A Wet‐and‐Dry Research Model: Unveiling Biomechanics in Odontoblast Polarization

ABSTRACT

Aim

Odontoblast polarization is key to the formation and regeneration of tubular dentine. However, the molecular mechanisms, particularly the mechanobiological cues that guide odontoblast polarization, remain unrecognized, hindering the advancement of regenerative endodontics. To address this, we implemented a Wet‐and‐Dry platform that integrates photolithographic micropatterning and mathematical modelling, aiming to characterize the mechanobiological features underlying odontoblast polarization at the single cell level.

Methodology

A combined wet‐and‐dry platform was designed. In the wet experiments, PEGDA‐based photolithography created adhesive micropatterns with defined areas and aspect ratios, onto which human dental pulp stem cells (hDPSCs) were seeded. Cell morphology, polarity metrics, odontogenic marker expression and mechano‐inductive protein levels were quantified by immunofluorescence and image analysis. Mechanisms were probed by disrupting the cytoskeleton, inhibiting Rho‐ROCK signalling and blocking YAP1 activity. Using an ex vivo mouse tooth germ model, the role of RhoA/ROCK in odontoblast polarization was confirmed under inhibitor treatment. In the dry experiments, a vertex‐based computational model simulated single hDPSCs adhesion and spreading on the micropatterns, and predicted the spatiotemporal force distribution during polarization, including deformation energy, adhesion strength and traction forces. The simulation outputs were validated against experimental measurements.

Results

Micropatterned substrates mimicking odontoblast‐like cell shapes provide an ideal platform for single‐cell mechanobiology. Patterns with 2700 μm² area and 1:4 aspect ratio most strongly promoted hDPSCs polarization and odontogenic differentiation. Polarized cells exhibited increased myosin II and increased nuclear YAP1. A vertex‐based model predicted that high aspect ratio increases traction and adhesion forces. Loss‐of‐function experiments confirmed that Rho‐dependent cytoskeletal remodelling and increased tension are required for geometry‐driven polarization and differentiation. Ex vivo tooth germ culture showed that inhibiting RhoA/ROCK disrupts odontoblast polarization and tissue organization, underscoring the importance of this signalling axis.

Conclusion

By integrating micropatterned substrates with in silico modelling and ex vivo tissue validation, we demonstrate that microenvironmental geometry directs hDPSCs' polarization and odontogenic differentiation via a Vinculin–RhoA–YAP signalling axis, which clarifies a core mechanobiological mechanism underlying odontoblast polarization.

1. Introduction

The rising prevalence of dental caries and pulp diseases poses an escalating threat to oral health. Dentine, the principal mineralized component of the tooth, comprises a hydroxyapatite‐collagen composite permeated by a hierarchical network of dentinal tubules, which together confer mechanical resilience, sensory transduction and immune defence. Although tissue engineering approaches have enabled the regeneration of dentine‐like constructs, these often present as bone‐like tissues with sparse dentinal tubules, disordered hydroxyapatite alignment and diminished elasticity, limiting their functional integration with native dentine (Azaryan et al. 2023). This mismatch in structural and mechanical properties remains a central obstacle in regenerative endodontics.

A critical determinant of tubular dentine regeneration is the establishment of odontoblast polarity. During odontogenesis, progenitor cells undergo cytoskeletal remodelling and organelle repositioning to become elongated, polarized odontoblasts that secrete matrix apically, initiating tubular dentine formation. Polarization entails spatial asymmetry in cytoskeletal reorganization, a process tightly regulated by biomechanical cues within the microenvironment (Kuhn et al. 2025). Notably, mesenchymal condensation during tooth development generates physiologically relevant compressive stress that triggers odontoblast polarization (Miyashita et al. 2017). Moreover, external mechanical stimuli such as tensile strain, compression and low‐intensity pulsed ultrasound also enhanced odontogenic differentiation of dental stem cells (Marrelli et al. 2018). In addition, substrate parameters such as geometric confinement or stiffness influence integrin‐mediated cell adhesion to the extracellular matrix, activating RhoA/ROCK and Yes‐associated protein/transcriptional coactivator with PDZ‐binding motif (YAP/TAZ) signalling to drive cytoskeletal tension and leading to altered cell behaviours (Wu et al. 2023).

Despite the mounting evidence that mechanical forces orchestrate odontoblast polarization via cytoskeletal modulation, the molecular pathways remain ambiguous, largely owing to the lack of appropriate study models. In vivo models involve intricate and interdependent signalling networks, while in vitro models generally fail to recapitulate the biochemical and biomechanical complexity of the native microenvironment. Thus, there remains a pressing need for an appropriate platform that integrates controlled polarity induction, defined mechanical stimulation and direct molecular readouts. Micropatterning technology, originally adapted from electronic‐surface modification, enables precise control over cell‐substrate interactions and microenvironmental mechanics (Li, Jiang, et al. 2023). Photolithographic patterning of PEGDA hydrogels yields non‐fouling substrates with well‐defined adhesive patterns, directing cell adhesion and morphology to predetermined geometries (Hakim Khalili et al. 2023). This approach offers streamlined design, tunable biomechanical properties, high reproducibility and compatibility with high‐resolution imaging, thus providing desirable platforms for systematic studies of cell polarity (Bril et al. 2023).

While micropatterning technology has provided significant insights into odontoblast‐related studies, it faces challenges in quantitatively elucidating molecular mechanisms and dynamically illustrating mechanical forces. The limited number of viable cells obtainable from such platforms poses substantial constraints on conventional molecular biology assays. Furthermore, this approach lacks the capability for real‐time and quantitative assessment of mechanical forces acting on targeted cells. To address these limitations, computational modelling has emerged as an essential complementary tool. By developing mechanistic models that incorporate established biophysical principles and biochemical interactions, it becomes feasible to systematically simulate the cellular behaviours under geometric constraints identical to those in experimental settings. Computational models that explicitly describe cell adhesion, signalling and mechanotransduction (Betorz et al. 2023; Xu et al. 2023) have played critical roles in illustrating the mechanobiological characteristics of multiple cells (Fang et al. 2021).

In this study, we present a Wet‐and‐Dry joint platform that integrates micropattern‐based experimental assays and a computational simulation‐based mathematics model to illustrate the mechanical mechanism underlying odontoblast polarization. Regarding the Wet platform, the micropatterns were generated via PEGDA and a photolithography technique, enabling the isolation of single hDPSCs and their transition toward odontoblasts. By confining single hDPSCs within defined microscale geometries, we showed that shape‐induced cytoskeletal tension activated RhoA/ROCK signalling and promoted Yes‐associated protein 1 (YAP1) nuclear translocation, ultimately leading to odontoblast‐like polarized morphology and upregulated odontogenesis. Regarding the Dry platform, we custom‐designed a computer simulation that recapitulated the morphological evolution of spreading hDPSCs and predicted the spatiotemporal features of stress distribution during cell polarization. The in silico data further verified the experimental results and collectively supported the force‐driven mechanism underlying odontoblast polarization. In summary, our findings uncover the biomechanical basis of odontoblast polarization and highlight the utility of such joint systems for probing undetected mechanisms in tooth development.

2. Materials and Methods

The current manuscript has described the main stages of the study according to the Preferred Reporting Items for Laboratory Studies in Endodontology (PRILE) 2021 guidelines (Nagendrababu, Murray, et al. 2021) (Figure S1) and the Preferred Reporting Items for Animal Studies in Endodontology (PRIASE) 2021 guidelines (Nagendrababu, Kishen, et al. 2021) (Figure S2).

2.1. Isolation, Cultivation and Identification of hDPSCs

Impacted third molars, which were extracted from patients as part of routine clinical care, were used to isolate hDPSCs. All patients provided written informed consent for the use of their discarded tissues for research purposes. The protocol was approved by the Ethics Committee of Jilin University, China (No. SJDKQ2024066). The teeth were removed solely for clinical indications and were not extracted for the purpose of this study. Extracted third molars were immediately placed in phosphate‐buffered saline (PBS) containing 1% penicillin/streptomycin (Sigma‐Aldrich, St. Louis, MO, USA) and processed promptly in the laboratory. After crown separation, the dental pulp was extracted, and 2 mm of the apical root pulp was removed. The tissue was rinsed twice in PBS containing 1% penicillin/streptomycin, followed by two additional washes in PBS. The pulp tissue was then transferred to a digestive solution containing 1 mg/mL collagenase I and 2.5 mg/mL neutral protease and incubated at 37°C for 30 min. During digestion, the centrifuge tube was gently shaken every 10 min to ensure thorough contact between the tissue and the enzymatic solution. Digestion was terminated by adding 5 volumes of Minimum Essential Medium Alpha (α‐MEM, Gibco, Grand Island, NY, USA) supplemented with 20% (v/v) fetal bovine serum (FBS, Haixing Biosciences, Suzhou, China) and 1% penicillin/streptomycin. The resulting cell suspension was centrifuged at 1200 rpm for 4 min, and the supernatant was discarded. The cell pellet was resuspended in α‐MEM containing 20% FBS and 1% penicillin/streptomycin, and seeded into T25 culture flasks. The cells were cultured in a humidified incubator at 37°C with 5% CO₂. After approximately 5 days, cell attachment was observed under an inverted microscope, and the culture medium was replaced. Thereafter, the medium was refreshed every 2–3 days. Cells at passages P3–P5 were used for experiments. Flow cytometry analysis was performed to identify surface markers of hDPSCs, including CD90, CD105, CD34 and CD45 (Invitrogen; Thermo Fisher Scientific, Carlsbad, CA, USA). The expression of these markers was assessed by flow cytometry using a MACSQuant Analyzer (Miltenyi Biotec, Bergisch Gladbach, Germany).

