Identification of differential modification sites in OSCC

Differential analysis of all lactylation sites with valid quantitative data was conducted using the screening criterion of a P-value<0.05, with significant upregulation defined by a fold change exceeding 1.5 and significant downregulation by a fold change less than 1/1.5. A total of 642 lactylation sites on 329 proteins were identified, comprising 554 sites with increased lactylation and 88 sites with decreased lactylation (Fig. 1i). This finding aligns with expectations, given the Warburg effect in tumor tissues, leading to lactate accumulation compared to normal tissues. Differential analysis results of protein expression and lactylation levels were visualized as volcano plots (Fig. 1j, k; Additional file 1: Supplemental Fig. 3). And the corresponding quantitative data were presented as heatmaps (Fig. 1l, m), revealing both upregulation and downregulation of protein expression levels in OSCC compared to NAT, with a notable widespread upregulation of lactylation levels.

Impact of lactylation on biological functions and relevant signaling pathways in OSCC

Subcellular localization analysis of differentially modified proteins indicated that 44.2% of lactylation occurred in the cytoplasm and 28.57% in the nucleus (Additional file 1: Supplemental Fig. 4a). Functional classification enrichment analysis demonstrated that lactylation impacted various metabolic activities, including glycolysis, lipid metabolism, and amino acid metabolism, potentially influencing the transport of metabolites and signaling molecules (Additional file 1: Supplemental Fig. 4b). Subsequent Kyoto Encyclopedia of Genes and Genomes (KEGG) classification enrichment analysis revealed a significant impact of lactylation on cell growth and cellular communication. Among the diseases influenced by lactylation, inflammation-related diseases exhibited the highest susceptibility, followed by tumors. Additionally, associations were observed between lactylation and neurodegenerative diseases, cardiovascular diseases, metabolic diseases, and immune-related diseases (Additional file 1: Supplemental Fig. 4c). Previous studies have underscored the significant regulatory role of lactylation in these diseases.

To comprehensively elucidate the role of lactylation in OSCC, the molecular mechanisms governed by lactylation were investigated. Functional enrichment and pathway analyses of upregulated and downregulated datasets from proteomic and lactylation omic analyses were performed. Results indicated that proteomic functional enrichment primarily encompassed gene expression and cellular functions, such as the negative regulation of gene expression, positive regulation of gene expression, cell migration, regulation of biological processes, and regulation of cellular processes (Fig. 2a). Conversely, lactylome enrichment predominantly included cellular metabolism, such as glucose metabolic processes, gluconeogenesis, and nicotinamide adenine dinucleotide (NAD+) metabolic processes. Additionally, lactylation significantly impacted signaling pathways and cellular activities, including the G protein-coupled receptor signaling pathway, signal transduction, cell surface receptor signaling pathway, regulation of cell growth, and cellular development (Fig. 2b). To determine the functional disparities among proteins with varying fold changes, these proteins were categorized into four groups based on their fold changes, denoted as Q1 to Q4 (Fig. 2c). Gene ontology (GO) and KEGG enrichment analyses were conducted for each group (Fig. 2d). Results revealed that, in addition to affecting metabolic activities, these proteins influenced mRNA activity, protein localization, signal transduction, and other processes.

Open in a separate window

Fig. 2

Proteome and lactylome enrichment analysis. a Biological functional enrichment of differential proteins within the proteome. b Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment of differential proteins in the proteome (one-way ANOVA, ns: P>0.05, *P<0.05, **P<0.01, ***P<0.001). c Differentially expressed proteins were categorized into four groups based on fold change. d Biological functions and KEGG enrichment results of proteins exhibiting various fold changes. e Biological functional enrichment of differential proteins within the lactylome. f KEGG pathway enrichment of differential proteins in the lactylome. g Differentially expressed sites were classified into four groups according to fold change. h Biological functions and KEGG enrichment results of sites displaying different fold changes (one-way ANOVA, ns: P>0.05, *P<0.05, **P<0.01, ***P<0.001)

