TTK Defines a High-Risk Oral Squamous Cell Carcinoma Subtype Through Dual mTORC1/NF-κB Activation.

For single-cell RNA sequencing, quality control was applied to the raw single-cell TPM (Transcripts Per Kilobase Million) expression matrix. Cells were flagged as outliers if they deviated by more than 5 MADs (Median Absolute Deviations) in log1p-transformed total TPMs, the number of genes expressed by TPMs, or the percentage of TPMs in the top 20 genes. Doublets were identified and removed using the "scDblFinder" package. The "scvi" method 33 was then used to integrate cells across samples in Seurat, 34 incorporating batch information. Cell clusters were identified with the "Leiden" algorithm, and robust cluster resolution was established. Cell types were annotated using marker genes generated by the "FindAllMarkers" function and validated against public marker databases. For array-based expression profiling, the raw expression matrix was normalized using the "Robust Multi-array Average" (RMA) algorithm implemented in the "oligo" package, 35 followed by filtering of absent probes. For datasets such as GSE20599 and GSE41116 that lacked annotation files, transcript annotation was generated by realigning to GRCh38 with the "Rsamtools" package and a transcript file. For The Cancer Genome Atlas (TCGA) data, RNA Sequencing (RNA-seq) raw count files, Whole Exome Sequencing (WES) mutation annotation format (MAF) files, copy number variation (CNV) segment files, and clinical information were downloaded and processed with the "TCGAbiolinks" 36 and "maftools" 37 packages. CNV scores were then computed using "GISTIC2" 38 software. For bulk RNA-seq data from SCC-15 and CAL-27 cell lines, raw FASTQ files were quality-checked with "fastqc" and trimmed using "trim_galore" software. The cleaned reads were aligned to the GRCh38 reference genome with "STAR," 39 and count files were generated using "featureCounts" software. The "Monocle3" 40 pipeline was used for trajectory analysis. A monocle object was created, preprocessing and dimensionality reduction were performed, and cells were clustered with the "cluster_cells" function, substituting the UMAP embeddings generated from Seurat. The trajectory graph was then constructed, and cells were ordered in pseudotime. For single-cell RNA-seq data, pathway activity scores were calculated with the "AUCell" 41 package. Rankings were generated using the "AUCell_buildRankings" function applied to the log2(TPM + 1) expression matrix. Pathway activity scores were computed with the "AUCell_calcAUC" function, incorporating gene sets from “MigSigDB.” For bulk RNA-seq data, genes expressed in fewer than ten samples were excluded. Pathway activity scores were computed on the log2(TPM + 1)-normalized matrix using the "GSVA" 42 package with the parameters "method = 'ssgsea',” "kcdf = 'Gaussian',” "abs.ranking = TRUE,” "min.sz = 15,” "max.sz = 500". Min-Max normalization was then applied to each gene set across samples. Univariate Cox regression was performed by integrating ssGSEA (single-sample Gene Set Enrichment Analysis) scores with clinical data through the "survival" package. Pearson correlation coefficients were then calculated between log2(TPM + 1)-normalized gene expression and min-max normalized ssGSEA scores. Differential gene expression analysis between groups was performed using the "DESeq2" 43 package after excluding low-expression genes. Genes with an absolute "log2FoldChange" greater than 1.5 and an adjusted p-value ("padj") less than 0.05 were defined as significantly differentially expressed. For single-cell marker genes, enrichment was analyzed using the "clusterProfiler," 44 "org.Hs.eg.db,” and "enrichplot" packages with gene sets from “MigSigDB” 45 . Enrichment categories were clustered and visualized with the "cnetplot" function. For differentially expressed genes from bulk RNA-seq, pathway enrichment was carried out using "Metascape." 46 ( https://metascape.org ) with the following parameters: Min Overlap = 3, P Value Cutoff = .01, Min Enrichment = 1.5. WES MAF files and GISTIC2 CNV results were analyzed using the "maftools" package. Samples were stratified into TTK-high and TTK-low groups based on TTK RNA expression. Mutation profiles were compared, and tumor mutational burden (TMB) was calculated. Comparative analyses included odds ratios and p-value calculations. Mutation patterns were extracted and mapped to COSMIC single-base substitution (SBS) signatures 47 using the "deconstructSigs" 48 package with the "whichSignatures" function. After excluding low-weight signatures, signature scores were recalculated, and normalized weights were compared using the Wilcoxon test. Total cosine similarity between groups was determined using signature extraction in "maftools." The human OSCC cell lines SCC-15 and CAL-27 were cultured in DMEM supplemented with 10% heat-inactivated fetal bovine serum and 1% penicillin–streptomycin. Cultures were maintained at 37°C in a humidified incubator with 5% CO₂. Stable knockdown cell lines were established using a lentiviral system with 3 shRNA sequences (shTTK1, shTTK2, shTTK3). Lentiviral packaging plasmids (psPAX2, pMD2.G) were transfected into 293T cells to generate viral supernatants. SCC-15 and CAL-27 cells were infected with viral particles and selected with puromycin (1 µg/mL) to obtain stable clones. Protein knockdown efficiency was confirmed by Western blot, with knockdown defined as a reduction of protein levels by more than 60% in at least 2 shRNA groups compared to controls. 