Unravelling the heterogeneity in oral cancer stem cells – A single-cell computational exploratory study

Unravelling the heterogeneity in oral cancer stem cells – A single-cell computational exploratory study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Unravelling the heterogeneity in oral cancer stem cells – A single-cell computational exploratory study Surendra Kumar Acharya, Guo Rean Wong, Yee Wen Choon, Mamata Rai, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7844179/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Cancer stem cells (CSCs) are key drivers of heterogeneity, recurrence, and therapy resistance in oral squamous cell carcinoma (OSCC). CD44, CD133, ALDH, and BMI1 have been highlighted as clinically relevant CSC markers, yet no single marker sufficiently defines CSC identity. This study aims to explore computationally CSC heterogeneity in Asian-derived ORL cell lines (ORL-48, ORL-115, ORL-174, ORL-214). Unsupervised clustering of single-cell qPCR datasets revealed enrichment of quadruple-positive (CD44⁺/CD133⁺/ALDH1A1⁺/BMI1⁺) subsets, alongside smaller heterogeneous fractions. PhenoGraph clustering and viSNE analyses of arcsinh-transformed flow cytometry datasets independently confirmed dominant CD44⁺ and BMI1⁺ fractions, smaller CD133⁺ and ALDH⁺ subsets, and rare CSC-enriched multi-marker clusters consistent with the single-cell qPCR analysis. In addition, sphere formation assays further validated the presence of self-renewing CSC-like populations across all lines. As this work was designed as an exploratory study, analyses were performed once per cell line without replicates; future studies will incorporate independent validations. Biological sciences/Cancer Biological sciences/Cell biology Biological sciences/Computational biology and bioinformatics Biological sciences/Stem cells Oral Cancer Stem Cells PhenoGraph Unsupervised clustering Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Introduction Oral squamous cell carcinoma (OSCC) remains one of the most prevalent malignancies of the head and neck, with persistently poor survival rates and frequent recurrence despite advances in treatment 1 . The presence of cancer stem cells (CSCs), a subpopulation of tumor cells with self-renewal capacity, therapy resistance, and tumor-initiating potential, is a major factor underlying this clinical outcome 2 . CSCs are increasingly recognized as key drivers in tumor heterogeneity, metastasis, and treatment failure, making their accurate identification and characterization a critical objective in CSC-targeted cancer therapies 3 . The challenge in CSC study lies in their rarity, plasticity, and reliance on surrogate marker expression. Markers such as CD44, CD133, ALDH, and BMI1 have been widely implicated in head and neck and oral cancers, yet no single marker defines CSC identity, and marker expression is often heterogeneous within tumors 4 , 5 . To investigate CSC heterogeneity in previously authenticated Asian-OSCC-derived ORL cell lines 6 , 7 , expression of three established CSC markers (CD44, CD133, and ALDH1A1) and a key regulator of stemness and resistance (BMI1) in ORL cell lines was profiled, aiming to capture overlapping subpopulations. Additionally, to avoid inter-operator variability and bias from reliance on previously reported OSCC-CSC subpopulations, computational methods were adopted to objectively and reproducibly explore the heterogeneous nature of oral CSCs in the current study 8 , 9 , 10 . Sphere formation ability, which has also been used as a CSC marker 11 , was also assessed as complementary validation of CSCs’ behaviour in ORL cell lines. Materials and Methods Cell culture Four authenticated and genetically-defined Asian OSCC cell lines (ORL-115, ORL-48, ORL-174, ORL-214) were grown as previously described 6 , 7 . Gene expression of CD44, CD133, ALDH1A1, BMI1 in single-cell assay To investigate the expression of CD44, CD133, ALDH1A1, and BMI1 at single-cell resolution , 70% sub-confluent ORL cells were used for specific target amplification (STA) qRT-PCR measurement of mRNA levels in individual cells using the C1 Single-Cell Autoprep System and BioMark HD system Standard BioTools Inc., California, USA). The priming and loading of cells of the Integrated Fluid Circuit (IFC) have been described elsewhere 12 . Briefly, cells were loaded on a C1 TM Single-Cell Preamp IFC, 17-25µm (Standard BioTools Inc., California, USA) using the Fluidigm C1 machine. A Leica DMI6000B (Leica, Heidelberg, Germany) fully automated inverted research microscope was used to identify wells containing single cells. Pre-amplified cDNA was then generated from each cell using the Ambione TM Single Cells-to-C T TM Kit (Thermo Fisher Scientific Inc., Massachusetts, USA), with pooled qPCR primers for CD44, CD133, ALDH1A1, and BMI1 (Thermo Fisher Scientific Inc., Massachusetts, USA) (100µM each) and Fluidigm STA reagents (Standard BioTools Inc., California, USA). The BioMark TM HD system was then used for qPCR of single-cell pre-amplified cDNA according to the manufacturer’s protocol (“Gene expression with the Flex Six IFC Using Fast/Standard Taqman assays”, Standard BioTools Inc., California, USA). Briefly, a 1.8 µL of each preamplified cDNA was mixed with 2.0 µL of 2X TaqMan Standard PCR Master Mix (Thermo Fisher Scientific Inc. Massachusetts, USA) and 0.2 µL GE Sample Reagent (Standard BioTools Inc., California, USA) and each sample mix was then pipetted into sample inlet in a FLEXsix TM IFC chip (Standard BioTools Inc., California, USA). Then, 2.0 µL of Individual 20X Taqman Gene Expression ( CD44, CD133, ALDH1A1, BMI1 ) was mixed with 2.0 µl Assay Loading Reagent (Standard BioTools Inc., California, USA), and pipetted into the assay inlet FLEXsix TM IFC chip. IFC Controller HX (Standard BioTools Inc., California, USA) was used to load the mixes into the FLEXsix TM IFC chip, and qPCR was performed on the BioMark TM HD real-time PCR reader (Standard BioTools Inc., California, USA) following the manufacturer's instructions. qPCR data were collected on Biomark TM HD (instrument ID BIOMARKHD193) using software version 4.1.3. Baseline correction was set to LINEAR, with automatic global thresholding applied across all assays (Ct Threshold Method = Auto Global). The software quality threshold was 0.65, and ROX was used as a passive reference. CT values were exported in CSV tabular format for downstream analyses (see Data processing and computational analysis) Flow cytometry analysis of CD44, CD133, ALDH, BMI1 expression To validate the expression of CD44, CD133, ALDH, and BMI1, 70% subconfluent ORL cells were harvested by trypsinization. Cells were washed in PBS and stained with respective antibodies. Due to the technical incompatibility of combining live-cell ALDH detection (which required an intact membrane) with BMI1 intracellular staining (which requires fixation and permeabilization), co-expression of ALDH and BMI1 could not be analyzed within the same sample. Instead, ALDH and BMI-1 expression were independently profiled with CD44 and CD133, allowing complementary assessment of these markers. Briefly, ALDEFLUOR™ Kit (Stem Cell Technologies, Vancouver, Canada, catalog #01700) was used to detect ALDH activity. Cells were suspended in ALDEFLUOR™ Kit buffer containing 5µl ALDH substrate (BAAA, 1 µmol/L per 10 6 cells) and incubated for 30 min at 37°C. As a negative control, an aliquot for each sample of cells was treated with 50 mmol/L diethylaminobenzaldehyde (DEAB), a specific ALDH inhibitor. For staining of BMI1, cells were treated with eBioscience TM Intracellular Fixation and Permeabilization Buffer Set (Thermo Fisher Scientific Inc., Massachusetts, USA) before incubation with 10 µL human BMI-1 APC-conjugated antibody (R&D Systems, Minneapolis, USA, clone# 384515, catalog number: IC33341a). ALDEFLUOR™- and BMI1-reacted cells were subsequently stained with 10µL BD Pharmigen TM PE-Cy TM 7 mouse anti-human CD44 antibody (BD Biosciences, California, USA, clone G44-26, material number: 559942) and 10µL monoclonal anti-human CD133/1-PE antibody (Miltenyi Biotec, Bergisch Gladbach, Germany, order no. 130-113-670) for 30 min on ice. After staining, the cells were washed in phosphate-buffered saline buffer (PBS), reconstituted with PBS buffer. Data were acquired using a BD FACSCanto TM II flow cytometer (BD Biosciences, California, USA) and exported as FCS files using BD FACSDiva software (version 6.1.3) for downstream analyses (see Data processing and computational analysis). Data processing and computational analysis This section outlines the computational workflow for identifying and interpreting clusters of marker-positive subpopulations in ORL cell lines. Dimensionality reduction and unsupervised clustering were performed first, followed by threshold-based marker annotation for biological interpretation. Computational analysis was performed using a combination of Python 3.9.6 libraries: pandas, numpy (data processing), matlibplot, seaborn (plots), scikit-learn (PCA, KMeans), Upsetplot (UpSet visualization) on MacOS Sonoma 14.7.8. qPCR single-cell Ct data Input data and marker panel The dataset was generated from Gene expression of CD44, CD133, ALDH1A1, and BMI1 in a single-cell assay and comprised per-cell Ct values for four markers: CD44, CD133, ALDH1A1, and BMI1. Each entry also included the corresponding cell line and a unique sample identifier. Violin plot visualization of Ct value distribution qPCR data for CD44, CD133, ALDH1A1, and BMI1 were aggregated as per-cell Ct values. Wells failing amplification (no Ct or Ct = 999) and Ct values > 35 (defined as non-detection by conventional Ct detection threshold < 35) were excluded from the kernel density estimate and visualization. A violin plot shows the mirrored kernel density of the Ct distribution for each gene within each cell line. Inside each violin, the median lines and interquartile ranges (IQR) (dashed) were shown, whiskers extended to 1.5xIQR. Raw, untransformed Ct values were used for these plots. Lower Ct indicated higher expression. Principal Component Analysis (PCA) and K-means clustering (unsupervised) Each ORL cell line was processed independently. CT values from the four markers were first log 2 -transformed. The resulting data matrix was standardized via z-score normalization using the Standard Scaler module in scikit-learn. PCA was applied to this normalized matrix, and the top two components (PC1, PC2) were retained. K-Means clustering (k = 3) was then performed on the PCA-reduced data to identify these unsupervised clusters pre-cell line. Cluster centroids were initialized using a random seed (random_state = 42) to ensure reproducibility. Marker positivity classification (post-clustering) To assign biological meaning to the unsupervised clusters, a separate marker classification step was performed using the original Ct matrix. For each marker in each cell line, a threshold was calculated using the heuristic: Threshold = Mean (Ct) = 2 x standard deviation (Ct). Cells with Ct values lower than the threshold were considered positive for that marker (binary = 1). This produced a marker positivity matrix (0/1) for each cell. Marker combination labelling and cluster interpretation Each cell was annotated with a marker combination label, such as CD133 + ALDH1A1 + CD44 - BMI1 + , based on its binary positivity status. A cross-tabulation was performed between cluster assignments and marker combination labels to examine which clusters were enriched in specific marker-positive populations. Visualization outputs For each cell line, the following visual outputs were generated: Violin plots of Ct distributions PCA scatter plots with cluster assignments Stacked bar charts showing marker combination composition within clusters UpSet plots - a scalable matrix-based set visualization method designed for quantitative analysis of multiple set intersections. Unlike Venn or Euler diagrams, which become intractable with more than three or four sets, UpSet provides an interpretable and scalable visualization of intersections and their cardinalities 13 . Flow cytometry computational analysis Raw flow cytometry standard files were acquired (see Flow cytometry analysis of CD44, CD133, ALDH, BMI1 expression ). Data were exported as standard .fsc files in Flow Cytometry Standard v3.1 format. Raw FCS files contained a vendor spillover key (SPILL or $SPILLOVER). Before analysis, matrices were read, and compensation to the fluorescence channels was applied. If channels were exported as compensated (prefixed Comp-) compensation was not applied. Channels recorded were: CD44-PE-Cy7-A, CD133-PE-A, ALDEDLUOR TM -ALDEFLUOR-A, and BMI-APC-1. Computational analysis was performed using a combination of Python 3.9.6 libraries (FlowCal, NumPy, pandas, matplotlib, scanpy/phenograph) and R 4.2.3 libraries (flowCore, Rtsne, ggplot2, ggraph) on MacOS Sonoma 14.7.8. Unsupervised clustering and visualization were performed using the PhenoGraph algorithm in combination with viSNE, which has been validated as a robust approach for analyzing high-dimensional flow cytometry data. DiGiuseppe et al. (2018) 9 demonstrated that this approach could reliably distinguish normal versus aberrant T-cell subsets in clinical flow cytometry without reliance on manual pre-gating, thereby minimizing operator bias. The strategy was adapted in this analysis. By applying PhenoGraph clustering followed by viSNE projection, CSC heterogeneity and visualization of distinct subpopulations could be objectively explored without imposing prior assumptions about marker co-expression patterns. Preprocessing and gating All samples were first pre-gated on forward and side scatter to remove debris and doublets. A three-step P1 gating strategy was explored: 1) a rectangular gate retaining the 5th-95th percentiles of FSC-A/SSC-A, followed by 2) a 97.5% confidence ellipse, and 3) a 97.5% Minimum Covariance Determinant (MCD) elliptical gate was also explored. A final step-1-and-2 gating strategy was decided and applied. This procedure retained 85-88% of input events per sample and ensured consistency across runs and inclusion of potential rare CSC subpopulations. P1 gate was learned on the unstained files and then applied unchanged to the matched stained files. No further manual gating was applied. All P1-gated cells were included in the downstream analyses. BMI1 analysis was carried out independently due to technical incompatibility with live ALDHFLUOR TM staining. Fluorescence intensity transformation Raw fluorescence intensities were arcsinh-transformed (with the cofactor (c) determined per marker by scaling the 99th percentile of the stained sample to 5) to linearize low-intensity signals while compressing high-intensity outliers, preserving dynamic range. Thresholding determination for marker positivity Marker positivity was defined using matched controls: a) DEAB-treated negative control (ALDH), unstained control (CD44, CD133, BMI1). Thresholds were computed as mean + 2xSD of the control distributions. Events with arcsinh-transformed values above the threshold were classified as marker-positive. Unsupervised clustering Phenograph clustering was performed on arcsinh-transformed marker intensities, constructing a shared nearest-neighbour (SNN) graph with Jaccard similarity and resolving communities via Louvain optimization, without predefining the number of clusters. Median arcsinh expression of each marker was calculated per cluster and visualized in a heatmap with hierarchical ordering. Cluster assignments were saved for downstream annotation. Marker classification and per-cell phenotype calls Each P1-gated, arcsinh-transformed cell was classified as positive or negative for CD44, CD133, ALDH, and BMI1 based on the thresholds above. Positivity calls were combined into per-cell phenotypes (e.g., CD44 + CD133 - BMI1 + , CD44 + CD133 - ALDH + ) Cluster annotation with Wilson score intervals Clusters were annotated using per-cell positivity calls. For each marker within the cluster, the Wilson 95% lower bound for the fraction of positive cells was computed. A cluster was labelled “positive” for a marker if this lower bound exceeded 0.5, ensuring conservative assignment. Each cluster was given a composite label (e.g., CD44 + CD133 - BMI1 + , CD44 + CD133 - ALDH + ). viSNE embedding and annotation viSNE was applied to the same transformed markers using Barnes-Hut t-SNE with perplexity set to 30 and seed fixed at 1979 to ensure reproducibility. Embedding coordinates were joined with cluster assignments and per-cell marker phenotypes. viSNE plots were generated in two formats: 1) colored by Phenograph cluster identity, and 2) colored by per-cell phenotype (single, double, or triple marker positivity) with percentages reported. Sphere formation assay To determine the sphere formation capability, 70% sub-confluent ORL cells were trypsinized, centrifuged, resuspended in culture medium, and counted with a hemocytometer. Cells were then seeded with a density of 1 x 10 3 cells in a 6-well Ultra-Low Attachment plate (Corning, New York, USA) in 2 mL culture medium and kept in 5% CO 2 at 37 º C for 10 days. 1 ml of culture medium was replenished every 3 days. The presence of spheres more than 20 µm visualized under a microscope at 40× magnification was counted as positive. Results As this work was designed as an exploratory study, each ORL cell line was analyzed once without technical or biological replicates. Reported values, therefore, represented single-run datasets with full disclosure of event counts and positive percentages in Supplementary Information. Replicated will be incorporated in future validation studies. Detection of CSC markers across ORL cell lines As no housekeeping gene was included in the study, results were presented as relative comparisons of Ct distributions across cell lines rather than normalized expression levels. Single-celled qPCR showed 100% detection of CD44, CD133, ALDH1A1, and BMI1 across ORL cell lines (Ct values < 35). CT values indicated that CD133 and ALDH1A1 transcripts were detected at much lower cycle thresholds compared to CD44 and BMI1 across all cell lines. The results suggest that CD133 and ALDH1A1 were strongly expressed relative to CD44 and BMI1. Violin plots revealed marker heterogeneity in expression levels (Figure 1). Violin plots show distributions of Ct values for CD133, ALDH1A1, CD44, and BMI1 measured by single-cell qPCR across ORL cell lines. Each violin depicts the density of per-cell measurements, with internal lines marking quartiles. The dashed red line at Ct = 35 indicates the conventional detection cutoff. CD133 