C57Bl/6 mice were handled according to Swiss Animal Welfare regulations. The study was approved by the Cantonal Veterinary Office in Zurich (licenses ZH197/2017 and ZH086/2021). To induce OSCC, mice were given drinking water containing 50 μg/mL 4-nitroquinoline 1-oxide (4NQO; #N8141, Sigma-Aldrich) from 6 to 22 weeks of age. Control mice received water with the same amount of Dimethyl sulfoxide (DMSO) (#D4540, Sigma-Aldrich). After 16 weeks of treatment, mice continued on normal water until euthanasia (32 weeks).

After tissue collection, samples were fixed in 4% paraformaldehyde (PFA) for 1 h at 4 °C, then cryoprotected in 30% sucrose. The tissues were embedded in cryo-embedding medium (Tissue-Tek O.C.T. Compound, Sakura) and stored at −80 °C until further processing. Sections of 10 µm thickness were cut using a cryostat (Leica CM3050S) and kept at −80 °C until staining.

For immunofluorescence analysis, cryosections were first permeabilized with 0.5% Triton X-100 in Phosphate-buffered saline (PBS) for 1 h at room temperature, followed by blocking in a solution containing 10% fetal bovine serum (FBS) and 0.1% Triton X-100. Sections were then incubated overnight at 4 °C with the primary antibody anti-FZD4 (#SAB2108140, Sigma-Aldrich; 1:200) diluted in blocking buffer. After washing with PBS, sections were incubated with the secondary antibody donkey anti-rabbit IgG Alexa Fluor 488 (1:1000; Invitrogen). Nuclei were counterstained with DAPI (4′,6-diamidino-2-phenylindole dihydrochloride; #D1306, Invitrogen). Finally, slides were mounted using Mowiol 4–88 (#81381 and #G-6279, Sigma-Aldrich). Fluorescent images were acquired using a Leica DM6000B fluorescence microscope.

The SCC-25 epithelial cell line, originating from the tongue of a 70-year-old OSCC patient, was obtained from ATCC/LGC (American Type Culture Collection, SCC-25CRL-1628™). SCC25 were maintained in Dulbecco’s Modified Eagle Medium: Nutrient Mixture F-12 (DMEM/F12) medium supplemented with 1.2 g/L sodium bicarbonate, 2.5 mM L-glutamine, 15 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 0.5 mM sodium pyruvate (#11330032, Gibco-ThermoFisher), and 10% FBS (#16000044, Gibco-ThermoFisher). Once cultures reached ∼80% confluence, cells were passaged using trypsin/Ethylenediaminetetraacetic acid (EDTA) (#15400054, Gibco-ThermoFisher) and cryopreserved at a density of 10⁶ cells/vial in 10% DMSO (#D4540, Sigma-Aldrich). For experimental treatments, low passage (<20 passages) SCC-25 cells were exposed to 3-Hydroxy-5-[[(2-Naphthalenylamino)carbonyl]amino]benzoic acid (FzM1) (10µM; # HY-116553, MedChemExp), FzM1+ Retinoic acid receptor inhibitor (RARi) (AGN 193109 100nM; #HY-U00449, MedChemExp) or vehicle control DMSO (#D8418, Sigma-Aldrich) for 5 days.

SCC-25 cells were seeded at a density of 20,000 cells per well in a 12-well plate with coverslips (Electron Microscopy Sciences #72196). Fixation was carried out with 4% PFA or cold 100% MetOH for 15 min followed by PBS washes. Cells were then permeabilized for 30 min with 0.5% Triton and subsequently blocked in PBS containing 10% FBS and 0.1% Triton for 1h RT. Primary antibodies were diluted in blocking solution and incubated overnight. For immunofluorescence, cells were probed with anti-FZD4 (#SAB2108140; Sigma-Aldrich, 1:100), anti-phospho Histone3 (pH3) (#9701, Cell Signaling, 1:300); anti-Ki67 (#ab16667, abcam, 1:100) and detected with donkey anti-rabbit IgG Alexa Fluor 488 or 546 (1:1000; Invitrogen). Nuclear staining was performed using DAPI (4′,6-diamidino-2-phenylindole, dihydrochloride; #D1306, Invitrogen). Samples were mounted using Mowiol 4–88 and glycerol (#81381 and #G-6279, Sigma-Aldrich). Immunofluorescence images were produced using a Leica DM6000B microscope. Quantifications were performed in five different fields of view for each sample (n = 3).