In pathway enrichment, notable differences emerged between proteomics and lactylomics. Unlike the proteomic enrichment results, lactylation enrichment results showed elevated rankings for pathways such as glycine, serine, and threonine metabolism, the Hippo signaling pathway, T helper cell 17 (Th17) cell differentiation, the interleukin 17 (IL-17) signaling pathway, glycolysis/gluconeogenesis, and the mitogen-activated protein kinase (MAPK) signaling pathway. These pathways were predominantly concentrated in regions with heightened lactylation modifications (Fig. 2e, f). Concerning differential results, the proportion of high fold change Q4 in lactylation modification results exhibited a significantly greater magnitude compared to other groups (Fig. 2g). Enrichment analysis also underscored functions and pathways associated with tumors (Fig. 2h). This suggests that increased lactylation may play a pivotal regulatory role in these pathways during the occurrence and development of OSCC.

Lactylation-associated epigenetic regulation

To comprehensively elucidate the potential biological functions of lactylation in OSCC, its upstream regulatory role was investigated. Initially, the OSCC cell line CAL27 was exposed to 10 mM lactate for 24 h, resulting in increased lactylation levels post-treatment (Additional file 1: Supplemental Fig. 5a). Subsequently, both treated and control group samples underwent CUT&TAG-seq with a pan-anti-lactylation modification antibody, as well as ATAC-seq and RNA-seq analysis. This approach provided insights into lactylation-related epigenetic regulation, changes in chromatin accessibility, and alterations in gene expression under high lactate conditions.

Compared to the control group, CUT&TAG-seq and ATAC-seq results in the lactate treatment group showed peak aggregation of peaks near transcription start sites (TSS) suggesting that lactate enhances transcriptional processes (Additional file 1: Supplemental Fig. 5b). Separate differential analyses of CUT&TAG-seq and ATAC-seq data revealed that differentially identified peaks were primarily concentrated in promoter and distal intergenic regions, indicating significant lactylation impact on transcriptional regulation (Additional file 1: Supplemental Fig. 5c). Functional enrichment and KEGG pathway analyses of differential CUT&TAG-seq results provided further insights into functions and signaling pathways influenced by lactylation-associated histones. Enriched peaks, whether upregulated or downregulated post-lactate treatment, were predominantly linked to intracellular metabolic activities (Additional file 1: Supplemental Fig. 6a, b) and processes such as biosynthesis and gene expression. Pathway enrichment analysis indicated that lactylation-associated histones significantly influenced not only metabolic-related signaling pathways but also the phosphatidylinositol-3-kinase/protein kinase B (PI3K/AKT), hypoxia-inducible factor-1 (HIF-1), and cancer-related signaling pathways (Additional file 1: Supplemental Fig. 6c, d).

To elucidate the genes associated with lactylation-related epigenetic regulation and chromatin accessibility, a comprehensive analysis integrating CUT&TAG-seq and ATAC-seq data from the untreated control group was conducted. Peaks located within a 3-kb range upstream and downstream of the TSS were defined as essential for gene transcription. Overlapping peaks between the three datasets were considered identical, and a Venn diagram visualized this overlap (Fig. 3a). The intersection revealed 14,518 peaks where lactylation positively correlated with chromatin accessibility. Peaks exclusively in the TSS and CUT&TAG-seq data but absent in the ATAC-seq data (2390 peaks) were regarded as negatively correlated with lactylation. Peaks near the TSS present only in ATAC-seq data totaled 18,164, indicating chromatin accessibility unrelated to lactylation, potentially influenced by other epigenetic mechanisms such as acetylation. These peaks were annotated with gene information. A similar joint analysis on lactate-treated samples revealed increased lactylation-driven chromatin accessibility, concomitant with the upregulation of other chromatin accessibility mechanisms (Fig. 3b). Analyzing CUT&TAG-seq and ATAC-seq data demonstrated that lactate treatment not only increased lactylation but also enhanced chromatin accessibility (Fig. 3c, Additional file 1: Supplemental Fig. 7a, b). Previous reports suggested that lactate treatment may also elevate acetylation levels, potentially contributing to increased chromatin accessibility.