and ALDH1A1 were consistently detected at low Ct values (indicative of strong expression), while CD44 and BMI1 exhibited higher Ct values (moderate to lower expression) and broader heterogeneity (Complete Ct datasets, Ct detection per-cell-line-per-gene summary, and Ct mean ± SD, median ± IQR, and detection rate tables are provided in Supplementary Information). Unsupervised clustering identified heterogeneous subpopulations Principal component analysis (PCA) of z-scored Ct values followed by k-means clustering revealed distinct subgroups of cells within individual cell lines (Figure 2). All ORL cancer cells were segregated into three transcriptionally distinct clusters, suggesting intra-line heterogeneity in CSC marker expression patterns (PCA/KMeans cluster assignment table is provided in Supplementary Information). Principal component analysis (PCA) was performed on z-scored Ct values for CD133, ALDH1A1, CD44, and BMI1 within each cell line. Shown are the first two principal components (PC1 and PC2), which together capture the largest sources of variance in the dataset. Points represent individual cells, coloured by unsupervised K-means cluster assignment (k = 3). Percent variance explained by each PC is indicated on the axes. The number of cells analyzed for each line is shown in the lower right corner (n = x). Distinct clustering patterns were observed, with ORL-48 cells separating into three subgroups (Clusters 0–2), while ORL-115 cells were more concentrated with fewer dispersed subclusters. Subpopulation composition by marker combinations To assess distributions of multi-marker subsets, cells were classified by binary positivity thresholds (mean ± 2SD). Cluster composition analysis showed a dominant quadruple-positive subsets with the majority of CD133 + ALDH1A1 + CD44 + BMI1 + cells, with smaller proportions of double-, and triple-positive combinations across ORL cell lines (Figure 3). (Per-cell-line thresholds and marker positivity tables are provided in Supplementary Information). Stacked bar plots show the proportion of unsupervised K-means clusters (k = 3) within each marker-positivity combination for single-cell qPCR data from ORL-48, ORL-115, ORL-174, and ORL-214. Marker combinations were defined using binary calls (positive/negative) for CD133, ALDH1A1, CD44, and BMI-1 based on a mean + 2×SD threshold. Each bar represents one marker combination, with stacked colors corresponding to the percentage contribution of each cluster. Across all lines, the CD133⁺/ALDH1A1⁺/CD44⁺/BMI1⁺ combination predominated, while additional rare combinations appeared at lower frequencies, reflecting underlying heterogeneity in subpopulation structure. UpSet visualization of multi-marker overlaps To visualize the intersection across markers, UpSet plots were generated for each line. Quadruple-positive subpopulations were dominant (85.9% - 92.9%) while smaller subsets (< 5%) exhibited double- and triple-positivity across ORL cell lines (Figure 4). UpSet plots show the distribution of single-cell gene expression combinations for CD133, ALDH1A1, CD44, and BMI-1 in ORL-115, ORL-174, ORL-214, and ORL-48. Each vertical bar represents the number of cells positive for the corresponding marker combination (filled circles), with counts and percentages annotated above the bars. Horizontal bars on the left indicate the overall frequency of positivity for each marker. In all cell lines, the quadruple-positive combination (CD133⁺/ALDH1A1⁺/CD44⁺/BMI1⁺) predominated, while additional rare subpopulations were detected at lower frequencies, highlighting heterogeneity in CSC marker co-expression. Quality control and single-cell event selection A sequential gating strategy was applied to unstained ORL cells. For example, in ORL-48, in Panel A for staining of CD44, CD133, and ALDH (Figure 5A), an initial rectangular trim (5–95th percentiles, black box) retained 88.07% of events. This was followed by a 97.5% ellipse gate to capture the main single-cell population, shown as a classical covariance ellipse (blue) retaining 85.81% of events and a robust MCD ellipse (red dashed) retaining 76.46% of events. The classical method was selected as the final P1 gate. This gating strategy was replicated for staining of CD44, CD133, and BMI1 in Panel B (Figure 5B) and across all ORL cell lines. This gating step excluded debris and aggregates, ensuring that only the primary single-cell population proceeded to downstream (for CD44_CD133_ALDH in Panel A and CD44_CD133_BMI1 in Panel B) analysis. (Absolute cell counts and percentages are provided in. Supplementary Information). Dot plot of FSC-A and SSC-A for unstained ORL cells illustrating the P1 gate. Events outside the 5-95th percentile rectangle were excluded, followed by a 97.5% covariance ellipse. Both classical (blue) and robust MCD (red dashed) are shown, with the classical method chosen as final. The percentage of retained events at each step is indicated. This quality control (QC) gate removed debris and doublets before downstream analysis (Panel A: CD44_CD133_ALDH, Panel B: CD44_CD133_BMI1) Flow cytometry reveals heterogeneous expression of CSC markers across ORL cell lines To complement the single-cell qPCR profiling, CD44, CD133, ALDH, and BMI1 expressions were assessed by flow cytometry in all ORL cell lines. Overlay histograms demonstrated consistently high CD44 and BMI1 expressions (> 90% cell positive) across all cell lines. In contrast, CD133 was expressed in only a minority of cells. ALDH activity, determined using DEAB-matched gating, was similarly restricted to small fractions (Figure 6). These results validated the qPCR-based findings, in which dominant CD44 + and BMI1 + subsets co-existed with rarer multi-marker combinations. Colored histograms of raw fluorescence intensity (log 10 scale) are shown for CD44, CD133, ALDH (Panel A), and CD44, CD133, BMI1 (Panel B) compared with the unstained control (gray). Each panel displays the density distribution of single-cell events, highlighting separation between negative and positive populations. CD44 (blue), CD133 (purple), ALDH-negative control - DEAB (red), and ALDH/BMI1 (green) signals were clearly shifted relative to the unstained baseline (grey), confirming robust marker detection. Shared axes across panels allow direct comparison of fluorescence intensity distributions. (Percentages of positive cells are displayed above each histogram). PhenoGraph clustering identifies discrete CSC subpopulations across ORL cell lines To resolve heterogeneity beyond single-marker gating, arcsinh-transformed flow cytometry data for CD44, CD133, ALDH, and BMI1 were subjected to PhenoGraph clustering using the Louvain method, with median marker intensities per cluster visualized as heatmaps (Figure 7). Across all ORL cell lines, PhenoGraph identified 24-26 and 14-18 distinct clusters per cell line for CD44, CD133, and ALDH (Figure 7A), and CD44, CD133, and BMI1 (Figure 7B) expression, respectively. (Scaled medians, thresholds for PhenoGraph, cluster assignments, and summaries are provided in Supplementary Information). Heatmaps showing median arcsinh-transformed expression values of CD44, CD133, ALDH (Panel A), and CD44, CD133, BMI1 (Panel B) across PhenoGraph-derived clusters (C1-Cn). Rows represent individual clusters and columns represent markers. Values within each cell denote the cluster median. Clusters are hierarchically ordered to highlight similarities in marker expression profiles. Warmer colors (yellow–red) indicate higher expression, while cooler colors (blue) indicate lower expression. This visualization reveals heterogeneity in CSC marker expression. viSNE visualization of PhenoGraph clusters To complement the heatmap analysis, PhenoGraph-defined Louvain clusters were projected onto viSNE maps for each ORL cell line (Figure 8). Each cluster was color-coded and distributed across the two-dimensional embedding, enabling visualization of discrete subpopulations within global cell populations. While the number of clusters varied, all showed clear separation between dominant clusters and smaller, rarer subpopulations. Panel A - CD44, CD133, ALDH expression. Panel B – CD44, CD133, BMI1 expression (Wilson’s phenotype embedding and clusters are provided in Supplementary Information). t-distributed stochastic neighbour embedding (t-SNE, viSNE) was applied to arcsinh-transformed marker intensities (Panel A – CD44, CD133, ALDH, Panel B – CD44, CD133, BMI1). Cells are displayed in two dimensions according to expression similarity and coloured by Louvain clusters (C1-Cn) identified by the PhenoGraph algorithm. Structural heterogeneity and how cells were grouped into discrete clusters were highlighted. viSNE overlay of per-cell phenotype The same viSNE embedding with per-cell marker positivity was annotated based on thresholding (mean ± 2xSD from controls) to visualize the distribution of CSC phenotypes (Figure 9). Across cell lines, the CD44 + and CD44 + BMI1 + subpopulation dominated, consistent with flow cytometry results. A smaller fraction of CD133 + and ALDH + subsets was observed, often interspersed within larger CD44 + and CD44 + BMI1 + clusters. Notably, rare triple-positive (CD44 + BMI1 + ALDH + ) subpopulations were detected in ORL-48 and ORL-214 (viSNE PhenoGraph embedding and overall phenotype counts are provided in Supplementary Information). The same viSNE embedding colored by binary phenotype assignments derived from control-based positivity thresholds (mean ± 2SD of controls). Distinct phenotypes, including major CD44 + (Panel A) and CD44 + BMI1 + (Panel B) subpopulations. Rarer CD44 + CD133 + ALDH + subpopulations are annotated with their relative proportions in the legend. Panel A and Panel B show how marker-defined phenotypes distribute across the viSNE landscape and align with PhenoGraph clustering. Functional validation by sphere formation assay Sphere formation assay confirmed the functional CSC capacity of ORL cell lines. All four cell lines consistently generated tumour spheres in low-attachment conditions (Figure 10). Representative images of tumour spheres derived from (A) ORL-48, (B) ORL-115, (C) ORL-174, and (D) ORL-214 cultures under non-adherent conditions. All cell lines generated compact spherical colonies, supporting the functional presence of self-renewing CSC-like subsets. Discussion The application of unsupervised clustering in the current study aligns with emerging evidence that machine learning and unsupervised computational methods are increasingly valuable for analyzing complex single-cell datasets. Recent work has shown that these approaches enable improved detection of rare and heterogeneous cell subsets while reducing reliance on subjective manual gating 8 . In particular, prior applications of PhenoGraph and viSNE to flow cytometry data demonstrated that unsupervised clustering can resolve phenotypic continua and visualize rare subsets that would otherwise remain obscured 9 . The current study is, to the best of the authors’ knowledge, the first study that demonstrated the feasibility of integrating single-cell microfluidic qPCR and flow cytometry with an unsupervised analytical pipeline to interrogate CSC marker heterogeneity in OSCC-derived cell lines. Conventional methods such as bulk transcriptomics average signals across thousands of cells, obscuring the diversity of CSC phenotypes and masking rare but functionally important subpopulations 14 , 15 , 16 . Microfluidic qPCR profiling in the current study captured transcriptional heterogeneity and revealed CSC markers’ co-expression patterns through unsupervised clustering. Flow cytometry, analyzed through arcsinh transformation, PhenoGraph clustering, and viSNE visualization, independently validated these findings at the protein level, confirming the predominance of CD44 + and BMI1 + subsets, the presence of smaller ALDH + fractions, and rare multi-marker CSC-enriched clusters. Prince et al. (2007) first demonstrated that CD44 + cells isolated from primary head-and-neck squamous cell carcinoma (HNSCC) tumours or cell lines could initiate tumours in mice and showed elevated BMI1 expression in the CD44 + population, with frequencies ranging from ~ 0.4 to 40% 17 . In contrast, our study found 100% of cells in all ORL cell lines positive for CD44 under the applied threshold, a finding that suggests a saturation of total CD44 expression in these cell lines. This observation raises the possibility that CD44 variant isoforms (CD44v) or standard isoform (CD44s) may be detected, both of which are splice variants of the same gene but differ only in exon composition 18 , 19 . In some models, CD44v8-10 has been shown to enhance migration and sphere formation similarly to CD44s, though its impact on metastasis may differ 20 . Understanding which spice variants are expressed in ORL cell lines (CD44s vs CD44v) will therefore be critical, particularly because therapeutic strategies targeting CD44 often do not distinguish between isoforms and may require isoform-specific targeting 4 . ALDH is a family of 19 intracellular enzymes involved in oxidizing aldehydes to carboxylic acids; among these, ALDH1 has been widely reported as a head and neck CSC marker, Clay et al. (2010) demonstrated that ALDH-high cells (representing ~ 1.0 to 7.8% of the tumour population) are highly tumorigenic: as few as 500 ALDH-high cells could form tumours in mice, whereas ALDH-low required much higher number. In Clay’s work, most ALDH-high cells co-expressed CD44 (50.6–74.4%), but only a small subset of CD44 + cells were ALDH+ (9.8–23.6%) 21 . The frequency of ALDH + cells in the current study ranged from 1.47–4.62% across ORL cell lines, and the percentage of CD44 + /ALDH + cells ranged from 0.03–20.71%. These data are broadly consistent with Clay et al. (), reinforcing that ALDH marks a rarer, more discriminatory CSC subset compared with CD44. Kulsum et al. (2017) further support this: inhibiting ALDH1A1 in HNSCC models reduced sphere formation, downregulated stemness and drug resistance genes, and increased chemosensitivity to cisplatin 22 . Findings from the current study suggest that ALDH activity identifies a minor but potentially functionally important CSC population in ORL cell lines, lending weight to targeting ALDH for therapy 23 . CD133 is a glycosylated membrane protein with 5 transmembrane domains and 2 large extracellular loops. It is a marker for hematopoietic stem cells and also a CSC marker for brain 24 . Chiou et al were the first to identify CD133 as a CSC marker for oral cancer. The method they used was different from others. They first grew oral cancer cells as spheres and showed that these spheres were more tumorigenic than parental cells. Next, they also found that expression of CD133 in 60% spheres generated from 2 OSCC cell lines. The frequency of CD133 positive cells from 5 primary tumours ranged from 0.8–4.2% in the parental cell population and was more enriched, ranging from 28.2–40.2% in spheres in their 25 . In the current study, the frequency of CD133-positive cells ranged from 3.93–21.59%, consistent with the reported frequency. Moreover, Yu et al demonstrated that silencing of CD133 reduced tumorigenicity and heightened drug sensitivity of OSCC cells 26 . BMI1 (polycomb group protein Moloney murine leukaemia virus insertion site 1) is a key regulator of stem cell self-renewal whose aberrant expression contributes to cancer initiation, metastasis, and treatment resistance. Chen et al. () used a 4-nitroquinoline-1-oxide (4-NQO) mouse model of HNSCC with lineage tracing (BMI1-CreER;Rosa-tdTomato) to identify BMI1 + cells as slow-cycling, tumour-initiating stem cells. These cells were also found to drive metastasis to the cervical lymph node, and combining cisplatin with PTC-209 – a small-molecule BMI1 inhibitor – significantly reduced tumour progression and the BMI1 + CSC pool in vivo 27 . In vitro, PTC-209 treatment of HNSCC cell lines (Cal27, FaDu) suppressed proliferation, induced apoptosis, reduced migratory and invasive capacity, and acted synergistically with cisplatin or 5-FU 28 . In the current study, nearly 100% of BMI1 + cells in ORL cell lines were under the applied thresholding criteria, raising the question of whether all BMI1 positivity reflects functional CSC potential or rather a general stemness-associated regulatory state. Given the promising preclinical activity of PTC-209 in HNSCC, its capacity to impair tumorigenicity, enhance chemosensitivity, and reduce CSC subpopulations makes BMI1 an especially attractive target in ORL models 29 . In addition, an investigation of phenotypic diversity in in silico simulation reported that 28 phenotypes had evolved in the smallest virtual tumour after 730 days. Moreover, the group simulated an in silico biopsy and found that the average frequency of CSC ranged from 0.7 to 10% and a single collected biopsy sample can be divided into 10 subpopulations with approximately 10,000 cells. These findings are corroborated by the discovery of 10–28 distinct clusters from PhenoGraph and viSNE clustering of the current study. The frequency of CD133 + and ALDH + cells from the current study also fell within the simulated range 30 . Sphere (or organoid) formation assays were originally developed in neural stem cell biology when Johansson et al. (1999) isolated adult neural stem cells from human brain ependyma and demonstrated that dissociated cells cultured in suspension without substrate formed free-floating clusters (“neurospheres”) which could later differentiate into neurons and glial cells upon attachment and appropriate stimulation 31 . In OSCC, Chen et al. (2012) applied sphere formation under non-adherent conditions to enrich for CSCs; tumour spheres derived this way displayed elevated expression of CSC markers (CD133, ALDH1), increased tumorigenic capacity in vivo, and resistance to chemotherapeutic agents 32 . More recently, Pozzi and colleagues showed that sphere assays can reliably enrich CSCs from HNSCC by culturing CD44 + /ALDH + sorted cells or by using marker-unsorted cell populations in ultralow attachment or soft agar settings, preserving CSC traits over multiple passages 33 . 