters, including
EGR1,
VEGFA, and
IL-10 (
Fig. 5(h)). Additional ChIP-qPCR revealed that EGR1 bound to the promoter regions of
ARG1 and
CD206, hence promoting their expression and triggering M2 macrophage polarization (
Fig. 5(i)). Correspondingly, RT-qPCR analysis of
EGR1,
JAK3,
VEGFA, and
IL-10 mRNA levels in PEP-treated MØ macrophages revealed a positive correlation with M2 gene expression (Figs. S5(d) and (e) in Appendix A). Furthermore, a lactic acid detection assay confirmed increased endogenous lactic acid production in PEP-treated macrophages (Fig. S5(f) in Appendix A). Given the emerging understanding of HDACs as erasers of histone lactylation
[27], we hypothesized that PEP might affect HDAC de-lactylation activity, thereby increasing H3K18la levels. This hypothesis was tested by expressing either wild-type or kinase-dead (H141A) HDAC1 in HSC3 and SCC9 cells. Western blot analysis revealed that H3K18la, CD206, and ARG1 protein levels were negligibly affected by the H141A HDAC1 mutation and that PEP addition did not increase their expression, suggesting that the regulation of H3K18la and macrophage polarization by PEP is dependent on HDAC1 kinase activity (Fig. S5(g) in Appendix A). Additionally, drug affinity responsive target stability (DARTS) analysis was applied to assess the direct binding of PEP to HDAC1 (Fig. S5(h) in Appendix A). To further investigate whether H3K18la is modulated by ENO2-associated PEP production, HSC3 and SCC9 cells were transfected with either ENO2
WT or ENO2
K394R. Subsequent Western blot analysis revealed elevated levels of H3K18la and macrophage polarization markers in macrophages treated with CM from ENO2
WT, whereas ENO2
K394R HSC3 and SCC9 cells restored the expression levels of H3K18la and macrophage polarization markers (Fig. S5(i) in Appendix A). H3K18la levels also increased in a dose-dependent manner with PEP treatment (Fig. S5(j) in Appendix A). Since we previously reported that PEP directly binds to H141 of HDAC1
[20], these findings collectively demonstrate that PEP may induce lactate production and inhibit HDAC1 activity, thereby increasing H3K18la levels and leading to M2 polarization in macrophages.
3.6. PEP-mediated polarized macrophages enhance HNSCC EMT and migration
To determine whether ENO2 promotes EMT and migration in HNSCC through its metabolite PEP, we transfected HSC3 cells with ENO2
K394R or ENO2
WT. Western blotting, which was conducted 24 h post-transfection, revealed a significant increase in the expression of EMT markers, including N-cadherin and vimentin, along with a significant decrease in the E-cadherin expression level in ENO2
WT-transfected cells. In contrast, transfection with ENO2
K394R significantly reduced the expression levels of these markers (Fig. S6(a) Appendix A). Similarly, Transwell assays revealed enhanced invasion and migration activities in HSC3 and SCC9 cells transfected with ENO2
WT, but transfection with ENO2
K394R significantly reduced their invasion and migration activities (
Figs. S6(b) and
(c) in Appendix A). Given that M2 macrophages in the TME promote tumor EMT, we further assessed the potential feedback regulation in HNSCC progression. CM from PEP-treated macrophages (PEP-CM) was collected and used to treat HNSCC cells for 24 h (Fig. S7(a) in Appendix A). Subsequent Western blot analysis revealed increased expression of the EMT markers N-cadherin, vimentin, EpCAM, and CD44 but decreased expression of E-cadherin compared with that in the control group (Fig. S7(b) in Appendix A). Additionally, we found that PEP-CM increased the migration capabilities of both HSC3
vector and HSC3
ENO2 cells, with a more pronounced effect observed in HSC3
ENO2 cells (Figs. S7(c) and (d) in Appendix A). To further demonstrate that macrophage polarization contributes to the EMT of HNSCC cells, we used siRNA to knock down
EGR1 in macrophages, effectively inhibiting M2 polarization. CM collected from PEP-treated
EGR1-knockdown macrophages significantly reduced HNSCC cell migration, suggesting that macrophage polarization is essential for HNSCC metastasis (Fig. S7(e) in Appendix A).