293T cells were chosen for lentiviral packaging because of their high transfection efficiency and robust viral production, ensuring effective delivery of shRNA constructs into OSCC cells. Total protein was extracted using RIPA buffer, and concentrations were measured with the BCA assay. Protein samples were mixed with 6 × loading buffer, heated at 100°C, and quantified against a standard curve generated with protein standards. Protein samples (30 µg) were separated on 10% SDS-PAGE gels and transferred onto NC membranes. Membranes were blocked with 5% non-fat milk and incubated overnight at 4°C with primary antibodies. Following incubation with HRP-conjugated secondary antibodies, protein bands were detected using ECL reagents and imaged with the ChemiDoc system. Total RNA was extracted from SCC-15 and CAL-27 cells, with 3 biological replicates per group (control and knockdown). RNA libraries were prepared and sequenced, generating a total of 12 samples. 293T cells were transfected with 6 µg Flag-TTK plasmid and lysed after 48 hours in IP buffer. Lysates were incubated with Flag-beads overnight at 4°C. Beads were washed, resuspended in loading buffer, and analyzed by Western blot. For mass spectrometry, a portion of the protein lysates was separated by SDS-PAGE and stained with Coomassie blue to confirm protein integrity before analysis. Cell proliferation was evaluated using the MTT assay. SCC-15 and CAL-27 cells were seeded in 96-well plates at a density of 5,000 cells per well. Cell viability was assessed at 0, 24, 48, 72, 96, and 120 hours by adding MTT solution, incubating for 3–4 hours, and dissolving the resulting formazan crystals in DMSO. Absorbance was measured at 490 nm. Cell migration and invasion were analyzed using Transwell chambers with or without Matrigel coating. A total of 40,000 cells were seeded in the upper chamber in serum-free medium, while medium containing 10% FBS was added to the lower chamber. After 24 hours, migrated or invaded cells were fixed with 4% paraformaldehyde, stained with crystal violet, and imaged microscopically for quantification. SCC-15 and CAL-27 cells were seeded in 6-well plates at a density of 800 cells per well. After 7 days, colonies were fixed with 4% paraformaldehyde, stained with crystal violet, and counted under a microscope. Cells were seeded in 96-well plates and exposed to a 10-point cisplatin concentration gradient with a maximum of 4 µM. Cell viability was evaluated using the MTT assay after 24 hours. IC50 values were determined by plotting cell viability against cisplatin concentrations using non-linear regression in GraphPad Prism. Four-week-old male immunodeficient nude mice were injected subcutaneously with 3 × 10⁶ SCC-15 cells mixed 1:1 with Matrigel. Tumor volumes were measured every 3 days, and cisplatin was administered intraperitoneally every other day once tumors reached 100–200 mm³. At 30 days post-injection, tumors were excised, weighed, and processed for further analysis. FFPE sections (4 µm) were deparaffinized, rehydrated, and subjected to heat-induced epitope retrieval (citrate pH 6.0 or EDTA pH 9.0). Endogenous peroxidase was quenched (3% H₂O₂, 25 min), slides were washed in PBS (pH 7.4), blocked (3% BSA, 30 min), and incubated with primary antibodies (dilutions in Materials), overnight at 4°C. After PBS washes, an HRP-polymer secondary antibody was applied (50 min), signal was developed with DAB, nuclei were counterstained with hematoxylin, and slides were dehydrated, cleared in xylene, and mounted. Immunostaining was quantified using the H-score (range 0-300). For each case, cells were classified by intensity as 0 (negative), 1 (weak), 2 (moderate), or 3 (strong), and the percentage of cells at each level (P0–P3; sum = 100%) was estimated. The H-score was calculated as H-score = 1 × P1 + 2 × P2 + 3 × P3. Antibodies used in this study: Anti-TTK (Abcam, ab11108), Anti-mTOR (CST, #2983), Anti-p-mTOR (CST, #5536), Anti-p65 (Proteintech, 66535-1-Ig), Anti-p-p65 (CST, #3033), Anti-S6K (CST, #34475), Anti-p-S6K (CST, #9234), Anti-4EBP1 (#9644), Anti-p-4EBP1 (Santa Cruz Biotechnology, sc-293124), Anti-p38 (CST, #8690), Anti-p-p38 (CST, #4511), Anti-JNK (CST, #9252), Anti-p-JNK (CST, #4668), Anti-TAK1 (Abcam, ab109526), Anti-p-TAK1 (CST, #4508), Anti-TAB1 (Proteintech, 27566-1-AP), Anti-TAB3 (Abcam, ab124723), Anti-IKKα (Abcam, ab32041), Anti-IKKβ (Abcam, ab124957), Anti-p-IKKα/β (CST, #2697), Anti-IKBα (CST, #4812), Anti-p-IKBα (CST, #2859), Anti-c-Myc (Santa Cruz Biotechnology, sc-42), Anti-Cyclin D1 (Proteintech, 26939-1-AP), Anti-E-cadherin (Proteintech, 20874-1-AP), Anti-N-cadherin (Proteintech, 22018-1-AP), Anti-ZEB1 (Proteintech, 21544-1-AP), Anti-TWIST (Proteintech, 25465-1-AP), Anti-MMP2 (Proteintech, 10373-2-AP), Anti-rabbit IgG (Santa Cruz Biotechnology, sc-2004), Anti-mouse IgG (Santa Cruz Biotechnology, sc-2005), For Western blotting, the primary antibody was used at a dilution of 1:1000 and the secondary antibody at 1:2500. For immunohistochemistry, the primary antibody was applied at a dilution of 1:200 and the secondary antibody at 1:300. Database used in this study: TCGA (OSCC, extracted from HNSC), GEO ( GSE103322 , GSE25099 , GSE41116 , GSE30784 ). We prioritized OSCC-specific cohorts with adequate sample size, raw data availability, and platform compatibility to enable harmonized preprocessing and cross-cohort comparability. HNSC samples not meeting OSCC criteria were excluded to avoid confounding.