3D models in OSCC confirm that tumorospheres (free-floating CSC sphere models) are widely employed for assessing stemness, drug sensitivity, and phenotypic heterogeneity 34 , 35 , 36 , 37 . Sphere formation was robust across all ORL cell lines in the current study, suggesting that sphere/organoid assays constitute a valuable functional complement to molecular and cytometric profiling of CSC markers. Conclusions In conclusion, the current study reveals pronounced CSC heterogeneity discovered through an autonomous computational method without prior assumption or subjective interference. While CD44 and BMI1 are nearly ubiquitously expressed, which is in line with the stemness regulatory program, ALDH and CD133 delineate rarer, more discriminative CSC subpopulations. The concordance between transcript-level Ct thresholding and protein-level clustering validates that rare multi-marker co-expressing subsets, though low in frequency, are consistently present across cell lines. Importantly, the functional sphere formation across ORL cell lines supports that these molecular markers reflect biologically relevant self-renewing capacity. Though exploratory without replicates, computational analysis such as this offers reproducibility and a basis for CSC targeting in future studies. Limitations This work was designed as an exploratory, proof-of-concept study. Each ORL cell line was analyzed once without technical or biological replicates, precluding statistical comparisons across runs. While this limits the generalizability of the frequency estimates, all event counts, gating percentages, and positive fractions are fully disclosed. Despite these constraints, this study demonstrates methodological feasibility and establishes a transparent baseline for reproducible CSC profiling. Future directions and translational significance Future studies will incorporate biological replicates, patient-derived samples, and expanded marker panels to validate and refine current findings. Integration of pathway-specific inhibitors may help dissect the functional role of subpopulations in OSCC therapy resistance. Advanced models such as organoids and lineage tracing will further strengthen causal interference. Importantly, by linking marker-defined CSC heterogeneity to canonical signalling pathways the future studies will identify pathway-specific vulnerabilities, paving the way for rational CSC-targeted therapies and improved clinical outcomes in OSCC. Declarations Funding This study was funded by the Faculty of Dentistry, University of Malaya, University of Malaya Research Grant (UMRG, RU012-2013) and Universiti Malaya Bantuan Kecil Penyelikan (BKP Grant - UM.TNC2/IPPP/UPGP/628). Acknowledgement We also extend our gratitude to Cancer Research Malaysia (CRM), a non-profit cancer research organization, for their generous help in providing ORL cell lines and technical assistance. Author contributions S.K.A. and Y.F.C. conceived the idea for this project. G.R.W. and Y.F.C. performed and acquired data from single-cell qPCR and flow cytometry experiments. Y.W.C., M.K.A., and I.F. performed qPCR computational analyses. Y.W.C., N.P.R., and P.G.S.D. performed flow cytometry computational analyses. S.K.A. M.R. and Y.F.C. performed a literature search and wrote the manuscript. S.K.A., M.R., and Y.F.C. worked on formatting the manuscript for publishing and made the final revisions. All authors have read and agreed to the published version of the manuscript. Competing interests The author(s) declare no competing interests. Supplementary Information and Data Availability The data that support the findings of this study are available on request from the corresponding author, HS. The data are not publicly available due to the restrictions on the use of vital statistics data. References Ghanaati, S.; Udeabor, S. E.; Winter, A.; Sader, R.; Heselich, A. Cancer recurrence in operated primary oral squamous cell carcinoma patients seems to be independent of the currently available postoperative therapeutic approach: a retrospective clinical study. Curr. Oncol. 32, 208. https://doi.org/10.3390/curroncol32040208 (2025). Article PubMed Google Scholar Lee, H.; Kim, B.; Park, J.; Park, S.; Yoo, G.; Yum, S.; Kang, W.; Lee, J.-M.; Youn, H.; Youn, B. Cancer stem cells: landscape, challenges and emerging therapeutic innovations. Signal Transduct. Target. Ther. 10, 248. https://doi.org/10.1038/s41392-025-02360-2 (2025). Article PubMed Google Scholar El-Tanani, M.; Rabbani, S. A.; Satyam, S. M.; Rangraze, I. R.; Wali, A. F.; El-Tanani, Y.; Aljabali, A. A. A. Deciphering the role of cancer stem cells: drivers of tumor evolution, therapeutic resistance, and precision medicine strategies. Cancers 17, 382. https://doi.org/10.3390/cancers17030382 (2025). Article PubMed Google Scholar Acharya, S. K.; Shai, S.; Choon, Y. F.; Lim, C. T.; Tan, W. L.; Yeap, S. K.; Singh, R. Cancer stem cells in oral squamous cell carcinoma: a narrative review on experimental characteristics and methodological challenges. Biomedicines 12, 2111. https://doi.org/10.3390/biomedicines12092111 (2024). Article PubMed Google Scholar Lan, L.; Behrens, A. Are there specific cancer stem cell markers? Cancer Res. 83, 170–172. https://doi.org/10.1158/0008-5472.CAN-22-2053 (2023). Article PubMed Google Scholar Hamid, S.; Lim, K. P.; Zain, R. B.; Ismail, S. M.; Lau, S. H.; Mustafa, W. M. Establishment and characterization of Asian oral cancer cell lines as in vitro models to study a disease prevalent in Asia. Int. J. Mol. Med. 6, 453–460. https://doi.org/ ... (2007). Article PubMed Google Scholar Fadlullah, M. H.; Chiang, I. K. N.; Dionne, K. R.; Pang, J. C.; Lo, P. H. Y.; Ng, W. K.; Tsang, C. M.; Tsao, S. W.; Wong, K. K.; Lung, M. L. Genetically defined novel oral squamous cell carcinoma cell lines for the development of molecular therapies. Oncotarget 7, 27802–27818. https://doi.org/ ... (2016). Article PubMed Google Scholar Diggins, K. E.; Ferrell, P. B.; Irish, J. M. Methods for discovery and characterization of cell subsets in high-dimensional mass cytometry data. Methods 82 , 55–63. https://doi.org/10.1016/j.ymeth.2015.05.008 (2015). Article PubMed Google Scholar DiGiuseppe, J. A.; Cardinali, J. L.; Rezuke, W. N.; Pe’er, D. PhenoGraph and viSNE facilitate the identification of abnormal T-cell populations in routine clinical flow cytometric data. Cytometry B Clin. Cytom. 94, 588–601. https://doi.org/ ... (2018). Article PubMed Google Scholar Spidlen, J.; Leif, R. C.; Moore, W.; Roederer, M.; Brinkman, R. R. Gating-ML: XML-based gating descriptions in flow cytometry. Cytometry A 73, 1151–1157. https://doi.org/10.1002/cyto.a.20637 (2008). Article PubMed Google Scholar Chiou, S. H.; Yu, C. C.; Huang, C. Y.; et al. Positive correlations of Oct-4 and Nanog in oral cancer stem-like cells and high-grade oral squamous cell carcinoma. Clin. Cancer Res. 14, 4085–4095. https://doi.org/10.1158/1078-0432.CCR-07-4404 (2008). Article PubMed Google Scholar DeLaughter, D. M. The use of the Fluidigm C1 for RNA expression analyses of single cells. Curr. Protoc. Mol. Biol. 122, e55. https://doi.org/10.1002/cpmb.55 (2018). Article PubMed Google Scholar Lex, A.; Gehlenborg, N.; Strobelt, H.; Vuillemot, R.; Pfister, H. UpSet: visualization of intersecting sets. IEEE Trans. Vis. Comput. Graph. 20, 1983–1992. https://doi.org/10.1109/TVCG.2014.2346248 (2014). Article PubMed Google Scholar Patel, A. P.; Tirosh, I.; Trombetta, J. J.; et al. Single-cell RNA-seq highlights intratumoral heterogeneity in primary glioblastoma. Science 344, 1396–1401. https://doi.org/10.1126/science.1254257 (2014). Article PubMed Google Scholar Lieberman, B.; Kusi, M.; Hung, C. N.; et al. Toward uncharted territory of cellular heterogeneity: advances and applications of single-cell RNA-seq. J. Transl. Genet. Genom. 5, 1–21. https://doi.org/10.20517/jtgg.2020.51(2021). Article PubMed Google Scholar Navin, N.; Kendall, J.; Troge, J.; et al. Tumour evolution inferred by single-cell sequencing. Nature 472, 90–94. https://doi.org/10.1038/nature09807 (2011). Article PubMed Google Scholar Prince, M. E.; Sivanandan, R.; Kaczorowski, A.; Wolf, G. T.; Kaplan, M. J. Identification of a subpopulation of cells with cancer stem cell properties in head and neck squamous cell carcinoma. Proc. Natl. Acad. Sci. USA 104, 973–978 (2007). Article PubMed Google Scholar Mesrati, M. H.; Syafruddin, S. E.; Mohtar, M. A.; Syahir, A. CD44: a multifunctional mediator of cancer progression. Biomolecules 11, 1850. https://doi.org/10.3390/biom11121850 (2021). Article PubMed Google Scholar Zhang, H.; Brown, R. L.; Wei, Y.; et al. CD44 splice isoform switching determines breast cancer stem cell state. Genes Dev. 33, 166–179. https://doi.org/10.1101/gad.319889.118 (2019). Article PubMed Google Scholar Hiraga, T.; Nakamura, H. Comparable roles of CD44v8-10 and CD44s in the development of bone metastases in a mouse model. Oncol. Lett. 12, 2962–2969. https://doi.org/10.3892/ol.2016.4985 (2016). Article PubMed Google Scholar Clay, M. R.; Tabor, M.; Owen, J. H.; et al. Single-marker identification of head and neck squamous cell carcinoma cancer stem cells with aldehyde dehydrogenase. Head Neck 32, 1195–1201. https://doi.org/10.1002/hed.21315(2010). Article PubMed Google Scholar Kulsum, S.; Sudheendra, H. V.; Pandian, R.; et al. Cancer stem cell mediated acquired chemoresistance in head and neck cancer can be abrogated by aldehyde dehydrogenase 1A1 inhibition. Mol. Carcinog. 56, 694–711. https://doi.org/10.1002/mc.22526 (2017). Article PubMed Google Scholar De Castro, L. R.; de Oliveira, L. D.; Milan, T. M.; et al. Up-regulation of TNF-α/NFkB/SIRT1 axis drives aggressiveness and cancer stem cell accumulation in chemoresistant oral squamous cell carcinoma. J. Cell Physiol. 239, e31164. https://doi.org/10.1002/jcp.31164 (2024). Article PubMed Google Scholar Singh, S. K.; Hawkins, C.; Clarke, I. D.; et al. Identification of human brain tumour initiating cells. Nature 432, 396–401 (2004). Article PubMed Google Scholar Chiou, S. H.; Yu, C. C.; Huang, C. Y.; et al. Positive correlations of Oct-4 and Nanog in oral cancer stem-like cells and high-grade oral squamous cell carcinoma. Clin. Cancer Res. 14, 4085–4095. https://doi.org/10.1158/1078-0432.CCR-07-4404 (2008). Article PubMed Google Scholar Yu, C. C.; Hu, F. W.; Yu, C. H.; Chou, M. Y. Targeting CD133 in the enhancement of chemosensitivity in oral squamous cell carcinoma-derived side population cancer stem cells. Head Neck 38, E231–E238. https://doi.org/10.1002/hed.23975 (2016). Article PubMed Google Scholar Chen, D.; Wu, M.; Li, Y.; et al. Targeting BMI1⁺ cancer stem cells overcomes chemoresistance and inhibits metastases in squamous cell carcinoma. Cell Stem Cell 20, 621–634.e6. https://doi.org/10.1016/j.stem.2017.02.003 (2017). Article PubMed Google Scholar Wang, Q.; Li, Z.; Wu, Y.; et al. Pharmacological inhibition of Bmi1 by PTC-209 impairs tumor growth in head and neck squamous cell carcinoma. Cancer Cell Int. 17, 107. https://doi.org/10.1186/s12935-017-0481-z (2017). Article PubMed Google Scholar Yang, D.; Liu, H. Q.; Yang, Z.; Fan, D.; Tang, Q. Z. BMI1 in the heart: novel functions beyond tumorigenesis. EBioMedicine 63, 103193. https://doi.org/10.1016/j.ebiom.2020.103193 (2021). Article PubMed Google Scholar Poleszczuk, J.; Hahnfeldt, P.; Enderling, H. Evolution and phenotypic selection of cancer stem cells. PLoS Comput. Biol. 11, e1004025. https://doi.org/10.1371/journal.pcbi.1004025 (2015). Article PubMed Google Scholar Johansson, C. B.; Momma, S.; Clarke, D. L.; et al. Identification of a neural stem cell in the adult mammalian central nervous system. Cell 96, 25–34. https://doi.org/10.1016/S0092-8674(00)80956-3 (1999). Article PubMed Google Scholar Chen, S. F.; Chang, Y. C.; Nieh, S.; et al. Nonadhesive culture system as a model of rapid sphere formation with cancer stem cell properties. PLoS One 7, e31864. https://doi.org/10.1371/journal.pone.0031864 (2012). Article PubMed Google Scholar Pozzi, V.; Sartini, D.; Rocchetti, R.; et al. Identification and characterization of cancer stem cells from head and neck squamous cell carcinoma cell lines. Cell Physiol. Biochem. 36, 784–798. https://doi.org/10.1159/000430138 (2015). Article PubMed Google Scholar Millen, R.; de Kort, W. W. B.; Koomen, M.; et al. Patient-derived head and neck cancer organoids allow treatment stratification and serve as a tool for biomarker validation and identification. Med. 4, 290–310.e12. https://doi.org/10.1016/j.medj.2023.04.003 (2023). Article PubMed Google Scholar Tanaka, N.; Osman, A. A.; Takahashi, Y.; et al. Head and neck cancer organoids established by modification of the CTOS method can be used to predict in vivo drug sensitivity. Oral Oncol. 87, 49 57. https://doi.org/10.1016/j.oraloncology.2018.10.018 (2018). Article PubMed Google Scholar Siquara da Rocha, L. d. O.; Souza, B. S. d. F.; Coletta, R. D.; et al. Mapping cell-in-cell structures in oral squamous cell carcinoma. Cells 12, 2418. https://doi.org/10.3390/cells12192418 (2023). Article PubMed Google Scholar Goričan, L.; Gole, B.; Potočnik, U. Head and neck cancer stem cell-enriched spheroid model for anticancer compound screening. Cells 9, 1707. https://doi.org/10.3390/cells9071707 (2020). Article PubMed Google Scholar Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7844179","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":540942502,"identity":"524ea626-4c91-4434-a9ee-27d450dd4448","order_by":0,"name":"Surendra Kumar Acharya","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA0ElEQVRIiWNgGAWjYFACHgjFDybZSNEi2UCyFoMDxGrhbz978HNlm5298Y3sBIYPZYcZdGck4NcicSYvWfJsW3Lithu5GxhnnDvMYHaDgBaGGzwGko1tzAlmQC3MvG1EaJG/wWP8s7Gt3t54BlDLX2K0GNzgMQPacphxgwRQCyMxWgzP5JhZNpw7njjjzNsNB3vOpfOYnXmAX4vc8TPGNxvKqu3523M3PvhRZi1ndpyALWDACI2OAwygaBIgRgvDH2QO/wFitIyCUTAKRsEIAgBfF0eKURCK9QAAAABJRU5ErkJggg==","orcid":"","institution":"University of Malaya","correspondingAuthor":true,"prefix":"","firstName":"Surendra","middleName":"Kumar","lastName":"Acharya","suffix":""},{"id":540942503,"identity":"6bd01457-02e5-4672-8ce8-ad0f4d711dae","order_by":1,"name":"Guo Rean Wong","email":"","orcid":"","institution":"Mahsa University","correspondingAuthor":false,"prefix":"","firstName":"Guo","middleName":"Rean","lastName":"Wong","suffix":""},{"id":540942504,"identity":"954d1561-117d-4200-b2dc-78abc9bc086a","order_by":2,"name":"Yee Wen Choon","email":"","orcid":"","institution":"XPETO Pte Ltd, 114, Lavender Street, 07-75, #2 CT Hub, 338729","correspondingAuthor":false,"prefix":"","firstName":"Yee","middleName":"Wen","lastName":"Choon","suffix":""},{"id":540942505,"identity":"b25e03c1-975f-4560-a9dd-5ca0c1ba9979","order_by":3,"name":"Mamata Rai","email":"","orcid":"","institution":"MMIHS, Tribhuvan University","correspondingAuthor":false,"prefix":"","firstName":"Mamata","middleName":"","lastName":"Rai","suffix":""},{"id":540942506,"identity":"85865319-a9d4-4473-bb76-22138c287cf4","order_by":4,"name":"Muhammad Khan Asif","email":"","orcid":"","institution":"Mahsa University","correspondingAuthor":false,"prefix":"","firstName":"Muhammad","middleName":"Khan","lastName":"Asif","suffix":""},{"id":540942508,"identity":"49459537-d38d-4082-84e7-f5ca8684a36c","order_by":5,"name":"Ismat Fakhra","email":"","orcid":"","institution":"Mahsa University","correspondingAuthor":false,"prefix":"","firstName":"Ismat","middleName":"","lastName":"Fakhra","suffix":""},{"id":540942509,"identity":"0fe84587-a645-4bea-8e91-de7aad096b5e","order_by":6,"name":"Narendra Prakash Rai","email":"","orcid":"","institution":"Mahsa University","correspondingAuthor":false,"prefix":"","firstName":"Narendra","middleName":"Prakash","lastName":"Rai","suffix":""},{"id":540942510,"identity":"9e740386-4935-4ace-a928-781411ec6f86","order_by":7,"name":"Pateel Deepak Gowda Sadashivappa","email":"","orcid":"","institution":"Mahsa University","correspondingAuthor":false,"prefix":"","firstName":"Pateel","middleName":"Deepak Gowda","lastName":"Sadashivappa","suffix":""},{"id":540942511,"identity":"dbe4d752-dabc-430c-9079-2958fa0b5162","order_by":8,"name":"Yee Fan Choon","email":"","orcid":"","institution":"Mahsa University","correspondingAuthor":false,"prefix":"","firstName":"Yee","middleName":"Fan","lastName":"Choon","suffix":""}],"badges":[],"createdAt":"2025-10-13 03:53:25","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7844179/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7844179/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":95527170,"identity":"796392b3-c4f9-42ae-b38d-984003efaecc","added_by":"auto","created_at":"2025-11-10 10:11:18","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":69573577,"visible":true,"origin":"","legend":"","description":"","filename":"singlecellrevised.docx","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/d98fe4e9d42300fbd05e9a39.docx"},{"id":95391332,"identity":"05c9c9a4-bb85-4a3d-aa8f-c388205196c2","added_by":"auto","created_at":"2025-11-07 14:06:57","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9882,"visible":true,"origin":"","legend":"","description":"","filename":"dacdf973f1674aeb8b8d8a0e3782301c.json","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/f41157d3e8ede3af890f48e0.json"},{"id":95391338,"identity":"ab5d9f5c-9c5c-4f4d-84ad-648480ca6a65","added_by":"auto","created_at":"2025-11-07 14:06:57","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":131873,"visible":true,"origin":"","legend":"","description":"","filename":"dacdf973f1674aeb8b8d8a0e3782301c1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/ec7a162751f42404f8b1f38b.xml"},{"id":95527143,"identity":"e924f7ec-2b53-484b-bf7d-ae49141926b2","added_by":"auto","created_at":"2025-11-10 10:11:10","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":132858,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/a8603e0401f7dacd65e6415e.png"},{"id":95526121,"identity":"ef86b751-994d-4c68-9b59-128cd90310ae","added_by":"auto","created_at":"2025-11-10 10:06:18","extension":"jpeg","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":461160,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage10.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/fd691d22d7064cdb32b52254.jpeg"},{"id":95525668,"identity":"2b35cd37-aefa-4c68-9f1e-643e9a1b6261","added_by":"auto","created_at":"2025-11-10 10:05:32","extension":"jpeg","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":390530,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage11.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/041e6de7793939398852050e.jpeg"},{"id":95526323,"identity":"7cd23313-212d-49e0-a1dd-fea2eebca681","added_by":"auto","created_at":"2025-11-10 10:06:46","extension":"jpeg","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":498457,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage12.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/9ac246df0baff7008230cde3.jpeg"},{"id":95391380,"identity":"7a0adcce-b883-4a75-af82-1257f3d1a1bc","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":462291,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage13.