Open in a separate window

Fig. 3

Lactylation-driven expression of an oncogene set. a A heatmap depicts alterations in lactylation-associated gene expression and chromatin accessibility pre- and post-lactate treatment in the oral squamous cell carcinoma (OSCC) cell line CAL27. b Correlation analysis between lactylation-driven gene expression levels and chromatin accessibility. c Variations in lactylation-driven genes and chromatin accessibility in CAL27 cells exposed to high lactate concentrations. d Enrichment analysis of lactylation-driven genes and gene sets positively and negatively correlated with chromatin accessibility via single-sample gene set enrichment analysis (ssGSEA). e Intersection analysis of lactylation-driven gene sets positively correlated with chromatin accessibility and differentially expressed genes, determined by cleavage under targets and tagmentation sequencing (CUT&TAG-seq), assay for transposase-accessible chromatin with high-throughput sequencing (ATAC-seq), and RNA sequencing (RNA-seq). f Validation of increased lactylation enrichment in the promoter regions of CASP3 and TPR in OSCC tissues through chromatin immunoprecipitation-quantitative polymerase chain reaction (ChIP-qPCR, two-tailed unpaired Student’s t-test, **P<0.01) and prediction of promoter sequences potentially affected by lactylation. g Validation of elevated expression of CASP3, HSP90AA1, HSP90B1, TPR, RAS, and CCND1 in OSCC tissues through reverse transcription-quantitative polymerase chain reaction (RT-qPCR, two-tailed paired Student’s t-test, *P<0.05; **P<0.01). Lac: lactate

Integrated multi-omics analysis: from epigenetics to transcription for identifying lactylation-driven oncogenes

To further investigate downstream functions governed by genes associated with lactylation and chromatin accessibility, we annotated the aforementioned positively and negatively correlated peaks to genes. Subsequently, ssGSEA was conducted based on the enrichment intensity of these genes in chromatin, allowing determination of their potential functions and relevant signaling pathways (Fig. 3d). Given the significantly higher number of positively correlated genes in CUT&TAG-seq and ATAC-seq compared to negatively correlated genes, and considering the likely predominant role of lactylation in promoting tumor progression in OSCC development, particular focus was placed on the mechanisms of oncogenes exhibiting increased chromatin accessibility driven by lactylation.

Enriched pathways of positively correlated genes exhibiting elevated scores in both CUT&TAG-seq and ATAC-seq samples were analyzed (Additional file 1: Supplemental Fig. 8). Lactylation-driven genes enriched in these pathways were retrieved and intersected with differential results from CUT&TAG-seq, ATAC-seq, and subsequent RNA-seq analyses. The results were visualized using an upset plot (Fig. 3e), revealing 217 lactylation-driven genes with increased expression, suggesting a potential oncogene set influenced by lactylation in OSCC.

Experimental validation of the aforementioned genes was performed. Chromatin immunoprecipitation-quantitative polymerase chain reaction (ChIP-qPCR) employing a pan-lactylation antibody was conducted on OSCC tissues and NAT to validate the transcriptional status of lactylation-driven genes (Fig. 3f). The results showed increased lactylation modifications proximal to the promoters of CASP3 and TPR in tumor tissues compared to NAT samples, consistent with sequencing results. Concurrently, motif sequence information of lactylation-driven gene expression was predicted. RNA extracted from OSCC tissues and paired NAT underwent RT-qPCR to validate transcriptional alterations of select potential oncogenes (Fig. 3g). The results demonstrated significant upregulation of CASP3, HSP90AA1, HSP90B1, TPR, RAS, and CCND1 in OSCC relative to NAT, aligning with previous analyses. This suggests these key genes may be influenced by lactylation, contributing to their upregulated expression.

Immunohistochemistry

Serial 4-μm thick tissue sections were stained using the Chem Mate Envision system (K5007, Dako). Antigen retrieval was conducted with 1 mM ethylene diamine tetraacetic acid (EDTA). The sections were then incubated with a pan anti-lactyllysine antibody (PTM-1401RM, PTM-Bio) at a 1:100 dilution. Immunocomplexes were visualized using the 3,3′-Diaminobenzidine Liquid Plus Substrate Plus Chromogen System (Dako, Glostrup, Denmark) and counterstained with hematoxylin for light microscopy examination. Two pathologists independently evaluated the immunostaining results, blinded to patient information.