2.2. Design, Fabrication and Characterization of Micropatterned Arrays

Micropatterns were designed to recapitulate key morphological features of odontoblasts. To emulate the columnar cell body and long cytoplasmic extension characteristic of odontoblasts, the patterns were designed with a combined geometry consisting of a rectangle (corresponding to the cell body) and a triangle (mimicking the odontoblastic process). A series of progressively increasing aspect ratios (1:1, 1:2, 1:3 and 1:4) was employed to approximate this polarization feature, acknowledging the practical limits of in vitro patterning to reproduce the extreme slenderness of native processes. Based on prior measurements of human dental pulp stem cells (hDPSCs) spreading on extracellular matrix (ECM)‐mimetic substrates, which was approximately 1900 μm² (Chang et al. 2021), three pattern areas were selected: 1800, 2700 and 3600 μm². The upper limit (3600 μm²) was chosen to preserve single‐cell occupancy and minimize multicellular adhesion that could confound phenotypic analysis. Following the design phase in Adobe Illustrator, a glass photolithography mask was fabricated. Glass slides were cut to a preset size, sequentially cleaned in absolute ethanol and deionized water, and air‐dried. Slides were then subjected to 10 min of plasma surface treatment to activate the surface, generating silanol groups for subsequent grafting. Treated slides were immersed in a silane solution—a mixture of absolute ethanol and silane coupling agent (SCA, Sigma)—for 30 min, then rinsed with ethanol and air‐dried for later use. The hydrogel solution was prepared using poly(ethylene glycol) diacrylate (PEGDA, Sigma) and the photoinitiator I2959 (Sigma). For fabrication, each slide was placed on the stage of a photolithography machine, and 50 μL of the prepolymer solution was dropped onto the surface. After covering with the photomask, ultraviolet (UV) exposure for 10 s resulted in the formation of the target PEG hydrogel patterns (Figure 1a). To verify the precision of micropattern fabrication, tetramethylrhodamine ethyl ester (TMRE, MedChem Express (MCE), Monmouth Junction, NJ, USA) was added to the PEGDA hydrogel solution, and the patterns were observed under fluorescence imaging to confirm their morphology. Additionally, the microstructure of the micropatterns was examined using scanning electron microscopy (SEM). For SEM observation, the fabricated PEG hydrogel micropatterns were lyophilized, sputter‐coated with platinum and imaged with a JEOL Ltd. (Tokyo, Japan). Subsequently, the surface topography and roughness of the fabricated micropatterns were characterized using atomic force microscopy (AFM; Bruker Corporation, Billerica, MA, USA) in ScanAsyst mode. The surface roughness parameters were analysed using NanoScope Analysis 3.00 software.

FIGURE 1.

Illustration of the Wet‐and‐Dry platform exploring the biomechanics of odontoblast polarization. (a) Schematic of the micropattern fabrication workflow: (I) oxygen plasma treatment of glass substrate; (II) silane modification; (III) application of PEGDA prepolymer mixed with photoinitiator; (IV) mask alignment; (V) ultraviolet exposure for polymerization; (VI) rinsing to remove non‐crosslinked prepolymer. (b) Representative fluorescence micrographs of rhodamine‐labelled PEGDA hydrogel micropatterns with different areas (1800, 2700 and 3600 μm²) and aspect ratios (1:1, 1:2, 1:3 and 1:4). The images show the fluorescence signal of rhodamine incorporated into the hydrogels, and no cells were cultured on the substrates. Scale bars: 100 μm. (C) Morphology of hDPSCs cultured on different micropatterns for 24 h. The cells were stained for F‐actin (phalloidin, red) and nuclei (DAPI, blue). The PEGDA hydrogels used for cell culture experiments were not labelled with fluorescent dye, thus eliminating any fluorescent background from the substrates. Scale bars: 100 μm. (D) Particle‐based simulation of cell‐substrate mechanics on an odontoblast‐shaped micropattern: (I) Substrate discretization: Fixed ligand particles (blue) define adhesive sites within the pattern (grey). (II) Vertex‐based cell modelling: Membrane vertices (cyan) interconnected by a triangular network (blue edges) to represent cytoskeleton. (III) Deformation forces: The triangular membrane network resists shape changes through elastic forces, including in‐plane shear (stretching along edges), out‐of‐plane bending and constraints on membrane area and cell volume, ensuring structural integrity during spreading. (IV) Adhesive forces: Membrane receptors bind substrate ligands, producing adhesion that resists membrane recoil. (V) Traction forces: Contractile springs within the cytoskeleton transmit tensile loads through membrane vertices to adhesive sites, driving outward spreading. (e) Simulation predictions of steady‐state cell morphologies on PEGDA micropatterns with aspect ratios of 1:1, 1:2, 1:3 and 1:4. The light blue plane marks the micropattern boundary, and the red surface depicts the simulated steady‐state cell shape. Cells progressively elongate along the pattern's long axis as aspect ratio increases.

2.3. Simulation Model

A vertex model is used to model the experimental adhesion‐to‐spreading transition on each micropattern geometry. The underlying substrate is discretized into a set of fixed particles representing adhesive sites. Cells are modelled using a vertex approach, where the cell membrane is defined by a set of moving vertices, and the cytoskeleton is represented by a triangular network formed by connecting the vertices. The motion of each vertex is governed by Newtonian second law, given by

$$\left\{\begin{array}{c}\frac{d{x}_i}{\mathit{dt}}=v_i,\\ m\frac{d{v}_i}{\mathit{dt}}=F_i^C+F_i^D+F_i^R+F_i^G+F_i^W+F_i^E+F_i^A+F_i^T,\end{array}\right.$$ (1)

where m, v _i , x _i are the mass, velocity and coordinate of the vertex i, $F_i^C$ is the conservative force accounting for fluid compressibility, $F_i^D$ is the dissipative force describing fluid viscosity, $F_i^R$ is the random force to maintain the fluid temperature, $F_i^G$ is the gravity force, $F_i^W$ is the repulsive force to repel the liquid and membrane vertices close to the substrate, $F_i^E$ is the membrane elastic deformation force, $F_i^A$ is the adhesion force and $F_i^T$ is the traction force. The velocity‐Verlet method, is used to solve Equation (1) is solved using the velocity‐Verlet method, a predictive‐corrective algorithm, to update x _i and v _i based on the given initial conditions $x_i^0$ and $v_i^0$. Detailed formulations of the force terms and the step‐by‐step implementation of the velocity‐Verlet method can be found in our previous work.

To accurately represent hDPSCs behaviour on PEGDA hydrogel substrates, the vertex model requires a set of mechanical and biophysical parameters. The key parameters characterizing cellular biomechanics are summarized as follows. The cell membrane shear modulus and bending modulus are set to 6 × 10⁻⁶ N/m and 2.0 × 10⁻¹⁹ J, respectively, which are consistent with values commonly adopted in established cellular mechanical models (Li et al. 2005; Pivkin and Karniadakis 2008). The kinetic parameters governing cytoskeletal dynamics are based on the classical Brownian ratchet model for traction force generation, with the free polymerization and depolymerization rates of actin filaments assigned as 5.16 × 10⁻⁸ m/s and 2.03 × 10⁻⁹ m/s, respectively (Döbereiner et al. 2004; Brevier et al. 2007; Fang et al. 2020). Furthermore, the interaction between cellular receptors and substrate‐bound ligands is defined by an association rate of 7.18 × 10⁴ s⁻¹ and a dissociation rate of 8.26 s⁻¹. These specific receptor‐ligand kinetic rates were calibrated in the present work to ensure that the simulated adhesive dynamics and the resultant adhesion‐to‐spreading transition faithfully replicate the experimental observations of hDPSCs on PEGDA substrates. All other relevant parameters, such as geometry, density, viscosity and time step, are detailed in our previous work (Ye et al. 2017; Wang et al. 2021; Xia et al. 2025) unless otherwise specified.

It should be noted that the vertex‐based model is designed as a phenomenological representation of cell mechanics. It captures the physical principles governing cell shape evolution and force distribution, but does not explicitly simulate the biochemical recruitment dynamics (e.g., vinculin accumulation) or resolve molecular‐scale structures (e.g., individual stress fibres). Furthermore, the current model treats the cell as a continuum with homogeneous material properties without a distinct nucleus. Consequently, while it enables direct comparison of morphological outcomes and robustly predicts trends in force generation, a strict quantitative comparison with molecular markers like vinculin intensity or stress fibre alignment angles remains beyond its scope. Instead, its primary value lies in providing mechanistic explanations that bridge the gap between geometric cues and the experimentally observed cellular responses.

