Microfabrication of Substrates with Microscale Stiffness Gradients to Guide Bone Marrow Stromal Cell Migration

Summary

This protocol details the fabrication of a bilayer polydimethylsiloxane (PDMS) substrate via soft lithography. The method utilizes geometrically modulated microstructures in a rigid underlayer combined with a soft flat top layer to generate microscale effective stiffness gradients, enabling cellular mechanobiology studies without topographic interference.

Abstract

Understanding how cells respond to mechanical signals is crucial for elucidating the mechanisms involved in tissue development and disease progression. However, existing in vitro cell culture substrates often fail to replicate the physiological stiffness gradients at the cellular scale while also eliminating confounding topographical cues. In this study, we present a decoupled stiffness model utilizing a bilayer polydimethylsiloxane (PDMS) substrate. This substrate consists of a soft, flat top film suspended over a rigid underlayer featuring ridge-and-groove microstructures. The top layer effectively transmits stiffness variations in the underlying structure while maintaining uniform topography and chemistry at the cell contact surface. Stiffness modulation is achieved by varying the width of the alternating ridges and grooves, which are spaced equally. Scanning electron microscopy confirmed the flat surface morphology and consistent contact topography. Atomic force microscopy demonstrated that stiffness variations were dependent on the microstructure: for 20 μm patterns, the elastic moduli were approximately 950 kPa for ridges and 850 kPa for grooves; for 50 μm patterns, these values were around 1070 kPa and 950 kPa, respectively. Mouse primary bone marrow cells adhered well and spread on the substrate, showing a preference for nuclear localization toward the stiffer ridge regions in the 50 µm pattern (61.49%, p < 0.05), thereby confirming effective cellular perception of the mechanical gradient. In summary, this protocol offers a reproducible method to construct a cell culture substrate with microscale stiffness gradients, minimizing the chemical or topography interference, enabling investigations into cell mechanotransduction.

Introduction

Cells have the ability to sense and respond to mechanical stimuli within their external microenvironment^1,2,3, which are crucial for regulating various cellular processes, including adhesion, migration, proliferation, and differentiation⁴. In vivo, the mechanical properties of the tissue microenvironment exhibit significant spatial heterogeneity, with variations in matrix stiffness observed both in physiological contexts - such as at the junctions of bone and muscle - and in pathological conditions, such as within tumors^5,6,7. A comprehensive investigation of these mechanical factors is essential for understanding the regulatory mechanisms governing cell behavior. These mechanisms are pivotal in processes such as tissue development, disease progression, and the process of regeneration and repair^8,9.

To simulate this complex mechanical microenvironment in vitro, hydrogels or elastomers are widely used to create substrates with varying stiffness by adjusting the degree of cross-linking or component ratio^{3,10,11,12,13}. However, these methods often result in discrete stiffness changes that fail to reproduce the continuous, microscale stiffness gradients that accurately mimic physiological conditions^{13,14,15,16,17}. Recent studies have explored techniques such as optical cross-linking (e.g., using a photomask to regulate UV exposure)¹⁸, microfluidic multi-channel mixing^19,20, and diffusion-driven means²¹ to construct stiffness gradients. Unfortunately, these approaches can be technically complex and introduce additional variables, such as mesh size²², swelling^23,24, and topography²⁵. Introduction of disturbance factors complicates the interpretation of cellular responses to stiffness. A bilayer PDMS system, utilizing micropillar arrays or checkerboard patterns as a support structure, has been adopted to create cell culture substrates with stiffness gradients^26,27. These approaches rely on discrete underlying supports beneath a chemically homogeneous membrane to achieve isotropic stiffness modulation, providing valuable platforms for investigating cell adhesion and spreading. However, the native cellular microenvironment in vivo is often highly anisotropic in its mechanical properties^28,29. This directional mechanical signaling is postulated to profoundly influence fundamental biological events. Therefore, current biofabrication strategies are limited in their ability to replicate the subtle cell-stiffness interactions observed in vivo, which depend on accurately mimicking microscale stiffness gradients and spatial heterogeneity within cell-loaded substrates.