To calculate the growth fraction upon treatment, SCC25 cells were plated on coverslips and treated with DMSO (control), FzM1 (treated) or FzM1+RARi (treated) for 5 days before kinetic analyses. Cells were exposed to 10 μM 5-Ethynyl-2′-deoxyuridine (EdU, MedChemExp, # HY-118411) for 1h, 4h, 8h, 10h, 12h, 14h, 16h, 20h, 24h, 38h, and 60 h. After incubation time, cells were fixed in 4% PFA and stained for EdU using the VF 488 Click-iT EdU Universal Cell Proliferation Detection Kit (MedChemExp). Nuclear staining was performed using DAPI (4′,6-diamidino-2-phenylindole, dihydrochloride; #D1306, Invitrogen). Samples were mounted using Mowiol 4–88 and glycerol (#81381 and #G-6279, Sigma-Aldrich). Immunofluorescence images were produced using a Leica DM6000B microscope. Quantifications were performed in five different fields of view for each sample (n = 3). Calculation of cell cycle length (Tc) and S-phase length (Ts) were extrapolated from the function obtained graphically of the growing phase (Nowakowski et al., 1989). Quantification was carried out using ImageJ and FIJI software (Schindelin et al., 2012; Schneider et al., 2012). Graphs and statistical analyses were performed using Prism software (GraphPad, GraphPad Software, San Diego California USA; http://www.graphpad.com/). All data were analyzed using the two-tailed Student's t-test. P < 0.05 were considered statistically significant.

Treated SCC-25 cells were harvested in Radioimmunoprecipitation assay (RIPA) buffer (#9806; Cell Signaling) supplemented with protease inhibitors (#11836170001; Roche) and homogenized following scraping. Proteins were precipitated by adding cold acetone to the lysates and incubating for 1 h at −20 °C. The samples were then centrifuged at 15,000 rpm for 10 min, and the resulting protein pellets were resuspended in a buffer mixture containing RIPA buffer, Laemmli Buffer (LB, #1610747, Bio-Rad), and dithiothreitol (DTT, #7016, Cell Signaling). For Sodium Dodecyl Sulfate-PolyAcrylamide Gel Electrophoresis (SDS-PAGE), 20 ug of protein lysate was loaded per lane and separated on 6%–10% Mini-PROTEAN TGX Gels using the Mini-PROTEAN system (#1658005, Bio-Rad) under 70 V constant voltage. Protein transfer was carried out with the Trans-Blot Turbo system (#1704150, Bio-Rad) using Trans-Blot Turbo Mini 0.2 µm Nitrocellulose Transfer Packs (#1704158, Bio-Rad), following the standard Turbo protocol for 30 min. Blocking of the membrane was performed in 5% non-fat dry milk in Tris-Buffered Saline (TBS) or 1% Bovine Serum Albumin (BSA) in TBS for 1h at room temperature. Primary antibody used were incubated overnight at 4 °C in a shaking condition. Antibodies used were: a- glyceraldehyde-3-phosphate dehydrogenase (GAPDH)- Horseradish Peroxidase (HRP) conjugated (Clone YA874, MEC, #HY-P80951A; 1:1000); a-Keratin10 (Krt10) (Covance, 1:1000). Membrane was washed in TBS-Tween® 20 (TBS-T, Tween20 #P9416; Sigma-Aldrich) and TBS for 1h at RT in shaking conditions. Species-specific secondary antibodies a-rb HRP conjugated (Abcam, #ab97051, 1:1000) were used.