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/beaef59703fc46749ad7f195.jpeg"},{"id":95391334,"identity":"761d438e-6c69-480d-969a-651d0233395d","added_by":"auto","created_at":"2025-11-07 14:06:57","extension":"jpeg","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1074,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage14.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/7d0554b559adc8848fe7ef75.jpeg"},{"id":95527176,"identity":"c4c0c94f-e9f1-4c99-99eb-bc5ae717d3c6","added_by":"auto","created_at":"2025-11-10 10:11:18","extension":"jpeg","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":359806,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage15.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/4e091583d24686567fc5cf08.jpeg"},{"id":95391357,"identity":"51ff1cc6-d775-47e9-b99c-3bc897078e76","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":350328,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage16.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/b3828479a5d55049594affe6.jpeg"},{"id":95391337,"identity":"2eae7197-c37a-4915-8841-49ccb2807c9f","added_by":"auto","created_at":"2025-11-07 14:06:57","extension":"jpeg","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5849,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage17.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/8db06e35705c9d4ed8da91be.jpeg"},{"id":95391409,"identity":"4ae36c95-713d-453e-aa23-678a44d7287d","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"jpeg","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":287161,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage18.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/699009b86628f5d2e9371674.jpeg"},{"id":95391404,"identity":"fb60f9ee-d023-43fd-ba5a-33c155d34101","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"jpeg","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":421861,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage19.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/c4ae709b18a4eb50baa7dadf.jpeg"},{"id":95526040,"identity":"7af32414-dc1a-4524-8f75-9b52ff38f8dc","added_by":"auto","created_at":"2025-11-10 10:06:07","extension":"jpeg","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":8741,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/106cd8e3862e680188ce4619.jpeg"},{"id":95391346,"identity":"86a29a11-9f8e-4e81-9a16-01980f089526","added_by":"auto","created_at":"2025-11-07 14:06:57","extension":"jpeg","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5621,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage20.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/61c4ef371ba323747254d54e.jpeg"},{"id":95391348,"identity":"96b0b57f-0070-42c4-aaf8-554e82ebad6d","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":322903,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage21.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/13bac7dd0069ee2cbf3fb30e.jpeg"},{"id":95526716,"identity":"5a4c33dd-34f3-4650-be2d-393dbf26898f","added_by":"auto","created_at":"2025-11-10 10:07:40","extension":"jpeg","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":247371,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage22.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/a229def7e7c5eb2d2fffb808.jpeg"},{"id":95391347,"identity":"1b63b8a7-25e9-47e1-94e3-7712edb7c7ee","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1074,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage14.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/a108a4333d0d0774dd84b284.jpeg"},{"id":95391342,"identity":"7fbb5420-7f12-4e1d-a838-e9312b12f516","added_by":"auto","created_at":"2025-11-07 14:06:57","extension":"jpeg","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9345,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage24.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/af0cdc8d60936ec55d5187e3.jpeg"},{"id":95391355,"identity":"36cf6558-54cd-46a7-90d8-d1c421e432af","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":725494,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage25.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/96e384f6d5f002c185c3542c.jpeg"},{"id":95525997,"identity":"16c5a1cc-9412-4a6a-a478-1104e8a146a8","added_by":"auto","created_at":"2025-11-10 10:06:01","extension":"jpeg","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":478819,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage26.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/66b4ef8cc3c5fa7debaa36f7.jpeg"},{"id":95526793,"identity":"c003d1e5-01c8-40c8-94be-1b7541cf6421","added_by":"auto","created_at":"2025-11-10 10:07:56","extension":"jpeg","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6366,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage27.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/ce2c4921dc47e487d867fa77.jpeg"},{"id":95526257,"identity":"51472639-e74e-4ed6-8eb8-e3cc50dd40e8","added_by":"auto","created_at":"2025-11-10 10:06:38","extension":"jpeg","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":565744,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage28.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/41d8131693957a30e3ee8c71.jpeg"},{"id":95391350,"identity":"bc5fda00-f80e-4e84-9284-c34d82fa4e91","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":25,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":520065,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage29.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/f0e4244659a553446bbf3619.jpeg"},{"id":95526319,"identity":"27a7f842-789a-4c11-a8de-9db5d9035ec0","added_by":"auto","created_at":"2025-11-10 10:06:46","extension":"jpeg","order_by":26,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":413667,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/69a429708aa2e13206e2fbaf.jpeg"},{"id":95391358,"identity":"4b1979d8-20a9-46cf-9b2b-ee270723aff4","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":27,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1074,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage14.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/81af2e036ca54e7250e6094c.jpeg"},{"id":95391351,"identity":"96bb89ee-314c-4c92-9c19-89b3164053cd","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":28,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5534,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage31.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/5324b7e0a1a93e1bbd220046.jpeg"},{"id":95526439,"identity":"960f2a9e-cfda-481c-aba6-28dfd1823456","added_by":"auto","created_at":"2025-11-10 10:06:59","extension":"jpeg","order_by":29,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":515259,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage32.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/ea3269097d2167706ab1b191.jpeg"},{"id":95391362,"identity":"33a5f398-79d5-45db-b617-50cda162a2a6","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":30,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":314164,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage33.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/39dbb88d92b3dcc57a08e66a.jpeg"},{"id":95391366,"identity":"d9c5c99b-24af-43ad-a54c-af577fc8cdc9","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":31,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5800,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage34.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/238dffbb2176542e6bca5ff9.jpeg"},{"id":95525544,"identity":"695a69f8-d879-452d-9bca-a976a7f5076e","added_by":"auto","created_at":"2025-11-10 10:05:16","extension":"jpeg","order_by":32,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":413212,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage35.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/06b42a3f56725ae0403971b3.jpeg"},{"id":95391365,"identity":"0665a4c6-babe-4afa-bf20-97a48749d448","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":33,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":402037,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage36.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/da2254e2cade1d84eb17167f.jpeg"},{"id":95526167,"identity":"9bb3ea12-346e-428f-8ec9-b8bb65e5dd11","added_by":"auto","created_at":"2025-11-10 10:06:26","extension":"jpeg","order_by":34,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1074,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage14.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/be586f914105ad7a1b38d6d0.jpeg"},{"id":95391359,"identity":"c5c8beb3-2687-41f0-9f16-68d6550dcbe8","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":35,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10981,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage38.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/58a898460c3d66c2fe2ef4f5.jpeg"},{"id":95391363,"identity":"678dff4c-7df9-4e96-8ea0-dfc2dae4f454","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":36,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":695836,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage39.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/d65a016903443d67c97835d3.jpeg"},{"id":95391387,"identity":"959a549e-e5a8-4076-85a9-bfae031dbcd6","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"jpeg","order_by":37,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10182,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/33bbc1dc30672c76724120fd.jpeg"},{"id":95525627,"identity":"f1168542-ec97-4bfd-bbd1-0d5da2bf6e11","added_by":"auto","created_at":"2025-11-10 10:05:28","extension":"jpeg","order_by":38,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":609661,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage40.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/9eb21b30d9a0667e43f07e04.jpeg"},{"id":95391367,"identity":"322c4b80-e8ed-4af3-82ff-e6f71ad87e33","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":39,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":346368,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/ddb3b72c07ae80e95175ba2b.jpeg"},{"id":95391371,"identity":"1aab0b55-b2e4-4604-921d-3b11d76b3c91","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":40,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":215890,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/5028526b7eb00dc18534302a.jpeg"},{"id":95527168,"identity":"6ce1c749-61cc-4b79-8872-637cee8a5ca9","added_by":"auto","created_at":"2025-11-10 10:11:18","extension":"jpeg","order_by":41,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":214408,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/60bf9112c6ad8ef9a2630f8e.jpeg"},{"id":95391364,"identity":"eb0843ac-061a-44c5-b3f9-dd528391e3e8","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":42,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1074,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage14.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/f10012eb8e36b633ae7f1c72.jpeg"},{"id":95391361,"identity":"2df7c047-f958-49ad-be3f-c5c4d74a50a4","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"jpeg","order_by":43,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6854,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage9.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/d463417a316697fe339736eb.jpeg"},{"id":95391397,"identity":"1eba80c4-dac7-401f-9022-6143b57172fc","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"png","order_by":44,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":40980,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/42764493f1874e9343c04913.png"},{"id":95391370,"identity":"c5309c72-97a5-4f6f-ab3d-0858913daa25","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"png","order_by":45,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":102042,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage10.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/95170a77df502d817346e198.png"},{"id":95391400,"identity":"192320c6-0c5e-4742-8c64-166d7f010b12","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"png","order_by":46,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":85395,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage11.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/b6c65e60d3bfd8264d892f56.png"},{"id":95391378,"identity":"90c0cd56-df57-4af1-80b2-d71fdb01d92a","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"png","order_by":47,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":100161,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage12.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/e8906b3fd04ffed2bac38fe3.png"},{"id":95526336,"identity":"44976c2c-5d48-422f-b90e-cabe04341ea6","added_by":"auto","created_at":"2025-11-10 10:06:48","extension":"png","order_by":48,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":90646,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage13.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/cce2f1922bc8e345d65323b7.png"},{"id":95526457,"identity":"10569751-878c-47a5-ad7a-eacc7c88e5fb","added_by":"auto","created_at":"2025-11-10 10:07:00","extension":"png","order_by":49,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":935,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage14.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/d26cecc49c4f2c526d39379d.png"},{"id":95391394,"identity":"07b85fa8-edb5-4d43-9ed5-a64589f068c4","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"png","order_by":50,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":54186,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage15.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/4f39d32195a640a8fd79e86b.png"},{"id":95526296,"identity":"10071ba0-b417-4afd-855e-f380cc0178b3","added_by":"auto","created_at":"2025-11-10 10:06:44","extension":"png","order_by":51,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":73960,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage16.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/a7c6efc4b7137782282e0252.png"},{"id":95526552,"identity":"b61014dd-cf14-4965-b056-9aa37b4df59c","added_by":"auto","created_at":"2025-11-10 10:07:14","extension":"png","order_by":52,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2001,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage17.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/33a5e8642c7c50c5f2ff68db.png"},{"id":95526725,"identity":"5b60d5f2-9d9f-4435-8041-cafa1aa21198","added_by":"auto","created_at":"2025-11-10 10:07:42","extension":"png","order_by":53,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":49590,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage18.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/00d9137bc45f409442ee20e8.png"},{"id":95391369,"identity":"1f583d92-c19a-4200-a0bc-3237440580c6","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"png","order_by":54,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":72023,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage19.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/2332ad15a4e14cebd4e79a96.png"},{"id":95525736,"identity":"123a78cc-bfdb-4ef8-9a21-96bdc56f3382","added_by":"auto","created_at":"2025-11-10 10:05:38","extension":"png","order_by":55,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2620,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/eb5ff0a1b754548b3ed1542b.png"},{"id":95391395,"identity":"b48a86c1-9d1f-4ef2-a00b-3de63e366ede","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"png","order_by":56,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1950,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage20.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/e365b36352a411400524ce85.png"},{"id":95526473,"identity":"d4483ba7-90e0-4ba8-aa24-390ead51a6b1","added_by":"auto","created_at":"2025-11-10 10:07:02","extension":"png","order_by":57,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":50756,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage21.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/47ccadfc8973d3c3ddc14613.png"},{"id":95526717,"identity":"cda2e1ff-5d0a-4526-999d-bc18fd10e6a5","added_by":"auto","created_at":"2025-11-10 10:07:41","extension":"png","order_by":58,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":38796,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage22.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/d81d8b4f26ad42c9093495cc.png"},{"id":95526936,"identity":"cb25aacd-b4c2-46a5-8130-e0056c7baa17","added_by":"auto","created_at":"2025-11-10 10:08:47","extension":"png","order_by":59,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":935,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage14.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/4077c44d8d1f732c70e0b3c9.png"},{"id":95391399,"identity":"b7f2b590-acad-431b-b33a-d482744db95b","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"png","order_by":60,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2230,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage24.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/6a4087633222ac96ed073fb8.png"},{"id":95526472,"identity":"65304649-5887-407f-b154-f90889e27d82","added_by":"auto","created_at":"2025-11-10 10:07:02","extension":"png","order_by":61,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":224619,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage25.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/20695ce4aa95361ff0d1b11b.png"},{"id":95391389,"identity":"d1f8ff92-11f5-46e7-b705-efe8bb183101","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"png","order_by":62,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":148568,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage26.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/3962e072efef29c4ff083a2b.png"},{"id":95525547,"identity":"3a53a6bf-2dac-4309-9dad-634d04d35195","added_by":"auto","created_at":"2025-11-10 10:05:16","extension":"png","order_by":63,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2000,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage27.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/8915437d8c950df39230f332.png"},{"id":95391420,"identity":"c20cc3c0-e5f1-4bf9-9eca-69c130c0742a","added_by":"auto","created_at":"2025-11-07 14:07:00","extension":"png","order_by":64,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":159083,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage28.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/3fa6178505d6e893c6f77b18.png"},{"id":95526260,"identity":"3d02987c-edce-4615-9fe9-2358dd0d6623","added_by":"auto","created_at":"2025-11-10 10:06:39","extension":"png","order_by":65,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":143535,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage29.