2.4. Seeding hDPSCs on Micropatterns

Micro‐patterned samples were placed in 6‐well plates, sterilized by soaking in 75% ethanol, and exposed to UV light for 30 min. Residual ethanol was thoroughly rinsed off with PBS, and the samples were immersed in complete culture medium for subsequent use. A cell suspension was seeded onto the sample surface at a density of 4 × 10⁴ cells per well. After 1 h, non‐adherent cells were washed away with PBS, fresh complete culture medium was added, and the samples were incubated at 37°C in a 5% CO₂ environment for further culture.

2.5. Inhibition Assays

The concentrations of inhibitors were determined based on CCK‐8 cytotoxicity assays and live/dead cell staining to ensure optimal cell spreading without inducing noticeable morphological changes (Figures S3–S5). All inhibitors were obtained from MCE unless otherwise specified, and were used at the following concentrations: Nocodazole 0.03 μM (Watson et al. 2023), Cytochalasin D 0.5 μg/mL (Ying et al. 2024), Y‐27632 8 μM (Srikawnawan et al. 2022), NSC23766 50 μM (Jiang et al. 2020), ML141 10 μM (Chaker et al. 2018), Blebbistatin 10 μM (Zheng et al. 2018) (Sigma) and Verteporfin 0.1 μM (Liu‐Chittenden et al. 2012). These inhibitors targeted microtubule polymerization, actin filament polymerization, ROCK, Rac1, cell division control protein 42 homologue (Cdc42), myosin II and YAP1, respectively. The inhibitors were applied for 1 day in polarization‐related experiments and for 6 days in differentiation‐related experiments.

2.6. Immunofluorescence Staining

After culturing hDPSCs on micropatterns for 48 h, cells were washed with PBS, fixed with 4% paraformaldehyde (Solarbio Life Sciences, Beijing, China) for 15 min, and permeabilized with 0.2% Triton X‐100 in PBS for 5 min. Cells were then blocked with 5% bovine serum albumin (BSA) at room temperature for 1 h. Primary antibodies against GM130 (BD Biosciences, Franklin Lakes, NJ, USA), Myosin II (Sigma), YAP1 (Proteintech Group Inc., Rosemont, IL, USA) and vinculin (Proteintech) were diluted 1:200 and incubated overnight at 4°C. For assessment of odontogenic marker expression, dentine sialophosphoprotein (DSPP) immunostaining was performed after the 7‐day differentiation protocol: cells were fixed after 7 days of induction and processed as above. Primary anti‐DSPP antibody (Proteintech) was applied at 1:200 and incubated overnight at 4°C. After thorough washing, all samples were incubated with fluorophore‐conjugated secondary antibodies (FITC‐conjugated goat anti‐mouse IgG or FITC‐conjugated goat anti‐rabbit IgG, both from Proteintech) diluted 1:200 for 1 h at room temperature. F‐actin was labelled with rhodamine‐phalloidin (Solarbio) for 30 min, and nuclei were counterstained with DAPI (Solarbio) for 7 min. Coverslips were mounted using antifade mounting medium (Invitrogen). Immunofluorescence images were captured on an OLYMPUS IX83 inverted fluorescence microscope (Olympus Corporation, Tokyo, Japan) using identical exposure settings for quantitative comparisons.

2.7. Cell Polarity Assays

Single hDPSCs that adhered to the micropatterns and matched their shapes were selected for staining and imaging. Polarization indicators were defined based on literature references.

Polarization indicators included:

1. Cell polarization angle (Maninová et al. 2013; Costa et al. 2021): A cell is considered polarized if the Golgi apparatus is located within a 120° sector originating from the nuclear centroid and oriented toward the tip of the micropattern.

2. Nucleus‐Golgi distance (Chang et al. 2020): The straight‐line distance between the nuclear centroid and the Golgi centroid. A greater distance indicates a higher degree of polarization.

3. Nuclear Displacement Index (Gomes et al. 2005): The displacement of the nuclear centroid along the cell's major axis toward the trailing edge (opposite the protrusive tip). A larger rearward shift reflects enhanced polarization.

4. Nucleus reorientation (Maninová and Vomastek 2016): The angle between the nuclear long axis and the cell's long axis. A smaller angle indicates greater alignment and thus stronger polarization.

5. Nuclear area (Lomakin et al. 2020): The projected two‐dimensional area of the nucleus (μm²). A larger area is associated with increased polarization.

6. Nuclear aspect ratio (AR) (Nie et al. 2025): The ratio of the nuclear long‐axis length to its short‐axis length. Values above 1 indicate nuclear elongation, with higher ratios reflecting greater polarization.

To quantify these parameters, fluorescence images were processed in ImageJ (National Institutes of Health, Bethesda, MD, USA) (Schindelin et al. 2012). The nucleus (DAPI), Golgi apparatus (GM130) and cell morphology (phalloidin) channels were segmented independently by threshold‐based selection of high‐intensity signal against a dark background. Centroids of the nucleus, Golgi apparatus and cell centroid were extracted from each segmented region using the Analyze Particles routine (centroid measurement enabled). The coordinate of the tip of the cellular protrusion was identified manually for each cell. Extracted centroid coordinates and ellipse‐fit nuclear parameters were then used to compute polarization metrics 1–3 from their relative spatial relationships, while nuclear orientation, aspect ratio (AR) and area (metrics 4–6) were obtained from the fitted nuclear ellipse.

2.8. ALP Activity and DSPP Quantification During hDPSCs Differentiation

To induce odontogenic differentiation in hDPSCs, the cells were cultured for 6 days in a medium containing 10% FBS, 50 μM ascorbic acid‐2‐phosphate, 10 mM β‐glycerophosphate and 100 nM dexamethasone. This induction was preceded by an initial day in a growth medium, making a total culture period of 7 days. To inhibit stem cell proliferation, hDPSCs were continuously exposed to 0.5 mg/mL Aphidicolin (MCE) (Yao et al. 2013). Alkaline phosphatase (ALP) activity was used as a marker for odontogenic differentiation (Wang et al. 2010). After washing the cells with PBS, they were fixed with 4% paraformaldehyde (Solarbio) for 15 min. ALP staining was performed as instructed, with colour developing after 3 h of incubation. To distinguish single cells adhered to micropatterns, the cells were also stained with DAPI (Solarbio). Images were captured using an inverted microscope for subsequent analysis.

To evaluate the effects of different shapes and areas on hDPSCs' odontogenic differentiation, ImageJ was used for statistical analysis of single cells adhered to the micropatterns. Two key parameters were assessed:

1. ALP‐positive cell ratio (Gunn et al. 2022): The percentage of single cells exhibiting positive ALP staining.

2. Average Optical Density (AOD) of ALP (Gou et al. 2014; Lu et al. 2022): Images were converted to grayscale in ImageJ and analysed using the ‘Uncalibrated OD’ setting to obtain average optical density (AOD). Mean AOD values were measured within defined regions of interest (ROIs) and compared across different pattern shapes and areas.

3. DSPP mean fluorescence intensity (MFI). Single‐cell DSPP quantification was performed in ImageJ. Regions of interest (ROIs) were drawn around each cell guided by F‐actin (phalloidin) and DAPI staining, and the mean grey value was recorded. Background intensity was measured in adjacent cell‐free regions and subtracted; values are reported as background‐subtracted MFI.

2.9. Cellular Biomechanics Analysis

To assess the spatial relationship between F‐actin and myosin II, fluorescence images stained with phalloidin (for F‐actin) and immunolabelled for myosin II were analysed using the Scatter J plugin in ImageJ. The Pearson correlation coefficient (PCC) was calculated to quantify the degree of colocalization between these two cytoskeletal components, with values ranging from −1 to 1, where values approaching 1 indicate a high degree of colocalization. The MFI of myosin II, YAP1 and vinculin was measured using ImageJ. Regions of interest (ROIs) were defined for individual cells, and average fluorescence intensities within these ROIs were extracted to compare protein expression levels under different experimental conditions. To evaluate YAP1 subcellular localization, ROIs were independently designated for the nucleus and cytoplasm. The fluorescence intensities of YAP1 in each compartment were measured, and the nuclear‐to‐cytoplasmic (N/C) ratio was calculated, providing a quantitative indicator of YAP1 translocation and cellular mechanotransduction status.

2.10. Ex Vivo Tooth Germ Culture and Pharmacological Inhibition

All animal experiments were performed in accordance with protocols approved by the Laboratory Animal Ethics Committee of the School of Basic Medical Sciences at Jilin University (No. 20250699). E14.5 mouse mandibular first molar tooth germs were cultured ex vivo with or without inhibitor treatment. Detailed culture conditions, inhibitor administration, tissue processing, and quantitative analysis of odontoblast polarization (cell aspect ratio, apical‐basal polarity index, and cell polarization angle, as illustrated in Figure S6) are described in Appendix S1.

