Applying microfluidic techniques for biomaterial and mechanical regulation of mesenchymal stem cells

Abstract

Mesenchymal stem cells (MSCs) hold great promise for tissue repair and regeneration. The therapeutic efficacy of MSCs depends on the precise regulation of their proliferation and differentiation. In vivo, this process is synergistically modulated by extracellular matrix-derived biochemical cues and mechanical stress-dominated biophysical stimuli. However, traditional two-dimensional culture systems fail to replicate the complex three-dimensional microenvironment. In addition, population-averaged assays are often confounded by cellular heterogeneity and paracrine effects, which notably limits our mechanistic understanding of MSC fate determination. Microfluidic techniques, with high throughput, high accuracy and integration, provide powerful tools to overcome these limitations. In this review, we summarize recent research progress in the use of microfluidic techniques to investigate the mechanisms of MSC regulation. We outline the roles of biomaterials and mechanical stress in MSC-mediated repair processes, and highlight typical applications of microfluidic techniques in cell sorting and heterogeneity analysis, precise loading of biomaterials and mechanical stress, and dynamic phenotypic tracking of MSCs. We then discuss the current challenges and future directions in stem cell therapy, providing a methodological framework for MSC investigation and accelerating clinical translation.

Key Words: Microfluidic techniques; Mesenchymal stem cells; Biomaterials; Mechanical stress; Single-cell analysis; Three-dimensional cell culture

Core Tip: This review organizes microfluidic applications for mesenchymal stem cell research into three sequential stages: Cell quality control to acquire pure subpopulations, microenvironment engineering to replicate the physiological microenvironment, and phenotypic monitoring to track real-time responses. These continuous stages consequently link raw cell preparation to functional readout. To accelerate the clinical adoption of stem cell therapies, we propose that integration, intelligence, and scalable manufacturing are crucial directions for microfluidic systems.

INTRODUCTION

Mesenchymal stem cells (MSCs) can be isolated from tissues like bone marrow, adipose tissue, and umbilical cord. Their multilineage differentiation potential and immunosuppressive effects[1] make them one of the most promising cell sources in regenerative medicine. MSCs have shown considerable therapeutic potential in a range of refractory diseases, including osteoarthritis, myocardial infarction, nerve injury, and graft-vs-host disease[2,3]. After systemic administration, MSCs home to sites of tissue damage and inflammation[4]. Once there, they rapidly respond to local microenvironmental signals, thereby initiating changes in gene expression and cellular phenotype[5]. Stable and precise regulation of MSC proliferation and differentiation is essential for sustained therapeutic efficacy. This process is coordinately regulated in vivo by biochemical and mechanical cues. On the biochemical side, growth factors, such as vascular endothelial growth factor, platelet-derived growth factor and epidermal growth factor (EGF), are essential in MSC regulation[6,7]. Vascular endothelial growth factor/platelet-derived growth factor receptor signaling makes an important contribution to MSC vascularization[6]. Soluble EGF expands MSCs without differentiation, while tethered EGF drives osteogenesis[7]. On the physical side, extracellular matrix (ECM) stiffness and mechanical stress (e.g., fluid shear stress and cyclic stretch) also govern MSC gene expression and phenotype through cytoskeletal reorganization and nuclear mechanotransduction[8-10]. Elucidating these regulatory mechanisms is critical for translating MSC therapy toward clinical use.

However, traditional two-dimensional (2D) culture systems cannot faithfully recreate the native three-dimensional (3D) microenvironment. They lack physiologically relevant cell-ECM interactions that occur in real tissues and fail to model the dynamic mechanical stimulation[11]. Furthermore, most existing studies treat large populations of cells as homogeneous entities, overlooking the confounding effects of cellular heterogeneity. Such shortcomings lead to the loss of critical biological features[12] and make it difficult to determine how specific physicochemical signals regulate MSC proliferation and differentiation. Microfluidic techniques, with their high throughput, precise microscale manipulation, and real-time monitoring, offer a wealth of opportunities to overcome these challenges[13,14]. They enable high-throughput sorting and single-cell capture of MSCs, providing purified subpopulations for single-cell analysis and effectively reducing the confounding influence of heterogeneity[15,16]. Moreover, these techniques can be integrated with 3D hydrogel scaffolds, mechanical loading modules, or real-time imaging systems, allowing in-depth investigation of how biophysical signals shape MSC fate[17,18].