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/c85483d7e0d9bb41b7086d9e.png"},{"id":95391415,"identity":"cb076a81-89b5-4ff2-b9ad-0116623859ff","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"png","order_by":66,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":88428,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/a70d8f31d1663e40c0a138cb.png"},{"id":95391379,"identity":"0959ec8b-366a-4c41-94e0-1a511a0820bb","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"png","order_by":67,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":935,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage14.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/5af82be13106f2468ad25bbf.png"},{"id":95391405,"identity":"92891b36-7445-453f-ab86-e1903848b444","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"png","order_by":68,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1868,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage31.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/40f09677eee74071449475a6.png"},{"id":95526779,"identity":"f1cd39bd-e7fd-4cb4-8c77-675bfa890e98","added_by":"auto","created_at":"2025-11-10 10:07:51","extension":"png","order_by":69,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":159914,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage32.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/532f8db0d997d2bbf1338044.png"},{"id":95526750,"identity":"417f15d5-3054-4c70-92a8-6794f38fcbc1","added_by":"auto","created_at":"2025-11-10 10:07:46","extension":"png","order_by":70,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":96877,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage33.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/d463b4ab8497ee314e47c13d.png"},{"id":95391417,"identity":"39ecacd7-39e7-4b2b-832b-f6cf36fc9f5e","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"png","order_by":71,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1974,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage34.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/d866d7ed6ff1eb8592e78329.png"},{"id":95391385,"identity":"5948126f-4d7d-4809-a0d0-f1cf8e8cbc4f","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"png","order_by":72,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":99634,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage35.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/194be20e178827d0658daae1.png"},{"id":95391407,"identity":"9105b782-846c-4434-aabb-f69c8de975b2","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"png","order_by":73,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":93152,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage36.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/6f2c31ab07a20b0b3674cca8.png"},{"id":95525491,"identity":"c3537a5e-b8dd-462c-96a9-01114babd0bf","added_by":"auto","created_at":"2025-11-10 10:05:08","extension":"png","order_by":74,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":935,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage14.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/a18e5586c08cb43f04d2f112.png"},{"id":95391391,"identity":"a50743e7-6478-4174-9dac-8715cfd48b11","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"png","order_by":75,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2390,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage38.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/254801390b345a461c5f229f.png"},{"id":95391401,"identity":"2cd5a64d-8e38-443e-8393-d7b29015acff","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"png","order_by":76,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":224980,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage39.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/6ba852cdf94748106304aa4d.png"},{"id":95391418,"identity":"a2617f88-7fe6-4a74-9f41-0a1a3f6ed2b3","added_by":"auto","created_at":"2025-11-07 14:07:00","extension":"png","order_by":77,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3032,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/072c968a8a8d05da90a8e220.png"},{"id":95391423,"identity":"7598c73c-db9f-471a-9988-e491ff349042","added_by":"auto","created_at":"2025-11-07 14:07:00","extension":"png","order_by":78,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":200372,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage40.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/5780ec999f7a7048f662634b.png"},{"id":95527180,"identity":"26dcc888-adeb-414a-b0a4-3231318eb312","added_by":"auto","created_at":"2025-11-10 10:11:18","extension":"png","order_by":79,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":34513,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/fa514e8e3f96243adac164c4.png"},{"id":95391421,"identity":"0a575899-0b42-4d4f-a52e-7a9352d56d37","added_by":"auto","created_at":"2025-11-07 14:07:00","extension":"png","order_by":80,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":94872,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/aa045eb37da1a6d56cb012b4.png"},{"id":95391398,"identity":"f2a9d478-4413-41b3-9d61-05be0f08e7db","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"png","order_by":81,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":95206,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/1c1f203ec7589b75839988a2.png"},{"id":95525884,"identity":"9cbd15e0-5ca9-4e40-9f20-edd25d5cea31","added_by":"auto","created_at":"2025-11-10 10:05:48","extension":"png","order_by":82,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":935,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage14.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/1210d154665b14e3a9d5ba01.png"},{"id":95526652,"identity":"c204b300-4485-41c5-9896-bc844146cdeb","added_by":"auto","created_at":"2025-11-10 10:07:31","extension":"png","order_by":83,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2063,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage9.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/b734db52d28910f56b1565f3.png"},{"id":95525684,"identity":"aa4d1d1c-7102-4f19-86ee-c71f679c3363","added_by":"auto","created_at":"2025-11-10 10:05:33","extension":"xml","order_by":84,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":130900,"visible":true,"origin":"","legend":"","description":"","filename":"dacdf973f1674aeb8b8d8a0e3782301c1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/b2e65ae1a68cdddcf1f59e93.xml"},{"id":95391396,"identity":"c9b77e30-9771-4103-b73b-5f6ecc5f8f31","added_by":"auto","created_at":"2025-11-07 14:06:59","extension":"html","order_by":85,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":148669,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/77d30860742d13039a5a9775.html"},{"id":95391329,"identity":"56dbf116-c66a-4395-aba5-91574d35e900","added_by":"auto","created_at":"2025-11-07 14:06:57","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":99730,"visible":true,"origin":"","legend":"\u003cp\u003eSingle-cell Ct distributions of CSC marker expression across ORL cell lines.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/cba782135c336554b2439cb9.png"},{"id":95391331,"identity":"ed6affdc-65a3-45cc-9295-7033c6ffa961","added_by":"auto","created_at":"2025-11-07 14:06:57","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":156316,"visible":true,"origin":"","legend":"\u003cp\u003ePCA with K-Means clustering of single-cell Ct profiles in ORL cell lines.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/f096ab3d2f8b6f037a1adaed.png"},{"id":95527162,"identity":"b1255790-c16a-4a88-a989-76b4f979a4e3","added_by":"auto","created_at":"2025-11-10 10:11:15","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":242188,"visible":true,"origin":"","legend":"\u003cp\u003eCluster composition by marker combination across ORL cell lines.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/35c2e15b23d7bcef42362e2b.png"},{"id":95391328,"identity":"24cde6bf-18e4-425f-821d-6ef785635259","added_by":"auto","created_at":"2025-11-07 14:06:57","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":104494,"visible":true,"origin":"","legend":"\u003cp\u003eMarker positivity combinations in ORL cell lines visualized by UpSet plots.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/87d2a47620d039eae0ac55cb.png"},{"id":95526147,"identity":"0edcacb3-c545-44e2-907e-a4a84ddc9ffb","added_by":"auto","created_at":"2025-11-10 10:06:22","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":847851,"visible":true,"origin":"","legend":"\u003cp\u003eForward and side scatter (FSC-A vs SSC-A) gating of unstained ORL cells for P1 population definition.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/fd56a3e817bdac1d753b230e.png"},{"id":95526743,"identity":"8013fd12-c709-4932-a7ad-ef234b0e0e73","added_by":"auto","created_at":"2025-11-10 10:07:46","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":307097,"visible":true,"origin":"","legend":"\u003cp\u003eOverlay histograms of marker expression in ORL cell lines.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/7acdb0745d6fde2474780120.png"},{"id":95526425,"identity":"c3d93188-92ef-4688-9309-ff821a1e46e5","added_by":"auto","created_at":"2025-11-10 10:06:58","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":427715,"visible":true,"origin":"","legend":"\u003cp\u003ePhenoGraph cluster heatmaps of CSC marker expression in ORL cell lines.\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/a859ca3609bde62b9038d856.png"},{"id":95527046,"identity":"9747ea05-e320-4b49-9912-388efa9ac093","added_by":"auto","created_at":"2025-11-10 10:09:14","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":1237568,"visible":true,"origin":"","legend":"\u003cp\u003eviSNE maps colored by Louvain clusters of ORL cell lines showing cluster-level organization.\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/40cb25017a95e13267ac5fd0.png"},{"id":95391375,"identity":"87904c01-78dd-499c-a9d6-24eb0e982fca","added_by":"auto","created_at":"2025-11-07 14:06:58","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":748083,"visible":true,"origin":"","legend":"\u003cp\u003eviSNE colored by per-cell phenotype.\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/c3ec23615a9075924c552792.png"},{"id":95391339,"identity":"64a4f88d-ee46-442f-b4a0-33bebe9c49fe","added_by":"auto","created_at":"2025-11-07 14:06:57","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":1273346,"visible":true,"origin":"","legend":"\u003cp\u003eSphere formation assay confirmed the presence of self-renewing CSC-like subpopulation across ORL cell lines.\u003c/p\u003e","description":"","filename":"10.png","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/5f3096567291c42121a9992e.png"},{"id":98289826,"identity":"c573af8d-c3ea-4a83-b416-396ec7e752a3","added_by":"auto","created_at":"2025-12-16 07:55:16","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":5977514,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7844179/v1/83f88f5a-34db-46e7-a5fc-70b0d84eeae5.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Unravelling the heterogeneity in oral cancer stem cells – A single-cell computational exploratory study","fulltext":[{"header":"Introduction","content":"\u003cp\u003eOral squamous cell carcinoma (OSCC) remains one of the most prevalent malignancies of the head and neck, with persistently poor survival rates and frequent recurrence despite advances in treatment\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. The presence of cancer stem cells (CSCs), a subpopulation of tumor cells with self-renewal capacity, therapy resistance, and tumor-initiating potential, is a major factor underlying this clinical outcome\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. CSCs are increasingly recognized as key drivers in tumor heterogeneity, metastasis, and treatment failure, making their accurate identification and characterization a critical objective in CSC-targeted cancer therapies\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eThe challenge in CSC study lies in their rarity, plasticity, and reliance on surrogate marker expression. Markers such as CD44, CD133, ALDH, and BMI1 have been widely implicated in head and neck and oral cancers, yet no single marker defines CSC identity, and marker expression is often heterogeneous within tumors\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/span\u003e,\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. To investigate CSC heterogeneity in previously authenticated Asian-OSCC-derived ORL cell lines\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/span\u003e,\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e, expression of three established CSC markers (CD44, CD133, and ALDH1A1) and a key regulator of stemness and resistance (BMI1) in ORL cell lines was profiled, aiming to capture overlapping subpopulations. Additionally, to avoid inter-operator variability and bias from reliance on previously reported OSCC-CSC subpopulations, computational methods were adopted to objectively and reproducibly explore the heterogeneous nature of oral CSCs in the current study\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/span\u003e,\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/span\u003e,\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. Sphere formation ability, which has also been used as a CSC marker\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e, was also assessed as complementary validation of CSCs\u0026rsquo; behaviour in ORL cell lines.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003e\u003cstrong\u003eCell culture\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFour authenticated and genetically-defined Asian OSCC cell lines (ORL-115, ORL-48, ORL-174, ORL-214) were grown as previously described\u003cu\u003e\u003csup\u003e6\u003c/sup\u003e\u003c/u\u003e\u003csup\u003e,\u003cu\u003e7\u003c/u\u003e\u003c/sup\u003e.\u003c/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eGene expression of \u003cem\u003eCD44, CD133, ALDH1A1, BMI1\u003c/em\u003e in single-cell assay\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo investigate the expression of \u003cem\u003eCD44, CD133, ALDH1A1, and BMI1\u0026nbsp;\u003c/em\u003eat single-cell resolution\u003cem\u003e,\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/em\u003e70% sub-confluent ORL cells were used for specific target amplification (STA) qRT-PCR measurement of mRNA levels in individual cells using the C1 Single-Cell Autoprep System and BioMark HD system Standard BioTools Inc., California, USA). The priming and loading of cells of the Integrated Fluid Circuit (IFC) have been described elsewhere\u003cu\u003e\u003csup\u003e12\u003c/sup\u003e\u003c/u\u003e. Briefly, cells were loaded on a C1\u003csup\u003eTM\u0026nbsp;\u003c/sup\u003eSingle-Cell Preamp IFC, 17-25\u0026micro;m (Standard BioTools Inc., California, USA) \u0026nbsp; using the Fluidigm C1 machine. \u0026nbsp;A Leica DMI6000B (Leica, Heidelberg, Germany) fully automated inverted research microscope was used to identify wells containing single cells. Pre-amplified cDNA was then generated from each cell using the Ambione\u003csup\u003eTM\u003c/sup\u003e Single Cells-to-C\u003csub\u003eT\u003c/sub\u003e\u003csup\u003eTM\u003c/sup\u003e Kit (Thermo Fisher Scientific Inc., Massachusetts, USA), with pooled qPCR primers for \u003cem\u003eCD44, CD133, ALDH1A1, and BMI1\u003c/em\u003e (Thermo Fisher Scientific Inc., Massachusetts, USA) (100\u0026micro;M each) and Fluidigm STA reagents (Standard BioTools Inc., California, USA). The BioMark\u003csup\u003eTM\u003c/sup\u003e HD system was then used for qPCR of single-cell pre-amplified cDNA according to the manufacturer\u0026rsquo;s protocol (\u0026ldquo;Gene expression with the Flex Six IFC Using Fast/Standard Taqman assays\u0026rdquo;, Standard BioTools Inc., California, USA). Briefly, a 1.8 \u0026micro;L of each preamplified cDNA was mixed with 2.0 \u0026micro;L of 2X TaqMan Standard PCR Master Mix (Thermo Fisher Scientific Inc. Massachusetts, USA) and 0.2 \u0026micro;L GE Sample Reagent (Standard BioTools Inc., California, USA) and each sample mix was then pipetted into sample inlet in a FLEXsix\u003csup\u003eTM\u003c/sup\u003e IFC chip (Standard BioTools Inc., California, USA). Then, 2.0 \u0026micro;L of Individual 20X Taqman Gene Expression (\u003cem\u003eCD44, CD133, ALDH1A1, BMI1\u003c/em\u003e) was mixed with 2.0 \u0026micro;l Assay Loading Reagent (Standard BioTools Inc., California, USA), and pipetted into the assay inlet FLEXsix\u003csup\u003eTM\u003c/sup\u003e IFC chip. IFC Controller HX (Standard BioTools Inc., California, USA) was used to load the mixes into the FLEXsix\u003csup\u003eTM\u003c/sup\u003e IFC chip, and qPCR was performed on the BioMark\u003csup\u003eTM\u003c/sup\u003e HD real-time PCR reader (Standard BioTools Inc., California, USA) following the manufacturer\u0026apos;s instructions. qPCR data were collected on Biomark\u003csup\u003eTM\u003c/sup\u003e HD (instrument ID BIOMARKHD193) using software version 4.1.3. Baseline correction was set to LINEAR, with automatic global thresholding applied across all assays (Ct Threshold Method = Auto Global). The software quality threshold was 0.65, and ROX was used as a passive reference. CT values were exported in CSV tabular format for downstream analyses (see Data processing and computational analysis)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFlow cytometry analysis of CD44, CD133, ALDH, BMI1 expression\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo validate the expression of CD44, CD133, ALDH, and BMI1,\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e70% subconfluent ORL cells were harvested by trypsinization. Cells were washed in PBS and stained with respective antibodies. Due to the technical incompatibility of combining live-cell ALDH detection (which required an intact membrane) with BMI1 intracellular staining (which requires fixation and permeabilization), co-expression of ALDH and BMI1 could not be analyzed within the same sample. Instead, ALDH and BMI-1 expression were independently profiled with CD44 and CD133, allowing complementary assessment of these markers. Briefly, ALDEFLUOR\u0026trade; Kit (Stem Cell Technologies, Vancouver, Canada, catalog #01700) was used to detect ALDH activity. \u0026nbsp; Cells were suspended in ALDEFLUOR\u0026trade; Kit buffer containing 5\u0026micro;l ALDH substrate (BAAA, 1 \u0026micro;mol/L per 10\u003csup\u003e6\u0026nbsp;\u003c/sup\u003ecells) and incubated for 30 min at 37\u0026deg;C. As a negative control, an aliquot for each sample of cells was treated with 50 mmol/L diethylaminobenzaldehyde (DEAB), a specific ALDH inhibitor. For staining of BMI1, cells were treated with eBioscience\u003csup\u003eTM\u003c/sup\u003e Intracellular Fixation and Permeabilization Buffer Set (Thermo Fisher Scientific Inc., Massachusetts, USA) before incubation with 10 \u0026micro;L human BMI-1 APC-conjugated antibody (R\u0026amp;D Systems, Minneapolis, USA, clone# 384515, catalog number: IC33341a). ALDEFLUOR\u0026trade;- and BMI1-reacted cells were subsequently stained with 10\u0026micro;L BD Pharmigen\u003csup\u003eTM\u003c/sup\u003e PE-Cy\u003csup\u003eTM\u003c/sup\u003e7 mouse anti-human CD44 antibody (BD Biosciences, California, USA, clone G44-26, material number: 559942) and 10\u0026micro;L monoclonal anti-human CD133/1-PE antibody (Miltenyi Biotec, Bergisch Gladbach, Germany, order no. 130-113-670) for 30 min on ice. After staining, the cells were washed in phosphate-buffered saline buffer (PBS), reconstituted with PBS buffer. Data were acquired using a BD FACSCanto\u003csup\u003eTM\u003c/sup\u003e II flow cytometer (BD Biosciences, California, USA) and exported as FCS files using BD FACSDiva software (version 6.1.3) for downstream analyses (see Data processing and computational analysis).