Tumors present significant therapeutic challenges due to their marked heterogeneity. Defining tumor evolutionary trajectories and identifying highly malignant subsets are key steps in developing effective therapeutic strategies, and advances in single-cell sequencing now provide powerful tools to achieve this. Single-cell RNA sequencing of 5,895 cells from 18 OSCC patients ( GSE103322 ) identified 15 distinct cell populations, including immune cells, stromal cells, and tumor cells ( Figure 1 A). Tumor cells were further classified into 5 subgroups (C1-C5), reflecting their intrinsic heterogeneity. Because tumor cells evolve asynchronously, genomic alterations and positive selection drive their divergent progression. By applying trajectory analysis, we defined the evolutionary relationships among heterogeneous tumor cell subpopulations. Pseudotime analysis showed that all tumor subgroups originated from C2 and subsequently branched into 2 distinct developmental trajectories ( Figure 1 B). Differential expression analysis revealed significant enrichment of pathways such as NF-κB, DNA repair, MYC targets, and mTORC1 signaling within subgroup C2 ( Figure 1 C). These pathways are commonly associated with increased tumor aggressiveness. Univariate Cox regression analysis in the TCGA cohort further confirmed the prognostic relevance of these pathways, particularly mTORC1 and NF-κB, which were linked to worse clinical outcomes ( Figure 1 D). Additional analysis showed that mTORC1 was primarily activated in C2 and C3, whereas NF-κB was predominantly activated in C1 and C2, consistent with the tumor development trajectory ( Figure 1 B, E–F). Notably, the concurrent activation of mTORC1 and NF-κB in subgroup C2 suggests an elevated potential for tumor proliferation, invasion, and metastasis ( Figure 1 G). Fig. 1 Single-cell RNA sequencing reveals a highly malignant OSCC subgroup with co-activation of mTOR and NF-κB signaling. (A) UMAP visualization of 5,895 single cells from 18 OSCC patients ( GSE103322 ), annotated into 15 distinct clusters based on transcriptomic profiles and canonical marker genes. Cell types include immune (CD8+ T, Tregs, macrophages), stromal (fibroblasts, endothelial), and malignant epithelial subsets (C1–C5). (B) Pseudotime trajectory inferred with Monocle3 shows that C2 serves as the developmental root of the epithelial lineage, giving rise to divergent tumor evolutionary branches. (C) Enrichment analysis of hallmark gene sets across tumor clusters reveals that C2 is uniquely enriched for NF-κB signaling, mTORC1 signaling, MYC targets, and unfolded protein response; gene set activity inferred using AUCell scores. Circle size represents the number of genes per pathway; color reflects adjusted P value. (D) Univariate Cox regression analysis using ssGSEA-derived pathway scores from TCGA OSCC cohort (n = 338) identifies mTORC1 and NF-κB signaling as significantly associated with worse survival; hazard ratios and 95% confidence intervals shown. (E–F) UMAP projections of AUCell scores indicate strong mTORC1 activation in C2/C3 (E) and NF-κB activation in C1/C2 (F). (G) Overlay of mTORC1 (red) and NF-κB (green) AUCell activity reveals robust co-activation predominantly in C2, defining a transcriptionally aggressive tumor subpopulation. Fig 1 Single-cell RNA sequencing reveals a highly malignant OSCC subgroup with co-activation of mTOR and NF-κB signaling. (A) UMAP visualization of 5,895 single cells from 18 OSCC patients ( GSE103322 ), annotated into 15 distinct clusters based on transcriptomic profiles and canonical marker genes. Cell types include immune (CD8+ T, Tregs, macrophages), stromal (fibroblasts, endothelial), and malignant epithelial subsets (C1–C5). (B) Pseudotime trajectory inferred with Monocle3 shows that C2 serves as the developmental root of the epithelial lineage, giving rise to divergent tumor evolutionary branches. (C) Enrichment analysis of hallmark gene sets across tumor clusters reveals that C2 is uniquely enriched for NF-κB signaling, mTORC1 signaling, MYC targets, and unfolded protein response; gene set activity inferred using AUCell scores. Circle size represents the number of genes per pathway; color reflects adjusted P value. (D) Univariate Cox regression analysis using ssGSEA-derived pathway scores from TCGA OSCC cohort (n = 338) identifies mTORC1 and NF-κB signaling as significantly associated with worse survival; hazard ratios and 95% confidence intervals shown. (E–F) UMAP projections of AUCell scores indicate strong mTORC1 activation in C2/C3 (E) and NF-κB activation in C1/C2 (F). (G) Overlay of mTORC1 (red) and NF-κB (green) AUCell activity reveals robust co-activation predominantly in C2, defining a transcriptionally aggressive tumor subpopulation. The newly identified tumor cell subgroups were present in varying proportions across patients. Their distinct pathway activation states and strong association with malignant proliferation prompted further investigation of these populations. Because pathway activation typically correlates with altered expression of upstream and downstream genes, we collected and analyzed multiple OSCC datasets totaling 691 cases ( GSE25099 , GSE41116 , GSE30784 , TCGA-OSCC). Genes positively correlated with mTORC1 and NF-κB scores were identified, yielding 20 genes consistently linked to mTORC1 ( Figure 2 A) and 33 to NF-κB ( Figure 2 B) across all datasets. Intersecting these lists revealed 17 shared genes ( Figure 2 C), which were analyzed for protein-protein interactions using STRING. This analysis identified TTK, DLGAP5, and CCNA2 as central hub genes ( Figure 2 D). Of note, TTK and DLGAP5 have previously been reported to associate with stemness in head and neck squamous cell carcinoma. These genes were also markers of the C2 subgroup ( Figure 2 G). Given the availability of clinical inhibitors for