Treated SCC-25 cells were pelleted and snap-frozen in liquid nitrogen prior to RNA isolation following the manufacturer’s protocol (RNeasy Mini Kit; Qiagen). Genomic DNA was removed via gDNA removal columns and further purified. Total RNA yield was quantified using a NanoDrop spectrophotometer (Thermo Fisher Scientific, Waltham, USA). For cDNA synthesis, 500 ng of RNA was reverse transcribed with the iScript™ cDNA Synthesis Kit (Bio-Rad Laboratories AG, Cressier FR, Switzerland) according to the supplier’s instructions.

Quantitative real-time PCR was conducted on the CFX Connect Real-Time System using CFX Manager software (Version 3.1). Gene expression analyses were carried out with SYBR® Green PCR Master Mix (Applied Biosystems, Carlsbad, CA, USA) together with the corresponding oligonucleotide primers (Supplementary Table S1). Exon-exon primers were designed using Primer3 software (Kõressaar et al., 2018). Hairpin, hetero-dimers and self-dimers formation was excluded (tollerance DeltaG >=-9 kcal/mol). Each reaction was run in technical duplicates across four biological replicates. Thermocycling conditions were as follows: initial denaturation at 95 °C for 10 min; 40 cycles of 95 °C for 15 s and 60 °C for 60 s. Melting curve analysis was performed under the following parameters: 95 °C for 15 s, 60 °C for 5 s, and 95 °C for 30 s. Relative expression levels were determined using the ΔΔCt method (2^−ΔΔCT) after normalization to the Ct values of the housekeeping gene. Graphs and statistical analyses were performed using Prism10 software (GraphPad, GraphPad Software, San Diego California USA; http://www.graphpad.com/). All data were considered statistically significant when p < 0.05.

Bulk RNA sequencing was performed as follows: Total RNA was isolated using the Qiagen RNeasy Mini Kit (Qiagen, #74104) with on-column genomic DNA elimination using RNase-Free DNase I (Qiagen, #79256), according to the manufacturer’s instructions. RNA quality was assessed prior to library preparation, and only samples with RNA integrity (RINe) scores >9.2 were used for downstream processing. High-quality RNA was reverse-transcribed into cDNA, followed by fragmentation and adapter ligation. Sequencing libraries were prepared using the Illumina TruSeq protocol and sequenced on an Illumina NovaSeq 6000 platform (200 million reads per sample). Raw sequencing data were quality-checked using FastQC and aligned to the reference genome with STAR. Gene expression levels were quantified by counting aligned reads with Kallisto and normalized as fragments per kilobase of transcript per million mapped reads (FPKM). Differentially Expressed Genes (DEG) were obtained from the statistical packages DEseq2 output (p.adj <= 0.01) (Supplementary Method 1). We determine pathway enrichment using the ReactomeGSA Bioconductor R package for pathway-based analysis (Johannes Griss, 2025) GSEA 4.0.3 (https://www.gsea-msigdb.org/gsea/index.jsp), Enrichr and Appyters (https://maayanlab.cloud/Enrichr/).

The datasets generated for this study can be found in the BioProject PRJNA1355448 https://dataview.ncbi.nlm.nih.gov/object/PRJNA1355448?reviewer=3ct113p5bca2ami318gd1hpel9.

To elucidate the potential contribution of FZD4 to Wnt signaling activity in the context of OSCC, its expression was investigated in vivo using a 4NQO–induced murine model of OSCC (Sagheer et al., 2021). The carcinogen 4NQO was administered through the drinking water for 16 weeks, followed by a 10-week washout phase (Figure 1A). This protocol reliably induced malignant transformation with uniform penetrance, resulting in exophytic lesions on the dorsal surface of the tongue. In contrast, DMSO-treated control animals do not exhibit any local or systemic pathological manifestations. Following tissue collection, the tongues were embedded and cryosectioned for analysis. In control specimens, fluorescent immunolabelling revealed FZD4 expression restricted to the basal layer of the tongue epithelium (Figure 1B), with positive signal on both dorsal and ventral surfaces (Figures 1B,a,c). Positive staining was also detected in the fungiform papillae containing taste buds (Figures 1B,b). In 4NQO-treated mice, FZD4 immunoreactivity persisted within the dysplastic epithelium of the lesions, showing a clustered distribution pattern (Figure 1C, overview and magnifications).