Previous research has demonstrated that cells can perceive the effective stiffness of rigid objects even when they are not in direct cellular contact³⁰. Based on this study, we introduce a novel bilayer polydimethylsiloxane (PDMS) cell culture substrate that generates controlled stiffness gradients at spatial scales ranging from 5 to 50 µm, aligning with the size range relevant to subcellular and single-cell mechanosensing. The substrate consists of two distinct layers: a rigid PDMS underlayer featuring tunable geometric microstructures and a uniform, flat soft PDMS overlayer that serves as the cell-contacting surface. This design allows for mechanical modulation solely through variations in substructure geometry, while maintaining consistent surface chemistry and topography. Consequently, it effectively decouples the stiffness signal from other physical and chemical cues, offering a more biologically relevant system for stiffness modulation.

In contrast to traditional hydrogel-based substrates, our approach enables single-variable mechanical studies without interference from swelling, mesh size, or surface topology. The substrate is straightforward to fabricate using standard soft lithography and is highly reproducible³¹. It is also compatible with high-resolution microscopy, which makes it particularly well-suited for live-time imaging and quantitative studies of cellular responses to microscale stiffness cues. This manuscript presents the complete fabrication protocol, details the parameter control strategy, and demonstrates the substrate's utility with representative cell culture experiments. Ultimately, this approach provides a robust and accessible in vitro tool for exploring cell responses to a microscale stiffness gradient.

Protocol

Primary mouse bone marrow stromal cells (mBMSCs) were isolated from the tibia and femur of mice (No. WP20230204, approved by the Ethics Committee of Wuhan University Center for Animal Experimentation).

1. Master mold preparation

NOTE: Our protocols are specific to epoxy-based negative photoresist used during this research.

1. Place two clean silicon wafers (50.8 ± 0.2 mm, <100> oriented, 525 ± 25 µm) on the hotplate for a few minutes to dehydrate.
2. Take one silicon wafer from the hotplate and cool it down to room temperature before proceeding with the spin coating. Apply 3-5 mL of photoresist to the wafer center.
3. To obtain a feature height of 20 µm, apply the following spin protocol: spin for 10 s at 7.1 × g, followed by spin at 410 × g for 30 s. To obtain a feature height of 30 µm, apply the following spin protocol: spin for 10 s at 7.1 × g, followed by spin at 170 × g for 30 s. Use a scalpel carefully to remove any edge bead that may form during the spin coating.
4. Soft bake in the following manner: set the hotplate to 95 °C, allow the wafer to sit on hot plate for 13 min. Remove the wafer from the hotplate and allow it to cool to room temperature naturally, avoiding a sudden temperature drop.
5. Align the photomask on the photoresist-coated wafer and expose to UV light with a total dose of 120 mJ/cm².
6. Postexposure bake in the following manner: using two hotplates (65 °C and 95 °C), allow the photoresist-coated wafer to sit on the 65 °C hotplate for 1 min before transferring it to the 95 °C hotplate and leaving it there for 5 min. Move the wafer onto a stack of microfiber tissues and allow it to cool down to room temperature.
7. Develop the wafer by transferring it to a Petri dish filled with PGMEA, and submerge the entire wafer in the developer. Accelerate the development by manual agitation.
   NOTE: PGMEA is flammable and irritant to the eyes and respiratory tract. Dispose of it as "flammable organic waste" in dedicated containers.
8. After development, take out the wafer from the developer and wash it in isopropanol. If a milky white flow forms in the isopropanol close to the sample surface (the development is not complete), immerse the sample back in the developer again to complete the development process. Wash the developed sample in isopropanol until no milky white flow forms close to the sample surface. Once fully developed, hard bake the photoresist by placing the wafer on a 150 °C hotplate for 15 min.
9. Fluorinate the wafer for the following soft lithography processes: expose the wafer to oxygen plasma under vacuum for 30 s (100 W) to activate the surface, and then place it in a vacuum desiccator. Drop 30 µL of trichloro(1H,1H,2H,2H-tridecafluoro-n-octyl) silane onto a glass slide placed inside the vacuum desiccator. Evacuate the desiccator for 2 min and maintain the vacuum condition for 30 min. Bake the wafer at 90 °C for 1 h to complete the silanization process.
   NOTE: Strictly control the ambient relative humidity below 60% throughout the process. Excessive moisture can induce quick hydrolysis of the silane, which generates white non-volatile polysiloxane precipitates. The precipitates will impair the uniformity of silane vapor distribution, resulting in a failed silanization. Carry out the treatment in a hood as trichloro(1H,1H,2H,2H-tridecafluoro-n-octyl) silane reacts violently with water, releasing toxic HCl fumes. Neutralize with dilute alkali before disposal as fluorinated/chlorinated organic waste. The silanization process achieves a 90% success rate.