2.11. Data Analysis

Statistical analyses and graphing were conducted using GraphPad Prism (version 9.0, La Jolla, CA, USA). All data were derived from at least three independent experiments and are presented as the mean ± standard error of the mean (SEM), unless otherwise stated. For each experimental condition, a minimum of 30 cells per group were analysed. Prior to analysis, normality and homogeneity of variance were assessed. Statistical comparisons between two groups were performed using an unpaired two‐tailed Student's t‐test, while comparisons among multiple groups were evaluated using one‐way analysis of variance (ANOVA), followed by Tukey's or Dunnett's post hoc test as appropriate. A p‐value of less than 0.05 was considered statistically significant. Detailed statistical information is provided in the figure legends.

3. Results

3.1. Evaluation of Single hDPSCs With Micropatterns and Computational Simulations

To validate our Wet‐and‐Dry platform for studying hDPSCs mechanobiology, we first confirmed the accuracy of the fabricated micropatterns. Fluorescence microscopy of TMRE‐conjugated PEGDA hydrogels confirmed the precise replication of the pre‐designed micropattern shapes and sizes across all areas and aspect ratios (Figure 1b). To further characterise the topographical features, we employed scanning electron microscopy (SEM) and atomic force microscopy (AFM) to analyse the fabricated PEG hydrogel micropatterns (Figure S7). The SEM images confirmed the structural integrity of the patterns (Figure S7a), while AFM analysis revealed that the average roughness (Ra) of the cell‐adhesive regions was significantly lower than that of the hydrogel regions and comparable to that of the glass substrate (Figure S7b,c). This indicates that the micropatterning process did not introduce additional nanotopography into the cell‐adhesive areas, thereby excluding potential nanotopographical effects on cell behaviour (Rahman et al. 2022; Kim et al. 2024). Prior to seeding, flow cytometry confirmed the mesenchymal origin of the hDPSCs used in this study (Figure S8). Upon seeding, hDPSCs adhered efficiently to the patterns, with seeding efficiencies of approximately 60% and single‐cell occupancy rates of approximately 51% across all pattern areas (Figure S9). This confirmed the platform's robust biocompatibility and suitability for single‐cell analyses. Moreover, F‐actin staining showed that single hDPSCs conformed precisely to the micropattern outlines and adopted typical odontoblast‐like polarized morphologies (Figure 1c), confirming the successful simulation of the in vivo odontoblast morphology.

Additionally, a step‐by‐step mathematical simulation was conducted to model the experimental adhesion‐to‐spreading transition on each micropattern geometry. The underlying substrate was discretized into a set of fixed particles representing adhesive sites (Figure 1d). Cells were modelled using a vertex approach, where the cell membrane was defined by a set of moving vertices. To simulate cellular mechanical properties, these vertices were connected into a triangular network to represent the cytoskeleton, with edges acting as elastic springs. The evolution of the cell shape during spreading was simulated by numerically solving the equations of motion governing the vertices, driven by the balance of deformation, traction and adhesion forces. This simulation successfully replicated the distinct cellular morphologies observed experimentally across all micropattern designs (Figure 1e). Collectively, both experimental and simulation data validated our micropatterning platform as an effective in vitro system for precise control of hDPSCs shape and mechanical environment.

3.2. Cellular Geometry Regulates hDPSCs Polarization and Differentiation

To quantitatively assess the polarization of single hDPSCs on micropatterns, we used six previously established morphological metrics (see Materials and Methods for detailed definitions) that collectively quantify nuclear positioning, Golgi orientation, and nuclear deformation (Figure 2a,b). Applying these metrics, we found that in cells cultured on 2700 μm² patterns, increasing the aspect ratio from 1:1 to 1:4 led to a progressive enhancement in all six polarization metrics. For instance, on 2700 μm² patterns, the nucleus‐Golgi distance increased from 8.02 ± 0.57 μm in the 1:1 group to 10.81 ± 0.47 μm in the 1:4 group (p < 0.01; Figure 2e–j). The polarization process was accompanied by a directional reorientation of the Golgi apparatus, with a significant increase in the proportion of cells meeting the cell polarization angle metric, suggesting that their secretory activity may be directed toward the predefined odontoblast‐like protrusion. Further observations revealed a coordinated spatial redistribution between the nucleus and the Golgi apparatus, characterized by an increased centroid‐to‐centroid distance between the two organelles. The nuclear displacement index indicated that the nucleus shifted away from the protrusion along the cell's long axis. This spatial configuration, with the Golgi biased toward the protrusion and the nucleus biased toward the opposite side, collectively defined the anteroposterior axis of the cell, providing a structural basis for the establishment of polarity. Concurrently, nuclear morphology adapted; the angle between the nuclear long axis and the cellular long axis decreased, while nuclear area and aspect ratio increased significantly, indicating nuclear elongation and alignment with the direction of cellular elongation. These changes collectively illustrate a geometry‐guided polarity‐establishment process, which was most fully realized on the 2700 μm² patterns and most pronounced at the 1:4 aspect ratio. In contrast, on 1800 μm² and 3600 μm² substrates, only nuclear orientation and nuclear area (metrics 4 and 5) exhibited significant responses to aspect ratio variations, suggesting that suboptimal pattern areas may limit the full development of cellular polarization.

FIGURE 2.

Polarization and differentiation of single hDPSCs on micropatterns. (a) Representative immunofluorescence micrographs of a non‐polarized (top) and a polarized (bottom) hDPSCs on micropatterns. Cells were stained for the Golgi apparatus (GM130, green), F‐actin (phalloidin, red) and nuclei (DAPI, blue). In the polarized cell, the Golgi apparatus is positioned toward the protrusive tip, and the nucleus is displaced rearward. Scale bar: 100 μm. (B) Schematic illustration of the six quantitative metrics used to assess cell polarization: (1) Cell polarization angle: The Golgi apparatus is considered polarized if located within a 120° sector from the nuclear centroid toward the pattern tip; (2) Nucleus‐Golgi distance: The straight‐line distance between the nuclear centroid and the Golgi centroid; (3) Nuclear displacement index: The rearward shift of the nuclear centroid relative to the tip of the cellular protrusion; (4) Nuclear reorientation: The angle between the nuclear long axis and the cell's long axis; (5) Nucleus area: The projected two‐dimensional area of the nucleus (μm²); (6) Nucleus aspect ratio: The ratio of nuclear long axis to short axis. (c) Representative phase‐contrast images of single hDPSCs cultured on micropatterns of three areas (1800, 2700 and 3600 μm²) and four aspect ratios (1:1, 1:2, 1:3 and 1:4) For 24 h. Cells progressively elongate and align with the pattern geometry as aspect ratio increases. Scale bar: 100 μm. (D) Representative immunofluorescence micrographs of single hDPSCs on micropatterns, stained for DSPP (green), F‐actin (red) and nuclei (blue). DSPP expression increases with aspect ratio, most notably on 2700 μm² patterns. Scale bar: 100 μm. (e–j) Quantification of the six polarization metrics defined in (b) for hDPSCs on micropatterns with varying areas and aspect ratios. (k) Percentage of ALP‐positive single hDPSCs on each micropattern type. (l) DSPP mean fluorescence intensity (MFI) in single hDPSCs. All data are presented as mean ± SEM. For the cell polarization angle (e) and ALP‐positive ratio (k): Data were calculated from each independent experiment and are shown as mean ± SEM of at least three independent experiments ($n\ge 3$), with each experiment comprising at least 30 cells per group. For all other quantitative data (f–j, l): $n\ge 40$ Cells per group from at least three independent experiments. Statistical significance was determined by one‐way ANOVA followed by Tukey's post hoc test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Measures of odontogenic differentiation paralleled the morphological changes, increasing progressively from 1:1 to 1:4. Within the 2700 μm² group, ALP positivity and AOD (Figure S10) were significantly higher on 1:3 and 1:4 patterns than on 1:1, with the largest difference observed between 1:4 and 1:1. For example, the ALP‐positive rate increased from 12.02% ± 1.89% in the 1:1 group to 25.04% ± 1.73% in the 1:4 group (p < 0.0001; Figure 2c,k). DSPP mean fluorescence intensity rose from 9.69 ± 0.43 in the 1:1 group to 14.03 ± 1.07 in the 1:4 group (p < 0.01; Figure 2d,l). Collectively, these data indicate that, among the patterns tested, 2700 μm² micropatterns most effectively promote geometry‐dependent polarization and odontogenic differentiation of single hDPSCs.