Furthermore, we sought to elucidate the mechanisms by which M2 macrophages induce EMT in tumor cells. PEP was added to MØ macrophage cultures, and cytokine and chemokine mRNA levels were measured by RT-qPCR after 24 h. We observed significant increases in the expression levels of transforming growth factor-β (TGF-β), IL-17, and C-C motif chemokine ligand 12 (CCL12) (Fig. S7(f) in Appendix A). TGF-β regulates glycolytic pathways and induces metabolic reprogramming; therefore, we further assessed the expression level of its receptor TGFβR1 in tumor cells and found a significant increase (Fig. S7(g) in Appendix A), suggesting cytokine–receptor interaction-mediated regulation of tumor EMT. Moreover, we treated HSC3 cells with the TGFβR1 inhibitor SB431542 while simultaneously exposing them to PEP-CM for 24 h. As expected, this treatment reduced the migration and invasion abilities of the HSC3 cells (Fig. S7(h) in Appendix A). RT-qPCR analysis further revealed a decrease in the expression of mesenchymal markers and an increase in epithelial marker expression (Fig. S7(i) in Appendix A). Collectively, these findings suggest that PEP induces macrophage polarization to release TGF-β, which then interacts with TGFβR1 on tumor cells, promoting EMT and migration, which may lead to tumor metastasis.
3.7. Pharmacological inhibition of ENO2 suppresses M2 macrophages and HNSCC metastasis
To further investigate the role of ENO2 in regulating the TME of HNSCC, we assessed the antimetastatic effect of the ENO2 enzymatic activity inhibitor POMHEX. HSC3vector and HSC3ENO2 cells were generated, and the CM of HSC3vector, HSC3ENO2, and HSC3ENO2 + POMHEX cells was collected, followed by the addition of MØ macrophages for 24 hours (Fig. S8(a) in Appendix A). Compared with those from HSC3ENO2 CM-treated macrophages, CM from HSC3ENO2 + POMHEX-treated macrophages presented significant decreases in the expression of M2 macrophage markers and corresponding increases in the expression of M1 macrophage markers (Fig. S8(a)). Flow cytometry analysis further validated these findings, indicating that POMHEX treatment effectively decreased the proportion of ARG1+CD206+ (M2) macrophages and increased the proportion of iNOS+CD80+ (M1) macrophages (Figs. S8(b)–(f) in Appendix A).
Next, we examined the antimetastatic effect of POMHEX
in vivo in a mouse model of LN metastasis
. HSC3
vector and HSC3
ENO2 cells were isolated from the abovementioned xenograft mice and reintroduced sublingually into BALB/c-nu mice to establish the model (
Fig. 6(a)). Five days post-inoculation, the mice received 20 mg∙kg
−1 POMHEX via intraperitoneal injection (i.p.) daily for two weeks, followed by IVIS analysis. Although HSC3
ENO2 inoculation led to severe LN metastasis, POMHEX treatment significantly inhibited this effect (
Figs. 6(b) and
(c)). Consistently, ultrasound was applied to detect the necks of the mice, and LN metastasis was observed only in the HSC3
ENO2 group (
Fig. 6(d)). Moreover, ENO2 overexpression dramatically promoted LN metastasis, whereas POMHEX treatment effectively delayed LN metastasis (
Fig. 6(e)). Furthermore, H&E staining indicated that POMHEX treatment effectively decreased the number of metastases and reduced the metastasis rate (
Figs. 6(f) and
(g)). The expression levels of EMT markers were also evaluated, which revealed that POMHEX restored the level of E-cadherin and reduced the expression of N-cadherin and vimentin (Fig. S8(g) in Appendix A). Moreover, immunofluorescence staining revealed a decrease in M2 macrophage (ARG1
+CD206
+) infiltration at the tumor margins of treated mice (
Figs. 6(h) and
(i)). Collectively, these findings indicate that pharmacological inhibition of ENO2 with POMHEX effectively suppresses EMT and M2 macrophage infiltration, thereby significantly reducing tumor metastasis in HNSCC. These findings suggest that POMHEX could be a potential therapeutic approach for managing and treating HNSCC metastasis.