2. Fabrication

NOTE: The soft lithography process used to fabricate the bilayer polydimethylsiloxane (PDMS) substrate can be divided into two distinct steps: 1) The preparation of both the structured underlayer and the top layer, 2) the bonding of the two PDMS layers. The overall manufacturing yield is over 80%.

1. PDMS preparation
   1. For the structured underlayer, prepare the PDMS precursor by mixing the prepolymer and curing agent in a ratio of 10:1 by weight, followed by degassing under vacuum. Pour 1.5 mL of the PDMS precursor (10:1) onto the SU8 master mold, covering the entire surface, followed by degassing under vacuum and resting for 30 min. Carefully move it to an oven and cure at 90 °C for 1 h. Demold the structured PDMS underlayer from the master mold.
   2. For the top layer, prepare the PDMS precursor by mixing the prepolymer and curing agent in a ratio of 30:1 by weight and degassing under vacuum. Pour the PDMS precursor on a fluorinated silicon wafer. To obtain a thickness of 9 µm, apply the following spin coating protocol: spin at 7.1 × g for 10 s, followed by a spin at 2,300 × g for 60 s.
   3. After spin coating, visually inspect the PDMS film under ambient light to check for uniformity -- specifically examining for any visible defects such as streaks, bubbles, or uneven edge buildup. If the film appears uniformly transparent with consistent reflectivity across the entire wafer surface, proceed to the next step; if not, repeat the spin coating process with a new wafer. Place the coated wafer on a level surface to ensure a homogeneous PDMS film thickness. Carefully move it to an oven and cure at 90 °C for 3 h.
      ​NOTE: Films with a thickness of 9.11 ± 0.34 µm were obtained.
2. Bonding
   1. Place the structured PDMS underlayer (10:1) with patterned side facing down onto the PDMS-top-layer film, ensuring good contact between the two layers without bubbles at the interface.
   2. Move the assembly into the oven and cure at 90 °C for 3 h.
   3. Allow the assembly to cool down to room temperature on a stack of microfiber tissues. Detach the final double-layer structure from the wafer, and store in a Petri dish for further use.
      NOTE: If bubbles form between the top layer and the underlayer during peeling, it typically indicates incomplete adhesion between the two layers. Ensure thorough degassing of the PDMS precursor before spin-coating, and verify that the film surface is free of dust or contaminants prior to the application of the PDMS underlayer for full contact. Ensure the silicon wafer is level during spin-coating, assembly, and curing. Film tearing may occur due to the uneven silanization of the wafer or non-uniform PDMS thickness. To solve the problem, strictly control the environmental humidity during silanization under 60% RH. The silanization process achieves a 90% success rate.