TTK, we focused on its functional role in OSCC. Immunohistochemical analysis of OSCC tissues and adjacent normal oral mucosa revealed a significantly elevated expression of TTK in tumors. Higher TTK expression correlated with advanced tumor grade ( Figure 2 E) and poorer patient survival ( Figure 2 F), underscoring its oncogenic potential. Fig. 2 TTK aligns with mTORC1/NF-κB activation across cohorts and associates with grade and outcome. (A–B) Venn diagrams showing genes positively correlated (Pearson r > 0.5) with mTORC1 (A) or NF-κB (B) pathway scores across 4 bulk RNA-seq OSCC datasets (TCGA, GSE25099 , GSE41116 , GSE30784 ). (C) Intersection analysis identifies 17 genes consistently co-correlated with both pathways, suggesting shared regulatory drivers. (D) STRING network of the 17 shared genes highlights TTK, DLGAP5, and CCNA2 as central hub nodes within the co-activation module. (E) TTK expression across tumor grades in TCGA OSCC cohort shows significant upregulation with increasing histological grade. (F) Kaplan–Meier survival curves stratified by TTK expression, showing significantly worse prognosis in TTK-high patients. (G) Heatmap of single-cell transcriptomes across tumor and microenvironmental populations shows elevated expression of TTK, DLGAP5, and CCNA2 predominantly in C2, confirming subtype-specific enrichment. Boxplots display data quartiles; ** P < .01, **** P  0.5) with mTORC1 (A) or NF-κB (B) pathway scores across 4 bulk RNA-seq OSCC datasets (TCGA, GSE25099 , GSE41116 , GSE30784 ). (C) Intersection analysis identifies 17 genes consistently co-correlated with both pathways, suggesting shared regulatory drivers. (D) STRING network of the 17 shared genes highlights TTK, DLGAP5, and CCNA2 as central hub nodes within the co-activation module. (E) TTK expression across tumor grades in TCGA OSCC cohort shows significant upregulation with increasing histological grade. (F) Kaplan–Meier survival curves stratified by TTK expression, showing significantly worse prognosis in TTK-high patients. (G) Heatmap of single-cell transcriptomes across tumor and microenvironmental populations shows elevated expression of TTK, DLGAP5, and CCNA2 predominantly in C2, confirming subtype-specific enrichment. Boxplots display data quartiles; ** P < .01, **** P < .0001. TTK serves as both a marker of the newly identified malignant subgroup and a potential therapeutic target. An effective therapeutic target not only regulates multiple critical signaling pathways but also provides therapeutic benefits across several levels. Genomic instability often creates opportunities for immunotherapy, and highly proliferative subgroups frequently display this feature. Aberrant activation of the spindle assembly checkpoint further contributes to increased genomic instability. We therefore systematically investigated the genomic characteristics of TTK-high tumors. Compared with TTK-low tumors, TTK-high tumors exhibited higher TP53 mutation rates (73% vs 62%) and harbored unique translation start site mutations ( Figure 3 A-B). Analysis of SBS patterns revealed enrichment of SBS2 in TTK-high patients, consistent with APOBEC enzyme activity ( Figure 3 C). TTK-high samples also demonstrated greater cosine similarity with SBS2 (73.4%) than TTK-low samples (71.2%) ( Figure 3 D). Furthermore, TTK-high tumors displayed significantly elevated TMB and persistent TMB (pTMB) ( Figure 3 E-F). The TTK-high group exhibited unique mutations in NALCN and ZNFX1, as well as significantly higher mutation frequencies in TP53, FMN2, KMT2D, and NSD1. Conversely, PKHD1 and RAC1 mutations occurred more frequently in the TTK-low group ( Figure 3 G). CNV analysis revealed distinct amplifications (e.g., Chr1p12, Chr3p13, Chr17q12) and deletions (e.g., Chr3p12.3, Chr14q13.2) in TTK-high tumors ( Supplementary Figure 1 A–B and Supplementary Figure 2 ), highlighting genes associated with these CNVs. Collectively, these findings suggest that tumors with high TTK expression exhibit greater TMB and genomic instability, underscoring their potential relevance for immunotherapeutic interventions. Fig. 3 TTK is associated with genomic instability in OSCC. (A–B) Waterfall plots of somatic mutations in TCGA-OSCC samples stratified by TTK expression (n = 167 per group). TTK-high tumors (B) exhibit increased TP53 mutations and unique alterations at translation start sites compared to TTK-low tumors (A). (C) Signature analysis reveals elevated APOBEC-associated SBS2 scores in TTK-high tumors. (D) Cosine similarity analysis of aggregate SBS profiles confirms higher SBS2 similarity in TTK-high versus TTK-low tumors. (E–F) Boxplots show significant increases in tumor mutational burden (E) and persistent TMB (F) in TTK-high tumors. (G) Forest plot comparing mutation frequencies between TTK-high and TTK-low tumors; genes with statistically significant enrichment (e.g., RAC1 in TTK-low; NALCN in TTK-high) are highlighted. * P < .05, ** P < .01. Fig 3 TTK is associated with genomic instability in OSCC. (A–B) Waterfall plots of somatic mutations in TCGA-OSCC samples stratified by TTK expression (n = 167 per group). TTK-high tumors (B) exhibit increased TP53 mutations and unique alterations at translation start sites compared to TTK-low tumors (A). (C) Signature analysis reveals elevated APOBEC-associated SBS2 scores in TTK-high tumors. (D) Cosine similarity analysis of aggregate SBS profiles confirms higher SBS2 similarity in TTK-high versus TTK-low tumors. (E–F) Boxplots show significant increases in tumor mutational burden (E) and persistent TMB (F) in TTK-high tumors. (G) Forest plot comparing mutation frequencies between TTK-high and TTK-low tumors; genes with statistically significant enrichment (e.g., RAC1 in TTK-low; NALCN in TTK-high) are highlighted. * P < .05, ** P < .01. Because TTK was identified as a marker and potential regulatory gene