To extend these findings to a human model, we analyzed SCC25 cells derived from the tongue of a 70-year-old male patient with squamous cell carcinoma (ATCC #CRL-1628) (Rheinwald and Beckett, 1981). Immunofluorescence confirmed FZD4 expression, displaying a membrane-associated, reticular pattern on the cell surface.

To assess the functional relevance of FZD4 in OSCC, we utilized FzM1, a negative allosteric modulator of FZD4 specifically inhibiting the FZD4-WNT signaling axis in a non-competitive manner (Generoso et al., 2015). SCC25 cells were treated with DMSO (control group) or FzM1 (treated group), and total RNA was subsequently extracted for downstream analyses. A transcriptomic library was generated for bulk RNA sequencing, revealing 834 differentially expressed genes (Supplementary Method 1).

The top 194 significant genes are visualized in the heatmap (Figure 2A). Pathway enrichment analysis using Enrichr Reactome Pathways and Kinase Enrichment Analysis (KEA) databases showed that most differentially expressed genes are associated with regulation of proliferation and cell cycle progression (Figures 2B,C). Gene Ontology (GO) analysis of biological processes further confirmed that the majority of affected categories relate to regulation of cellular kinetic, highlighting the transcriptional impact of FzM1 treatment on SCC25 proliferation (Figure 2D). Gene Set Enrichment Analysis (GSEA) consistently supported this observation, revealing altered expression of genes involved in the regulation of cell cycle progression (Figure 2E). To characterize these changes in greater detail, we analyzed the expression of individual transcripts by RT-PCR. Cyclin-dependent kinase 1 (CDK1) expression was markedly reduced upon FzM1 treatment, whereas other CDKs, such as Cyclin-dependent kinase 2 (CDK2), appeared mostly unchanged. Among the differentially expressed cyclins, FzM1 treatment led to decreased expression of cyclin A family members (A1 and A2) and cyclin E (E2) (Figure 2F).

We then sought to phenotypically characterize the effects of FzM1 on proliferating OSCCs cells. Firstly, we investigated whether FzM1 induces cell cycle exit or quiescence by analyzing the expression of the proliferation marker Ki67. Since Ki67 is expressed during all active phases of the cell cycle (late G1, S, G2, and M) but absent in resting cells (G0), we quantified Ki67-positive cells following FzM1 treatment (Figures 3A,B). No significant changes were observed, indicating that FzM1 does not induce quiescence.

We then examined the cell cycle phase most affected by FzM1. Mitotic figures corresponding to metaphase, anaphase, and telophase were readily observed in both control and treated cells (Figure 3C). Quantification revealed a significant reduction in the number of cells undergoing active mitotic division upon FzM1 treatment (Figure 3D). Consistently, immunofluorescence for phospho-H3, a mitotic marker expressed from late G2 through advanced mitosis, showed fewer labeled cells in treated samples compared to controls (Figures 3E,F).

To further validate these observations, we performed GSEA to assess gene expression changes related to mitosis. FzM1 treatment was negatively correlated with gene sets associated with mitotic spindle assembly and metaphase-anaphase transition (Figure 3G left panels; Supplementary Figure S1). Additionally, GSEA identified differential expression of genes regulating S-phase progression and cell cycle progression mediated by E2F transcription factors (Figure 3G, right panels; Supplementary Figure S1). These findings suggest that cells are kept in cycle during FzM1 treatment, with several phases of the cycle being affected by FzM1 exposure.