In this review, we systematically outline representative applications of microfluidic techniques in investigating how biomaterials and mechanical stress regulate MSC proliferation and differentiation: (1) Cell quality control: Enabling high-throughput sorting and heterogeneity analysis to obtain purified MSC subpopulations for quality control of raw materials; (2) Microenvironment engineering: Precisely applying biomaterials and mechanical stress to construct biomimetic 3D microenvironments that recapitulate physicochemical signals in vivo; and (3) Phenotypic monitoring: Enabling real-time tracking of proliferation and differentiation to detect cellular responses and elucidate underlying regulatory mechanisms (Figure 1). Finally, we discuss future directions for applying microfluidic techniques in stem cell therapy, aiming to provide a theoretical foundation for understanding the regulation of MSC proliferation and differentiation and thereby facilitate the clinical translation of stem cell therapies.

Figure 1 Schematic overview of microfluidic techniques for investigating the regulation of mesenchymal stem cell fate by biomaterials and mechanical stress. Key applications include: (1) Cell quality control via single-cell sorting and heterogeneity analysis; (2) Microenvironment engineering through hydrogel-based three-dimensional culture and controlled mechanical stress; and (3) Phenotypic monitoring integrated with live-cell imaging. This system enables precise recapitulation of physiological microenvironments and dynamic tracking of mesenchymal stem cell response, supporting mechanistic studies and clinical translation of stem cell therapies (selected illustrations were created with BioRender.com). 3D: Three-dimensional; MSCs: Mesenchymal stem cells.

MICROFLUIDIC TECHNIQUES FOR MSC SORTING AND HETEROGENEITY ANALYSIS

The inherent heterogeneity of MSCs is a major obstacle to their widespread clinical application[19]. Even when MSCs are obtained from the same donor and tissue source, different clones may exhibit significant variations in colony size and multilineage differentiation capacity[20]. Traditional assays that average signals across millions of cells often mask this heterogeneity at the population level, diluting the regulatory effects of biomaterials and mechanical stimuli[21]. Therefore, in order to decipher how these cues regulate MSC proliferation and differentiation, we need to isolate highly purified MSC subpopulations and profile their heterogeneity at single-cell resolution. Microfluidic techniques meet this need by enabling the sorting of MSCs based on physical (e.g., size, density, dielectric characteristics) or biological (e.g., surface markers) properties. They support both label-free and affinity-based isolation (Table 1) and subsequent single-cell analysis[22,23].

Table 1 Microfluidic techniques for sorting and heterogeneity analysis of mesenchymal stem cells.

Category	Method	Isolation principle	Discrimination parameters	Throughput	Advantages	Limitations	Ref.
Label-free sorting	Inertial microfluidics	Inertial lift and Dean drag forces	Cell size, shape	High	Label-free	Limited resolution for subtle biophysical differences	[14,16,25,28,29,31]
	Deterministic lateral displacement	Lateral deflection using micropillar arrays	Cell size, shape, deformability	High	Label-free	Limited resolution for subtle biophysical differences	[26,32,33]
	Dielectrophoresis	Non-uniform electric field	Dielectric properties, cell size	Low to medium	Label-free	Limited resolution for subtle biophysical differences	[27,34,36,37]
	Magnetophoresis	Magnetic field gradient	Magnetic susceptibility	Medium	Label-free	Limited resolution for subtle biophysical differences	[38,39]
	Acoustophoresis	Acoustic radiation forces	Cell density, compressibility, size	Medium	Label-free, gentle on cells	Limited resolution for subtle biophysical differences	[38,40,41]
Affinity-based sorting	Fluorescence-activated cell sorting	Fluorescent labeling & optical detection	Surface markers, fluorescence intensity	Relatively low	High specificity, multiparameter	Requires labeling	[44,47-49]
	Magnetic-activated cell sorting	Magnetic bead affinity	Surface marker expression	Medium	High specificity	Requires labeling	[43,50,51]
	Affinity capture	Surface ligand-receptor adhesion	Ligand binding kinetics	Very low	High specificity	Ligand activity loss	[52-54]

Label-free sorting for MSC heterogeneity analysis

Label-free sorting relies on physical properties and requires no antibody labeling. As a result, it avoids the cost and cell disturbance associated with labeling. This approach has become a mainstream strategy in microfluidic cell sorting[24]. Several techniques have been developed based on different physical properties, including inertial microfluidics[25], deterministic lateral displacement (DLD)[26], and dielectrophoresis (DEP) (Table 1)[27].

Inertial microfluidics: This technique harnesses inertial lift forces and Dean drag forces within spiral microchannels to achieve size-based cell separation[28]. It has been widely applied to fractionate MSC subpopulations, confirming that MSCs of different sizes possess diverse capacities[14,29]. For example, among bone marrow-derived MSCs (BMSCs), cells of intermediate size (17-21 μm) display enhanced proliferative capacity and chondrogenic potential[30]. In contrast, cells that are too large (> 40 μm) or too small (< 10 μm) show significantly reduced differentiation capacity[14,31]. Studies on dental pulp-derived MSCs show a similar pattern: Different size ranges correspond to distinct functional capacities. The subpopulations from 18.6 μm to 20.5 μm exhibit superior immunomodulatory capacity to suppress T cell proliferation and reverse M1 macrophage polarization, providing experimental evidence for the application of MSCs in cartilage regeneration and the treatment of immune-mediated diseases[16].