4. Discussion
HNSCC is among the most commonly diagnosed cancers worldwide, and its incidence is increasing
[28]. Human papillomavirus (HPV) is a leading cause of HNSCC
[29], and both local and metastatic disease are important because of the effect of tumor progression within the TME
[7]. Consequently, immune checkpoint inhibitors (ICIs) targeting the TME have been approved for treating HNSCC
[30]. Unfortunately, only 15%–20% of patients benefit from such therapies, highlighting the critical need to comprehensively investigate the “cellular ecosystem” of HNSCC
[31]. Our research identified the glycolytic enzyme ENO2 as a crucial regulator of HNSCC metastasis that modulates the interaction between tumor cells and the TME. Although primarily identified as a glycolytic enzyme, ENO2, like its isoform ENO1, is expressed in a variety of tumors and plays diverse roles in tumorigenesis. For example, ENO1 has been shown to promote programmed cell death ligand 1 (PD-L1) degradation and enhance antitumor immune responses
[32]. Interestingly, we observed here that the ENO1 expression level remained similar in HNSCC nonmetastatic and metastatic patients. Here, we demonstrate that ENO2 facilitates macrophage polarization in a catalytic activity-dependent manner and facilitates HNSCC metastasis, suggesting that ENO2, a previously underestimated enolase, may play a critical role in tumor progression.
Tumor-associated macrophages (TAMs) play a crucial role within the TME by maintaining cellular plasticity and homeostasis. Furthermore, interactions between cells in the TME rely heavily on extracellular metabolites such as lactate, which are often released by glycolytic tumor cells and can drive immune cell polarization
[33]. For example, succinate secreted by tumor cells can modulate macrophage polarization
[34]. Although the role of ENO2 in the metabolic reprogramming of the TME is not fully understood, our findings suggest that excessive PEP accumulation due to ENO2 activation promotes macrophage polarization. We observed that CM from ENO2-expressing HSC3 cells induces the polarization of MØ macrophages toward the M2 phenotype, whereas enzymatically inactive ENO2 does not. Moreover, the direct addition of PEP to MØ macrophage cultures also results in M2 polarization. LC–MS analysis revealed increased PEP uptake by M2 macrophages after treatment. Interestingly, markers associated with M2 macrophages were significantly correlated with both the PEP level and enzymatic activity of ENO2, whereas M1 macrophage markers were not fully restored by the K394R mutation. This could be attributed to the various sources of M2 macrophages, including resident tissue macrophages, monocyte-derived macrophages and stem cell-derived macrophages
[35]. These results underscore the underappreciated role of the ENO2 metabolite PEP in immune cell polarization regulation, necessitating further investigation into whether other enolase-derived PEPs could exert similar effects. Moreover, the role of macrophage polarization in HNSCC metastasis requires further investigation in transgenic mice lacking macrophages.
Thus far, the mechanisms by which ENO2 regulates macrophage polarization remain unclear. Therefore, we investigated the mechanism underlying macrophage polarization. Given the pivotal role of ENO2 in glycolysis and its regulation of lactate uptake, we investigated whether lactate accumulation influences histone modification in macrophages. Recent studies have shown that histone lactylation increases during the late stages of M1 macrophage polarization, leading to ARG1 expression
[17]. We found that ENO2-mediated H3K18la, a less studied histone modification involved in macrophage polarization, significantly impacts TME remodeling. Studies have linked H3K18la-mediated VCAM1 expression to gastric cancer progression and metastasis
[36]. Additionally, H3K18la has been found to increase monocyte–macrophage transition in the early post-myocardial infarction stage, with target genes including LRG1, VEGFA, and IL-10
[37]. The identification of similar gene enrichments in this study suggests promising directions for further exploration of downstream mechanisms involving macrophage H3K18la in HNSCC. Epigenetic modifications play a major role in macrophage polarization; therefore, various modifications contribute sequentially to this process. The continuous increase in H3K18 lactylation levels is consistent with our observations
[23]. Other histone modifications, such as H3K27ac, decrease at M2-related genes during M1 polarization, facilitating the shift toward a proinflammatory phenotype
[38].