Protein extraction

Tissues stored at – 80 °C were placed on ice, cut into 1 mm³ pieces using sterile scissors, and added to a 1% Triton buffer in a 4:1 ratio containing 1×protease inhibitor cocktail and 1×phenylmethanesulfonyl fluoride. The tissues were homogenized, incubated on ice for 30 min, sonicated, and centrifuged at 12,000 g for 10 min at 4 °C. For cell samples, the medium was discarded, cells were washed twice with phosphate-buffered saline (PBS), lysed in an appropriate amount of lysis buffer with 1×protease inhibitor cocktail, and incubated on ice for 30 min. Cells were then transferred to a centrifuge tube and centrifuged at 12,000 g for 10 min at 4 °C. The resulting supernatant was the tissue protein extract, and its protein concentration was determined using a bicinchoninic acid assay kit (P0010, Beyotime). The extract was stored at – 80 °C for future experiments.

Trypsin digestion

Equal amounts of each protein sample were enzymatically digested [62]. The volume was adjusted to match that of the lysis buffer, followed by the gradual addition of 20% trichloroacetic acid and thorough vortexing. The mixture was precipitated at 4 °C for 2 h, centrifuged at 4500 g for 5 min, and the supernatant discarded. The precipitate underwent two to three washes with pre-cooled acetone. After air-drying, 200 mM triethylammonium bicarbonate was added, and the mixture was sonicated to disperse the precipitate. Trypsin was introduced at a 1:50 (enzyme to protein) ratio for overnight digestion. Dithiothreitol was then added to a final concentration of 5 mM, and the mixture was reduced at 56 °C for 30 min. Iodoacetamide was subsequently added to a final concentration of 11 mM, and the mixture was incubated at 25 °C in the dark for 15 min.

Liquid chromatography-mass spectrometry (LC–MS) analysis

Peptide segments were dissolved in mobile phase A of liquid chromatography and separated using the Easy-nLC1000 Ultra-High-Performance Liquid Chromatography System. Mobile phase A comprised an aqueous solution containing 0.1% formic acid and 2% acetonitrile, while mobile phase B consisted of acetonitrile–water with 0.1% formic acid. The liquid phase gradient was programmed as follows: 0–70 min, transitioning from 9 to 25% B; 70–82 min, from 25 to 35% B; 82–86 min, from 35 to 90% B; and 86–90 min, at 90% B, maintaining a flow rate of 450 nL/min. Following separation using the ultra-high-performance liquid chromatography system, the peptide segments were ionized through the capillary ion source for ionization and subsequently analyzed with the TimsTOF Pro mass spectrometer.

Modification enrichment

Peptide segments were dissolved in an immunoprecipitation buffer (containing 100 mM NaCl, 1 mM EDTA, 50 mM Tris–HCl, 0.5% NP-40, pH 8.0). The supernatant was transferred to pre-washed resin (PTM1404, PTM-Bio) and placed on a rotating shaker at 4 °C for gentle overnight incubation. The resin was then sequentially washed four times with IP buffer and twice with deionized water. Peptides bound to the resin were eluted with 0.1% trifluoroacetic acid solution, with three elutions performed. The eluate was collected and subjected to vacuum freeze-drying. After drying, desalting was performed following the C18 ZipTips instructions. A final round of vacuum freeze-drying was done, and the samples were prepared for LC–MS analysis.

Database search

Secondary mass spectrometry data was analyzed using MaxQuant [63] (v1.6.15.0). The database utilized was Homo_sapiens_9606_SP_20230103.fasta, containing 20,389 sequences. To assess the false discovery rate from random matches, a decoy database was employed. Additionally, a common contaminant database was included to mitigate the impact of contaminating proteins on the identification results. The mass tolerance for first-level precursor ions for both the initial and primary searches was set at 20 ppm, as was the mass tolerance for second-level fragment ions.

ATAC-seq

Nuclei were extracted from the samples, and the nuclei pellet was reconstituted using a Tn5 transposase reaction mixture. Equimolar amounts of Adapter 1 and Adapter 2 were added post-transposition, followed by PCR amplification of the library. The Illumina Novaseq platform was then utilized for library sequencing, producing 150-bp paired-end reads [64].