DLD: This technique leverages microfabricated pillar arrays to deflect larger or more rigid cells laterally, while smaller or more deformable cells continue along the fluid stream, with separation efficiencies exceeding 90%[26,32]. For instance, an inverted L-shaped pillar array enables the separation of MSCs from blood cells in bone marrow aspirates. A multi-chip design can process 2.5 mL of raw bone marrow aspirate within 20 ± 5 minutes, achieving an MSC recovery rate twice that of conventional centrifugation - rendering it highly suitable for clinical sample processing[33]. Moreover, DLD sorters can be cascaded with inertial microfluidics, yielding enriched MSC subpopulations with a six-fold increase in expansion capacity, thereby providing technical support for the scalable production of stem cell products[26].

DEP: When cells are placed in a non-uniform electric field, their dielectric properties result in distinct migration trajectories, thus enabling efficient sorting[27]. Early work by Vykoukal et al[34] employed dielectrophoretic field-flow fractionation to enrich NG2-positive stem cell populations from adipose tissue, with no significant impact on cell viability. DEP systems can also distinguish the electrical properties of MSCs derived from different sources. For instance, although adipose-derived MSCs (ADMSCs) and BMSCs share similar membrane capacitance, they differ significantly in cytoplasm conductivity and transient slope[35]. These biophysical properties correlate with differences in differentiation outcomes between these two cell types[35]. In vivo studies also demonstrated the heterogeneity between ADMSCs and BMSCs. For example, ADMSCs are superior to BMSCs in promoting angiogenesis and resisting hypoxia-induced apoptosis and oxidative stress-mediated senescence[36]. In addition, DEP techniques can continuously sort MSCs and their differentiated progeny, with purities exceeding 80%[37].

In addition to the methods mentioned above, magnetophoresis and acoustophoresis are critical complementary techniques for label-free MSC sorting, as summarized in Table 1[38]. Together, these approaches highlight the intrinsic link between physical phenotypes of MSCs and their functional heterogeneity. For magnetophoresis, different MSC subpopulations exhibit specific magnetic susceptibility, enabling their enrichment in high-gradient magnetic fields[39]. Acoustic radiation forces generated within microchannels allow sorting of MSCs based on cell size, density, and compressibility. Such acoustic sorting further supports that even MSCs derived from the same donor may possess substantial heterogeneity[40,41].

Affinity-based sorting for MSC heterogeneity analysis

The core principle of affinity-based microfluidic sorting is antigen-antibody interactions. In practice, antibodies or other affinity ligands are labeled with detectable tags like fluorophores. When these probes bind to specific receptors on MSCs, the microfluidic chip will then identify and isolate target subpopulations through integrated optical, electrical, or magnetic modules[42,43]. These approaches achieve high specificity and targeting efficiency; thus, researchers can directly correlate the sorting results with MSC heterogeneity - including stemness, differentiation potential, and immunomodulatory capacity[44]. Representative techniques are fluorescence-activated cell sorting (FACS), magnetic-activated cell sorting, and affinity capture, as summarized in Table 1[43,45,46].

FACS is a key component of affinity-based sorting[44]. In a microfluidic chip, it is possible to integrate modules for cell culture, washing, fixation, and antigen-antibody reactions. This integration thereby enables sequential sorting and analysis of MSC surface markers (e.g., CD15, CD34, CD44, CD45, CD73, and HLA-DR). In these procedures, cell viability and sorting purity both exceed 90%[47,48]. Compared with conventional flow cytometry, microfluidic droplet encapsulation and integrated design prevent aerosol generation and sample loss, significantly increasing the throughput of single-cell analysis[49]. In addition to FACS, magnetic bead labeling offers another powerful strategy. It not only amplifies physical properties (e.g., size, density, or stiffness), but also directly separates target cells in magnetic fields[43]. Emerging microfluidic platforms (such as IM-MIS and MagSculptor) can maintain high cell viability while performing multi-step sorting of low marker expression subpopulations. These advances lay the groundwork for subsequent analysis of how biomaterials and mechanical stimuli affect MSCs[50,51]. Another strategy is affinity capture, which functionalizes microchannel surfaces with affinity ligands (e.g., antibodies, peptides, aptamers)[52]. For instance, an E7 peptide-modified collagen substrate has been shown to significantly improve MSC enrichment efficiency within microfluidic chips[53]. Notably, the loss of protein activity during immobilization is the core challenge to overcome in affinity capture[54].

