Soft Pneumatic Device Designed to Mimic the Periosteal Environment for Regulating the Fate of Mesenchymal Stem Cells

Abstract

Replicating the complex mechanical forces of muscle movement and fluid flow in in vitro cell culture systems is crucial for understanding cell differentiation and development. However, previous research focused on cell differentiation on static micro/nanotextures without a force field or flat 2‐dimensional substrates under a continuous in‐plane mechanical force. In this study, cell differentiation is reported using a spatial geometric platform that can periodically modulate complex mechanical forces through a custom‐made soft pneumatic device (SPD) to mimic the interfaces between periosteum and interstitial fluid. To elucidate fluidic dynamics and cell fates relevant to bone physiology, the platform exhibited distinct functional responses based on mechanical force levels: low mechanotransduction induced mesenchymal stem/progenitor cells differentiation into osteoprogenitor cells (≈1.5‐fold increase in osteo‐differentiation), while high mechanotransduction resulted in structural disruptions resembling cell detachment without protein degradation (≈2‐fold increase in effective cell detachment). Numerical simulations of SPD elucidated the principal mechanical components for programmable cell differentiation and detachment by deconvoluting the in‐plane and out‐of‐plane mechanical forces of the SPD complex mode. This study offers comprehensive and novel insights into the correlation between mechanical forces and cell differentiation, recovery, and injury in organisms.

This study presents a custom‐made soft pneumatic device (SPD) that mimics complex mechanical forces and fluid flow to enhance in vitro cell differentiation. It shows that low mechanotransduction promotes MSC differentiation into osteoprogenitor cells, while high mechanotransduction causes structural disruptions and cell detachment, providing new insights into mechanobiology.

1. Introduction

Devices designed to induce cell differentiation are garnering attention as effective tools for developing specific cell lineages in a human body‐like environment.^{[ 1 ]} These systems are pivotal in understanding the mechanisms underlying diseases and their regeneration by aiding in regulating the biological characteristics of cell differentiation.^{[ 2} , ^{3 ]} Recent research has resulted in the development of various platforms to induce cell differentiation, which target induced pluripotent stem cells, adult stem cells, and cancer cells.^{[ 4} , ⁵ , ⁶ , ⁷ , ^{8 ]} However, replicating the mechanical forces present near mammalian organs is challenging for cell differentiation devices. Furthermore, understanding the relationship between mechanical loading and the cellular microenvironment is crucial for promoting cell differentiation and gaining insights into movement, recovery, and injury in organisms.

In mammalian movement, muscle expansion and contraction exert shear stress and strain on various cell tissues, along with fluid flow.^{[ 9} , ^{10 ]} While the impact on bones is expected to be minimal, the periosteum, situated between bones and muscles, experiences the movement of interstitial and extracellular fluids within the lacunar–canalicular porosity network inside bones due to eccentric muscle contraction.^{[ 11} , ¹² , ^{13 ]} When the periosteum is stressed, the periosteal fibrotic layer shifts away from or toward the periosteal osteogenic layer, exerting tension to promote bone formation.^{[ 14 ]} Although mild mechanical forces promote bone recovery,^{[ 15 ]} repetitive and excessive loads can lead to stress fractures and microfractures in bones.^{[ 16 ]} These fracture sites undergo bone regeneration through mesenchymal progenitor cells within the cambium layer of the periosteum and their differentiated cells, including osteoblasts or periosteal cells.^{[ 17 ]}

Mammalian mesenchymal stem/progenitor cells (MSCs) can differentiate into various cell types.^{[ 18 ]} Previous studies have shown that shear stress can induce the differentiation of mesenchymal cells into cardiomyocytes.^{[ 19} , ²⁰ , ^{21 ]} Considering the heart is one of the organs exposed to strong mechanical forces, these studies have proposed a close relationship between mechanical forces and cell differentiation. While shear stress has been extensively studied, research on the effects of other mechanical forces on cell differentiation remains limited.^{[ 11} , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ^{24 ]} Specifically, despite the importance of understanding the bone fluidic forces related to the differentiation and depletion of mesenchymal progenitor cells, observing their interaction in vivo is complicated. However, no in vitro systems have been implemented to replicate multiple mechanical forces experienced repetitively during mammalian movement. Additionally, the previous systems were limited to a planar extension and texture‐based microenvironment mimicking systems.^{[ 25} , ²⁶ , ^{27 ]}

In this study, we aim to design an in vitro soft pneumatic device (SPD) under specific conditions (Figure 1 ). First, the proposed device should replicate the osteo‐differentiation process of MSCs within the osteogenic layer of the periosteum, simulating the mechanical forces experienced during mammalian movement. We particularly focus on shear stress and strain resulting from muscle contraction and interstitial fluidics, which act on MSCs attached to rigid bones in the osteogenic layer. Second, we aim to explore how increased mechanical force levels could induce structural disruptions, similar to bone fractures caused by excessive mechanical loading. Our platform is designed to showcase the functional changes in MSCs when subjected to two levels of mechanical force within the same system: low mechanotransduction (LM) induces MSC differentiation into osteoprogenitor cells, while high mechanotransduction (HM) mimics structural disruptions, such as cell detachment. The SPD generates both a 2D in‐plane mode and an 3D out‐of‐plane shear component, resulting in complex mechanical modes at each coordinate. We examine cellular differentiation based on these complex modes using mechanical force simulation. Moreover, compared to conventional methods that degrade proteins during cell detachment, HM detachment preserves the extracellular matrix (ECM) and adhesion‐related proteins. Understanding the effects of complex mechanical forces provides new insights into the bone environment, positioning the SPD as a promising platform for cell therapy, equipped with both differentiation and detachment modes.