\u003c/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eData processing and computational analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis section outlines the computational workflow for identifying and interpreting clusters of marker-positive subpopulations in ORL cell lines. Dimensionality reduction and unsupervised clustering were performed first, followed by threshold-based marker annotation for biological interpretation. Computational analysis was performed using a combination of Python 3.9.6 libraries: pandas, numpy (data processing), matlibplot, seaborn (plots), scikit-learn (PCA, KMeans), Upsetplot (UpSet visualization) on MacOS Sonoma 14.7.8.\u0026nbsp;\u003c/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eqPCR single-cell Ct data\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInput data and marker panel\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe dataset was generated from \u003cstrong\u003eGene expression of \u003cem\u003eCD44, CD133, ALDH1A1, and BMI1\u003c/em\u003e in a single-cell assay\u0026nbsp;\u003c/strong\u003eand comprised per-cell Ct values for four markers: CD44, CD133, ALDH1A1, and BMI1. Each entry also included the corresponding cell line and a unique sample identifier.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eViolin plot visualization of Ct value distribution\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eqPCR data for CD44, CD133, ALDH1A1, and BMI1 were aggregated as per-cell Ct values.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eWells failing amplification (no Ct or Ct = 999) and Ct values \u0026gt; 35 (defined as non-detection by conventional Ct detection threshold \u0026lt; 35) were excluded from the kernel density estimate and visualization. \u0026nbsp;A violin plot shows the mirrored kernel density of the Ct distribution for each gene within each cell line. Inside each violin, the median lines and interquartile ranges (IQR) (dashed) were shown, whiskers extended to 1.5xIQR. Raw, untransformed Ct values were used for these plots. Lower Ct indicated higher expression.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePrincipal Component Analysis (PCA) and K-means clustering (unsupervised)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEach ORL cell line was processed independently. CT values from the four markers were first log\u003csub\u003e2\u003c/sub\u003e-transformed. The resulting data matrix was standardized via z-score normalization using the Standard Scaler module in scikit-learn. PCA was applied to this normalized matrix, and the top two components (PC1, PC2) were retained. K-Means clustering (k = 3) was then performed on the PCA-reduced data to identify these unsupervised clusters pre-cell line. Cluster centroids were initialized using a random seed (random_state = 42) to ensure reproducibility.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMarker positivity classification (post-clustering)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo assign biological meaning to the unsupervised clusters, a separate marker classification step was performed using the original Ct matrix. For each marker in each cell line, a threshold was calculated using the heuristic: Threshold = Mean (Ct) = 2 x standard deviation (Ct). Cells with Ct values lower than the threshold were considered positive for that marker (binary = 1). This produced a marker positivity matrix (0/1) for each cell.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMarker combination labelling and cluster interpretation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEach cell was annotated with a marker combination label, such as \u0026nbsp;CD133\u003csup\u003e+\u003c/sup\u003eALDH1A1\u003csup\u003e+\u003c/sup\u003eCD44\u003csup\u003e-\u003c/sup\u003eBMI1\u003csup\u003e+\u003c/sup\u003e\u003cstrong\u003e,\u0026nbsp;\u003c/strong\u003ebased on its binary positivity status. A cross-tabulation was performed between cluster assignments and marker combination labels to examine which clusters were enriched in specific marker-positive populations.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eVisualization outputs\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFor each cell line, the following visual outputs were generated:\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003eViolin plots of Ct distributions\u003c/li\u003e\n \u003cli\u003ePCA scatter plots with cluster assignments\u003c/li\u003e\n \u003cli\u003eStacked bar charts showing marker combination composition within clusters\u003c/li\u003e\n \u003cli\u003eUpSet plots - a scalable matrix-based set visualization method designed for quantitative analysis of multiple set intersections. Unlike Venn or Euler diagrams, which become intractable with more than three or four sets, UpSet provides an interpretable and scalable visualization of intersections and their cardinalities\u003cu\u003e\u003csup\u003e13\u003c/sup\u003e\u003c/u\u003e.\u003c/li\u003e\n\u003c/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eFlow cytometry computational analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRaw flow cytometry standard files were acquired (see\u003cstrong\u003e\u0026nbsp;Flow cytometry analysis of CD44, CD133, ALDH, BMI1 expression\u003c/strong\u003e). Data were exported as standard .fsc files in Flow Cytometry Standard v3.1 format. Raw FCS files contained a vendor spillover key (SPILL or $SPILLOVER). Before analysis, matrices were read, and compensation to the fluorescence channels was applied. If channels were exported as compensated (prefixed Comp-) compensation was not applied. Channels recorded were: CD44-PE-Cy7-A, CD133-PE-A, ALDEDLUOR\u003csup\u003eTM\u003c/sup\u003e-ALDEFLUOR-A, and BMI-APC-1. Computational analysis was performed using a combination of Python 3.9.6 libraries (FlowCal, NumPy, pandas, matplotlib, scanpy/phenograph) and R 4.2.3 libraries (flowCore, Rtsne, ggplot2, ggraph) on MacOS Sonoma 14.7.8. Unsupervised clustering and visualization were performed using the PhenoGraph algorithm in combination with viSNE, which has been validated as a robust approach for analyzing high-dimensional flow cytometry data. DiGiuseppe et al. (2018)\u003cu\u003e\u003csup\u003e9\u003c/sup\u003e\u003c/u\u003e demonstrated that this approach could reliably distinguish normal versus aberrant T-cell subsets in clinical flow cytometry without reliance on manual pre-gating, thereby minimizing operator bias. The strategy was adapted in this analysis. By applying PhenoGraph clustering followed by viSNE projection, CSC heterogeneity and visualization of distinct subpopulations could be objectively explored without imposing prior assumptions about marker co-expression patterns.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePreprocessing and gating\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll samples were first pre-gated on forward and side scatter to remove debris and doublets. A three-step P1 gating strategy was explored: 1) a rectangular gate retaining the 5th-95th percentiles of FSC-A/SSC-A, followed by 2) a 97.5% confidence ellipse, and 3) a 97.5% Minimum Covariance Determinant (MCD) \u0026nbsp; elliptical gate was also explored. A final step-1-and-2 gating strategy was decided and applied. This procedure retained 85-88% of input events per sample and ensured consistency across runs and inclusion of potential rare CSC subpopulations. P1 gate was learned on the unstained files and then applied unchanged to the matched stained files. No further manual gating was applied. All P1-gated cells were included in the downstream analyses. BMI1 analysis was carried out independently due to technical incompatibility with live ALDHFLUOR\u003csup\u003eTM\u0026nbsp;\u003c/sup\u003estaining.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFluorescence intensity transformation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRaw fluorescence intensities were arcsinh-transformed (with the cofactor (c) determined per marker by scaling the 99th percentile of the stained sample to 5) to linearize low-intensity signals while compressing high-intensity outliers, preserving dynamic range.\u0026nbsp;\u003c/p\u003e\n\n\n\u003cp\u003e\u003cstrong\u003eThresholding determination for marker positivity\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMarker positivity was defined using matched controls: a) DEAB-treated negative control (ALDH), unstained control (CD44, CD133, BMI1). Thresholds were computed as mean + 2xSD of the control distributions. Events with arcsinh-transformed values above the threshold were classified as marker-positive.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eUnsupervised clustering\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePhenograph clustering was performed on arcsinh-transformed marker intensities, constructing a shared nearest-neighbour (SNN) graph with Jaccard similarity and resolving communities via Louvain optimization, without predefining the number of clusters. Median arcsinh expression of each marker was calculated per cluster and visualized in a heatmap with hierarchical ordering. \u0026nbsp; Cluster assignments were saved for downstream annotation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMarker classification and per-cell phenotype calls\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEach P1-gated, arcsinh-transformed cell was classified as positive or negative for CD44, CD133, ALDH, and BMI1 based on the thresholds above. Positivity calls were combined into per-cell phenotypes (e.g., CD44\u003csup\u003e+\u003c/sup\u003eCD133\u003csup\u003e-\u003c/sup\u003eBMI1\u003csup\u003e+\u003c/sup\u003e, \u0026nbsp;CD44\u003csup\u003e+\u003c/sup\u003eCD133\u003csup\u003e-\u003c/sup\u003eALDH\u003csup\u003e+\u003c/sup\u003e)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCluster annotation with Wilson score intervals\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eClusters were annotated using per-cell positivity calls. For each marker within the cluster, the Wilson 95% lower bound for the fraction of positive cells was computed. A cluster was labelled \u0026ldquo;positive\u0026rdquo; for a marker if this lower bound exceeded 0.5, ensuring conservative assignment. Each cluster was given a composite label (e.g., CD44\u003csup\u003e+\u003c/sup\u003eCD133\u003csup\u003e-\u003c/sup\u003eBMI1\u003csup\u003e+\u003c/sup\u003e, \u0026nbsp;CD44\u003csup\u003e+\u003c/sup\u003eCD133\u003csup\u003e-\u003c/sup\u003eALDH\u003csup\u003e+\u003c/sup\u003e).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eviSNE embedding and annotation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eviSNE was applied to the same transformed markers using Barnes-Hut t-SNE with perplexity set to 30 and seed fixed at 1979 to ensure reproducibility. Embedding coordinates were joined with cluster assignments and per-cell marker phenotypes. viSNE plots were generated in two formats: 1) colored by Phenograph cluster identity, and 2) colored by per-cell phenotype (single, double, or triple marker positivity) with percentages reported.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSphere formation assay\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo determine the sphere formation capability, 70% sub-confluent ORL cells were trypsinized, centrifuged, resuspended in culture medium, and counted with a hemocytometer. Cells were then seeded with a density of 1 x 10\u003csup\u003e3\u0026nbsp;\u003c/sup\u003ecells in a 6-well Ultra-Low Attachment plate (Corning, New York, USA) in 2 mL culture medium and kept in 5% CO\u003csub\u003e2\u0026nbsp;\u003c/sub\u003eat 37\u003csup\u003e\u0026ordm;\u003c/sup\u003eC for 10 days. 1 ml of culture medium was replenished every 3 days. The presence of spheres more than 20 \u0026micro;m visualized under a microscope at 40\u0026times; magnification was counted as positive.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eAs this work was designed as an exploratory study, each ORL cell line was analyzed once without technical or biological replicates. Reported values, therefore, represented single-run datasets with full disclosure of event counts and positive percentages in Supplementary Information. Replicated will be incorporated in future validation studies.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDetection of CSC markers across ORL cell lines\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAs no housekeeping gene was included in the study, results were presented as relative comparisons of Ct distributions across cell lines rather than normalized expression levels. Single-celled qPCR showed 100% detection of CD44, CD133, ALDH1A1, and BMI1 across ORL cell lines (Ct values \u0026lt; 35). CT values indicated that CD133 and ALDH1A1 transcripts were detected at much lower cycle thresholds compared to CD44 and BMI1 across all cell lines. The results suggest that CD133 and ALDH1A1 were strongly expressed relative to CD44 and BMI1. Violin plots revealed marker heterogeneity in expression levels (Figure 1).\u003c/p\u003e\n\u003cp\u003eViolin plots show distributions of \u0026nbsp;Ct values for CD133, ALDH1A1, CD44, and BMI1 measured by single-cell qPCR across ORL cell lines. Each violin depicts the density of per-cell measurements, with internal lines marking quartiles. The dashed red line at Ct = 35 indicates the conventional detection cutoff. CD133 and ALDH1A1 were consistently detected at low Ct values (indicative of strong expression), while CD44 and BMI1 exhibited higher Ct values (moderate to lower expression) and broader heterogeneity (Complete Ct datasets, Ct detection per-cell-line-per-gene summary, \u0026nbsp;and Ct mean \u0026plusmn; SD, median \u0026plusmn; IQR, and detection rate tables are provided in Supplementary Information).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eUnsupervised clustering identified heterogeneous subpopulations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePrincipal component analysis (PCA) of z-scored Ct values followed by k-means clustering revealed distinct subgroups of cells within individual cell lines (Figure 2). All ORL cancer cells were segregated into three transcriptionally distinct clusters, suggesting intra-line heterogeneity in CSC marker expression patterns (PCA/KMeans cluster assignment table is provided in Supplementary Information).\u003c/p\u003e\n\u003cp\u003ePrincipal component analysis (PCA) was performed on z-scored Ct values for CD133, ALDH1A1, CD44, and BMI1 within each cell line. Shown are the first two principal components (PC1 and PC2), which together capture the largest sources of variance in the dataset. Points represent individual cells, coloured by unsupervised\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eK-means cluster assignment (k = 3).\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003ePercent variance explained by each PC is indicated on the axes. The number of cells analyzed for each line is shown in the lower right corner (n = x). Distinct clustering patterns were observed, with ORL-48 cells separating into three subgroups (Clusters 0\u0026ndash;2), while ORL-115 cells were more concentrated with fewer dispersed subclusters.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSubpopulation composition by marker combinations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo assess distributions of multi-marker subsets, cells were classified by binary positivity thresholds (mean \u0026plusmn; 2SD). Cluster composition analysis showed a dominant quadruple-positive subsets with the majority of \u0026nbsp;CD133\u003csup\u003e+\u003c/sup\u003eALDH1A1\u003csup\u003e+\u003c/sup\u003eCD44\u003csup\u003e+\u003c/sup\u003eBMI1\u003csup\u003e+\u003c/sup\u003e cells, with smaller proportions of double-, and triple-positive combinations across ORL cell lines (Figure 3). (Per-cell-line thresholds and marker positivity tables are provided in Supplementary Information).\u003c/p\u003e\n\u003cp\u003eStacked bar plots show the proportion of unsupervised K-means clusters (k = 3) within each marker-positivity combination for single-cell qPCR data from ORL-48, ORL-115, ORL-174, and ORL-214. Marker combinations were defined using binary calls (positive/negative) for CD133, ALDH1A1, CD44, and BMI-1 based on a mean + 2\u0026times;SD threshold. Each bar represents one marker combination, with stacked colors corresponding to the percentage contribution of each cluster. Across all lines, the CD133⁺/ALDH1A1⁺/CD44⁺/BMI1⁺ combination predominated, while additional rare combinations appeared at lower frequencies, reflecting underlying heterogeneity in subpopulation structure.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eUpSet visualization of multi-marker overlaps\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo visualize the intersection across markers, UpSet plots were generated for each line. Quadruple-positive subpopulations were dominant (85.9% - 92.9%) while smaller subsets (\u0026lt; 5%) exhibited double- and triple-positivity across ORL cell lines\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e(Figure 4).