Although microfluidic techniques are considered gentle, high-velocity fluid flow within microchannels still exerts considerable shear forces on cells. Such shear stress can still cause cellular stress and damage, such as structural disruption and abnormal gene expression[55]. These problems can be mitigated by modifying the surface materials, typically polydimethylsiloxane (PDMS), and optimizing the design of microchannels. For example, increased MSC growth has been observed after immobilizing collagen type 1 on PDMS[56]. Multichannel designs and the inclusion of buffer zones can also help achieve more uniform flow rates, which is beneficial for improving cell viability[57]. Overall, despite these challenges, microfluidic techniques continue to hold great promise for obtaining high-purity MSC subpopulations, serving as a critical foundation for subsequent research based on biomaterials and mechanical stress.

MICROFLUIDIC TECHNIQUES FOR 3D MICROENVIRONMENT MODELING

Microfluidic techniques are characterized by microscale and precise controllability. Following the isolation of highly purified MSC subpopulations, they allow researchers to tailor the microenvironment at the single-cell or multi-cell level[58,59]. In this section, we focus on three major applications of microfluidic techniques in 3D microenvironment recreation and MSC regulation: Hydrogel-based 3D biomimetic culture, precise mechanical stimulation, and cell co-culture and organ-on-a-chip. The coupling of biomaterials and mechanical stress is also discussed in this section.

Hydrogel-based 3D biomimetic culture

Hydrogels are hydrophilic polymers with 3D network structures. They possess excellent biocompatibility and can support cell growth. Such properties make them ideal for mimicking the native ECM[60,61]. Hydrogels can be divided into natural hydrogels (e.g., collagen, gelatin, and chitosan) and synthetic hydrogels [e.g., poly(ethylene glycol) and polyacrylamide]. Both categories have been used individually or in combination within microfluidic systems[62,63]. In this subsection, we focus on two miniaturized forms of hydrogels: Microspheres and microfibers (Figure 2) and discuss their applications in 3D MSC culture and for investigating the regulatory mechanisms of MSCs.

Figure 2 Microfluidic fabrication of hydrogel microspheres and microfibers. A: Production of hydrogel microspheres using droplet microfluidics (e.g., T-junction, Y-junction, flow focusing, and co-flow), followed by solidification via UV, thermal, or ionic cross-linking; B: Generation of hydrogel microfibers via coaxial laminar flow (core flow and sheath flow), enabling the formation of ordered, helical, and “bamboo-like” structures for three-dimensional cell culture and tissue engineering. Selected illustrations were created with BioRender.com.

Hydrogel microspheres: Droplet microfluidics represents one of the most prevalent approaches for fabricating hydrogel microspheres, typically utilizing PDMS or capillary microfluidic devices. Driven by shear forces between immiscible phases (such as aqueous and oil phases), uniformly sized droplets are generated through strategies including T-junction, Y-junction, flow focusing, and co-flow[64], and subsequently solidified via UV, thermal, or ionic cross-linking to form microspheres (Figure 2).

Hydrogel microspheres can potentiate MSC differentiation together with growth factors[65-67]. A co-delivery system for transforming growth factor-β1 and dental MSCs was established via microfluidics within arginine-glycine-aspartic acid-conjugated alginate hydrogel microspheres. The synergistic combination effectively enhanced the expression of chondrogenic genes and ectopic cartilage regeneration[68]. When encapsulated together with growth factors in gelatin methacryloyl (GelMA) microspheres, BMSCs exhibited early upregulation of alkaline phosphatase. After four weeks of culture, the calcium deposition rate significantly increased (from 40% to 70%), indicating osteogenic differentiation of BMSCs[69]. These findings provide promising strategies for bone regeneration.

Additionally, optimizing the microsphere structures or functionalizing their surfaces can further enhance culture efficacy. For instance, the pore size of GelMA porous microcarriers can be modulated by the freezing temperature (-20 °C, -60 °C, -196 °C). As a result, GelMA porous microcarriers with a pore size of 25.3 ± 3.2 μm (frozen at -60 °C) showed the highest cell attachment efficiency (90.2%) and the largest cell spreading area, which substantially outperformed the commercial microcarrier Cytodex-1[70]. Manipulating the molecular weight of alginate allows precise tuning of hydrogel microsphere shell stiffness (E = 574 Pa vs 214 Pa). Experiments demonstrated that MSCs encapsulated in hard-shell microspheres displayed improved proliferative activity and longer retention in mice (23 days vs 15 days for soft-shell counterparts)[71]. Concurrently, computational modeling revealed an increase in local tension at the integrin-ligand bonds within hard hydrogels, which promotes binding and inhibits dissociation. This may account for the enhanced cell adhesion[71]. These findings provide a valuable reference for regulating MSC subpopulations through hydrogel stiffness to facilitate long-term targeted therapy. Meanwhile, when tyramine and dopamine are grafted onto hyaluronic acid backbones, the resulting hydrogel microspheres gain the ability to scavenge reactive oxygen species. After this operation, MSCs can be protected from oxidative stress while maintaining high viability[72]. Transcriptomic analysis indicates that arginine-glycine-aspartic acid-functionalized silk fibroin-DNA hydrogels can markedly promote the expression of genes associated with cell adhesion and upregulate glycosaminoglycan biosynthesis pathways, inducing chondrogenic regeneration of MSCs[73].