of the newly discovered highly proliferative subgroup, we first evaluated its role in tumor cell proliferation. SCC-15 and CAL-27 cell lines, widely used OSCC models, were chosen to generate stable TTK knockdown lines with a lentiviral system. Western blotting confirmed a marked reduction in TTK protein levels in knockdown clones, and, compared with control cells, proliferation, migration, and invasion related protein markers were also significantly decreased. ( Figure 4 A). TTK knockdown significantly reduced cell proliferation, as shown by MTT assays and colony formation experiments ( Figure 4 B-E). In addition, Transwell assays demonstrated reduced migration and invasion of TTK-knockdown cells compared with controls ( Figure 4 F-H). Overall, these findings indicate that TTK actively drives OSCC proliferation, migration, and invasion. Fig. 4 TTK knockdown suppresses OSCC proliferation, migration, and invasion. (A) Western blot confirms TTK protein knockdown and reduction of proliferation and EMT markers in SCC-15 and CAL-27 cells. (B–C) MTT assays show that TTK silencing reduces cell proliferation over 96 hours in SCC-15 (B) and CAL-27 (C). (D) Colony formation is markedly suppressed by TTK knockdown. (E) Quantification of colony number per group. (F) Transwell assays show reduced migration and invasion in TTK-silenced cells. (G-H) Quantification of migrating (G) and invading (H) cells confirms functional impairment in TTK-depleted OSCC cells. * P < .05, ** P < .01, *** P < .001, **** P < .0001. Fig 4 TTK knockdown suppresses OSCC proliferation, migration, and invasion. (A) Western blot confirms TTK protein knockdown and reduction of proliferation and EMT markers in SCC-15 and CAL-27 cells. (B–C) MTT assays show that TTK silencing reduces cell proliferation over 96 hours in SCC-15 (B) and CAL-27 (C). (D) Colony formation is markedly suppressed by TTK knockdown. (E) Quantification of colony number per group. (F) Transwell assays show reduced migration and invasion in TTK-silenced cells. (G-H) Quantification of migrating (G) and invading (H) cells confirms functional impairment in TTK-depleted OSCC cells. * P < .05, ** P < .01, *** P < .001, **** P < .0001. Based on our prior analyses, TTK was identified as a key regulator of the mTORC1 and NF-κB pathways, with in vitro experiments showing its role in promoting tumor cell proliferation, migration, and invasion. To confirm this regulatory function and investigate the underlying molecular mechanisms, RNA-seq was performed on TTK-knockdown and control cells. Differentially expressed genes were enriched in the mTORC1 and NF-κB pathways, consistent with activation observed in bulk RNA-seq datasets ( Figure 2 C and Figure 5 A-B). Western blotting showed reduced phosphorylation of S6K and 4EBP1 after TTK knockdown, indicating suppression of the mTOR pathway ( Figure 5 C). Similarly, phosphorylation of TAK1, IKKα/β, IκBα, and p65 was decreased in knockdown cells, confirming inhibition of the NF-κB pathway ( Figure 5 D). Notably, the molecular mechanism by which TTK regulates NF-κB activation has not been previously reported. To further investigate how TTK influences NF-κB activation, we examined proteins that interact with TTK. Flag-TTK was ectopically expressed in 293T cells, and immunoprecipitation followed by mass spectrometry identified 442 candidate interactors. Cross-referencing these proteins with known NF-κB activators revealed 3 key candidates: TAK1, TAB1, and TAB3 ( Figure 5 E–F; Supplementary Figure 3 A-C). Subsequent immunoprecipitation and Western blot analyses confirmed interactions between TTK and all 3 proteins, as they co-immunoprecipitated with Flag-TTK ( Figure 5 G). Moreover, TAK1 immunoprecipitates showed that its interaction with TAB1 was strengthened in the presence of Flag-TTK ( Figure 5 H). These findings demonstrate that TTK regulates NF-κB activation by binding to the TAK1–TAB complex and enhancing the TAK1–TAB1 association. Fig. 5 TTK knockdown suppresses mTORC1 and NF-κB pathways and regulates NF-κB activation via the TAK1-TAB complex. (A) Bubble plot of downregulated hallmark pathways in TTK-silenced SCC-15 and CAL-27 cells from RNA-seq; mTORC1 and NF-κB signaling among top-ranked pathways. (B) Heatmap of differentially expressed genes associated with mTORC1 and NF-κB signaling after TTK knockdown. (C) Western blot shows reduced phosphorylation of mTOR, S6K, and 4EBP1, indicating suppression of mTORC1 signaling. (D) Canonical NF-κB pathway is suppressed as evidenced by reduced phosphorylation of TAK1, IKKα/β, IκBα, and p65 in TTK-knockdown cells. (E) Mass spectrometry identifies TAK1, TAB1, and TAB3 as TTK-interacting proteins. (F) Schematic diagram of proposed TTK–TAK1–TAB complex mediating NF-κB pathway activation. (G–H) Co-immunoprecipitation confirms physical interaction between TTK and TAK1, TAB1, and TAB3 (G), and enhanced TAK1–TAB1 binding in the presence of TTK (H). Fig 5 TTK knockdown suppresses mTORC1 and NF-κB pathways and regulates NF-κB activation via the TAK1-TAB complex. (A) Bubble plot of downregulated hallmark pathways in TTK-silenced SCC-15 and CAL-27 cells from RNA-seq; mTORC1 and NF-κB signaling among top-ranked pathways. (B) Heatmap of differentially expressed genes associated with mTORC1 and NF-κB signaling after TTK knockdown. (C) Western blot shows reduced phosphorylation of mTOR, S6K, and 4EBP1, indicating suppression of mTORC1 signaling. (D) Canonical NF-κB pathway is suppressed as evidenced by reduced phosphorylation of TAK1, IKKα/β, IκBα, and p65 in TTK-knockdown cells. (E) Mass spectrometry identifies TAK1, TAB1, and TAB3 as TTK-interacting proteins. (F) Schematic diagram of proposed TTK–TAK1–TAB complex mediating NF-κB pathway activation. (G–H) Co-immunoprecipitation confirms physical interaction between TTK and TAK1, TAB1, and TAB3 (G), and enhanced TAK1–TAB1 binding in