Figure 1.

Soft pneumatic device designed to mimic the interstitial fluid dynamics for regulating the fate of mesenchymal progenitor cells. a) Biomechanics of bone and mechanical forces affecting the osteogenic layer of periosteum and its cellular state. b) Spatial mechanotransduction system to mimic the periosteal environment between muscle and bone. c) Cell culture mechanism and potential applications of LM and HM states in the SPD. [Correction added on 24 March 2025 after first online publication: Figure 1 has been replaced.]

2. Results

2.1. Conceptualization of SPD Inspired by Muscle Contraction

Through the SPD system (Figure 1a,b), the interstitial fluid is simulated as cell fluid around cells, and the MSC layer in the periosteum is modeled as the MSC layer attached on top of the SPD. Additionally, the pressure induced by muscle movement is simulated using pneumatic pressure, and the mechanical strength of PDMS was adjusted to mimic the fibrous layer within the periosteum, allowing us to examine the MSCs’ differentiation capacity. Inspired by the perpetual shear stress and strain experienced by the periosteum during muscle contraction, a spatial pneumatic soft robot is fabricated to provide MSCs with a continuous curvilinear stimulus at a specific frequency.

To achieve continuous spatial soft robotic stimulation, flexible PDMS (Sylgard 184, Dow) was prepared and utilized to maintain elasticity without fracturing. The ratio of the precursor and curing agent was adjusted from the typical 10:1 ratio to 15:1 to soften the PDMS (Figure S1, Supporting Information). The Young's modulus was controlled to be ≈450 kPa, which falls between the modulus of mammalian skeletal muscle (5–170 kPa) and periosteum (≈1930 kPa) (Figure S2, Supporting Information).^{[ 28} , ^{29 ]} Assuming the same strain, according to Laplace's Law (𝑃 = 2(𝐸⋅𝜖⋅ℎ)/𝑟)), a soft material with ≈1930 kPa must be supplied with ≈4.29 times the internal pressure compared to one with ≈450 kPa. As a result, it was not feasible to achieve a higher Young's modulus while maintaining the same expansion rate in the SPD system using our pneumatic device. For example, as shown in Figure S2 (Supporting Information), 10:1 PDMS had a Young's modulus of ≈1039 kPa. However, since it was not an optimized condition for applying complex stress, we proceeded with a material that had a lower Young's modulus. Furthermore, according to a previous study,^{[ 30 ]} when the Young's modulus was below 150 kPa (skeletal muscle), chondrogenesis‐related factors such as GAG and collagen were significantly increased. In our SPD system, the softened PDMS exhibited osteogenic‐inducible mechanical stiffness (Figure S3, Supporting Information). During cell culture on the SPD, the culture fluidic and cell culture areas must match the pneumatic stimulation areas. To ensure this, we designed a stainless‐steel frame to prevent unwanted actuation of the SPD by establishing boundary conditions for the soft robot (Figure 2a). While periosteal forces can resemble a linear gradient in natural settings, the circular distribution of equiaxial forces in our system was designed to mimic the dynamic interaction between muscle expansion and periosteal stimulation. Muscle contraction leads to radial expansion, which plays a crucial role in force transmission to the periosteum through the extracellular matrix (ECM) and tendons.^{[ 31} , ³² , ^{33 ]} This radial expansion means that forces generated within the muscle are distributed in a circular pattern around a central point. Therefore, air was injected into the center of the pneumatic soft robot through an internal tube and exited back into the tube due to the elasticity of the PDMS.

Figure 2.

Preparation of soft pneumatic device. a) Pneumatic soft robot system. b) Schematic illustration of the pneumatic soft robot system. c) SPD shape morphing of each stage with cross‐section schemes and side view optical images. d) Optimization of the HM system for cell detachment. Scale bar: 200 µm. ^**** p ≤ 0.0001, compared to LM. (n = 6, unpaired two‐tailed t‐test; mean +/− SD). Mechanobiological stem cell growth state of the e) inlet, f) outlet, g) height profile including experimental and FEM simulation. Detachment of maturated stem cell state of the h) inlet, i) outlet, j) height profile including experimental and FEM simulation. Scale bar: 5 mm.

In the SPD, the amounts of air determines the magnitude of stimulation applied to the MSCs based on the following mechanotransduction stages (Figure 2b). To understand the bone environment influenced by mechanical forces from the stressed fibrotic layer and bone fluidic flow to cells, we established the LM and HM state in SPD. The height profile of the SPD provides the schematic cross‐section view and an optical side view during inflation and deflation in both LM and HM states at each position (Figure 2c; Movie S1, Supporting Information) from the initial stage with nonmechanotransduction (NM) state.