\u003c/p\u003e\n\u003cp\u003eUpSet plots show the distribution of single-cell gene expression combinations for CD133, ALDH1A1, CD44, and BMI-1\u003cem\u003e\u0026nbsp;\u003c/em\u003ein ORL-115, ORL-174, ORL-214, and ORL-48. Each vertical bar represents the number of cells positive for the corresponding marker combination (filled circles), with counts and percentages annotated above the bars. Horizontal bars on the left indicate the overall frequency of positivity for each marker. In all cell lines, the quadruple-positive combination (CD133⁺/ALDH1A1⁺/CD44⁺/BMI1⁺) predominated, while additional rare subpopulations were detected at lower frequencies, highlighting heterogeneity in CSC marker co-expression.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eQuality control and single-cell event selection\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA sequential gating strategy was applied to unstained ORL cells. For example, in ORL-48, in Panel A for staining of CD44, CD133, and ALDH (Figure 5A), an initial\u0026nbsp;rectangular trim (5\u0026ndash;95th percentiles, black box)\u0026nbsp;retained 88.07% of events. This was followed by a\u0026nbsp;97.5% ellipse gate\u0026nbsp;to capture the main single-cell population, shown as a\u0026nbsp;classical covariance ellipse (blue)\u0026nbsp;retaining 85.81% of events and a\u0026nbsp;robust MCD ellipse (red dashed)\u0026nbsp;retaining 76.46% of events. The\u0026nbsp;classical method\u0026nbsp;was selected as the final P1 gate. This gating strategy was replicated for staining of CD44, CD133, and BMI1 in Panel B (Figure 5B) and across all ORL cell lines. This gating step excluded debris and aggregates, ensuring that only the primary single-cell population proceeded to downstream (for CD44_CD133_ALDH in Panel A and CD44_CD133_BMI1 in Panel B) analysis.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e(Absolute cell counts and percentages are provided in. Supplementary Information).\u003c/p\u003e\n\u003cp\u003eDot plot of \u0026nbsp;FSC-A and SSC-A for unstained ORL cells illustrating the P1 gate. Events outside the 5-95th percentile rectangle were excluded, followed by a 97.5% covariance ellipse. Both classical (blue) and robust MCD (red dashed) are shown, with the classical method chosen as final. The percentage of retained events at each step is indicated. This quality control (QC) gate removed debris and doublets before downstream analysis (Panel A: CD44_CD133_ALDH, Panel B: CD44_CD133_BMI1) \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFlow cytometry reveals heterogeneous expression of CSC markers across ORL cell lines\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo complement the single-cell qPCR profiling, CD44, CD133, ALDH, and BMI1 expressions were assessed by flow cytometry in all ORL cell lines. Overlay histograms demonstrated consistently high CD44 and BMI1 expressions (\u0026gt; 90% cell positive) across all cell lines. In contrast, CD133 was expressed in only a minority of cells. ALDH activity, determined using DEAB-matched gating, was similarly restricted to small fractions (Figure 6). These results validated the qPCR-based findings, in which dominant CD44\u003csup\u003e+\u003c/sup\u003e and BMI1\u003csup\u003e+\u003c/sup\u003e subsets co-existed with rarer multi-marker combinations. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eColored histograms of raw fluorescence intensity (log\u003csub\u003e10\u003c/sub\u003e scale) are shown for CD44, CD133, ALDH (Panel A), and CD44, CD133, BMI1 (Panel B) compared with the unstained control (gray). Each panel displays the density distribution of single-cell events, highlighting separation between negative and positive populations. CD44 (blue), CD133 (purple), ALDH-negative control - DEAB (red), and ALDH/BMI1 (green) signals were clearly shifted relative to the unstained baseline (grey), confirming robust marker detection. Shared axes across panels allow direct comparison of fluorescence intensity distributions. (Percentages of positive cells are displayed above each histogram).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePhenoGraph clustering identifies discrete CSC subpopulations across ORL cell lines\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo resolve heterogeneity beyond single-marker gating, arcsinh-transformed flow cytometry data for CD44, CD133, ALDH, and BMI1 were subjected to PhenoGraph clustering using the Louvain method, with median marker intensities per cluster visualized as heatmaps (Figure 7). Across all ORL cell lines, PhenoGraph identified 24-26 and 14-18 distinct clusters per cell line for CD44, CD133, and ALDH (Figure 7A), and CD44, CD133, and BMI1 (Figure 7B) expression, respectively. (Scaled medians, thresholds for PhenoGraph, cluster assignments, and summaries are provided in Supplementary Information).\u003c/p\u003e\n\u003cp\u003eHeatmaps showing median arcsinh-transformed expression values of CD44, CD133, ALDH (Panel A), and CD44, CD133, BMI1 (Panel B) across PhenoGraph-derived clusters (C1-Cn). Rows represent individual clusters and columns represent markers. Values within each cell denote the cluster median. Clusters are hierarchically ordered to highlight similarities in marker expression profiles. Warmer colors (yellow\u0026ndash;red) indicate higher expression, while cooler colors (blue) indicate lower expression. This visualization reveals heterogeneity in CSC marker expression.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eviSNE visualization of PhenoGraph clusters\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo complement the heatmap analysis, PhenoGraph-defined Louvain clusters were projected onto viSNE maps for each ORL cell line (Figure 8). Each cluster was color-coded and distributed across the two-dimensional embedding, enabling visualization of discrete subpopulations within global cell populations. While the number of clusters varied, all showed clear separation between dominant clusters and smaller, rarer subpopulations. Panel A - CD44, CD133, ALDH expression. Panel B \u0026ndash; CD44, CD133, BMI1 expression (Wilson\u0026rsquo;s phenotype embedding and clusters are provided in Supplementary Information).\u003c/p\u003e\n\u003cp\u003et-distributed stochastic neighbour embedding (t-SNE, viSNE) was applied to arcsinh-transformed marker intensities (Panel A \u0026ndash; CD44, CD133, ALDH, Panel B \u0026ndash; CD44, CD133, BMI1). Cells are displayed in two dimensions according to expression similarity and coloured by Louvain clusters (C1-Cn) identified by the PhenoGraph algorithm. Structural heterogeneity and how cells were grouped into discrete clusters were highlighted.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eviSNE overlay of per-cell phenotype\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe same viSNE embedding with per-cell marker positivity was annotated based on thresholding (mean \u0026plusmn; 2xSD from controls) to visualize the distribution of CSC phenotypes (Figure 9). Across cell lines, the CD44\u003csup\u003e+\u003c/sup\u003e and CD44\u003csup\u003e+\u003c/sup\u003eBMI1\u003csup\u003e+\u003c/sup\u003e subpopulation dominated, consistent with flow cytometry results. A smaller fraction of CD133\u003csup\u003e+\u003c/sup\u003e and ALDH\u003csup\u003e+\u003c/sup\u003e subsets was observed, often interspersed within larger CD44\u003csup\u003e+\u003c/sup\u003e and CD44\u003csup\u003e+\u003c/sup\u003eBMI1\u003csup\u003e+\u003c/sup\u003e clusters. Notably, rare triple-positive (CD44\u003csup\u003e+\u003c/sup\u003eBMI1\u003csup\u003e+\u003c/sup\u003eALDH\u003csup\u003e+\u003c/sup\u003e) subpopulations were detected in ORL-48 and ORL-214 (viSNE PhenoGraph embedding and overall phenotype counts are provided in Supplementary Information).\u003c/p\u003e\n\u003cp\u003eThe same viSNE embedding colored by binary phenotype assignments derived from control-based positivity thresholds (mean \u0026plusmn; 2SD of controls). Distinct phenotypes, including major CD44\u003csup\u003e+\u003c/sup\u003e (Panel A) and CD44\u003csup\u003e+\u003c/sup\u003eBMI1\u003csup\u003e+\u003c/sup\u003e (Panel B) subpopulations. Rarer CD44\u003csup\u003e+\u003c/sup\u003eCD133\u003csup\u003e+\u003c/sup\u003eALDH\u003csup\u003e+\u003c/sup\u003e subpopulations are annotated with their relative proportions in the legend. Panel A and Panel B show how marker-defined phenotypes distribute across the viSNE landscape and align with PhenoGraph clustering.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunctional validation by sphere formation assay\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSphere formation assay confirmed the functional CSC capacity of ORL cell lines. All four cell lines consistently generated tumour spheres in low-attachment conditions (Figure 10).\u003c/p\u003e\n\u003cp\u003eRepresentative images of tumour spheres derived from (A) ORL-48, (B) ORL-115, (C) ORL-174, and (D) ORL-214 cultures under non-adherent conditions. All cell lines generated compact spherical colonies, supporting the functional presence of self-renewing CSC-like subsets.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe application of unsupervised clustering in the current study aligns with emerging evidence that machine learning and unsupervised computational methods are increasingly valuable for analyzing complex single-cell datasets. Recent work has shown that these approaches enable improved detection of rare and heterogeneous cell subsets while reducing reliance on subjective manual gating\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. In particular, prior applications of PhenoGraph and viSNE to flow cytometry data demonstrated that unsupervised clustering can resolve phenotypic continua and visualize rare subsets that would otherwise remain obscured\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eThe current study is, to the best of the authors\u0026rsquo; knowledge, the first study that demonstrated the feasibility of integrating single-cell microfluidic qPCR and flow cytometry with an unsupervised analytical pipeline to interrogate CSC marker heterogeneity in OSCC-derived cell lines. Conventional methods such as bulk transcriptomics average signals across thousands of cells, obscuring the diversity of CSC phenotypes and masking rare but functionally important subpopulations\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/span\u003e,\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/span\u003e,\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. Microfluidic qPCR profiling in the current study captured transcriptional heterogeneity and revealed CSC markers\u0026rsquo; co-expression patterns through unsupervised clustering. Flow cytometry, analyzed through arcsinh transformation, PhenoGraph clustering, and viSNE visualization, independently validated these findings at the protein level, confirming the predominance of CD44\u003csup\u003e+\u003c/sup\u003e and BMI1\u003csup\u003e+\u003c/sup\u003e subsets, the presence of smaller ALDH\u003csup\u003e+\u003c/sup\u003e fractions, and rare multi-marker CSC-enriched clusters.\u003c/p\u003e\u003cp\u003ePrince et al. (2007) first demonstrated that CD44\u0026thinsp;+\u0026thinsp;cells isolated from primary head-and-neck squamous cell carcinoma (HNSCC) tumours or cell lines could initiate tumours in mice and showed elevated BMI1 expression in the CD44\u0026thinsp;+\u0026thinsp;population, with frequencies ranging from ~\u0026thinsp;0.4 to 40%\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e17\u003c/span\u003e\u003c/sup\u003e. In contrast, our study found 100% of cells in all ORL cell lines positive for CD44 under the applied threshold, a finding that suggests a saturation of total CD44 expression in these cell lines. This observation raises the possibility that CD44 variant isoforms (CD44v) or standard isoform (CD44s) may be detected, both of which are splice variants of the same gene but differ only in exon composition\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/span\u003e,\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. In some models, CD44v8-10 has been shown to enhance migration and sphere formation similarly to CD44s, though its impact on metastasis may differ\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. Understanding which spice variants are expressed in ORL cell lines (CD44s vs CD44v) will therefore be critical, particularly because therapeutic strategies targeting CD44 often do not distinguish between isoforms and may require isoform-specific targeting\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. ALDH is a family of 19 intracellular enzymes involved in oxidizing aldehydes to carboxylic acids; among these, ALDH1 has been widely reported as a head and neck CSC marker, Clay et al. (2010) demonstrated that ALDH-high cells (representing\u0026thinsp;~\u0026thinsp;1.0 to 7.8% of the tumour population) are highly tumorigenic: as few as 500 ALDH-high cells could form tumours in mice, whereas ALDH-low required much higher number. In Clay\u0026rsquo;s work, most ALDH-high cells co-expressed CD44 (50.6\u0026ndash;74.4%), but only a small subset of CD44\u003csup\u003e+\u003c/sup\u003e cells were ALDH+ (9.8\u0026ndash;23.6%)\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e21\u003c/span\u003e\u003c/sup\u003e. The frequency of ALDH\u003csup\u003e+\u003c/sup\u003e cells in the current study ranged from 1.47\u0026ndash;4.62% across ORL cell lines, and the percentage of CD44\u003csup\u003e+\u003c/sup\u003e/ALDH\u003csup\u003e+\u003c/sup\u003e cells ranged from 0.03\u0026ndash;20.71%. These data are broadly consistent with Clay et al. (), reinforcing that ALDH marks a rarer, more discriminatory CSC subset compared with CD44. Kulsum et al. (2017) further support this: inhibiting ALDH1A1 in HNSCC models reduced sphere formation, downregulated stemness and drug resistance genes, and increased chemosensitivity to cisplatin\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. Findings from the current study suggest that ALDH activity identifies a minor but potentially functionally important CSC population in ORL cell lines, lending weight to targeting ALDH for therapy\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eCD133 is a glycosylated membrane protein with 5 transmembrane domains and 2 large extracellular loops. It is a marker for hematopoietic stem cells and also a CSC marker for brain\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. Chiou et al were the first to identify CD133 as a CSC marker for oral cancer. The method they used was different from others. They first grew oral cancer cells as spheres and showed that these spheres were more tumorigenic than parental cells. Next, they also found that expression of CD133 in 60% spheres generated from 2 OSCC cell lines. The frequency of CD133 positive cells from 5 primary tumours ranged from 0.8\u0026ndash;4.2% in the parental cell population and was more enriched, ranging from 28.2\u0026ndash;40.2% in spheres in their\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. In the current study, the frequency of CD133-positive cells ranged from 3.93\u0026ndash;21.59%, consistent with the reported frequency. Moreover, Yu et al demonstrated that silencing of CD133 reduced tumorigenicity and heightened drug sensitivity of OSCC cells\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eBMI1 (polycomb group protein Moloney murine leukaemia virus insertion site 1) is a key regulator of stem cell self-renewal whose aberrant expression contributes to cancer initiation, metastasis, and treatment resistance. Chen et al. () used a 4-nitroquinoline-1-oxide (4-NQO) mouse model of HNSCC with lineage tracing (BMI1-CreER;Rosa-tdTomato) to identify BMI1\u0026thinsp;+\u0026thinsp;cells as slow-cycling, tumour-initiating stem cells. These cells were also found to drive metastasis to the cervical lymph node, and combining cisplatin with PTC-209 \u0026ndash; a small-molecule BMI1 inhibitor \u0026ndash; significantly reduced tumour progression and the BMI1\u003csup\u003e+\u003c/sup\u003e CSC pool in vivo\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. In vitro, PTC-209 treatment of HNSCC cell lines (Cal27, FaDu) suppressed proliferation, induced apoptosis, reduced migratory and invasive capacity, and acted synergistically with cisplatin or 5-FU\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e28\u003c/span\u003e\u003c/sup\u003e. In the current study, nearly 100% of BMI1\u003csup\u003e+\u003c/sup\u003e cells in ORL cell lines were under the applied thresholding criteria, raising the question of whether all BMI1 positivity reflects functional CSC potential or rather a general stemness-associated regulatory state. Given the promising preclinical activity of PTC-209 in HNSCC, its capacity to impair tumorigenicity, enhance chemosensitivity, and reduce CSC subpopulations makes BMI1 an especially attractive target in ORL models\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eIn addition, an investigation of phenotypic diversity in \u003cem\u003ein silico\u003c/em\u003e simulation reported that 28 phenotypes had evolved in the smallest virtual tumour after 730 days. Moreover, the group simulated an \u003cem\u003ein silico\u003c/em\u003e biopsy and found that the average frequency of CSC ranged from 0.7 to 10% and a single collected biopsy sample can be divided into 10 subpopulations with approximately 10,000 cells. These findings are corroborated by the discovery of 10\u0026ndash;28 distinct clusters from PhenoGraph and viSNE clustering of the current study. The frequency of CD133\u0026thinsp;+\u0026thinsp;and ALDH\u0026thinsp;+\u0026thinsp;cells from the current study also fell within the simulated range\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eSphere (or organoid) formation assays were originally developed in neural stem cell biology when Johansson et al. (1999) isolated adult neural stem cells from human brain ependyma and demonstrated that dissociated cells cultured in suspension without substrate formed free-floating clusters (\u0026ldquo;neurospheres\u0026rdquo;) which could later differentiate into neurons and glial cells upon attachment and appropriate stimulation\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. In OSCC, Chen et al. (2012) applied sphere formation under non-adherent conditions to enrich for CSCs; tumour spheres derived this way displayed elevated expression of CSC markers (CD133, ALDH1), increased tumorigenic capacity in vivo, and resistance to chemotherapeutic agents\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. More recently, Pozzi and colleagues showed that sphere assays can reliably enrich CSCs from HNSCC by culturing CD44\u003csup\u003e+\u003c/sup\u003e/ALDH\u003csup\u003e+\u003c/sup\u003e sorted cells or by using marker-unsorted cell populations in ultralow attachment or soft agar settings, preserving CSC traits over multiple passages\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. 