Another significant finding of this study is that ENO2 regulates macrophage polarization by inhibiting the catalytic activity of HDAC1. Specifically, PEP, derived from tumor cell ENO2, binds directly to the active site of HDAC1
[20], thereby inactivating its delactylase activity. HDAC1, known as a new “eraser” for histone lactylation
[27], has been found to promote H3K18la by deacetylating lysine 18 of histone 3
[39]. We observed that kinase-inactivated HDAC1 does not increase H3K18la expression in macrophages treated with high-ENO2-expressing tumor cells. Therefore, histone modification represents a complex mechanism with diverse biological functions. In HNSCC, HDAC1 inhibition by PEP enhances H3K18la enrichment within the promoter region of M2 macrophage marker genes, including EGR1, VEGFA, and IL-10.
Although only approximately 10% of HNSCC patients exhibit distant metastasis, those with metastasis are rarely cured, and a significant proportion of patients experience recurrence posttreatment
[40]. Therefore, substantial efforts are underway to improve the clinical outcomes of these patients. Our results demonstrate that pharmacological inhibition of ENO2 effectively abolishes macrophage polarization and reverses the pro-metastatic TME, significantly reducing metastasis in a mouse model. POMHEX, an enolase inhibitor with selectivity for ENO2, has shown efficacy as a prodrug for ENO1
−/− glioma
[41]. Treatment with POMHEX inhibits EMT and macrophage infiltration in mouse model tumors, suggesting that further investigations of whether POMHEX could improve the prognosis of HNSCC patients are warranted.
5. Conclusions
In conclusion, our study elucidates the role of ENO2 in modulating macrophage polarization, thereby contributing to HNSCC metastasis. Through further mechanistic studies, we reveal that PEP, a metabolite of ENO2, inhibits HDAC1 activity, leading to increased H3K18la enrichment within the promoter region of M2 macrophage genes. M2 macrophages, in turn, promote tumor cell migration and metastasis. Importantly, our findings further suggest that pharmacological inhibition of ENO2 impedes M2 macrophage polarization and curtails HNSCC LN metastasis.
CRediT authorship contribution statement
Chenran Wang: Writing – review & editing, Writing – original draft, Visualization, Project administration, Methodology, Funding acquisition, Formal analysis, Data curation, Conceptualization. Lin Tan: Methodology, Formal analysis, Data curation. Maohua Huang: Writing – review & editing, Writing – original draft, Funding acquisition, Conceptualization. Yuning Lin: Visualization, Methodology, Formal analysis, Data curation. Minxiang Cai: Funding acquisition, Conceptualization. Lijuan Deng: Data curation. Xinpeng Hu: Formal analysis, Data curation. Shenghui Qiu: Formal analysis, Data curation. Xiaoting Chen: Data curation. Yiming Zhang: Data curation. Xiaomei Luo: Data curation. Changzheng Shi: Conceptualization. Minfeng Chen: Funding acquisition, Conceptualization. Wencai Ye: Funding acquisition, Conceptualization. Junqiu Zhang: Writing – review & editing, Writing – original draft, Funding acquisition, Data curation, Conceptualization. Dongmei Zhang: Writing – review & editing, Writing – original draft, Funding acquisition. Xiangning Liu: Writing – review & editing, Funding acquisition, Conceptualization.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This study was supported by grants from the National Natural Science Foundation of China (82204428, U24A20815, 82304526, 82204427, 82201001, 82430108, 82293681(82293680), and 82273941); the National High-level Personnel of Special Support Program (to Dongmei Zhang and Minfeng Chen); the Natural Science Foundation of Guangdong Province (2023A1515010361 and 2022A1515011813); the Guangdong Basic and Applied Basic Research Foundation (2024B1515020098); the Science and Technology Program of Guangzhou (SL2024A04J00410, SL2024A04J00374, and SL2024A04J00280); the Fundamental Research Funds for The Central Universities (21624103); the Science and Technology Projects in Guangzhou (2023A03J1030, 202201010173, and 202102070001); and the Clinical Frontier Technology Program of the First Affiliated Hospital of Jinan University, China (JNU1AF-CFTP-2022-a01210). We thanked College of Pharmacy Public Research Platform (Jinan University) for technical support.
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.eng.2024.11.036.
PDF
(7194KB)