the presence of TTK (H). Abnormal activation of the mTOR and NF-κB pathways has been associated with cisplatin resistance, and our findings show that TTK inhibition suppresses these pathways, suggesting its potential to enhance cisplatin sensitivity. Cisplatin remains a standard chemotherapy for OSCC, but resistance significantly limits its efficacy. To assess whether TTK knockdown influences cisplatin response, we performed MTT assays, which revealed markedly lower cisplatin IC50 values in TTK-knockdown cells compared with controls ( Figure 6 A-B). Colony formation assays demonstrated greater cisplatin-induced growth inhibition in TTK-knockdown cells ( Figure 6 C-E). In vivo, we conducted a cell-derived xenograft (CDX) experiment in nude mice using SCC-15 cells with or without TTK knockdown. TTK knockdown alone significantly reduced tumor growth compared with vehicle controls, and in cisplatin-treated groups, it further amplified cisplatin’s tumor-suppressive effect ( Figure 6 F-G). These results indicate that targeting TTK may improve cisplatin efficacy in OSCC. Fig. 6 TTK knockdown enhances OSCC sensitivity to cisplatin. (A–B) Dose-response curves show reduced cisplatin IC50 in TTK-knockdown versus control SCC-15 (A) and CAL-27 (B) cells. (C) Colony formation assays under vehicle or cisplatin treatment show enhanced growth suppression in TTK-silenced cells. (D-E) Quantification of colony numbers under vehicle (D) and cisplatin (E) conditions. (F) Images of tumor xenografts in nude mice bearing SCC-15 cells with or without TTK knockdown and treated with or without cisplatin. (G) Tumor volume curves show that TTK knockdown augments cisplatin-mediated tumor growth suppression in vivo. * P < .05, ** P < .01, *** P < .001. Fig 6 TTK knockdown enhances OSCC sensitivity to cisplatin. (A–B) Dose-response curves show reduced cisplatin IC50 in TTK-knockdown versus control SCC-15 (A) and CAL-27 (B) cells. (C) Colony formation assays under vehicle or cisplatin treatment show enhanced growth suppression in TTK-silenced cells. (D-E) Quantification of colony numbers under vehicle (D) and cisplatin (E) conditions. (F) Images of tumor xenografts in nude mice bearing SCC-15 cells with or without TTK knockdown and treated with or without cisplatin. (G) Tumor volume curves show that TTK knockdown augments cisplatin-mediated tumor growth suppression in vivo. * P < .05, ** P < .01, *** P < .001. To enable direct histological comparison of these key proteins including TTK, TAK1, TAB1 and NFκB (p65), we assembled a uniformly formatted panel of low- and high-magnification images from oral mucosa and OSCC tissue microarrays (TMA) ( Figure 7 A). Across representative fields, TTK showed minimal signal in normal mucosa but prominent cytoplasmic/membranous staining in tumor nests, particularly at invasive fronts. In contrast, TAK1 and TAB1 exhibited broader epithelial background staining with variable cytoplasmic intensity in both tissues, while NF-κB presented nuclear enrichment in the basal layer of mucosa and focally within tumor epithelium and stromal mononuclear cells—patterns consistent with context-dependent pathway activity rather than bulk overabundance. Quantification by blinded H-score analysis demonstrated a significant increase of TTK in tumors versus mucosa (2-sided Wilcoxon rank-sum test, P < .05), whereas TAK1, TAB1, and p65 did not differ significantly between groups (all P ≥ .05) ( Figure 7 B). These data substantiate that TTK is the most discriminatory IHC marker at the protein level for OSCC, whereas TAK1/TAB1/p65 primarily report localization and microenvironmental context rather than a consistent cross-tissue difference in aggregate abundance. Notably, although TTK expression was globally higher in OSCC than in oral mucosa, substantial inter-tumoral heterogeneity persisted, supporting IHC-based stratification of patients into a TTK-high subtype to inform consideration of cisplatin chemotherapy combined with a TTK inhibitor. Fig. 7 Comparative histology and IHC in oral mucosa and OSCC. (A) Representative tissue microarrays (TMA) cores from oral mucosa (left) and oral squamous cell carcinoma (right) stained for TTK, TAK1, TAB1, and NF-κB (p65). For each marker, a low-magnification overview (5 ×; scale bar, 200 μm) is shown with a region of interest (ROI) indicated, alongside a matched high-magnification field (40 ×; scale bar, 20 µm). TTK displays minimal staining in mucosa and strong cytoplasmic/membranous staining in tumor nests; TAK1 and TAB1 show variable epithelial cytoplasmic staining in both tissues; p65 shows nuclear enrichment in basal mucosa and focally at tumor invasive fronts. Images were acquired under matched scanning parameters and white-balance, and formatted with uniform magnification and scale bars. (B) Quantification of IHC H-scores for each marker in mucosa versus tumor. Boxplots display data quartiles; * P < 0.05, ns not significant. Fig 7 Comparative histology and IHC in oral mucosa and OSCC. (A) Representative tissue microarrays (TMA) cores from oral mucosa (left) and oral squamous cell carcinoma (right) stained for TTK, TAK1, TAB1, and NF-κB (p65). For each marker, a low-magnification overview (5 ×; scale bar, 200 μm) is shown with a region of interest (ROI) indicated, alongside a matched high-magnification field (40 ×; scale bar, 20 µm). TTK displays minimal staining in mucosa and strong cytoplasmic/membranous staining in tumor nests; TAK1 and TAB1 show variable epithelial cytoplasmic staining in both tissues; p65 shows nuclear enrichment in basal mucosa and focally at tumor invasive fronts. Images were acquired under matched scanning parameters and white-balance, and formatted with uniform magnification and scale bars. (B) Quantification of IHC H-scores for each marker in mucosa versus tumor. Boxplots display data quartiles; * P < 0.05, ns not significant.