Excessive stress on the bone can lead to microfractures, which can exacerbate soft and bone tissue loss at the site.^{[ 34 ]} As the frequency and angle of muscle stimulation affecting the periosteum can vary based on the type and intensity of the activity, we have focused our research on the changes in bone cells within the periosteal environment. We aimed to develop a design adhering to three fundamental rules of mechanical stimulation of bone^{[ 35 ]}: First, the system is activated by dynamic loading instead of static loading. Second, only a brief period of mechanical loading is needed to trigger an adaptive response. Third, bone cells adapt to a habitual mechanical loading environment, which reduces their sensitivity to regular loading signals. Our system incorporates these principles. Typically, bone recovery at low force takes days to weeks.The adaptive response begins almost immediately after the mechanical stimulus is applied, but the full remodeling process takes longer to complete. Microfractures can occur almost instantaneously with the application of high force. Hence, we optimized the HM state to inhibit cell growth and detach most undifferentiated or differentiated cells from the PDMS surface within 3 min while ensuring that most cells remain well‐attached in the LM state up to 12 days (Figure 2d). By alternating the inflation and deflation processes, perpetual curvilinear actuation is initiated to induce differentiation by providing specific stimuli to MSCs for 0.6 s/cycle in the LM state (Figure 2e–g). In the HM state, increasing the inflation time to enhance strain results in the detachment of the differentiated MSCs due to the increased stimulation for 1 s/cycle (Figure 2h–j). The human fibrous layer, mainly composed of collagen, can safely undergo strain up to ≈25%, beyond which the risk of fractures increases. Therefore, we designed the system to maintain a strain rate below 20% in the LM condition, while the HM condition exhibits a much higher strain rate, ≈40%, to reflect these mechanical differences.^{[ 36 ]}

2.2. SPD Actuation

The operation of the SPD was divided into an inlet process for air injection (Figure 2e,h) and an outlet process for air discharge (Figure 2f,i). The initial surface height of the pneumatic PDMS patch system was defined as 0 in its undeformed state. The time‐resolved simulation results are shown in steps of 0.01 s in LM state and 0.05 s in HM state, with a color line illustrating the surface height change of the SPD corresponding to the time of air injection and discharge at each stage. The experimental results of the surface height change are presented in steps of 0.01 s in LM state and 0.05 s in HM state with a dotted line. The simulation and experimental data exhibit similar trends, with rapid height changes observed at the beginning of each stage when air is injected and discharged. However, the rate of change gradually decreases over time, eventually reaching saturation (Figure 2e,h). This phenomenon occurs when the stress generated during air injection into the SPD and the physical elastic energy of the pneumatic patch reach equilibrium, stabilizing the deformation of the device at a certain level. Correspondingly, there is a rapid decrease in height changes at the initial stage, followed by saturation of the decrement, ultimately returning to the initial stage (height = 0 mm) (Figure 2f,i). In the LM state, the maximum surface height reached 1.9 mm in the experiment and 1.8 mm in the simulation. However, in the HM state, the maximum surface height increased to 4.9 mm in the experiment and 5.3 mm in the simulation.

The height of the central part over time is illustrated, with experimental values represented by dot symbols and simulation data represented by solid lines (Figure 2g,j). The trend shows a rapid increase initially, followed by saturation, and after the inflation period, it rapidly decreases again, returning to the initial state. The time scale appears slightly delayed due to local signal errors within the pneumatic air controller. However, the uniformity of the rising height and the trend of the graph tend to align. In the SPD, the variation in expansion rate determines whether cells can grow or detach (Figure 2). Therefore, we investigated the differences in mechanical forces resulting from the LM and HM states.

2.3. SPD Simulation for Understanding Multiple Mechanical Forces

Following the general relationship between stress and strain, the stress induced on the surface, determined by the degree of air injection, will be proportional to the change in the surface height of the SPD, i.e., the strain. This was validated through simulation of the stress distribution occurring on the surface of the SPD (Figure 3 ). To understand the impact of muscle movement on bone flow fluidics and periosteum expansion, we focused on fluidic shear stress and surface strain (Figure 3a).^{[ 9} , ¹⁰ , ^{14 ]} To explore the relationship between shear stress and strain, we divided the SPD surface into three zones: Center, 1/3R–1/2R and 3/4R–R (Figure 3b). The maximum shear stress was highest in the 1/3R–1/2R zone and lowest at the center (Figure 3b). The average shear stress values showed no significant difference between the 1/3R–1/2R and 3/4R–R zones (Figure 3c). Fluid velocity was observed to be higher in the HM state (Figure 3d), and according to the equation

$$Shearstress=\mu \frac{du}{dy}$$ (1)

when considering shear stress in the direction perpendicular to the patch, it becomes evident that the shear stress value is highest at the 1/3R point, as the center of the SPD is close to 0 and significant movement occurs in the perpendicular direction along its sides. For the SPD strain change rate, excluding fluid, the center of the SPD exhibited the highest values, followed by 3/R, 2/R, and 3/4R (Figure 3e; Figure S4 and Movie S2, Supporting Information). After examining the curves of strain change rate and shear stress, we hypothesized that one or both forces may exert a predominant effect on the cells.