3.3. Cellular Geometry Controls Single‐Cell Biomechanics

To investigate the influence of cellular aspect ratio on biomechanical properties, hDPSCs were immunostained for vinculin, Myosin II and YAP1 to evaluate cell adhesion, cytoskeletal tension, and mechanotransduction, respectively. Immunofluorescence showed significantly higher vinculin intensity on 1:4 micropatterns (18.59 ± 0.85) than on 1:1 substrates (10.94 ± 1.06; p < 0.0001; Figure 3a,b). Subcellularly, vinculin was diffusely distributed in the cytoplasm with punctate enrichment at cell extremities (Figure S11), consistent with its dual role as an inactive cytoplasmic reservoir and an activated component of mature focal adhesions (Case et al. 2015; Bays and DeMali 2017). These findings suggest that high aspect ratio geometries may enhance both vinculin expression and its recruitment to focal adhesions, thereby potentiating local mechanotransduction. To quantitatively assess the spatial colocalization between F‐actin and Myosin II, we performed PCC analysis. This analysis revealed strong co‐distribution of the two proteins across all groups (PCC > 0.80; Figure S12), consistent with their incorporation into stress fibres. Despite this ubiquitous colocalization, only cells cultured on 1:4 micropatterns developed prominent longitudinal stress fibres and showed a significant increase in Myosin II MFI (from 26.57 ± 1.38 in the 1:1 group to 31.24 ± 1.34 in the 1:4 group, p < 0.05; Figure 3c,d). Furthermore, YAP1 accumulated in the nuclei of cells on 1:4 micropatterns, as indicated by a higher nuclear‐to‐cytoplasmic MFI ratio (from 1.83 ± 0.07 in the 1:1 group to 2.17 ± 0.05 in the 1:4 group, p < 0.001; Figure 3e,f), whereas it was retained predominantly in the cytoplasm in the 1:1 group. Collectively, these findings demonstrate that higher aspect ratios enhance intracellular tension and promote YAP1 nuclear translocation.

FIGURE 3.

Mechanical analysis of single hDPSCs. (a) Representative immunofluorescence micrographs of single hDPSCs cultured on micropatterns with aspect ratios of 1:1, 1:2, 1:3 and 1:4 (All at 2700 μm² area). Cells were stained for Vinculin (FITC, green), F‐actin (phalloidin, red) and nuclei (DAPI, blue). Scale bar: 100 μm. (b) Quantification of mean Vinculin fluorescence intensity (MFI). Vinculin expression was significantly higher in cells on the 1:4 Micropatterns compared to those on the 1:1, 1:2 and 1:3 patterns. (c) Representative immunofluorescence micrographs of single hDPSCs stained for Myosin II (FITC, green), F‐actin (phalloidin, red) and nuclei (DAPI, blue). Prominent longitudinal stress fibres co‐localized with Myosin II were observed on the 1:4 pattern. Scale bar: 100 μm. (d) Quantification of mean Myosin II fluorescence intensity (MFI). Myosin II expression was significantly higher in cells on the 1:4 micropatterns compared to other aspect ratio groups, indicating elevated intracellular tension. (e) Representative immunofluorescence micrographs of single hDPSCs stained for YAP1 (FITC, green), F‐actin (phalloidin, red) and nuclei (DAPI, blue). Nuclear YAP1 accumulation was more pronounced in cells on the 1:4 pattern. Scale bar: 100 μm. (f) Quantification of YAP1 nuclear‐to‐cytoplasmic (N/C) fluorescence ratio (left) and mean fluorescence intensity (right). Cells on the 1:4 pattern exhibited a significantly higher N/C ratio, indicating enhanced nuclear translocation of YAP1. (g) Shape evolution of cell spreading in computer simulations on micropatterns with increasing aspect ratios (1:1–1:4). The simulated cell progressively elongated, closely matching the experimental morphologies shown in (a, c, e). (h) Contour plots of single‐particle forces from the particle‐based simulation. The top row shows adhesion forces on substrate ligand particles; the bottom row shows traction forces on membrane vertices. Columns represent aspect ratios of 1:1, 1:2, 1:3 and 1:4. Colour bars indicate force magnitude, ranging from low (purple) to high (yellow). Traction forces were concentrated at the cell tips, most prominently at the highest aspect ratio. (i) Quantification of total adhesion and traction forces exerted by the simulated cell. Both adhesion and traction forces increased significantly with increasing aspect ratio, reaching the highest levels in the 1:4 group. All quantitative data are presented as mean ± SEM, with n ≥ 30 cells per group from three independent experiments. Statistical significance was determined by one‐way ANOVA followed by Tukey's post hoc test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Having established that geometric cues modulate cytoskeletal tension and mechanotransduction at the protein level, we developed a customized in silico model to further validate and extend our understanding of the mechanical behaviour of hDPSCs on micropatterned substrates. First, we assessed the model's ability to recapitulate cell spreading dynamics. As shown in Figure S13, the vertex‐based model successfully recapitulated the key morphological changes of hDPSCs on 1:4 micropatterns at 1, 2, 4, 24 and 48 h post‐seeding, including initial spreading (1–2 h), elongation along the pattern axis (4–24 h) and stabilization of the polarized morphology (48 h). This qualitative consistency between simulated and experimental morphologies, a widely accepted criterion for validating phenomenological cell mechanics models, confirms that our model captures the principal biophysical processes governing cell spreading.

Leveraging this model, we next predicted the distribution of cellular traction and adhesion forces. The results revealed that traction force peaks were concentrated at the cell periphery, most prominently at the tips of elongated cells. In contrast, adhesion forces originating from the cell‐matrix interface exhibited a broader and more uniform distribution. Both traction and adhesion forces increased significantly with the aspect ratio of the micropatterns (Figure 3g–i). Importantly, the spatial distribution of high traction force regions corresponded closely to areas rich in myosin II‐positive longitudinal stress fibres, as observed by immunofluorescence. This finding indicates that the shape‐driven recruitment of myosin II and assembly of stress fibres constitute the primary mechanism responsible for generating highly localized active contractile forces at the cell tips. The model further predicted that these localized mechanical signals are transmitted via stress fibre–nuclear connections, leading to enhanced nuclear deformation. This finding elucidates a plausible mechanotransduction pathway for the enhanced YAP1 nuclear translocation observed in cells on high‐aspect‐ratio patterns.

3.4. Cytoskeletal Integrity Drives hDPSCs Polarization and Differentiation

Polarization marks a clear morphological transition of dental mesenchymal cells into odontoblasts, accompanied by cytoskeletal reorganization and organelle repositioning. To assess the contribution of cytoskeletal integrity to this process, hDPSCs confined to micropatterns were treated for 24 h with cytochalasin D (CytoD; 0.5 μg/mL) or nocodazole (NC; 0.03 μM) to selectively disrupt F‐actin or microtubule dynamics, respectively. Representative immunofluorescence images show cytoskeletal labeling (top rows: phalloidin for F‐actin; α‐tubulin‐FITC) and corresponding DSPP staining in the same fields (bottom rows; FITC). Nuclei were stained with DAPI (Figure 4a,b). CytoD fragmented F‐actin into punctate structures, whereas NC inhibited microtubule assembly. Quantitatively, both inhibitors significantly reduced all six polarization metrics compared with the control. For example, nuclear area was reduced from 233.9 ± 13.4 μm² in the control (1:4) group to 191.8 ± 9.6 μm² in the CytoD group (p < 0.01; Figure 4c–h), confirming that disruption of cytoskeletal integrity impairs the polarized morphology. They also decreased the fraction of ALP‐positive cells, from 22.84% ± 1.30% in the control (1:4) group to 14.44% ± 2.22% in the CytoD group and 9.42% ± 1.18% in the NC group (CD: p < 0.05; NC: p < 0.001; Figure 4i), consistent with suppression of early odontogenic activity, and reduced the mean DSPP fluorescence intensity, from 13.05 ± 0.85 in the control (1:4) group to 10.71 ± 0.57 in the CytoD group (p < 0.05; Figure 4j), indicating impaired progression toward an odontoblastic phenotype. Collectively, these findings indicate that intact actin and microtubule networks are essential for micropattern‐driven polarization and odontogenic differentiation of hDPSCs.

FIGURE 4.

Effects of actin‐ and microtubule‐disrupting drugs on cytoskeletal organization and DSPP immunostaining intensity in single hDPSCs. (a) Representative immunofluorescence micrographs of untreated hDPSCs (Control) and cells treated with Cytochalasin D (CD, 0.5 μg/mL) for 24 h on 1:4 micropatterns. The top row shows F‐actin (phalloidin, red) staining, and the bottom row shows DSPP immunostaining (FITC, green) under different conditions. Nuclei were stained with DAPI (blue). CD treatment fragments F‐actin into punctate structures and visibly reduces DSPP expression. Scale bar: 25 μm. (b) Representative immunofluorescence micrographs of untreated cells and cells treated with Nocodazole (NC, 0.03 μM). The top row shows α‐tubulin (FITC, green), and the bottom row shows DSPP (FITC, green). Nuclei were stained with DAPI (blue). NC treatment disrupts the organized microtubule network. Scale bar: 25 μm. (c–h) Quantification of the six polarization metrics (as defined in Figure 2b) in CD‐ and NC‐treated cells relative to the untreated control. Both inhibitors significantly reduced all polarization metrics. (i) Quantification of the percentage of ALP‐positive hDPSCs after CD or NC treatment. Both inhibitors significantly reduced the proportion of ALP‐positive cells. (j) Quantification of DSPP mean fluorescence intensity (MFI) in CD‐ and NC‐treated cells. Both inhibitors significantly decreased DSPP expression. All data are presented as mean ± SEM. For the cell polarization angle (c) and ALP‐positive ratio (i): Data were calculated from each independent experiment and are shown as mean ± SEM of at least three independent experiments (n ≥ 3), with each experiment comprising at least 30 cells per group. For all other quantitative data (d–h, j): N ≥ 30 cells per group from at least three independent experiments. Statistical significance was determined by one‐way ANOVA followed by Dunnett's post hoc test. *p < 0.05, **p < 0.01, ***p < 0.001. CD, Cytochalasin D; NC, Nocodazole.