We reasoned that these alterations might collectively lead to a net increase in cell cycle duration, thereby slowing the kinetic progression of SCC25 cells treated with FzM1. To test this hypothesis, we performed a population kinetic analysis comparing the cell cycle length of DMSO- and FzM1-treated cells. Using thymidine analog cumulative labeling (Nowakowski et al., 1989; Martí-Clúa, 2023), cells were continuously exposed to a saturating dose of EdU, followed by staining and quantification. A short EdU pulse revealed comparable labeling indices (43.6% ± 1.6 for DMSO-treated cells and 43.4% ± 0.98 for FzM1-treated cells) indicating that a similar fraction of cells was actively cycling under both conditions (Figures 3H,I).

Over time, cumulative EdU incorporation showed that the proportion of labeled cells remained constant for approximately 10 h in DMSO-treated samples (Ts = 10h), whereas the same labeling index persisted for up to 16 h in FzM1-treated cells (Figures 3H,I). The prolonged maintenance of an unchanged EdU + fraction indicates an extended S-phase in FzM1-treated cells (Ts = 16 h). Consistently, the saturation point at which all cycling cells become EdU-positive, was also delayed following FzM1 exposure. Because this value corresponds to the total cell cycle length minus the duration of S-phase (Tc–Ts), we were able to mathematically estimate the total cell cycle duration as 31 h for DMSO-treated SCC25 cells and 42 h for FzM1-treated cells (Figures 3I,L), (Nowakowski et al., 1989; Martí-Clúa, 2023).

This significant prolongation of the cell cycle highlights a clear alteration in cell cycle regulatory mechanisms, which could substantially affect the proliferative capacity and expansion dynamics of OSCC cells.

To elucidate the molecular mechanisms by which FzM1 delays cell cycle progression in OSCC cells, we performed GO enrichment analysis of Biological Processes (BP) using our transcriptomic dataset. This analysis revealed significant enrichment of pathways related to retinoic acid (RA) metabolism and response to vitamin A (Figure 4A). Consistently, GSEA demonstrated strong enrichment of pathways associated with retinol metabolism and cytochrome P450–mediated oxidation (Figure 4B). Volcano plot analysis revealed marked downregulation of UGT and CYP genes involved in RA catabolism (Figure 4C) (Rowbotham et al., 2010; Ross and Zolfaghari, 2011), suggesting that FzM1 promotes RA accumulation by inhibiting its metabolic breakdown, thereby stabilizing intracellular RA levels (Figure 5).

Since RA is known to regulate the expression of adhesion complexes in various epithelial contexts (Aldehlawi et al., 2020; Chateau and Boehm, 1996), we investigated whether transcriptomic changes upon FzM1 treatment reflect RA-mediated alterations in the SCC25 phenotype. In line with previous reports (Miyazono et al., 2020; Kautsky et al., 1995), our data reveal a transcriptional signature marked by differential expression of genes known to be influenced by RA signaling: such as focal adhesion, adherens junctions, integrin signaling, keratin production, and collagen remodeling (Supplementary Figures S2A–C).

We next investigated the potential link between prolonged RA signaling and the delayed cell cycle progression observed following FZD4 inhibition. We reasoned that blocking RA receptor activity would reverse the effects of FzM1 and restore SCC25 cell cycle kinetics to control levels. To test this, we treated cells for 5 days with FzM1 in combination with a RA receptor antagonist (RARi). As expected, RARi completely abolished the kinetic changes observed after FzM1 treatment: Phenotypic analysis of cell-cycle dynamics revealed that FzM1+RARi–treated cells exhibit a proportion of mitotic figures comparable to DMSO controls (3.049% ± 0.45) and phospho-H3 levels restored to baseline (8.7% ± 1.69), with no statistically significance when compared to DMSO values. We then quantify cell cycle length in FzM1+RARi–treated cells. Consistently with a rescued phenotype, EdU incorporation curves revealed a S-phase duration of 10 h (Ts = 10 h) and a total cell cycle length of 31 h (Tc = 31 h), matching the values observed in DMSO-treated cells (Figures 4D,E). Through these findings, we inferred that the proliferation and cell-cycle alterations induced by FzM1 require RA pathway activation downstream of FZD4 signaling, supporting a mechanistic model in which WNT and RA pathways cooperatively regulate cell-cycle progression and OSCC cell growth.