3. Characterization

1. Scanning electron microscopy (SEM)
   1. Cut the sample with a sharp blade to obtain clean edges.
   2. Mount the samples on 90° and 0° SEM stages using conductive carbon tape to enable cross-sectional and surface characterization, respectively.
   3. Sputter coat the samples with gold for 120 s to enhance SEM imaging quality and collect SEM images of the samples.
2. Atomic Force Microscopy (AFM)
   NOTE: Select ScanAsyst-Air probe for quantitative nanomechanical mapping of modulus.
   1. Position the AFM cantilever tip at the center of the view field. Position the laser spot onto the very end of the cantilever by aligning the optical lever system.
   2. Prior to PeakForce Quantitative Nanomechanical Mapping (PeakForce QNM) measurements, calibrate the AFM cantilever to determine the following essential parameters: Cantilever Spring Constant: 0.40135 N/m; Deflection Sensitivity: 28.993 nm/V (determined by performing a force curve on a rigid reference sample); Tip Radius: ~21 nm (calibrated using the relative method against a reference sample with known modulus).
   3. Within the Scan panel of the Scan Parameters List, configure the following initial scan settings: Set the Scan Size to 70 µm for samples featuring periodic stripes with 50 µm spacing. Set the Scan Size to 50 µm for samples featuring periodic stripes with 20 µm spacing. Set Samples/Line and Lines to 256 (resulting in a 256 x 256 pixel image).
   4. Open Engage Settings in the Microscope and set the Instrument's default Peak Force Engage Setpoint in Engage Parameters.
   5. Click the Engage icon on the Workflow Toolbar and continue to monitor the real-time signals during the automated approach.
   6. Ensure ScanAsyst Auto Control is set to Individual mode. Adjust the PeakForce Setpoint to an appropriate value to achieve the desired Deformation range: for the top layer, PeakForce Setpoint is 0.020 nN; for the structured underlayer, PeakForce Setpoint is 5 nN.
   7. Initiate the scan by clicking the top to bottom or bottom to top buttons.
   8. Force Curve Monitoring and Correction: If force curves appear distorted or synchronization issues are observed (e.g., irregular shapes, inconsistent baselines), right-click within the Force Monitor window and select Auto Config to automatically correct the force curve alignment and sensitivity.
   9. Within the modulus calculation channel settings, select DMT model to fit the retraction portion of the force curve for the calculation of the elastic modulus.
   10. Click the capture button to acquire and save the high-resolution topography and nanomechanical property map.
   11. Perform post-acquisition analysis using the AFM image processing and analysis software. Open the saved AFM data files and extract modulus values by selecting regions across the sample surface; calculate average modulus, standard deviation, and distribution profiles for both the top layer and underlayer using the software's built-in statistical analysis tools. Export processed modulus maps and quantitative data.
   12. Number of AFM scans per sample: Perform three independent full-area scans for each sample. Ensure each scan covers a region matching the sample's structural feature size, 70 µm × 70 µm for samples with 50 µm-spaced periodic stripes, and 50 µm × 50 µm for those with 20 µm-spaced stripes. For modulus measurements of the underlying structure (10:1 PDMS) and the top layer film (30:1 PDMS), implement the following sampling strategy to ensure data reliability.
       1. Top layer film (30:1 PDMS): measure the modulus 2x independently with each measurement consisting of two non-overlapping 5 µm × 5 µm scans.
       2. Underlying structure (10:1 PDMS): Given its two distinct periodic stripe configurations (20 µm and 50 µm spacing), measure the modulus 2x independently for each configuration.

4. In vitro cell culture experiment

NOTE: PDMS is a hydrophobic material that needs to be coated to promote cell adhesion before cell culture, and the appropriate coating method should be selected according to the cell type. Ethical approval must be obtained if primary cells are used for the cellular experiment.

1. Poly-D-lysine (PDL) coating
   1. Sterilize the substrate by alcohol immersion for 10 min; then wash for 3 x 5 min with PBS.
   2. Dissolve PDL (molecular weight 150,000 to 300,000 Da) using PBS at a concentration of 0.1 mg/mL. For the 24-well plate size substrate, add 1 mL of PDL working solution and place it in a 37 °C incubator for 4 h to fully coat the substrate surface.
   3. After the coating is completed, remove the PDL working solution, and remove any residual PDL by washing 3x with PBS. Use the substrate for subsequent cell experiments. If the coating is not used immediately, dry it and store at 4 °C.
      NOTE: A successful coating presented a homogeneous, featureless surface under phase-contrast microscopy, with no visible crystalline precipitates or drying artifacts. Substrates with any signs of crystallization or non-uniform coating were discarded.
2. Isolation and culture of primary mouse bone marrow stromal cells (mBMSCs)
   1. Isolate primary mouse bone marrow stromal cells (mBMSCs) from the tibia and femur of mice.
   2. Culture the cells in a growth medium consisting of α-MEM supplemented with 10% fetal bovine serum (FBS), 1% antibiotic-antimycotic, and 1% sodium pyruvate.
   3. Seed cells at passage 3 onto the substrates obtained in step 4.1.3 at a density of 5,000 cells/cm² to minimize potential confounding effects from cell-cell contact.
3. Immunofluorescence staining and imaging
   1. Following 16 h of culture (37 °C, 5% CO₂), fix the cells with 4% paraformaldehyde for 10 min at room temperature and wash 3x with PBS. Then, permeabilize the cells with 0.1% Triton X-100 for 15 min at room temperature and wash 3x with PBS.
   2. For cytoskeletal visualization, stain the F-actin using phalloidin (1:500 dilution) for 20 min at room temperature, and wash 3x with ample PBS to remove any residual dye.
   3. Mount the substrate with simultaneous staining of cell nuclei. Apply a drop of glycerin with DAPI to cover the cells. Thereafter, carefully place a coverslip on top of the substrate and seal the margins using nail polish. Avoid bubbles as much as possible.
   4. Acquire fluorescence images capturing cellular morphology and nuclear positioning using an inverted fluorescence microscope.