Figure 3.

SPD simulation for multiple mechanical forces. a) Multiple forces derived from the pneumatic soft robot system. b) Simulation data on the distribution of shear stress, with the SPD divided into three zones: center (green line), 1/3R–1/2R (yellow line), and 3/4R–R (red line). c) Single‐cycle simulation of average shear stress. d) Fluid velocity magnitude and vector during SPD expansion. e) Simulation data on the strain change rate for LM and HM states over time. f) Correlation between strain and shear stress at each zone. g) Cell morphology after 12 d of culture in the LM state of the SPD. Scale bar: 500 µm. h) Average cell area at each zone. i) Correlation between the number of nuclei and cell areas at each zone.

2.4. MSCs Differentiation Through Multiple Mechanical Forces

We investigated the morphology of cells in each zone after 12 d in the LM state (Figure 3g–I; Figures S5 and S6, Supporting Information). The durability of the SPD was confirmed after 12 d (Figure S7, Supporting Information). The area occupied by aligned cells was highest in the 1/3R–1/2R zone and lowest at the center (Figure 3g,h). This trend mirrored the graph of average shear stress values (Figure 3c). However, the area occupied by aligned cells was small in the 3/4R zone, where strain was low and average shear stress was similar to 1/3R–1/2R (Figure 3h). Therefore, we anticipated that strain is crucial for cell differentiation. Consequently, we compared cell gene expression using a system that provides continuous shear stress only (Figure S8, Supporting Information). A decrease in bone differentiation marker, OCN, was observed compared to cells cultured in the SPD. Additionally, cell sizes are proportional to the number of cell nuclei, indicating a state similar to aligned osteoblasts rather than polyploidy in a cell (Figure 3i).^{[ 37 ]} Examining the experimental results where cells were grown reveals that both the magnitude of the stress generated on the surface of the SPD and that of the strain affect cell growth and differentiation (Figure S9, Supporting Information). Regarding the mold structure in our system, the lines formed by the 3D printed molds are located on the interior surface, which is necessary for pneumatic expansion and not on the surface where cells adhere (Figure 3g). We maintain the integrity of the cell culture environment, thereby avoiding the potential issues related to mechanical forces, cell adherence, and distribution from the impart roughness.

Next, we analyzed how shear stress and strain impact cellular mechanisms in the LM state (Figure 4a; Figure S10, Supporting Information). Since critical force sensors such as PIEZO1 and PIEZO2 are required in mammalian bone development,^{[ 38 ]} we analyzed the mechanosensitive channel PIEZO1 and PIEZO2 (Figure 4a; Figure S10, Supporting Information). Although PIEZO2 and other ECM‐related genes showed no statistically significant differences, it was confirmed that upregulation of PIEZO1 increased the expression of ECM‐related genes FN1 and bone differentiation marker OPN (Figure 4a). Furthermore, after shPIEZO1 plasmid transfection, FN1 and OPN were significantly decreased (Figure 4a).

Figure 4.

Alteration of cellular fate exposed to the SPD. a) Genetic analysis for MSCs under LM differentiation. ^* p ≤ 0.05, ^** p ≤ 0.01, ^*** p ≤ 0.001, and ^**** p ≤ 0.0001 compared to NT (n = 4, unpaired two‐tailed t‐test; mean +/− SD). b) Schematic illustration of the conventional method (CM) and HM detachment for efficient reattachment. Immunofluorescent staining of c) human–fibronectin (green) and d) COL1 (green) in reattached MSCs after either CM or HM detachment. CM cells seeded five times higher than HM for cell density matching. Scale bar: 250 µm. e) Morphology of cells reattached after either CM or HM detachment after 3 d of culture on the nonmechanically forced SPD (NM). Scale bar: 50 µm. f) Analysis of average cell viability and cell area after CM or HM detachment. ^** p ≤ 0.01 compared to cell area of NT group (n = 3, unpaired two‐tailed t‐test; mean +/− SD). g) Morphology of cells reattached after either CM or HM detachment after 12 d of culture on the NM. Scale bar: 50 µm.

2.5. SPD Application Combining Both LM and HM States

We analyzed the HM state, which was developed for cell detachment (Figure 4b). In terms of shear stress and strain, the values of the HM state were significantly higher than those of the LM state in all zones, facilitating enzyme‐free cell detachment within 3 min (Figure 3b,c,e). We conducted experiments to compare the conventional cell detachment method (CM) with HM state‐based cell detachment after cell culture for a minimum of 3 d up to 12 d. After detaching the cells with CM or the HM state, we analyzed the cell ECM, which can be degraded by conventional trypsin treatment (Figure 4b–d). Despite seeding five times more cells from the CM to match cell density with the HM state, fibronectin and collagen Type 1 expression was significantly degraded in the CM group (Figure 4c,d). Additionally, while cell viability did not differ significantly, the cell area was over twice as large in the HM state compared to the CM group, indicating a significant decrease in the degradation rate of ECM and attachment‐related proteins (Figure 4e–g). We aim to demonstrate not only cell viability but also the improvement in cell adhesion due to the ECM retention achieved by our HM method, which contributes to maintaining a larger cell area (Figure 4f). As ECM plays a pivotal role in bone regeneration, we devised a concept that can combine both the LM state for MSC differentiation and the HM state for ECM‐sustainable cell detachment.