To address biocompatibility concerns associated with conventional oil-phase fabrication, a centrifugal microfluidic strategy has been developed for oil-free microsphere production using an air-gap design[74]. Transcriptomic profiling revealed significant upregulation of key adipogenic markers, demonstrating that these microspheres effectively support MSC culture and adipogenic differentiation[74]. Meanwhile, to satisfy the requirements for large-scale MSC manufacturing, poly(N-isopropylacrylamide) was covalently grafted onto GelMA microspheres to generate a thermoresponsive microcarrier (BrushGel). This system afforded a 5.3-fold expansion of MSCs within 5 days, with expanded cells retaining typical surface marker expression and 34% T-cell suppression capacity, fulfilling clinical requirements[75].

Hydrogel microfibers: In nature, liquid protein filaments extruded by spiders solidify upon exposure to air to form microscale fibers. The preparation of microfluidic hydrogel fibers follows a similar principle, typically relying on coaxial laminar flow. A pre-gel solution as the core flow and an immiscible cross-linking solution as the sheath flow are delivered through separate microchannels; upon contact, cross-linking occurs, yielding hydrogel microfibers (Figure 2)[64,76].

Emerging evidence indicates that the topological architecture and mechanical properties of hydrogel microfibers can direct the growth orientation of encapsulated cells[64,77]. For example, coaxial wet-spinning has been applied to fabricate core-shell hydrogel microfibers, embedding MSCs and human umbilical vein endothelial cells. Following three-week co-culture, both cell types were found to align along the fiber axis and express the endothelial-specific marker CD31, indicating the formation of microvascular networks. This method offers critical insights into the recreation of anisotropic native tissues, such as microvasculature, in vitro[77]. In addition, compared with 2D culture systems, hydrogel microfibers significantly increased the expression of insulin and Pdx-1 in MSCs, suggesting effective differentiation into insulin-secreting cells[78]. When transplanted into diabetic rats, these differentiated cells stably reduced blood glucose levels[78]. Interestingly, Yu et al[79] integrated droplet microfluidics with wet spinning to biomimetically construct a hybrid fibrous material with a “bamboo-like” arrangement. This configuration can be digitally encoded through a microfluidic system to confer flexible biological functionalities, indicating that microfluidic hydrogel techniques can serve not only as a bioreactor for large-scale stem cell expansion but also as a transitional spatial scaffold for 3D tissue formation in stem cell transplantation.

Although micro-hydrogels have demonstrated significant advantages in 3D culture of MSCs, issues such as mechanical strength, stability, and potential immunogenicity continue to impede their clinical applications. The combination of dual-network hydrogels, nanocomposites, and tissue-specific ECM components can solve these problems to some degree[66]. At the same time, batch-to-batch variations in hydrogels can significantly affect the reproducibility of experimental results; resolving this issue depends on the introduction of advanced biomanufacturing technologies.

Precise mechanical stimulation in microchannels

Under physiological conditions, MSCs encounter various mechanical cues in addition to ECM interactions. These include fluid shear stress in bone marrow cavity and blood vessels, cyclic stretch induced by respiration and heartbeat, and periodic compressive stress experienced by articular cartilage. Such forces profoundly influence MSC proliferation and differentiation[80-82]. Microfluidic chips can integrate structures such as microvalves, elastic membranes, and air chambers to deliver controlled mechanical stimulation to MSCs[83,84], offering unique advantages for elucidating the regulatory mechanisms by which MSCs respond to mechanical stress.