Results

Photolithography was used to fabricate reusable silicon master molds featuring parallel ridge-and-groove microstructures. Two designs were created, featuring groove width (ridge width is identical to groove width) of 20 and 50 μm, respectively. Each patterned area measured 7 mm × 7 mm. The precision and reusability of the silicon molds significantly simplified the process of repeated substrate fabrication. The master mold was then used to replicate the microstructures in PDMS. A hard PDMS formulation was molded against the silicon to create the underlayer, which was subsequently bonded with a thin, soft, and flat PDMS film on top, resulting in the bilayer system.

Figure 1 illustrates the fabricated bilayer substrate, which includes a central square region containing the stiffness gradient, surrounded by flat areas that serve as stiffness controls. SEM images confirmed the fidelity of structure transfer and revealed a flat, topographically featureless upper surface in the gradient region (Figure 2). This design ensures that cells cultured on the substrates experience only stiffness variations, without interference from topographical cues.

To validate the mechanical properties of the system, AFM was conducted to measure the elastic moduli of both layers independently. The soft top layer exhibited a modulus of 899 ± 50 kPa, while the rigid underlayer reached 2,289 ± 693 kPa, confirming a substantial stiffness difference between the two layers (Figure 3). Further AFM indentation tests on the assembled bilayer structure demonstrated that the effective modulus varied according to the underlying microstructure. In the 20 µm design, the effective modulus measured approximately 950 kPa over ridges and 850 kPa over grooves. In the 50 µm design, these values increased to 1,070 kPa (ridges) and 950 kPa (grooves), respectively (Figure 4). These results indicate that the stiffness gradient can be precisely tuned by modulating substructure dimensions, while maintaining a uniform top surface. This confirms the successful establishment of a robust and reproducible fabrication protocol for a bilayer PDMS substrate featuring a microstructured stiff base layer and a thin, flat soft upper film.

Next, primary mBMSCs isolated from mouse long bones were seeded onto the patterned substrates at a density of 5,000 cells/cm² and cultured under standard conditions (37 °C, 5% CO₂) for 16 h. After fixation and permeabilization, immunofluorescent staining was performed to visualize cell morphology and nuclear position. Fluorescent microscopy confirmed that the cells adhered well and exhibited spreading across both types of microstructured substrates (Figure 5). Experiments were performed with cells derived from a pool of at least three mice. Data were collected from three independent substrates (biological replicates), with four random fields (1481.48 μm × 1111.11 μm) of view imaged per substrate. Approximately 100 cells were quantified per field of view. On the 50 µm patterned substrates, a spatial preference of nuclei positioning was observed: 61.49% of nuclei were located over the ridge regions (stiffer support), while 38.51% were located over the corresponding groove areas, suggesting that mBMSCs are able to sense and respond to underlying stiffness variations transmitted through the flat soft top layer.

To assess the effect of our substrate on cell migration, we tracked the trajectories of individual cells over time. As shown in Figure 5C, within the same time frame, the cells on the 50 µm patterned substrate (n = 19) showed a substantially longer trajectories than the ones on the control (n = 17). Such observations suggest that cells cultured on the micropatterned substrates showed enhanced motility.

In conclusion, cells seeded on the substrates exhibit a higher motility and a nuclear positioning preference rather than full directional migration.

Figure 1: Fabrication Scheme of bilayer PDMS substrate. (A) Flow chart illustrating the soft lithography process. (B) Schematic representation of the bilayer substrate for cell culture experiment. (C) Overview of the substrate, highlighting the central square region with a stiffness gradient and the surrounding flat areas that serve as stiffness controls. Please click here to view a larger version of this figure.

Figure 2: Surface topography of the substrates. (A-D) Scanning electron microscopic images of the substrate patterned with (A,B) 20 µm and (C,D) 50 µm features. Scale bars = 100 µm (A,C), 10 µm (B,D). Please click here to view a larger version of this figure.