We evaluated the effectiveness of this system in maximizing the efficiency of bone differentiation through LM. The HM group was used alongside the NM and LM groups because we had already demonstrated that the HM method was superior to CM in terms of cell detachment efficiency. Therefore, we incorporated the HM method into the LM and NM groups to test whether the efficient cell detachment observed with HM could also be applied to these groups (Figure 5a). After confirming that differentiated osteoblasts could reattach following HM detachment, we examined the differentiation efficacy under bone differentiation conditions in LM and HM states (Figure 5b–e). Compared to NM, LM significantly increased alkaline phosphatase (ALP) on Day 3 after the operation (Figure 5b). Furthermore, on Day 12, Alizarin red staining showed attachment of differentiated cells after LM, with a significant increase in osteo‐differentiation (Figure 5c). Moreover, on Day 12, immunocytochemistry staining results for OCN and OPN showed that LM increased bone differentiation (Figure 5d,e). The expression of osteo‐differentiation markers significantly increased from Day 3 to 12 (Figure S11, Supporting Information). F‐actin staining revealed that the cells were not integrated but existed individually (Figure 5d,e). In this study, we devised an in vitro system to evaluate and harness the influence of interstitial fluid resulting from movement, guiding the differentiation of MSCs and further application for osteogenesis.

Figure 5.

Combining LM and HM for an effective therapeutic system. a) Schematic of combining LM and HM states for cell culture conditions. b) ALP staining of reattached cells following HM detachment in the NM‐ or LM‐cultured MSCs on Day 3. c) Alizarin red staining of reattached osteoblasts following HM detachment in the NM‐ or LM‐cultured cells on Day 12. Scale bar: 200 µm. ^*** p ≤ 0.0001, compared to NM+HM group. (n = 6, unpaired two‐tailed t‐test; mean +/− SD). d) Immunofluorescent staining for human–OCN (green) in reattached osteoblasts after HM detachment for the NM‐ or LM‐cultured cells on Day 12. Scale bar: 50 µm. e) Immunofluorescent staining for human–OPN (sky blue) in reattached osteoblasts after HM detachment for the NM‐ or LM‐cultured cells on Day 12. Scale bar: 50 µm.

Although cell culture devices are available for differentiating specific cell lineages under an in vivo mimic environment, more is needed to explore the impact of multiple forces periodically resulting from mammalian movement. When the periosteum is stressed, the periosteal fibrotic layer shifts away from or toward the periosteal osteogenic layer. Consequently, the osteogenic layer can be exposed to fluidic shear stress and surface strain. In this study, we successfully mimic the environment of repetitive muscular actuation to stimulate the osteogenic layer of the periosteum for bone regeneration or microfractures through interstitial fluid and surface strain. Consequently, we have newly adopted the spatial stretchable pneumatic system based on the PDMS elastomer to induce the in vitro shear stress and strain for specific cell fate control. In the SPD, specific stages are divided into the LM state for cell differentiation and the HM state for cell detachment through controlled air volume and time. The key findings are as follows: i) Shear stress from the interstitial fluid and strain resulting from the stressed fibrotic layer have been shown to play significant roles during osteoblast differentiation of MSCs within the periosteum. Studies supporting the beneficial effects of appropriate physiological movements on bone regeneration further validate this finding.^{[ 39} , ⁴⁰ , ^{41 ]} Moreover, osteoblast alignment within the periosteum indicates a strong relationship between this orientation and mechanical forces.^{[ 42 ]} Furthermore, numerical simulations were conducted to elucidate the dominant mechanical factors governing programmable cell differentiation and detachment by dissecting the in‐plane and out‐of‐plane mechanical forces of the SPD complex mode. The commercial FLEXCELL system is designed to create 2D equiaxial tension, focusing on shear stress and 2D applications. In our system, z‐axis component is induced for deformation direction enabling a more thorough 3D analysis of the combined effects of strain and stress by repeating inflation and deflation of 3D soft actuator. Moreover, the formation of curvature along the Z‐axis allows for more efficient cell detachment. ii) These movements have been found to increase ECM secretion and osteogenic gene expression. Specifically, fibronectin production by differentiating osteoblasts contributes to bone formation, with fibronectin accumulating at the osteogenic site.^{[ 43 ]} These enhanced cellular mechanisms primarily rely on PIEZO1 rather than PIEZO2, which is dominant in sensory tissues.^{[ 44 ]} Previous in vivo studies substantiating the importance of PIEZO1 for bone formation validate the suitability of the SPD as an in vivo mimic system.^{[ 45} , ⁴⁶ , ^{47 ]} iii) Potent mechanical forces have been confirmed to induce cell detachment originating from microfracture. We applied this phenomenon to the cell detachment method. A significant issue encountered in traditional cell therapy involves the loss of therapeutic efficacy due to ECM depletion during the detachment of osteoblasts cultured in vitro. We applied an enzyme‐free detachment method through the HM state to address this. As the surface strain increases in SPD, the interaction between the cell adhesion proteins and the SPD surface weakens, facilitating effective cell detachment by fluidic shear stress. Since this method can be applied within the same system as the LM state, we propose the detachment of enhanced osteo‐differentiated cells from the LM state to the HM state as a potential therapeutic approach. The significance of the HM condition is related to the maintenance of the ECM after the detachment of cells which is novelty of this study. When detaching cells using conventional methods, significant loss and degradation of the ECM often occur, which lowers the viability of cells in abnormal environments. Notably, ECM components such as collagen type I play a crucial role in bone regeneration, even by their mere presence.^{[ 48 ]} In addition, fibronectin can help the cell adhesion and cell migration in vivo.^{[ 49 ]} iv) We optimized Young's modulus (≈450 kPa) based on the previous papers and our experiments. In the SPD system, since air serves as the output, the shear stress between the fluid and the SPD is also significant correlation to maturate the stem cell. The SPD relies on the elasticity of PDMS for shape recovery when it outputs air. If the modulus is too low, the recovery speed will be slow, resulting in reduced shear stress with the fluid, which may hinder proper stem cell maturation. On the other hand, if the modulus is too high, the recovery speed will be too fast, potentially leading to cell detachment rather than maturation. Therefore, an appropriate modulus was expected to be necessary to balance the modulus for shear stress and detachment. Based on the internal pressure of the device, as well as stress and strain considerations, we selected the optimal condition.