3D models in OSCC confirm that tumorospheres (free-floating CSC sphere models) are widely employed for assessing stemness, drug sensitivity, and phenotypic heterogeneity\u003csup\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/span\u003e,\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/span\u003e,\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/span\u003e,\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/span\u003e\u003c/sup\u003e. Sphere formation was robust across all ORL cell lines in the current study, suggesting that sphere/organoid assays constitute a valuable functional complement to molecular and cytometric profiling of CSC markers.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eIn conclusion, the current study reveals pronounced CSC heterogeneity discovered through an autonomous computational method without prior assumption or subjective interference. While CD44 and BMI1 are nearly ubiquitously expressed, which is in line with the stemness regulatory program, ALDH and CD133 delineate rarer, more discriminative CSC subpopulations. The concordance between transcript-level Ct thresholding and protein-level clustering validates that rare multi-marker co-expressing subsets, though low in frequency, are consistently present across cell lines. Importantly, the functional sphere formation across ORL cell lines supports that these molecular markers reflect biologically relevant self-renewing capacity. Though exploratory without replicates, computational analysis such as this offers reproducibility and a basis for CSC targeting in future studies.\u003c/p\u003e"},{"header":"Limitations","content":"\u003cp\u003eThis work was designed as an exploratory, proof-of-concept study. Each ORL cell line was analyzed once without technical or biological replicates, precluding statistical comparisons across runs. While this limits the generalizability of the frequency estimates, all event counts, gating percentages, and positive fractions are fully disclosed. Despite these constraints, this study demonstrates methodological feasibility and establishes a transparent baseline for reproducible CSC profiling.\u003c/p\u003e\n\u003ch3\u003eFuture directions and translational significance\u003c/h3\u003e\n\u003cp\u003eFuture studies will incorporate biological replicates, patient-derived samples, and expanded marker panels to validate and refine current findings. Integration of pathway-specific inhibitors may help dissect the functional role of subpopulations in OSCC therapy resistance. Advanced models such as organoids and lineage tracing will further strengthen causal interference. Importantly, by linking marker-defined CSC heterogeneity to canonical signalling pathways the future studies will identify pathway-specific vulnerabilities, paving the way for rational CSC-targeted therapies and improved clinical outcomes in OSCC.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was funded by the Faculty of Dentistry, University of Malaya, University of Malaya Research Grant (UMRG, RU012-2013) and Universiti Malaya Bantuan Kecil Penyelikan (BKP Grant - UM.TNC2/IPPP/UPGP/628).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe also extend our gratitude to Cancer Research Malaysia (CRM), a non-profit cancer research organization, for their generous help in providing ORL cell lines and technical assistance.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;S.K.A. and Y.F.C. conceived the idea for this project. G.R.W. and Y.F.C. performed and acquired data from single-cell qPCR and flow cytometry experiments. Y.W.C., M.K.A., and I.F. performed qPCR computational analyses. \u0026nbsp;Y.W.C., N.P.R., and P.G.S.D. performed flow cytometry computational analyses. S.K.A. M.R. and Y.F.C. performed a literature search and wrote the manuscript. S.K.A., M.R., and Y.F.C. worked on formatting the manuscript for publishing and made the final revisions. All authors have read and agreed to the published version of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe author(s) declare no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSupplementary Information and\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eData Availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data that support the findings of this study are available on request from the corresponding author, HS. The data are not publicly available due to the restrictions on the use of vital statistics data.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eGhanaati, S.; Udeabor, S. E.; Winter, A.; Sader, R.; Heselich, A. Cancer recurrence in operated primary oral squamous cell carcinoma patients seems to be independent of the currently available postoperative therapeutic approach: a retrospective clinical study. Curr. Oncol. 32, 208. https://doi.org/10.3390/curroncol32040208 (2025). Article PubMed Google Scholar\u003c/li\u003e\n \u003cli\u003eLee, H.; Kim, B.; Park, J.; Park, S.; Yoo, G.; Yum, S.; Kang, W.; Lee, J.-M.; Youn, H.; Youn, B. Cancer stem cells: landscape, challenges and emerging therapeutic innovations. Signal Transduct. Target. Ther. 10, 248. https://doi.org/10.1038/s41392-025-02360-2 (2025). Article PubMed Google Scholar\u003c/li\u003e\n \u003cli\u003eEl-Tanani, M.; Rabbani, S. A.; Satyam, S. M.; Rangraze, I. R.; Wali, A. F.; El-Tanani, Y.; Aljabali, A. A. A. Deciphering the role of cancer stem cells: drivers of tumor evolution, therapeutic resistance, and precision medicine strategies. Cancers 17, 382. https://doi.org/10.3390/cancers17030382 (2025). Article PubMed Google Scholar\u003c/li\u003e\n \u003cli\u003eAcharya, S. K.; Shai, S.; Choon, Y. F.; Lim, C. T.; Tan, W. L.; Yeap, S. K.; Singh, R. Cancer stem cells in oral squamous cell carcinoma: a narrative review on experimental characteristics and methodological challenges. Biomedicines 12, 2111. https://doi.org/10.3390/biomedicines12092111 (2024). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eLan, L.; Behrens, A. Are there specific cancer stem cell markers? Cancer Res. 83, 170\u0026ndash;172. https://doi.org/10.1158/0008-5472.CAN-22-2053 (2023). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eHamid, S.; Lim, K. P.; Zain, R. B.; Ismail, S. M.; Lau, S. H.; Mustafa, W. M. Establishment and characterization of Asian oral cancer cell lines as in vitro models to study a disease prevalent in Asia. \u003cem\u003eInt. J. Mol. Med.\u003c/em\u003e 6, 453\u0026ndash;460. \u003cstrong\u003ehttps://doi.org/\u003c/strong\u003e\u003cstrong\u003e...\u003c/strong\u003e (2007). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eFadlullah, M. H.; Chiang, I. K. N.; Dionne, K. R.; Pang, J. C.; Lo, P. H. Y.; Ng, W. K.; Tsang, C. M.; Tsao, S. W.; Wong, K. K.; Lung, M. L. Genetically defined novel oral squamous cell carcinoma cell lines for the development of molecular therapies. \u003cem\u003eOncotarget\u003c/em\u003e 7, 27802\u0026ndash;27818. \u003cstrong\u003ehttps://doi.org/\u003c/strong\u003e\u003cstrong\u003e...\u003c/strong\u003e (2016). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eDiggins, K. E.; Ferrell, P. B.; Irish, J. M. Methods for discovery and characterization of cell subsets in high-dimensional mass cytometry data. \u003cem\u003eMethods\u003c/em\u003e\u003cstrong\u003e82\u003c/strong\u003e, 55\u0026ndash;63. https://doi.org/10.1016/j.ymeth.2015.05.008 (2015). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eDiGiuseppe, J. A.; Cardinali, J. L.; Rezuke, W. N.; Pe\u0026rsquo;er, D. PhenoGraph and viSNE facilitate the identification of abnormal T-cell populations in routine clinical flow cytometric data. \u003cem\u003eCytometry B Clin. Cytom.\u003c/em\u003e 94, 588\u0026ndash;601. \u003cstrong\u003ehttps://doi.org/\u003c/strong\u003e\u003cstrong\u003e...\u003c/strong\u003e (2018). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eSpidlen, J.; Leif, R. C.; Moore, W.; Roederer, M.; Brinkman, R. R. Gating-ML: XML-based gating descriptions in flow cytometry. \u003cem\u003eCytometry A\u003c/em\u003e 73, 1151\u0026ndash;1157. https://doi.org/10.1002/cyto.a.20637 (2008). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eChiou, S. H.; Yu, C. C.; Huang, C. Y.; et al. Positive correlations of Oct-4 and Nanog in oral cancer stem-like cells and high-grade oral squamous cell carcinoma. \u003cem\u003eClin. Cancer Res.\u003c/em\u003e 14, 4085\u0026ndash;4095. https://doi.org/10.1158/1078-0432.CCR-07-4404 (2008). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eDeLaughter, D. M. The use of the Fluidigm C1 for RNA expression analyses of single cells. \u003cem\u003eCurr. Protoc. Mol. Biol.\u003c/em\u003e 122, e55. https://doi.org/10.1002/cpmb.55 (2018). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eLex, A.; Gehlenborg, N.; Strobelt, H.; Vuillemot, R.; Pfister, H. UpSet: visualization of intersecting sets. \u003cem\u003eIEEE Trans. Vis. Comput. Graph.\u003c/em\u003e 20, 1983\u0026ndash;1992. https://doi.org/10.1109/TVCG.2014.2346248 (2014). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003ePatel, A. P.; Tirosh, I.; Trombetta, J. J.; et al. Single-cell RNA-seq highlights intratumoral heterogeneity in primary glioblastoma. \u003cem\u003eScience\u003c/em\u003e 344, 1396\u0026ndash;1401. https://doi.org/10.1126/science.1254257 (2014). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eLieberman, B.; Kusi, M.; Hung, C. N.; et al. Toward uncharted territory of cellular heterogeneity: advances and applications of single-cell RNA-seq. \u003cem\u003eJ. Transl. Genet. Genom.\u003c/em\u003e 5, 1\u0026ndash;21. https://doi.org/10.20517/jtgg.2020.51(2021). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eNavin, N.; Kendall, J.; Troge, J.; et al. Tumour evolution inferred by single-cell sequencing. \u003cem\u003eNature\u003c/em\u003e 472, 90\u0026ndash;94. https://doi.org/10.1038/nature09807 (2011). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003ePrince, M. E.; Sivanandan, R.; Kaczorowski, A.; Wolf, G. T.; Kaplan, M. J. Identification of a subpopulation of cells with cancer stem cell properties in head and neck squamous cell carcinoma. \u003cem\u003eProc. Natl. Acad. Sci. USA\u003c/em\u003e 104, 973\u0026ndash;978 (2007). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eMesrati, M. H.; Syafruddin, S. E.; Mohtar, M. A.; Syahir, A. CD44: a multifunctional mediator of cancer progression. \u003cem\u003eBiomolecules\u003c/em\u003e 11, 1850. https://doi.org/10.3390/biom11121850 (2021). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eZhang, H.; Brown, R. L.; Wei, Y.; et al. CD44 splice isoform switching determines breast cancer stem cell state. \u003cem\u003eGenes Dev.\u003c/em\u003e 33, 166\u0026ndash;179. https://doi.org/10.1101/gad.319889.118 (2019). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eHiraga, T.; Nakamura, H. Comparable roles of CD44v8-10 and CD44s in the development of bone metastases in a mouse model. \u003cem\u003eOncol. Lett.\u003c/em\u003e 12, 2962\u0026ndash;2969. https://doi.org/10.3892/ol.2016.4985 (2016). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eClay, M. R.; Tabor, M.; Owen, J. H.; et al. Single-marker identification of head and neck squamous cell carcinoma cancer stem cells with aldehyde dehydrogenase. \u003cem\u003eHead Neck\u003c/em\u003e 32, 1195\u0026ndash;1201. https://doi.org/10.1002/hed.21315(2010). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eKulsum, S.; Sudheendra, H. V.; Pandian, R.; et al. Cancer stem cell mediated acquired chemoresistance in head and neck cancer can be abrogated by aldehyde dehydrogenase 1A1 inhibition. \u003cem\u003eMol. Carcinog.\u003c/em\u003e 56, 694\u0026ndash;711. https://doi.org/10.1002/mc.22526 (2017). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eDe Castro, L. R.; de Oliveira, L. D.; Milan, T. M.; et al. Up-regulation of TNF-\u0026alpha;/NFkB/SIRT1 axis drives aggressiveness and cancer stem cell accumulation in chemoresistant oral squamous cell carcinoma. \u003cem\u003eJ. Cell Physiol.\u003c/em\u003e239, e31164. https://doi.org/10.1002/jcp.31164 (2024). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eSingh, S. K.; Hawkins, C.; Clarke, I. D.; et al. Identification of human brain tumour initiating cells. \u003cem\u003eNature\u003c/em\u003e 432, 396\u0026ndash;401 (2004). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eChiou, S. H.; Yu, C. C.; Huang, C. Y.; et al. Positive correlations of Oct-4 and Nanog in oral cancer stem-like cells and high-grade oral squamous cell carcinoma. \u003cem\u003eClin. Cancer Res.\u003c/em\u003e 14, 4085\u0026ndash;4095. https://doi.org/10.1158/1078-0432.CCR-07-4404 (2008). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eYu, C. C.; Hu, F. W.; Yu, C. H.; Chou, M. Y. Targeting CD133 in the enhancement of chemosensitivity in oral squamous cell carcinoma-derived side population cancer stem cells. \u003cem\u003eHead Neck\u003c/em\u003e 38, E231\u0026ndash;E238. https://doi.org/10.1002/hed.23975 (2016). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eChen, D.; Wu, M.; Li, Y.; et al. Targeting BMI1⁺ cancer stem cells overcomes chemoresistance and inhibits metastases in squamous cell carcinoma. \u003cem\u003eCell Stem Cell\u003c/em\u003e 20, 621\u0026ndash;634.e6. https://doi.org/10.1016/j.stem.2017.02.003 (2017). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eWang, Q.; Li, Z.; Wu, Y.; et al. Pharmacological inhibition of Bmi1 by PTC-209 impairs tumor growth in head and neck squamous cell carcinoma. \u003cem\u003eCancer Cell Int.\u003c/em\u003e 17, 107. https://doi.org/10.1186/s12935-017-0481-z (2017). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eYang, D.; Liu, H. Q.; Yang, Z.; Fan, D.; Tang, Q. Z. BMI1 in the heart: novel functions beyond tumorigenesis. \u003cem\u003eEBioMedicine\u003c/em\u003e 63, 103193. https://doi.org/10.1016/j.ebiom.2020.103193 (2021). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003ePoleszczuk, J.; Hahnfeldt, P.; Enderling, H. Evolution and phenotypic selection of cancer stem cells. \u003cem\u003ePLoS Comput. Biol.\u003c/em\u003e 11, e1004025. https://doi.org/10.1371/journal.pcbi.1004025 (2015). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eJohansson, C. B.; Momma, S.; Clarke, D. L.; et al. Identification of a neural stem cell in the adult mammalian central nervous system. \u003cem\u003eCell\u003c/em\u003e 96, 25\u0026ndash;34. https://doi.org/10.1016/S0092-8674(00)80956-3 (1999). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eChen, S. F.; Chang, Y. C.; Nieh, S.; et al. Nonadhesive culture system as a model of rapid sphere formation with cancer stem cell properties. \u003cem\u003ePLoS One\u003c/em\u003e 7, e31864. https://doi.org/10.1371/journal.pone.0031864 (2012). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003ePozzi, V.; Sartini, D.; Rocchetti, R.; et al. Identification and characterization of cancer stem cells from head and neck squamous cell carcinoma cell lines. \u003cem\u003eCell Physiol. Biochem.\u003c/em\u003e 36, 784\u0026ndash;798. https://doi.org/10.1159/000430138 (2015). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eMillen, R.; de Kort, W. W. B.; Koomen, M.; et al. Patient-derived head and neck cancer organoids allow treatment stratification and serve as a tool for biomarker validation and identification. \u003cem\u003eMed.\u003c/em\u003e 4, 290\u0026ndash;310.e12. https://doi.org/10.1016/j.medj.2023.04.003 (2023). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eTanaka, N.; Osman, A. A.; Takahashi, Y.; et al. Head and neck cancer organoids established by modification of the CTOS method can be used to predict in vivo drug sensitivity. \u003cem\u003eOral Oncol.\u003c/em\u003e 87, 49 57. https://doi.org/10.1016/j.oraloncology.2018.10.018 (2018). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eSiquara da Rocha, L. d. O.; Souza, B. S. d. F.; Coletta, R. D.; et al. Mapping cell-in-cell structures in oral squamous cell carcinoma. \u003cem\u003eCells\u003c/em\u003e 12, 2418. https://doi.org/10.3390/cells12192418 (2023). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n \u003cli\u003eGoričan, L.; Gole, B.; Potočnik, U. Head and neck cancer stem cell-enriched spheroid model for anticancer compound screening. \u003cem\u003eCells\u003c/em\u003e 9, 1707. https://doi.org/10.3390/cells9071707 (2020). Article\u0026emsp;PubMed\u0026emsp;Google Scholar\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Oral Cancer Stem Cells, PhenoGraph, Unsupervised clustering","lastPublishedDoi":"10.21203/rs.3.rs-7844179/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7844179/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eCancer stem cells (CSCs) are key drivers of heterogeneity, recurrence, and therapy resistance in oral squamous cell carcinoma (OSCC). CD44, CD133, ALDH, and BMI1 have been highlighted as clinically relevant CSC markers, yet no single marker sufficiently defines CSC identity. This study aims to explore computationally CSC heterogeneity in Asian-derived ORL cell lines (ORL-48, ORL-115, ORL-174, ORL-214). Unsupervised clustering of single-cell qPCR datasets revealed enrichment of quadruple-positive (CD44⁺/CD133⁺/ALDH1A1⁺/BMI1⁺) subsets, alongside smaller heterogeneous fractions. PhenoGraph clustering and viSNE analyses of arcsinh-transformed flow cytometry datasets independently confirmed dominant CD44⁺ and BMI1⁺ fractions, smaller CD133⁺ and ALDH⁺ subsets, and rare CSC-enriched multi-marker clusters consistent with the single-cell qPCR analysis. In addition, sphere formation assays further validated the presence of self-renewing CSC-like populations across all lines. As this work was designed as an exploratory study, analyses were performed once per cell line without replicates; future studies will incorporate independent validations.\u003c/p\u003e","manuscriptTitle":"Unravelling the heterogeneity in oral cancer stem cells – A single-cell computational exploratory study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-07 14:06:52","doi":"10.21203/rs.3.rs-7844179/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"45c9c0a1-e78e-4819-86da-2f594f5c1b1f","owner":[],"postedDate":"November 7th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":57544801,"name":"Biological sciences/Cancer"},{"id":57544802,"name":"Biological sciences/Cell biology"},{"id":57544803,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":57544804,"name":"Biological sciences/Stem cells"}],"tags":[],"updatedAt":"2025-12-16T07:54:47+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-07 14:06:52","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7844179","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7844179","identity":"rs-7844179","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}