Figure 3: Elastic moduli of the top layer and the structured underlayer in the substrate. At a 10:1 ratio, the elastic modulus E = 2289 ± 693 kPa, while at a 30:1 ratio, E = 899 ± 50 kPa. Please click here to view a larger version of this figure.

Figure 4: Effective moduli of the substrate. Stiffness map of the substrate patterned with (A,B) 20 µm and (C,D) 50 µm features, measured by Atomic Force Microscopy. Please click here to view a larger version of this figure.

Figure 5: Cellular response to different patterned substrates. (A) Fluorescence microscopic image of fixed cells at the flat control (left), 20 µm patterned substrate (middle), and 50 µm patterned substrate (right). Cell nuclei were stained blue (DAPI), and the cytoskeleton was stained red (phalloidin). Scale bars = 200 µm. (B) Quantification of nuclear position preference on the 5  µm patterned substrate, presented as average mean ± SD. (C) Tracking trajectories of single cells migrating the flat control (left) and 50 µm patterned substrate (right). All cells are shifted to a (0, 0) origin for visual comparison. Please click here to view a larger version of this figure.

Discussion

This study presents a novel approach to engineer a bilayer PDMS-based substrate designed to introduce localized stiffness variations while maintaining a topographically uniform featureless surface. This approach allows for mechanical modulation solely through variations in substructure geometry, while maintaining consistent surface chemistry and topography. Consequently, it effectively decouples the stiffness signal from other physical and chemical cues, offering a more biologically relevant system for stiffness modulation.

In this study, stiffness modulation is achieved by adjusting the geometry of the stiff bottom PDMS layer, and a thin, softer PDMS layer on top ensures a flat contact interface, which ensures broad compatibility with various cell types and imaging systems. Compared to conventional methods that rely on fully crosslinked hydrogels or bulk PDMS substrates with discrete stiffness levels, this approach provides higher spatial resolution and continuity. It allows for multiple stiffness regions on a single substrate, improving experimental comparability and minimizing batch variability. This substrate enables precise tuning of stiffness across a range that effectively encompasses the mechanical properties of granulation tissue (~0.99 MPa) during early bone healing and the stiffness difference between normal and scarred skin^32,33. Therefore, it is highly suitable for investigating stiffness-dependent cellular behaviors in these contexts. However, due to the intrinsic modulus of the PDMS³⁴, the base material, our system is limited in studying the extremely soft microenvironments, such as those of neural tissues or other soft parenchyma³⁵.

One of the most critical fabrication challenges is the removal of the spin-coated PDMS film from the silicon wafer. PDMS tends to strongly adhere to the wafer surface after curing, complicating the detachment of thin films without causing tearing, wrinkling, or leaving residues. To address this issue, we explored various surface treatment strategies. Sacrificial layers, while effective, require reapplication before each PDMS casting and can involve toxic chemicals or solvents that may cause the polymer to swell³⁶. Alternatively, anti-adhesion strategies can be used to reduce PDMS bonding to the substrate. Among these, silanization has been widely applied to modify surface properties, such as reducing surface energy and increasing hydrophobicity, thereby minimizing PDMS adhesion. For example, trichloromethylsilane has been used to prevent bonding between ultrathin PDMS films and silicon molds³⁷. In this study, we selected vapor-phase silanization using trichloro(1H,1H,2H,2H-tridecafluoro-n-octyl)silane as the optimal anti-adhesion treatment. This method forms smooth, uniform monolayers, reduces the need for silane purification, and offers improved reproducibility and surface control compared to solution-phase silanization^38,39.

Another parameter requiring optimization is the thickness of the top PDMS layer. If too thick, it dampens the mechanical effect transmitted from the underlying microstructure; if too thin, it becomes susceptible to wrinkling or tearing. Our protocol employs spin coating to control thickness, but the minimum achievable film thickness is limited by the maximum speed of the spin coater and the viscosity of the PDMS solution. Previous work using a similar bilayer structure⁴⁰ has demonstrated successful peeling of ultrathin ~2 µm PDMS membranes at a 60:1 ratio, but that study was conducted in a different research setting, with access to optimized fabrication equipment and an experienced operator. In our laboratory, limitations in available spin-coating equipment and technical training made it challenging to reproduce such thin and soft membranes with acceptable yield and consistency. Consequently, we adopted a more robust 30:1 formulation with a target thickness of ~9 µm to ensure reliable fabrication and downstream handling. This choice balances membrane integrity, manufacturability, and mechanical responsiveness, emphasizing the need for effective surface pretreatment on the silicon substrate to ensure clean and consistent membrane transfer.