3.5. Rho GTPases Regulate hDPSCs Polarization, Differentiation and Tension

To determine whether geometric cues activate RhoA upstream of cytoskeletal reorganization, we compared RhoA activity in hDPSCs on 1:1 versus 1:4 micropatterns. Immunofluorescence revealed significantly higher RhoA activity on elongated patterns (Figure S14), indicating that geometric cues trigger early RhoA signalling. To further dissect the functional role of Rho GTPases, single hDPSCs cultured on 1:4 micropatterns were treated with inhibitors targeting specific family members: Y‐27632 (8 μM; RhoA/ROCK inhibitor), NSC23766 (50 μM; Rac1 inhibitor) and ML141 (20 μM; Cdc42 inhibitor) (Figure 5). Y‐27632 treatment abolished stress fibres and Myosin II bundling and significantly reduced all six polarization metrics, Myosin II density, YAP1 nuclear localization, ALP activity and DSPP fluorescence intensity compared with untreated controls. For instance, Myosin II MFI decreased from 31.67 ± 1.22 in the control (1:4) group to 24.05 ± 0.93 in the Y‐27632‐treated group (p < 0.001). The YAP1 nuclear‐to‐cytoplasmic ratio was also significantly reduced from 2.24 ± 0.05 to 1.93 ± 0.05 (p < 0.01), and DSPP MFI dropped from 13.86 ± 0.98 in the control (1:4) group to 10.51 ± 0.65 in the Y‐27632‐treated group (p < 0.01). These results confirm that RhoA/ROCK signalling is required for generating the cytoskeletal tension necessary for polarization and downstream odontogenic differentiation. NSC23766 and ML141 each reduced four of the six polarization metrics (with the exception of nuclear reorientation angle and nuclear aspect ratio), lowered Myosin II density and YAP1 nuclear translocation and decreased both ALP‐positive rates and DSPP immunofluorescence. For instance, DSPP MFI was significantly reduced from 13.86 ± 0.98 in the control (1:4) group to 11.08 ± 0.55 in NSC23766‐treated cells and 11.18 ± 0.54 in ML141‐treated cells (p < 0.05). These data indicate that RhoA/ROCK signalling is critical for generating cytoskeletal tension and driving polarization and odontogenic differentiation in micropattern‐confined hDPSCs, with Rac1 and Cdc42 playing supportive roles. To validate these findings in a more physiologically relevant context, we employed an ex vivo culture model of E14.5 mouse molar tooth germs, which preserves the native three‐dimensional architecture of the developing tooth organ (Alfaqeeh and Tucker 2013). Having established the critical role of RhoA/ROCK‐mediated cell tension in odontoblast polarization, we further confirmed its involvement with the ex vivo model. The results indicated that treatment with the RhoA/ROCK inhibitor Y‐27632 severely disrupted tooth germ development in a dose‐dependent manner, resulting in significantly reduced tissue dimensions and a disorganized odontoblast layer (Figure S15a–d). Quantitative immunofluorescence analysis further revealed that Y‐27632 treatment almost completely abolished odontoblast polarization, as evidenced by significant reductions in cell aspect ratio, apical‐basal polarity index and the proportion of polarized cells (as determined by the cell polarization angle) with correctly positioned Golgi apparatus (Figure S15e–g; see figure legend for detailed statistics). These ex vivo results corroborate our micropattern findings, confirming that RhoA/ROCK signalling is essential for tissue organization and odontoblast polarization in the tooth germ microenvironment.

FIGURE 5.

Effects of Rho GTPase inhibitors on hDPSCs polarization, cellular tension and differentiation. (a) Representative immunofluorescence micrographs of hDPSCs cultured on 1:4 micropatterns and treated with Rho GTPase inhibitors: Y‐27632 (ROCK inhibitor, 8 μM), NSC23766 (Rac1 inhibitor, 50 μM), and ML141 (Cdc42 inhibitor, 20 μM). Cells were stained for GM130, Myosin II, YAP1 and DSPP (all FITC, green), F‐actin (phalloidin, red) and nuclei (DAPI, blue). Y‐27632 treatment abolished stress fibres and Myosin II bundling. NSC23766 and ML141 caused partial disruption of F‐actin organization. All inhibitors reduced YAP1 nuclear localization and DSPP expression. Scale bar: 100 μm. (b–g) Quantification of the six polarization metrics in inhibitor‐treated hDPSCs relative to the untreated control. Y‐27632 significantly reduced all six metrics, while NSC23766 and ML141 reduced four of the six metrics. (h) Quantification of mean Myosin II fluorescence intensity (MFI). All three inhibitors significantly reduced myosin II expression. (i) Quantification of YAP1 nuclear‐to‐cytoplasmic (N/C) fluorescence ratio. All inhibitors significantly decreased YAP1 nuclear translocation. (j) Quantification of mean YAP1 fluorescence intensity (MFI). (k) Quantification of the percentage of ALP‐positive hDPSCs. All inhibitors significantly reduced ALP activity. (l) Quantification of DSPP mean fluorescence intensity (MFI). All inhibitors significantly decreased DSPP expression. All data are presented as mean ± SEM. For the cell polarization angle (b) and ALP‐positive ratio (k): Data were calculated from each independent experiment and are shown as mean ± SEM of at least three independent experiments ($n\ge 3$), with each experiment comprising at least 30 cells per group. For all other quantitative data (c–j, l): $n\ge 50$ Cells per group from at least three independent experiments. Statistical significance was determined by one‐way ANOVA followed by Dunnett's post hoc test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

3.6. Verteporfin‐Mediated YAP Inhibition Abolishes Micropattern‐Induced Polarization and Odontogenic Differentiation in hDPSCs

Having established the pivotal role of RhoA in mediating geometric cues, we next investigated whether YAP1, a key downstream mechanotransducer identified in our 2D studies, mediates these geometry‐induced cellular responses. To determine whether YAP1 mediates shape‐dependent responses, single hDPSCs confined to 1:4 micropatterns were treated with the YAP1 inhibitor verteporfin (VP) (Figure 6). VP markedly decreased YAP1 nuclear localization and mean fluorescence intensity, with the nuclear‐to‐cytoplasmic ratio falling from 2.24 ± 0.05 in the control to 1.90 ± 0.09 in the VP‐treated group (p < 0.01), confirming effective YAP1 inhibition. All six polarization metrics declined significantly after VP treatment, indicating a substantial reduction in cellular polarization. Functionally, VP reduced both the proportion of ALP‐positive cells (from 26.45% ± 2.71% in the control to 6.88% ± 0.47% in the VP‐treated group; p < 0.01) and the mean optical density of ALP staining, consistent with suppression of early odontogenic activity; DSPP fluorescence intensity was also significantly decreased versus controls (from 12.78 ± 0.86 to 10.42 ± 0.37, p < 0.01), demonstrating impaired maturation toward an odontoblastic lineage. Collectively, these results indicate that YAP1 activity is required for micropattern‐induced hDPSCs polarization and odontogenic differentiation. To further validate the upstream regulatory role of cytoskeletal tension in this process, hDPSCs were treated with Blebbistatin, a specific inhibitor of myosin II. The results showed that inhibiting myosin II not only disrupted myosin II aggregation itself but also significantly suppressed YAP1 nuclear translocation, as well as subsequent polarization and differentiation markers (Figure S16). This finding strongly supports the conclusion that myosin II‐mediated cytoskeletal tension acts upstream of YAP1 nuclear translocation. Combined with the results from RhoA inhibition and YAP1 inhibition experiments, these findings collectively provide a complete chain of experimental evidence for the key signalling cascade by which geometric constraints drive cell polarization and differentiation through RhoA activation, myosin II‐mediated cytoskeletal tension and subsequent YAP1 nuclear translocation.

FIGURE 6.

YAP1‐mediated control of hDPSCs polarization and odontogenic differentiation. (a) Representative immunofluorescence micrographs of hDPSCs cultured on 1:4 micropatterns and treated with the YAP1 inhibitor Verteporfin (VP, 0.1 μM). Cells were stained for YAP1, GM130 and DSPP (all FITC, green), F‐actin (phalloidin, red) and nuclei (DAPI, blue). Scale bar: 100 μm. (b) Quantification of YAP1 nuclear‐to‐cytoplasmic (N/C) fluorescence ratio. (c) Quantification of mean YAP1 fluorescence intensity (MFI). (d–i) Quantification of the six polarization metrics in VP‐treated hDPSCs relative to the untreated control. (j) Quantification of the percentage of ALP‐positive hDPSCs. (k) Quantification of DSPP mean fluorescence intensity (MFI). All data are presented as mean ± SEM. For the cell polarization angle (b) and ALP‐positive ratio (j): Data were calculated from each independent experiment and are shown as mean ± SEM of at least three independent experiments ($n\ge 3$), with each experiment comprising at least 30 cells per group. For all other quantitative data (c–i, k): $n\ge 40$ Cells per group from at least three independent experiments. *p < 0.05, **p < 0.01 compared to the untreated control.