CUT&TAG-seq

Cells were immobilized onto Concanavalin A-coated magnetic beads, and digitonin was used to permeabilize the cellular membrane. The pA-Tn5 transposase enzyme, guided by a corresponding antibody, precisely bound to the DNA sequence near the target protein, resulting in factor-specific tagmentation. An anti-pan-lactyllysine antibody (PTM-Bio, PTM-1401RM) was used in this experiment. DNA sequences were tagmented, with adapters affixed to both ends. These tagged sequences were then amplified through PCR to generate sequencing-ready libraries. The libraries were purified using AMPure beads, and library quality was assessed using the Agilent Bioanalyzer 2100 System [65].

RNA-seq

Total RNA served as the starting material for RNA sample preparations. In brief, mRNA was isolated from the total RNA using magnetic beads conjugated to poly-T oligonucleotides. Complementary DNA (cDNA) products were generated via reverse transcription. PCR was performed using Phusion High-Fidelity DNA polymerase, universal PCR primers, and Index Primer. Finally, PCR products were purified using the AMPure XP System, and library quality was assessed using the Agilent Bioanalyzer 2100 System. The raw data in fastq format (raw reads) underwent preprocessing using custom Perl scripts.

Single sample gene set enrichment analysis (ssGSEA)

To elucidate the biological roles of gene sets influenced by lactylation-driven expression and chromatin accessibility, ssGSEA was performed [66]. This method computes a score for each sample by correlating it with the relevant gene set, converting the gene expression profile into an enrichment profile. This transformation allows for the evaluation of the significance of various functional signaling pathways. For this analysis, the “c2.cp.kegg.v6.2.-symbols” gene set was sourced from the MSigDB database [67].

Chromatin immunoprecipitation (ChIP)

Tissue samples were weighed, with each ChIP experiment utilizing 25 mg of tissue. The samples were kept at a low temperature to prevent protein degradation and finely minced with sterile scissors. After rinsing with PBS, formaldehyde was added to a final concentration of 1%, followed by a 10-min incubation at 37 °C. ChIP sequencing (ChIP-seq) was conducted using a ChIP-seq kit (9005S, Cell Signaling Technology) according to the provided protocol. An antibody against pan-lactyllysine (PTM-Bio, PTM-1401 RM) was employed in the experiment. Finally, DNA was purified using a DNA purification kit and analyzed via quantitative PCR (qPCR) using specific primers. Detailed primer information for ChIP-qPCR is provided in Additional file 2: Supplemental Table 2.

Real-time fluorescence qPCR (RT-qPCR)

Tissues were maintained at low temperatures and finely minced. RNA was extracted and purified using the RNA-Quick Purification Kit (RN001, ESscience) following the manufacturer’s guidelines. The RNA was then reverse transcribed into cDNA using the RT Reagent Kit with a gDNA Eraser (RR047A, Takara). qRT-PCR was performed using the Bio-Rad CFX96 Real-Time PCR Detection System with SYBR Green PCR Master Mix (A25741, Thermo Fisher Scientific). Primer sequences for qRT-PCR are listed in Additional file 2: Supplemental Table 3.

Cell culture

The OLK cell line LEUK1, the OSCC cell line CAL27, and HEK293T cells were obtained from the laboratory of Peking University School and Hospital of Stomatology. LEUK1 cells were cultured in 1640 medium, while CAL27 and HEK293T cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM). All cells were supplemented with 10% fetal bovine serum and incubated at 37 °C in a cell culture incubator with 5% carbon dioxide. In the treatment group, cells were exposed to 10 mM lactate for 24 h. The cell lines used have all been authenticated and are free from mycoplasma contamination.

Lentiviral transduction and transfection

Gene knockout experiments were conducted using short hairpin RNAs (shRNAs) delivered via lentiviral transduction. The PLKO.1 plasmid, containing the shRNA-DHX9 targeting sequence, was co-transfected with psPAX2 and PMD2.G plasmids in a 2:2:1 ratio into HEK293T cells, producing the shRNA-DHX9 lentivirus. Concurrently, co-transfection of pCDH-DHX9-WT/K146R/K146T, psPAX2, and PMD2.G plasmids (in a 2:2:1 ratio) into HEK293T cells generated a lentivirus overexpressing DHX9-WT/K146R/K146T genes. Post-transfection, cell supernatant containing lentiviral particles was collected, filtered through a 0.45-mm filter, and used to infect target cells. Polybrene was added at 8 μg/mL, followed by selection with 5 μg/mL puromycin to identify stable transfectants. The shRNA target sequences employed are detailed in Additional file 2: Supplemental Table 4.