Our system has the following limitations. First, it was challenging to create a system that accurately represents all of these aspects due to the variations in muscle movements and interactions between muscles and bones in different body parts. Additionally, when simulating the periosteum layer between bone and muscle, we constructed it in a 2D state, considering Young's modulus and stiffness suitable for bone differentiation. As 3D structures are more complex than the 2D state, our volumetric system could offer a better understanding of shear stress and strain. However, due to the size constraints of the SPD, obtaining a sufficient number of cells was challenging, making it difficult to conduct animal experiments for bone regeneration. Therefore, we aim to develop a device capable of spatial osteoblast differentiation and conduct animal experiments that entirely use the functionality of the current system by developing larger‐sized devices. In our study, we purposely used oxygen plasma treatment instead of a hydrophilic elastomer, as both adhesion and detachment play a significant role. According to previous studies,^{[ 50 ]} superhydrophilicity gradually transitions to a more hydrophobic state over a period of more than 10 days. Therefore, we leveraged this time‐dependent gradual adhesion changes for securing initial adhesion and final cell detachment without cell damage originating from strong adhesion when using a hydrophilic elastomer. Building upon the SPD system, further advancements toward temporal modulation of hydrophilicity and hydrophobicity would enhance measurement precision.

In conclusion, we have developed a system to understand how mechanical forces derived from muscle contraction contribute to bone regeneration. This approach is anticipated to apply to various tissues due to the modulation of Young's modulus and stiffness that can affect on the cell viability and differentiation.^{[ 51 ]} This study demonstrated the feasibility of investigating the relationship between various mechanical forces occurring within the body and cell fate using soft materials. Additionally, leveraging our understanding of the influence of mechanical forces on cells, we integrated the fields of soft materials and regenerative medicine, presenting novel approaches for cell culture and therapeutic applications. For real‐world applications, we believe that the SPD system could serve as bridge between in vitro stem cell research and clinical practice. This SPD system has potential in bone tissue engineering, orthopedic and dental implants, as well as bone‐related drug screening and disease modeling under controlled mechanotransduction.

3. Experimental Section

Materials for PDMS Patch and Pneumatic System

PDMS (Sylgard 184, Dow) was utilized with a 15:1 weight ratio of precursor to curing agent. The precursor mixture was poured into molds designed with a 3D printer (3DP‐310FB, Cubicon) using polylactic acid filaments. The molds were divided into inflation and substrate parts to obtain free volume inside the patches. Subsequently, the precursor mixture was cured at 60 °C for 4 h and the mold was removed. These replicated PDMS patches were attached to the PDMS precursor mixture and fully cured again at 60 °C for 4 h to obtain the final products (Figure S12, Supporting Information). Before the cell maturation experiment, the surface of the PDMS was cleaned. The dimensions of the mold, PDMS patch, and stainless‐steel frame are indicated in Figure S12 (Supporting Information). A pneumatic controller system was manually constructed, incorporating motors and Arduino chips. In the LM stage, the air injection time was regulated to 0.1 s with Motor A, which delivered air at a rate of 0.3 L min⁻¹, and maintained the deflation time at 0.5 s. In the HM stage, the air injection time was adjusted to 0.5 s while maintaining a 0.5 s deflation time. The experimental height profile of the SPD was measured by converting the grayscale of the image into height every 0.01 s.