4. Discussion

As a requisite to tubular dentine formation, odontoblast polarization marks a distinct morphological transition from symmetrical mesenchymal cells to high columnar odontoblasts. This transmission inevitably accompanies transitions in the cytoskeleton network and intracellular mechanics. However, due to the lack of an appropriate platform, the underlying mechanism remains uncertified. Traditional scratch‐wound assays and microchannel devices permit visualization of polarity markers but lack three‐dimensional mechanical context (Mazalan et al. 2024). Spheroid cultures maintain cell–cell contacts yet suffer from uneven nutrient diffusion and heterogeneity (Białkowska et al. 2020). GelMA‐based electrospun scaffolds we previously designed induced odontoblast‐like cell morphology and alignment, but the complex fabrication limited scalability and hindered high‐throughput molecular analyses (Chang et al. 2020). Herein we established a Wet‐and‐Dry platform combining a robust PEG‐micropatterned microarray chip with a vertex‐based simulation model. The Wet‐lab micropatterning chip provides an odontoblast‐mimicking microenvironment, enabling direct visualization of hDPSCs in response to cell geometry, cytoskeletal breakdown and cell tension disruption. Meanwhile, we developed an in silico model that dynamically recapitulates hDPSCs adhesion and spreading, and more importantly, quantifies biomechanical changes both intracellularly and at the cell‐matrix communicative interface. This Dry‐lab model further provides an effective tool that enables quantitative evaluation of the cytoskeletal dynamics and relevant biomechanical behaviours within single hDPSCs on the pre‐designed micropatterns.

During the terminal differentiation of odontoblasts, the establishment of a cellular polarity axis provides the structural basis for the directional secretion of dentine matrix. The six morphological metrics employed in this study collectively characterize the integrity of this process from multiple dimensions. Polarity establishment begins with basal nuclear migration (nuclear displacement index), which creates space for the repositioning of other organelles (Sasaki and Garant 1996). Subsequently, the Golgi apparatus relocates to the supranuclear region, and its position can be quantitatively assessed by the nucleus–Golgi distance and polarization angle, marking the functional reorganization of the secretory machinery. In differentiated odontoblasts, the Golgi apparatus serves as a central hub for the processing and transport of dentine matrix proteins (Kupfer et al. 1982). Its polarized localization reflects the functional demands of the cell and also represents a reliable marker for evaluating cell polarity (Chang et al. 2020). As the nucleus migrates basally and the Golgi repositions apically, a polarity axis anchored by the nucleus and extending along the cell process is established, thereby facilitating directional transport of secretory vesicles and ensuring polarized secretion of matrix components (Weinstock and Leblond 1974; Arana‐Chavez and Massa 2004). Concurrently, under tension exerted by the cytoskeleton via the linker of nucleoskeleton and cytoskeleton (LINC) complex, the nucleus undergoes reorientation and elongation (Dahl et al. 2008; Chang et al. 2019), aligning its long axis with the apico‐basal axis of the cell (Maninová et al. 2013; Lele et al. 2018). This morphological remodelling optimizes the spatial organization of the apical cytoplasmic region and minimizes physical obstacles to polarized vesicle trafficking (Chang et al. 2019). Moreover, the increase in nuclear area and associated morphological changes themselves constitute a mechanosensing mechanism (Niethammer 2021), which can influence chromatin structure and regulate the activity of transcription factors such as YAP1 (Emon et al. 2024), thereby converting extracellular mechanical cues into chemical signals that modulate the expression of genes involved in dentine matrix synthesis.

In summary, the six morphological indicators employed in this study systematically capture a continuum of functional states in odontoblasts, ranging from polarity axis establishment and secretory apparatus reorganization to the determination of secretory direction and intranuclear mechanotransduction. Together, they provide a comprehensive morphological basis for assessing and predicting the directional secretory capacity of odontoblasts. Our results showed that when hDPSCs were cultured on micropatterns with an area of 2700 μm² but varying aspect ratios, all six indicators progressively increased with geometric elongation. Notably, the 1:4 micropattern most strongly promoted Golgi–nucleus separation, nuclear basal displacement, nuclear alignment and nuclear elongation, exhibiting the most pronounced polarization phenotype. This outcome was expected, as this geometry most closely mimics the naturally elongated morphology of odontoblasts, thereby providing a favourable geometric template for cell polarization.

Furthermore, the size of the confinement area itself may significantly influence the polarization process. An excessively small area may fail to support adequate cytoskeletal spreading and tension accumulation (McBeath et al. 2004; Dupont et al. 2011), whereas an excessively large area could weaken geometric boundary constraints, thereby hindering the guidance of cells toward a unified polarization axis (Théry et al. 2006). In this context, the 2700 μm² micropattern appears to achieve a favourable balance; it provides sufficient spreading area for cells to establish a high aspect ratio morphology, while maintaining effective boundary constraints that concentrate directional tension. This balance, in turn, may potentiate the geometrically induced polarization of hDPSCs.

Notably, this platform is not intended to represent a specific stage of odontoblast differentiation. It captures the intertwined and synergistic processes of polarization and differentiation, which tend to occur concurrently during odontoblast maturation. This view is supported by the coordinated changes in polarization metrics and differentiation markers observed in the present study. Classic studies on mouse tooth development have demonstrated a temporal overlap between the appearance of polarized odontoblast morphology and the initial expression of differentiation markers (Sreenath et al. 1999), suggesting that these two processes occur in parallel under physiological conditions. This close relationship is further supported by findings that the expression of polarity‐related proteins increases during odontoblast differentiation (Yin et al. 2021). Thus, although polarization and differentiation describe distinct aspects of cellular behaviour, they are tightly intertwined and occur concurrently. The present study demonstrates that mechanical signals can concurrently induce cell polarization and activate the expression of differentiation‐related proteins, reflecting the coordinated nature of these two processes rather than a unidirectional causal relationship.

Mechanotransduction from focal adhesions (FAs) to nuclear effectors provides an integrative axis by which geometric cues direct hDPSCs polarization and differentiation. In this work, cells on 1:4 micropatterns exhibited significantly higher vinculin enrichment at FAs than cells on 1:1 micropatterns. This was further verified by our vertex‐based computational model that the force‐cloud maps identified regions of elevated traction at the cell tips, coinciding with sites of maximal vinculin accumulation and stress‐fibre anchorage. Consistent with previous reports that anisotropic shapes enhanced FA maturation (Cabezas et al. 2019), we propose that cell elongation redistributes traction forces toward focal adhesions, promoting their maturation and the anchoring of stress fibres that drive cytoskeletal remodelling. Meanwhile, phalloidin and myosin II staining showed that cells on 1:4 micropatterns developed denser, more aligned stress fibres and exhibited higher myosin II density than the 1:1 group, indicating elevated intracellular tension. As reported previously, Src activation resulting from integrin‐mediated stress fibre contractility inhibits Hippo pathway kinases such as large tumour suppressor kinase (LATS), thus promoting YAP/TAZ dephosphorylation and nuclear import (Si et al. 2017). Our results showed hDPSCs on 1:4 micropatterns displayed the highest nuclear‐to‐cytoplasmic YAP1 ratios, reflecting a tension‐driven YAP1 enrichment. This was also corroborated by the computational model that the traction forces at the cell periphery on 1:4 micropatterns were significantly greater than those on 1:1 micropatterns.

To verify the above hypothesis, we started with cytoskeleton modulation assays. Microfilaments are essential to the structure and function of the Golgi apparatus by interacting with its matrix proteins GM130 and GRASP65 to support cisternal stacking. Besides, actomyosin contractility generates forces that promote membrane trafficking (Chakrabarti et al. 2021) and facilitate nuclear positioning and orientation, which are all essential for cell polarization (Khuntia et al. 2022). Our results demonstrated that Cyto D treatment abolished stress fibre formation, displaced the Golgi apparatus and sharply reduced all six polarization metrics as well as ALP activity and DSPP intensity, particularly on 1:4 micropatterns, further underscoring the role of an intact F‐actin network in odontogenic polarization. Meanwhile, disruption of microtubule dynamics with nocodazole produced similarly impaired effects. Microtubules participate in Golgi alignment via dynein and kinesin motors, and nucleus relocation via LINC complexes (Bouzid et al. 2019). Besides, nocodazole–induced depolymerization released guanine nucleotide exchange factor H1 (GEF‐H1), activated RhoA‐dependent contractility and prevented recruitment of +TIP proteins necessary for cortical actin assembly (Joo and Olson 2021), which may be another possible mechanism underlying the disrupted hDPSCs mechanotransduction.