Fluid shear stress is a crucial mechanical cue in the native body. Microfluidic techniques for recapitulating physiological and pathological flow in vitro can be categorized as passive or active. Passive systems generate stable fluid shear stress at a fixed location when the inlet flow stays constant. In comparison, active systems enable dynamic modulation of shear stress via external devices (such as syringe pumps) (Table 2)[85]. Extensive studies have demonstrated that fluid shear stress can regulate the differentiation of MSCs. This effect is primarily mediated by the Yes-associated protein (YAP)/transcriptional coactivator with PDZ-binding motif signaling pathway[86-88]. For example, fluid shear stress upregulates YAP expression in MSCs, which promotes osteogenesis and inhibits adipogenesis[86]. In addition, when MSCs were exposed to extremely low shear stresses (10^-5-10^-2 dyn/cm²), they showed enhanced expression of filamentous actin[88]. The nuclear localization of transcriptional coactivator with PDZ-binding motif then increased through the Rho/Rho-associated protein kinase pathway, thereby promoting osteogenic differentiation of MSCs[87,88]. By contrast, shear stresses of 7.8-13.7 dyn/cm² are most favorable for endothelial differentiation[89], while obvious adipogenic differentiation only occurs at high shear stress levels (15 dyn/cm²)[90]. Thus, the magnitude of shear stress determines the direction of MSC differentiation: Toward osteogenic (low stress), endothelial (medium), and adipogenic (high) fates.

Table 2 Microfluidic techniques for mechanical stimulation of mesenchymal stem cells.

Category	Microfluidic implementation	Key effects on MSCs	Ref.
Fluid shear stress	Passive (constant flow) or active (externally modulated) systems	(1) Low shear stress (10-5-10-2 dyn/cm2): Osteogenesis; (2) Medium shear stress (7.8-13.7 dyn/cm2): Endothelial differentiation; and (3) High shear stress (15 dyn/cm2): Adipogenesis	[85,87-90,99]
Cyclic stretch	Elastic membrane deformation	(1) Directs alignment; (2) Promotes proliferation, osteogenesis, and angiogenesis; and (3) High stretch (RMD > 3.5%) suppresses adipogenesis	[84,91,92]
Compressive stress	Fluidic pressure, confined microchannels, or mechanoactive hydrogels	(1) Promotes osteogenesis; and (2) Elevates vesicle secretion	[17,93-96,100]

MSCs: Mesenchymal stem cells; RMD: Relative membrane deformation.

Beyond fluid shear stress, cyclic stretch is widely used to replicate mechanical forces generated by breathing, heartbeat, and body movements[91]. In microfluidic systems, cells are cultured on an elastic membrane. An external actuator periodically deforms this membrane, and the resulting bending is converted into cyclic tensile stimulation applied to the cells (Table 2). For example, Zhou et al[91] placed an elastic membrane above a microchannel 20-500 μm wide. Hydrostatic pressure applied from underneath makes the membrane bulge, creating a periodic circumferential strain of up to 20% on adherent MSCs. At strains exceeding 10%, SMAD2 expression increases significantly and the MSCs undergo marked directional alignment. Short-term stimulation induces nuclear accumulation of β-catenin, whereas long-term stimulation (> 6 hours) restores it to normal levels[91]. These results suggest that β-catenin responds early to mechanical cues. In another study, negative pressure was used to deform a PDMS membrane sandwiched between two microfluidic chips, thereby applying tensile stress to cultured MSCs[84]. Higher tensile stress (relative membrane deformation > 3.5%) substantially enhanced MSC proliferation and osteogenic differentiation while suppressing adipogenic commitment[84]. In addition, multiaxial stretch strain upregulated self-renewal-related genes in ADMSCs, and their proliferative capacity was markedly improved[92].

Typical methods to exert compressive stress include applying hydrostatic pressure around cells or using microprobes to deliver direct compressive stress (Table 2)[93,94]. For instance, İyisan et al[94] found that periodic hydrostatic pressure (200 kPa, 0.5 Hz, 30 minutes, 21 days) induced nuclear YAP localization, followed by elevated expression of the early osteogenic markers Runt-related transcription factor 2 and alkaline phosphatase. Enhanced collagen deposition and extensive mineralization further confirmed that compressive stress directs MSC osteogenic differentiation[94]. In addition, confined microchannels themselves can apply lateral compressive stress on migrating cells. This compressive stress increases the permeability of the cell membrane and stimulates the secretion of small extracellular vesicles, reaching approximately four times the level observed in conventional culture[95]. As a result, wound healing is accelerated.

Interestingly, a phenomenon called “mechanical memory” is widespread in MSCs. In other words, MSCs retain sustained morphological and functional changes even after mechanical stimulation is removed[96]. Mechanical memory is gradually emerging as a potential tool for MSC therapy, and epigenetic modifications may play a central role. For example, stretching can long-term remodel the cytoskeleton and mediate MSC metabolic homeostasis by histone H3 lysine 27 trimethylation[97]. Microfluidic techniques have also proven highly effective in research into mechanical memory. For instance, following release from 3 μm-wide microchannels, MSCs retain persistent nuclear deformation and elevated migration for several days. This mechanical memory may be mediated by epigenetic modifications such as the histone H3 lysine 9 acetylation[96].