Co-immunoprecipitation (co-IP)

Cell lysates were prepared using a lysis buffer consisting of 150 mM sodium chloride, 20 mM HEPES (pH 7.4), 2 mM EDTA, 1.5 mM magnesium chloride, and 0.5% NP-40. A protease inhibitor cocktail was added to the lysates, and co-immunoprecipitation (co-IP) was performed using an anti-DHX9 antibody (ab26271, Abcam). The immunocomplexes were incubated with protein A/G-sepharose (Invitrogen) at 4 °C for 6 h, followed by six washes with IP buffer, and eluted at 95 °C for 15 min. The eluted protein samples were then subjected to western blot analysis.

Western blot

Protein samples were combined in a 4:1 ratio with 5×loading buffer for sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and heated at 95 °C for 10 min. The proteins were separated via SDS-PAGE, transferred onto a nitrocellulose filter membrane, and blocked with non-fat milk. The membrane was probed with primary antibodies (anti-DHX9, ab26271, Abcam, diluted 1:1000; pan-anti-lactyllysine, PTM-1401RM, PTM-Bio, diluted 1:1000; anti-β-actin, 81115–1-RR, Proteintech, diluted 1:5000) overnight. Following rinsing, the blots were incubated with secondary antibodies and finally exposed.

Cell counting kit-8 (CCK8)

Cell proliferation was quantified using a CCK8 assay kit (CK001, Lablead). Cells were seeded in 96-well plates at a density of 4×10³ cells per well, with 100 μL of complete culture medium added to each well. At specified intervals, the medium was replaced with 100 μL of DMEM containing 10 μL of CCK8 dye. Optical density at 450 nm was measured using a microplate reader (Biotec). This experiment was performed in triplicate, with statistical analysis and visualization conducted using R package software (version 4.0.2).

Cloning formation assay

Cells from various groups were seeded in six-well plates at a density of 1×10³ cells per well and incubated for 2 weeks. Subsequently, the culture medium was discarded, and cells were fixed with 4% paraformaldehyde for 10 min, followed by staining with 0.5% crystal violet. Each experiment was independently replicated three times. The results were analyzed using ImageJ software, and quantitative data were further subjected to statistical analysis and visualization with R package software (version 4.0.2).

Wound healing assay

Cells designated for testing were seeded in 96-well plates and cultured to form a confluent monolayer. Following pre-treatment with serum-free medium for 12 h, a straight-line scratch was created in the cell monolayer using a sterile pipette tip. Cells were washed twice with PBS to remove debris and then incubated in a serum-free medium. Cell migration was monitored at 0, 12, and 24 h under an inverted microscope, with images captured at each time point. The cell migration rate (%) was calculated using the formula: (wound distance at 0 h−wound distance at 12 h/24 h)/wound distance at 0 h. The experiment was recorded and analyzed using the IncuCyte ZOOM system from Essen BioScience.

Transwell assay

In invasion experiments, 6 μL of matrix gel combined with 54 μL of serum-free culture medium was pre-incubated in a chamber at 37 °C for 6 h. The tested cells were suspended in a serum-free culture medium and introduced into the chambers (5×10⁴ cells per well). The lower chamber was supplemented with a medium containing 10% serum. After 48 h of incubation, cells on the upper surface were removed with a cotton swab, while those on the lower surface were fixed with 4% paraformaldehyde, stained with Giemsa, and quantified under a microscope to assess the number of cells that traversed the membrane. Each experiment was conducted independently in triplicate. Quantitative data obtained from the experimental results were analyzed using ImageJ software, followed by statistical analysis and visualization with R software (version 4.0.2).

The authors express their gratitude to Jingjie PTM BioLab Incoporated Company (Hangzhou) for their assistance with quantitative proteomic and lactylomic analysis.