This study identifies TTK as a central driver of a high-risk oral squamous cell carcinoma subtype, defined by concurrent activation of the mTORC1 and NF-κB pathways. Elevated TTK expression is linked to increased genomic instability, higher TP53 mutation rates, APOBEC mutational signatures, and elevated TMB, underscoring its critical role in OSCC progression and potential vulnerability to immunotherapy. Mechanistically, TTK activates the NF-κB pathway through direct interaction with the TAK1-TAB protein complex and promotes mTORC1 signaling by phosphorylating key downstream targets. TTK knockdown effectively suppressed these pathways, significantly reducing OSCC cell proliferation, invasion, and migration. Notably, TTK inhibition also enhanced OSCC sensitivity to cisplatin both in vitro and in vivo, highlighting its therapeutic relevance in overcoming chemoresistance. In summary, TTK functions as a pivotal mediator of oncogenic signaling and genomic instability in OSCC, making it a compelling therapeutic target. Future clinical studies evaluating TTK inhibitors are essential to improve treatment outcomes in aggressive OSCC.

Previous studies by Curtis R. Pickering and Yue Xue et al. have advanced understanding of the genomic and epigenomic landscapes of OSCC, identifying key alterations such as CDKN2A and FAT1 copy number losses, amplifications of CCND1, EGFR, MYC, and TP63, and hypermethylation of transcriptional start site CpG islands in genes including USP44, WT1, CDX2, and IRF4. 49 These bulk-level sequencing analyses provided critical foundations for elucidating OSCC pathogenesis. Building on the single-cell sequencing study of Sidharth V. Puram et al., which characterized major cellular compartments of OSCC tumors, including tumor cells, stromal cells, and immune infiltrates, 50 our work identifies a distinct OSCC molecular subtype defined by concurrent hyperactivation of the mTORC1 and NF-κB pathways at single-cell resolution ( Figure 1 B, E-G). Prior studies by J. Silvio Gutkind and C.H. Chung et al. have demonstrated that genomic alterations in the mTOR pathway are critical drivers of head and neck squamous cell carcinoma (HNSCC), promoting tumor proliferation and metabolic dysregulation. 51 Although overactivation of the mTOR pathway has been well established, our single-cell analysis identifies a distinct OSCC subset with simultaneous activation of both mTOR and NF-κB signaling ( Figure 1 G). Integrating transcriptomic data from 709 OSCC patients, we further identified TTK as a key regulator mediating this dual pathway activation ( Figure 2 ). Under physiological conditions, TTK acts as a mitotic checkpoint kinase essential for maintaining genomic integrity by ensuring accurate chromosome alignment during metaphase. However, dysregulated mitotic checkpoints in cancers significantly contribute to aneuploidy and chromosomal instability. 25 Our findings show that OSCC patients with elevated TTK expression exhibit pronounced genomic instability, marked by increased TMB ( Figure 3 E-F), higher frequencies of TP53 mutations, unique translation start site mutations ( Figure 3 A-B), and copy number amplifications in regions such as Chr1p12, Chr3p13, Chr3p14.3, Chr12q13.12, Chr17q12, and Chr21q21.3 ( Supplementary Figure 1 A). Moreover, enrichment of APOBEC mutational signatures further defines the genomic instability profile of the TTK-high subgroup ( Figure 3 C-D). To clarify the functional role of TTK in OSCC, RNA sequencing was performed in TTK-knockdown cell lines, which revealed significant downregulation of both mTOR and NF-κB signaling pathways after TTK suppression ( Figure 5 A-B). Functional assays confirmed that TTK knockdown strongly inhibited OSCC cell proliferation and invasion ( Figure 4 B-H). These results align with prior reports in other malignancies, where TTK was shown to activate the Akt/mTOR pathway in esophageal cancer, 52 regulate autophagy in ovarian carcinoma, 53 and drive NF-κB signaling in endometriosis. 54 However, the precise mechanism through which TTK activates NF-κB signaling in OSCC remains undefined. Co-immunoprecipitation combined with mass spectrometry identified several TTK-interacting proteins, notably TAK1, TAB1, and TAB3 ( Figure 5 E-F), which are key mediators of NF-κB pathway activation. Immunoprecipitation assays further validated direct interactions between TTK and the TAK1-TAB1-TAB3 complex ( Figure 5 G-H). Given its kinase activity, we hypothesize that TTK may directly phosphorylate components of this complex, thereby activating the NF-κB pathway. Future studies will aim to identify specific phosphorylation targets and evaluate TTK as a central mediator linking mTOR and NF-κB signaling pathways. Overall, our findings identify TTK as a key driver of concurrent mTOR and NF-κB pathway activation in OSCC and show that it is overexpressed in the majority of OSCC specimens ( Figure 7 ), with overexpression correlating with pronounced genomic instability. These findings highlight TTK’s potential as a therapeutic target, particularly given the link between mTOR and NF-κB pathways and resistance to conventional therapies. To assess the therapeutic relevance of targeting TTK, we examined its effect on cisplatin sensitivity in OSCC models. Functional analyses, including IC50 determinations ( Figure 6 A-B), colony formation assays ( Figure 6 C-E), and in vivo xenograft studies ( Figure 6 F-G), demonstrated that TTK inhibition markedly enhances cisplatin efficacy. Given the ongoing clinical trials of TTK inhibitors, our findings highlight their potential as a novel therapeutic strategy for OSCC. A key limitation of this study, however, is the absence of clinical efficacy data from patient samples. To address this gap and support clinical translation, future research should incorporate patient-derived tumor organoid models along with in vivo testing of TTK inhibitors. Moreover, applying spatial transcriptomics to OSCC tissues could clarify the context-specific roles of TTK and its downstream pathways within the tumor microenvironment, providing deeper insight into therapeutic response and resistance.