Notably, microfluidic techniques can also incorporate multiple mechanical stimuli or combine mechanical stimuli with biomaterials, thereby generating a more complex microenvironment for MSCs. Zheng et al[98] developed a microfluidic chip capable of applying fluid shear stress and cyclic stretch simultaneously or independently. They found that the organization of cellular stress fibers varies depending on cell type and loading pattern. In addition, under the same mechanical stimulation, MSCs cultured on different biomaterials exhibited various differentiation outcomes[99]. For example, under the same shear force, MSCs cultured on multi-walled carbon nanotube reinforced material exhibited a more pronounced osteogenic differentiation compared to the group reinforced by nanohydroxyapatite[99]. Novel mechanoactive hydrogels can also enable in-situ mechanical loading in response to biological stimuli. For instance, after incorporating photothermal nanoactuators into alginate hydrogel microspheres, near-infrared irradiation triggers isotropic compressive strains of up to 15% (generating forces of ~400 nN) at the single-cell level[100]. These techniques make it possible to probe MSC fate regulation under complex conditions.

However, current in vitro mechanical models still simplify the dynamic and multi-axial mechanical microenvironments in vivo. Therefore, results obtained from these platforms should be interpreted with caution. At the same time, research on the co-optimization of biomaterials and mechanical stress parameters remains in early stages. Detailed mechanisms underlying this co-optimization remain to be explored.

Cell co-culture and organ-on-a-chip

The in vivo microenvironment is a complex, dynamic network shaped by the interplay of biophysical signals and cell-cell interactions[101]. Microfluidic chips enable the construction of cell co-culture systems and organ-on-a-chip platforms. These systems create a multi-faceted coupling of biomaterials, mechanical stimuli, and cellular interactions. As a result, they realistically mimic in vivo conditions and deepen our understanding of how MSCs are regulated[102].

Vascularization is essential for cell survival and function in tissue engineering. In a microfluidic co-culture study, MSCs and human umbilical vein endothelial cells were cultured in either separate or shared channels[103]. Cells cultured in separate channels underwent pronounced inward vascular remodeling, which depends on α6 integrin-dependent interactions between endothelial cells and pericytes[103]. These findings highlight the pericyte-like supportive role of MSCs during vascular network formation[103]. In addition, the 3D assembly of microfluidic hydrogel microfibers offers a powerful strategy for constructing tubular tissues and biomimetic vascular networks in vitro[64]. For example, alginate-collagen composite microfibrous vessels of tunable dimensions can be fabricated by adjusting microfluidic flow rates. This method allows endothelial cells to align and form endothelialized luminal structures[104]. When these endothelialized microvessels were subsequently co-cultured with MSCs, they markedly enhanced osteogenic differentiation of MSCs compared with acellular scaffolds[104].

Osteogenic differentiation and bone repair are also important in MSC regulation. These processes rely on biophysical signals and multicellular crosstalk between MSCs and different types of osteocytes[105]. Vis et al[106] developed a “bone-on-a-chip” that establishes an osteoblast-osteoclast co-culture system. It offers controllable shear stress, continuous perfusion, convenient medium exchange, and live-cell imaging capabilities. On this platform, MSCs undergo osteogenic differentiation and self-assemble into scaffold-free osteoid tissue, reproducing the morphology of trabecular bone and supporting the adhesion and fusion of monocytes into multinucleated osteoclasts[106]. In addition, a biomimetic “cartilage-on-a-chip” was established by Liu et al[107]. They developed a biohybrid hydrogel containing neodymium magnets. Under mechanical loading by remote magneto-control, the expression of inflammation-related genes and collagen secretion of chondrocytes increased, mimicking the osteoarthritic phenotype[107]. This platform offers a novel tool for studying osteoarthritis and related pathologies. Multi-organ systems, also called “body-on-a-chip”, further help bridge the gap between in vitro models and native bodies[64]. For example, projection-based 3D printing has been used to fabricate modular paper-based microfluidic chips. These chips allow the assembly of multiple biomimetic tissue structures, such as bone, nasal, ear, and vascular tissues, holding great promise for future applications[108].

Together, the above studies on co-culture systems and organ-on-a-chip platforms show that microfluidic techniques can effectively probe MSC response in complex microenvironments. In particular, such approaches make it possible to dissect the integration of biomaterials, mechanical stress, and cell-cell interactions.