Fluid‐Structure FEM Simulations

The numerical modelling of the PDMS patch was established to analyze the morphological deformation of SPD and the accompanying stress and strain distribution according to the time. COMSOL Multiphysics v6.0 (COMSOL AB, Sweden) was used to describe the deformable PDMS patch which is capable of performing lamina flow and solid mechanics modules with fluid‐structure interaction (FSI) Mutiphysics interface module. The geometry of the SPD was designed as a 2D axisymmetric model, including the fluid chamber and PDMS patch. The fluid flow in the chamber was considered laminar flow, governed by the incompressible Navier–Stokes equations. And the displacement and deformation of the patch follow the governing equations of linear elastodynamics. All analyzes were performed with the fully coupled solid mechanics and lamina flow modules with FSI application, and SPD was implemented by applying the two parameter Mooney‐Rivlin hyperelastic model. The parameters C10 and C01 used in the two parameter Mooney‐Rivlin hyperelastic model, which is used as the boundary load of the patch, are 0.0013 and 0.0878 MPa respectively. Including these two parameters, the physical parameters of the PDMS and liquid in the chamber used in the numerical analysis were all data from the COMSOL material library. In the experiment, the area of SPD fixed by the stainless‐steel frame was approximated by setting it as a fixed constraint with no‐slip boundary conditions to simulate the pneumatic actuation, and a time‐dependent solver was used to analyze the phenomenon of air being injected and released again according to the time.

The mesh design parameters for each applied geometry domain are presented in the table below, and the optimal parameter conditions for the simulation solution to be converged were inductively derived.

	Mesh type	Maximum element size (cm)	Minimum elements size (cm)	Maximum element growth rate	Curvature factor	Resolution of narrow regions
	Tetrahedral	0.903	0.067	1.5	0.6	0.5
	Tetrahedral	0.406	0.0105	1.35	0.3	0.85
	Tetrahedral	0.08	0.00141	1.3	0.2	1

	Options	Parameter	Value
	Corner refinement	Minimum angle between boundaries	240 deg
		Element size scaling factor	0.35
	Boundary layers	Handling of sharp edges	Trimming
		Minimum angle for trimming	240 deg
		Maximum angle for trimming	50 deg
		Maximum layer decrement	2
		Number of layers	2
		Stretching factor	1.2
		Thickness adjustment factor	5
	Moving mesh	Mesh smoothing type	Yeoh
		Stiffening factor	10

Second, the modeling was constructed based on the basic formulas applied COMSOL Multiphysics. The governing equations for each applied laminar flow module and solid mechanics module, and the fluid‐structure interaction coupling equation is as follows.

Laminar flow module	${\displaystyle \frac{\partial u_{fluid}}{\partial t}}+\rho (u_{fluid}·\nabla )u_{fluid}=\nabla ·[-\rho I+K]+F+\rho g$ ρ∇ · ufluid = 0
Solid mechanics module	$\rho {\displaystyle \frac{{\partial }^2{u}_{solid}}{\partial t^2}}=\nabla ·{(FS)}^T+F_v$ F = I + ∇usolid
Fluid‐structure interaction	FA = [−ρI + K] · n $u_{tr}={\displaystyle \frac{\partial u_{solid}}{\partial t}}$

For the liquid and interfaces interaction, the physics interface employs an arbitrary Lagrangian‐Eulerian (ALE) method to integrate the fluid flow, which is formulated using an Eulerian description and deformed frame, with solid mechanics, which is formulated using a Lagrangian description and the PDMS membrane. The fluid flow in the chamber was considered laminar flow, governed by the incompressible Navier–Stokes and mass continuity equations.

$${\rho }_l\frac{D\vec{V}}{Dt}={\rho }_l\vec{g}+\mu {\nabla }^2\vec{V}-\nabla P$$ (2)

$$\frac{\partial {\rho }_l}{\partial t}+\left(\vec{V}·\nabla \right){\rho }_l=0$$ (3)

where, ρ _l is the density of the liquid, V is the velocity field, P is the pressure, and μ is the viscosity of the liquid. From the provided solution for $\vec{V}$ using above equation, the total force exerted by the fluid on the solid boundary is negative to the reaction force on the fluid, as shown below.

$$\vec{f}=\vec{n}·\left[-P\vec{I}+\left(\mu \left(\nabla \vec{V}+{\left(\nabla \vec{V}\right)}^T\right)-\frac{2}{3}\mu \left(\nabla ·\vec{V}\right)\vec{I}\right)\right]$$ (4)

where, $\vec{n}$ is the outward normal to the boundary, and $\vec{I}$ denotes the identity matrix. And the transformation of the force is

$$\vec{F}=\vec{f}·\frac{ds}{dS}$$ (5)

where ds and dS are the mesh element scale factors for the deformed and PDMS membrane frame, respectively. The structural velocity (∂v_solid /∂t) which represents a moving wall for the fluid domain were equal to to zero owing to the setting as the coupling type is the fluid loading on the structure. From these equations in COMSOL Multiphysics, therefore, the mesh within the fluid domains is free to move and adapts to the motion of the solid walls, and this geometric change of the fluid domain is automatically accounted for by using the Arbitrary Lagrangian‐Eulerian (ALE) method.