Our targeted inhibition of Rho GTPases revealed their unique yet complementary contributions to regulating hDPSCs behaviour. Inhibition of the RhoA–ROCK signalling axis with Y‐27632 induced collapse of stress fibres and aggregation of myosin II, coinciding with a decreased nuclear‐to‐cytoplasmic YAP1 ratio, implying reduced intracellular tension. This aligns with RhoA's established contribution in generating actin‐cytoskeletal tension that promotes YAP1 nuclear localization (García‐García et al. 2022). Meanwhile, Rac1 sustains cytoplasmic tension by promoting Myosin II heavy‐chain phosphorylation via protein kinase C beta II (PKCβII) (Pasapera et al. 2015), inhibiting myosin light‐chain phosphatase through CPI‐17/MYPT1 (Shibata et al. 2015) and remodelling perinuclear actin to enlarge nuclear pores and engage YAP1. Thus, Rac1 is critical for coupling cytoskeletal force to nuclear mechanotransduction (Sgarzi et al. 2023). Our data showed that NSC23766‐mediated Rac1 inhibition partially disrupted F‐actin bundles and markedly reduced nuclear import of YAP1, consistent with previous reports. Besides, Cdc42 inhibition disrupted Golgi positioning and rearward nuclear movement (Lee et al. 2025), and abolished MRCK‐mediated MLC phosphorylation, thereby preventing proper nucleus–Golgi spacing and lowering YAP1 levels (Gomes et al. 2005; Bourdais et al. 2023). These findings are consistent with our data that ML141 reduced the polarization and differentiation of single hDPSCs. Therefore, we propose that RhoA generates the primary contractile force required for actin‐mediated YAP1 activation, Rac1 refines nuclear‐pore mechanics to transmit tension to the nucleus, and Cdc42 orchestrates organelle positioning and auxiliary contractile pathways, and the interplay among all three GTPases establishes the intracellular force landscape that drives hDPSCs polarization and odontogenic differentiation.

To further validate these findings in a more physiologically relevant context, we employed an ex vivo culture model of E14.5 mouse molar tooth germs. The results showed that treatment with Y‐27632 significantly disrupted tooth germ development, leading to reduced tissue dimensions and a disorganized odontoblast layer. Quantitative analysis further revealed that Y‐27632 markedly decreased multiple polarization metrics of odontoblasts. These ex vivo results not only corroborate our micropattern findings but also provide further evidence that RhoA/ROCK signalling likely serves as a key regulatory factor driving odontoblast polarization in the native tooth germ microenvironment.

In our study, we observed that inhibition of Rho GTPase signalling markedly disrupted the perinuclear clustering of the Golgi apparatus but did not completely abrogate odontoblast process formation. These findings suggest that although both events are subject to regulation by Rho GTPases, they may not be linked by a strict causal relationship. They likely represent parallel and cooperative processes that collectively contribute to the establishment of cell polarity. The positioning of the Golgi apparatus is known to be tightly regulated by microtubule motors, which are in turn modulated by Rho GTPases such as Cdc42 and RhoA (Hehnly et al. 2010; Egea et al. 2015; Quassollo et al. 2015). In contrast, although Rho GTPases are key regulators of actin dynamics and process formation (Nobes and Hall 1995), inhibiting individual family members did not completely abolish odontoblast processes. This partial retention likely reflects functional redundancy and compensatory mechanisms within the Rho GTPase family, as exemplified by persistent filopodia formation in Cdc42‐deficient fibroblasts (Czuchra et al. 2005). Collectively, these observations suggest that Golgi positioning and process formation likely function as parallel and cooperative events that jointly orchestrate the establishment of cell polarity.

Having established the critical role of Rho GTPases, we next checked how the cytoplasmic tension got transmitted into the nucleus and regulated hDPSCs polarization and differentiation. Previous studies have proved that under the regulation of Rho GTPases, contractile actomyosin bundles convey force to the nucleus through the perinuclear actin cap and the LINC complex (Ma et al. 2024), increasing nuclear envelope tension and nuclear pore complex permeability and thereby facilitating YAP1 nuclear import (Andreu et al. 2022). Consistent with this, we further confirmed using the myosin II inhibitor Blebbistatin that disrupting actomyosin contractility directly blocks YAP1 nuclear translocation and subsequent polarization and differentiation, thereby identifying myosin II‐mediated cytoskeletal tension as a critical link in transmitting mechanical signals from RhoA to YAP1. Nuclear YAP1 primarily binds to the TEA domain transcription factor (TEAD) family of transcription factors (Heng et al. 2021), activating the expression of effectors associated with odontogenic differentiation (Li, Xu, et al. 2023) and thereby establishing a direct physical link from geometric sensing to transcriptional response. Our data validated that inhibition of Rho GTPases, particularly RhoA, decreased the YAP1 nuclear translocation efficiency, and more importantly, direct inhibition of YAP1 significantly decreased geometry‐induced cell polarization and differentiation, thereby confirming the role of YAP1 as a key transcriptional effector in linking mechanical input to the odontogenic gene program. Collectively, we defined a vinculin–RhoA–YAP signalling cascade that connects extracellular geometry to fate decisions and provides a mechanistic basis for physics‐based dental pulp regeneration strategies. Although this platform induces cell polarization primarily within the xy‐plane, diverging from the vertical elongation of odontoblasts in vivo, it successfully captures key polarization features at the single‐cell level, including basal nuclear migration and Golgi reorientation (Gundersen and Worman 2013; Chang et al. 2019). The findings are consistent with our previous three‐dimensional model and further enable mechanistic dissection of the underlying mechanotransduction pathways. Thus, this system serves as a high‐resolution platform for visualizing geometric cue‐driven mechanotransduction, functioning as a mechanobiological bridge that connects in vivo complexity with in vitro tractability and laying the groundwork for future three‐dimensional models incorporating vertical polarization.

5. Conclusion

With the integrated experimental and in silico platform, we define a detailed mechanobiological framework in which micropattern‐imposed geometric cues guide hDPSCs' polarization and odontogenic differentiation through coordinated adhesion, cytoskeletal, and nuclear signalling. Although our Wet‐and‐Dry joint platform offers new insights into the exploration of hDPSCs mechanobiology, it has limitations. Our future work will incorporate live‐cell tension sensors and more physiologically relevant patterns to fully capture force‐dependent signalling in the developmental niche. These advances will refine next‐generation mechano‐therapeutic design strategies for functional dentine regeneration and clinical translation.

Author Contributions

Huen Li conceived and designed the study, performed data acquisition, analysis and interpretation, and drafted and critically revised the manuscript. Nianzuo Yu contributed to study design and data analysis and critically revised the manuscript. Haofeng Liu, Ziwei Zhu, Xiheng Li, Chao Si, and Yi Li assisted in data acquisition and critically revised the manuscript. Xiaoduo Tang, Junhu Zhang, Ting Ye, and Bei Chang contributed to study conception and design, data interpretation, and both drafted and critically revised the manuscript. Hongchen Sun contributed to study conception and data interpretation and critically revised the manuscript. All authors gave final approval and agree to be accountable for all aspects of the work.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: National Natural Science Foundation of China (82470941, 82101075, 22275071 and 82301044), Young Elite Scientists Sponsorship Program by CAST (YESS20210407 and 2024QNRC001), Science and Technology Project of Jilin Provincial Department of Finance (jcsz2023481‐1).

Ethics Statement

Sound human third molars were collected from healthy adult donors as part of necessary clinical procedures requiring extraction. Written informed consent was obtained from all donors, and the protocol was approved by the Ethics Committee of Jilin University, China (No. SJDKQ2024066). The protocol for the animal experiment was approved by the Laboratory Animal Ethics Committee of the School of Basic Medical Sciences at Jilin University (No. 20250699).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: PRILE 2021 flowchart for the design and procedure of this study.

Figure S2: PRIASE 2021 flow chart of the study design and results.

Figure S3: CCK‐8 assay for determining the safe concentrations of different inhibitors (Y‐27632, NSC23766, ML141 and Nocodazole) in hDPSCs.

Figure S4: Effects of different inhibitor concentrations on hDPSCs viability.

Figure S5: Effects of Verteporfin, Blebbistatin and Cytochalasin D on hDPSCs viability.

Figure S6: Schematic diagram of odontoblast polarization parameters.

Figure S7: Characterization of micropattern topography.

Figure S8: Flow cytometry analysis of hDPSCs and immunofluorescence staining of DSPP after odontogenic induction.

Figure S9: Cell seeding efficiency and single‐cell seeding efficiency on micropatterns of varying area.

Figure S10: ALP activity in single hDPSCs cultured on micropatterned substrates.

Figure S11: Subcellular distribution of vinculin in hDPSCs on micropatterns.

Figure S12: Colocalization analysis of F‐actin and myosin II.

Figure S13: Computational model recapitulating the time‐series morphology of cell spreading.

Figure S14: Immunofluorescence staining of RhoA in hDPSCs cultured on micropatterns with aspect ratios of 1:1 and 1:4.

Figure S15: Effects of RhoA/ROCK inhibition on tooth germ development and odontoblast polarization in ex vivo culture.

Figure S16: Immunofluorescence staining of Myosin II, YAP1, GM130 and DSPP in hDPSCs under control (Con) and blebbistatin‐treated (+Bleb) conditions.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.