Oral squamous cell carcinoma (OSCC) is the most common malignancy of the oral and maxillofacial region, originating from the mucosal lining of the oral cavity. 1 Representing more than 90% of oral cancers, OSCC remains a major public health challenge because of its heterogeneous histology and etiology. 2 Globally, approximately 380,000 new OSCC cases and 180,000 deaths were reported in 2022. 3 From 2015 to 2019, the incidence of OSCC increased by 2–3% annually. 4 Despite advances in multidisciplinary treatment, the 5-year survival rate has stagnated at less than 50%. 5 Nearly half of patients experience local recurrence or distant metastasis, resulting in poor overall treatment outcomes that substantially impair quality of life and increase mortality risk. 6 , 7 , 8 These realities underscore the urgent need for innovative, targeted therapeutic strategies to improve both survival and patient quality of life. 9 Emerging research has identified several potential molecular targets in OSCC. About 63% of cases show mitosis-related gene alterations, including activating mutations in HRAS, PIK3CA , and BRAF , as well as amplification of mitogenic pathway components such as epidermal growth factor receptor (EGFR), PIK3CA, RPS6KB1, MYC , and AKT1 . Up to 80% of OSCC tumors carry at least 1 actionable genetic alteration (e.g., EGFR, ALK, SRC, VEGFA , or MTOR ), many of which are already targets of existing cancer therapies. 10 Among these, EGFR is frequently overexpressed in OSCC and is associated with poor survival. 11 Although the EGFR-targeting drug cetuximab has been approved since 2006 as the only molecular therapy for OSCC, only 15% of patients exhibit EGFR amplification, leaving most without effective targeted options. 12 Genomic alterations in tumors often cause abnormal activation of signaling pathways essential for cancer progression. The mTOR (mechanistic target of rapamycin) pathway, a key regulator of cell proliferation, is commonly hyperactivated in many cancers. mTOR operates within 2 multiprotein complexes, mammalian target of rapamycin complex 1 (mTORC1) and mammalian target of rapamycin complex 2 (mTORC2), driving growth signals by phosphorylating downstream targets such as p70S6K and 4EBP1 to promote translation and protein synthesis. 13 Overactivation of the PI3K/AKT/mTOR pathway has been associated with chemotherapy resistance. 14 The nuclear factor kappa B (NF-κB) pathway, which regulates immune responses and inflammation, is likewise aberrantly activated in many tumors. Canonical NF-κB signaling is initiated by phosphorylation of IκB by the IKK complex (IKKα, IKKβ, and NEMO), which triggers IκB degradation and subsequent nuclear translocation of NF-κB dimers (p65/p50) to activate target gene transcription. Abnormal activation of NF-κB contributes to tumor proliferation, survival, and resistance to chemotherapy, making it an important therapeutic target. 15 , 16 , 17 , 18 Surgical resection, often combined with neck dissection, remains the primary treatment for both early and advanced OSCC. For patients with high-risk pathological features, adjuvant chemotherapy or chemoradiotherapy is recommended. 19 Platinum-based drugs such as cisplatin remain the cornerstone of therapy for locally advanced head and neck cancers. However, prolonged cisplatin exposure frequently induces drug resistance, leading to treatment failure and poor prognosis. 20 Overcoming cisplatin resistance is essential for improving therapeutic outcomes in OSCC. In this study, we identified a distinct OSCC subtype marked by concurrent activation of the mTOR and NF-κB pathways, potentially regulated by threonine tyrosine kinase (TTK; also known as Mps1), a dual-specificity serine/threonine kinase. TTK is essential for the spindle assembly checkpoint (SAC), where it ensures accurate chromosome segregation and preserves genomic stability during mitosis. 21 , 22 , 23 , 24 Dysregulation or overexpression of the mitotic checkpoint can result in aneuploidy and genomic instability, both of which are hallmarks of cancer progression. 25 Overexpression of TTK has been reported in multiple malignancies, including glioblastoma, breast, liver, prostate, lung, and pancreatic cancers, where it correlates with early relapse and poor prognosis. 26 Several small-molecule inhibitors of TTK have been developed, with some currently undergoing clinical trials. 27 , 28 , 29 , 30 , 31 , 32 Given the urgent need for improved therapeutic strategies in OSCC, identifying molecular markers that both define aggressive subtypes and guide treatment decisions is critically important. TTK overexpression may serve as a clinically relevant biomarker for a subset of OSCCs with concurrent activation of the mTOR and NF-κB pathways. Targeting this kinase could improve the efficacy of conventional cisplatin-based chemotherapy, providing a rational combination strategy to enhance treatment response. This study was designed to evaluate the translational significance of TTK as both a stratification marker and a therapeutic target, with the goal of informing more precise and effective interventions for patients with high-risk OSCC.

None disclosed.