Cell Culture

Human mesenchymal stem cells (hMSCs) were purchased from Lonza (Basel, Switzerland) and cultured in Dulbecco's modified Eagle's medium (Gibco BRL) supplemented with 10% (v/v) fetal bovine serum (Gibco BRL) and 1% (v/v) penicillin/streptomycin (Gibco BRL). The cells were incubated at 37 °C with 5% CO₂ saturation. The medium was changed every 2 d, and cells within eight passages were used for the experiments. For osteogenesis, hMSCs were incubated in an MSC osteogenic differentiation medium (promo cell) with or without the SPD. For the shPIEZO1 MSC group, hMSCs were transfected with PIEZO1 shRNA lentivirus (Santa Cruz) using Polybrene (5 ug mL⁻¹)/media. Cell adhesion on PDMS was enhanced through O2 plasma treatment, which significantly improves the surface hydrophilicity. By exposing the treated PDMS to the cell medium for ≈10 min, high hydrophilicity was maintained, promoting better cell attachment.

Immunohistochemistry Staining

For immunohistochemical staining, cells were detached using the HM state and reattached to the dish. After fixing the cells with PFA for 10 min, OPN (Abcam, ab8448), OCN (Abcam, ab93876), fibronectin (Abcam, ab2413), and COL1 (Abcam, ab308455) antibodies were used. Positive signals were visualized using fluorescein‐isothiocyanate‐conjugated secondary antibodies (Jackson Immuno Research Laboratories, West Grove, PA). The cells were counterstained with DAPI and examined using fluorescence microscopy (IX71, Olympus, Tokyo, Japan).

qRT‐PCR

qRT‐PCR was employed to quantify the relative gene expression levels of OCN, OPN, PIEZO1, PIEZO2, FN1, RUNX2, CX43, SOX9, and PPARG. Total ribonucleic acid (RNA) was extracted from samples (105 cells per sample) using 1 mL TRIzol reagent (Life Technologies, Inc., Carlsbad, CA, USA) and 200 µL chloroform. After centrifugation at 12 000 rpm for 10 min at 4 °C, RNA pellets were washed with 75% (v/v) ethanol in water, dried, and dissolved in RNase‐free water. For qRT‐PCR, the SsoAdvanced Universal SYBR Green Supermix kit (Bio‐Rad, Hercules, CA, USA) and the CFX Connect real‐time PCR detection system (Bio‐Rad) were used according to the manufacturer's instructions.

DiI Staining

Cellular membrane and cell adhesion were assessed using DiI (Sigma–Aldrich) staining. Cells detached using HM or CM methods were reattached cells to dishes for 1 d, followed by treatment with a DiI solution (6.25 µm) and incubation for 30 min at 37 °C. After washing twice with PBS, cells were fixed with a 4% paraformaldehyde solution for 10 min, washed in PBS, and stained with 4′,6‐diamidino‐2‐phenylindole (DAPI, Vector Laboratories, Burlingame, USA), DiI fluorescence was measured using a fluorescence microscope (IX71, Olympus, Tokyo, Japan).

Alizarin Red Staining

Following differentiation in an osteogenic medium for 3 d under either NM or LM conditions, hMSCs that were reattached following CM or HM treatment were stabilized using 4% paraformaldehyde for 15 min at ambient temperature. Subsequently, they were dyed with a 0.2% alizarin red S solution for 30 min and rinsed thrice with PBS, enabling the observation of orange–red calcium deposits.

Alkaline Phosphatase Staining

MSCs were cultured under LM or NM conditions for 12 d. Subsequently, the differentiated cells were reattached to dishes under CM or HM states. The cells were then stained using an Alkaline Phosphatase Staining Kit (Abcam), following the manufacturer's instructions.

Statistical Analyses

All statistical analyses were conducted using GraphPad Prism v.8 software (GraphPad Software Inc.). No exclusion criteria were applied for any analysis. All data were presented as mean ± standard deviation of the mean. Comparisons between multiple groups were assessed by ANOVA, followed by Bonferroni's post‐test analysis. An unpaired two‐tailed Student's t‐test was used for comparisons between two groups. Statistical significance was set at p < 0.05.

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

G.‐B.I. and J.G.L. contributed equally to this work. G.‐B.I., J.G.L., J.J.W., and S.H.B. conceived and designed the project. G.‐B.I., J.G.L., H.L., J.‐W.L., H.S.P., Y.K., N.A., and H.‐R. K. conducted the experimental work. H.L. and J.‐W. L. performed the simulation. All authors discussed and analyzed the data and edited the results. G.‐B.I., J.G.L., J.J.W., and S.H.B. wrote the manuscript.

Supporting information

Supporting Information

Supplemental Movie 1

Supplemental Movie 2

Im G.‐B., Lee J. G., Lim H., Lee J.‐W., Park H. S., Kim Y., Asad N., Kim H.‐R., Wie J. J., Bhang S. H., Soft Pneumatic Device Designed to Mimic the Periosteal Environment for Regulating the Fate of Mesenchymal Stem Cells. Adv. Healthcare Mater. 2025, 14, 2403229. 10.1002/adhm.202403229

Data Availability Statement

We can share after accepting.