Surface treatment is essential for ensuring proper cell adhesion, as PDMS is inherently hydrophobic, and post-curing surfaces are generally non-conducive to cell attachment⁴¹. Standard coating strategies, such as fibronectin, collagen, or poly-D-lysine, were employed to improve cell adhesion^42,43,44. In addition, oxygen plasma treatment can temporarily increase PDMS hydrophilicity, improving protein adsorption and cell attachment efficiency, while minimally altering surface morphology and roughness^45,46. The obtained substrates can be integrated into multiwell plates and are compatible with high-content live-cell imaging systems. To facilitate simultaneous multi-position imaging with minimal z-axis adjustment, the overall height of each substrate must be consistent, reducing focus correction steps and improving temporal resolution during time-lapse acquisition. In our setup, we achieved consistent film thickness by carefully controlling the volume of PDMS mixture dispensed for the bottom layer.

In summary, our approach is well-suited for modeling mechanical transition zones in pathological or developmental contexts, such as granulation tissue during early fracture healing. Combined with fluorescent labeling, live imaging, or traction force microscopy, our approach provides a precise and reproducible tool for dissecting how cells sense and respond to microscale stiffness gradients.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
4% paraformaldehyde	Servicebio Technology	G1101
Acetone	Shanghai Aladdin	10000418
Anti-Anti	Gibco	15240-062	Antibiotic-Antimycotic
Antifade Mounting Medium with DAPI	Beyotime Biotechnology	P0131	glycerin with DAPI
Cell Culture Plate	NEST	702001	24 wells
Desiccator	Sichuan Shubo(Group) Co.,Ltd	913354-30SN	Inner diameter=180mm
Fetal Bovine Serum	HyClone	SV30208.02
Inverted fluorescence microscope	Olympus	DP74
Isopropanol	Shanghai Aladdin	80109218	GC≥99.7%
Mask aligner	Sichuan Nanguang Vacuum Technology Co.,Ltd	H94-37
MEM Alpha modification	HyClone	SH30265.01
Nanoscope Analysis	Bruker	Version 3.0	AFM image processing and analysis software.
Opera Phenix Plus High Content Imaging System	Perkin Elmer	2400L23248	Equipped with 10x air objective
PGMEA	Suzhou NanoMicro Technology Co.,Ltd	171011-1	Developer Negative Photoresist; SPEC: UL; Percent:min99.7%
Phalloidin	Uelandy	YP0052L	594-Phalloidin
Phosphate Buffered Saline solution	HyClone	SH30256.01
Plasma Cleaner	Plasma technology (Germany)	Flecto 10
Poly-D-lysine	Beyotime Biotechnology	ST508	CAS Number 27964-99-4
Scanning Probe Microscope	Bruker	Dimension Icon	Uses Nanoscope control software, including PeakForce Quantitative Nanomechanical Mapping (PF-QNM)and Point-and-Shoot Ramping experimental workspaces.
Silicon Tip on Nitride Lever	Bruker	SCANASYST-AIR	Probe features a single, V-shaped, Al reflex coated cantilever; k= 0.4 N/m (nominal), f0 = 70 kHz.
Silicon wafers	Shenzhen Rigorous Technology Co.,Ltd	-	50.8±0.3mm diameter, (100)oriented, 525±25 μm thickness
Sodium Pyruvate	Gibco	11360-070
Spin coater	Institute of Microelectronics of the Chinese Academy of Sciences	KW-4A
SU-8 3025	Suzhou NanoMicro Technology Co.,Ltd	-	Negative Photoresist; 72.3% Solids; Viscosity 4400; Density 1.143g/ml
SYLGARD 184 Silicone Elastomer Kit	The Dow Chemical Company	01673921
Trichloro(1H,1H,2H,2H-tridecafluoro-n-octyl)silane	Shanghai Aladdin	78560-45-9
Triton X-100	BioFroxx	1139ML100	0.1% Triton X-100 for permeabilization
Trypsin-EDTA	Gibco	25200-072	0.25% Trypsin-EDTA (1x)
Vacuum Oven	Memmert (Germany)	VO 200