REAL-TIME MONITORING OF MSC PHENOTYPES

MSCs respond to biophysical cues in a dynamic and continuous manner. Therefore, real-time tracking of their phenotypic changes is critical for us to understand how biomaterials and mechanical stress regulate their fate. Traditional cell profiling methods are often invasive or give only endpoint analysis[109]. As a result, they severely limit the ability to capture real-time cellular dynamics. Instead, microfluidic techniques can be integrated with optical, electrochemical, or impedance-based sensors to facilitate continuous, non-invasive monitoring of cellular behavior[110,111].

For optical sensing, Perottoni et al[112] built a miniaturized dynamic culture platform to replicate the perivascular stem cell niche. Real-time metabolic monitoring was achieved via fluorescence lifetime imaging microscopy. Under high-oxygen and low-fluid-shear-stress conditions, MSCs showed increased oxidative phosphorylation. In contrast, opposite conditions enhanced glycolysis, establishing an oxidative phosphorylation-glycolysis metabolic gradient that agrees with in vivo observations[112]. Additionally, pre-staining MSCs with calcium ion-specific fluorescent probes can achieve real-time visualization of calcium signaling under compressive stress[100].

Microfluidic techniques can also incorporate electrochemical impedance sensing for label-free monitoring of MSC behavior[113]. By combining a flexible-probe scanning electrochemical microscope with a detachable microfluidic device, researchers directly assessed osteogenic differentiation in MSC spheroids inside microwell arrays without labeling[114]. Droplet microfluidics has also been used to build impedance-based measurement and analysis systems, allowing investigation of osteogenic differentiation at the single-cell level[113]. These studies revealed substantial heterogeneity among individual cells at distinct differentiation stages. They also showed that proliferative and differentiation potential decline with increasing passage number[113]. In a related study, Fois et al[115] designed an advanced microfluidic bioreactor that supports long-term culture and non-destructive monitoring of MSC spheroids using non-Faradaic electrochemical impedance spectroscopy. Collectively, these features make microfluidic techniques powerful tools for the dynamic monitoring of MSC phenotypic transitions. They provide critical technical support for dissecting cell fate decisions within complex physiological microenvironments.

LIMITATIONS AND PERSPECTIVES

Limitations

As mentioned above, microfluidic techniques have demonstrated unique advantages in cell sorting and quality control[22], 3D microenvironment reconstruction[59], and real-time phenotypic monitoring[111]. These capabilities provide robust technical support to explore how MSCs respond to biomaterials and mechanical stress. For biomaterial engineering, microfluidic techniques produce hydrogels with high uniformity and controllable properties. Such hydrogels markedly improve cell attachment and proliferation[116]. Composite bioactive scaffolds made from these hydrogels further elevate MSC viability and tissue repair efficacy[117]. For mechanical stimulation, microfluidic techniques have been applied to integrate multiaxial mechanical cues. This capability allows them to replicate dynamic microenvironments in vitro[118]. One example is a cartilage-on-a-chip model. In this model, fluid shear stress and the hydrogel substrate work together to promote chondrogenic differentiation of MSCs[119]. This synergy offers a new way to dissect the pathological mechanisms of osteoarthritis.

However, despite the significant advances, several limitations must be acknowledged. For example, hydrogel encapsulation or high-velocity fluid flow inside microchannels can exert mechanical stress on MSCs. Li et al[71] performed a pseudotime analysis of transcriptomic data, and the results showed that MSCs followed distinct differentiation lineages. While this finding demonstrates the inherent heterogeneity of MSCs, it also suggests that unmeasured mechanical stimuli within the microchannels may influence the experimental results. In addition, it remains unclear how different subpopulations of sorted MSCs respond to the same biophysical signals. Cell culture still calls for synergistic optimization of biomaterials and mechanical stress parameters. Although some scholars have called for standardization of microfluidic techniques, progress in this area remains limited[120,121]. Currently, most on-chip research stays at the laboratory stage, and experimental results often lack reproducibility across different laboratories[122]. Thus, microfluidic systems still have a long way to go before they can be used reliably in clinical practice. Standardized protocols and long-term stability testing will be essential for future development.

Perspectives

Advances in materials science, micro/nanofabrication, and artificial intelligence (AI) have made integration, intelligence, and scalable manufacturing the key directions for microfluidic techniques. Integration is a primary trend in microfluidic techniques. Miniaturization and multi-module integration give microfluidic techniques a unique advantage that enables a “sample-in, product-out” workflow[123]. Such systems minimize operator-dependent variability and lower risks associated with manual operations. More importantly, integrated on-chip systems can combine biomaterial fabrication with mechanical stress application, allowing the synergistic regulation of MSCs. For instance, Loutherback et al[123] developed a closed-loop microfluidic platform that integrates sorting, counting, characterization, and culture. This platform efficiently separates peripheral blood mononuclear cells and provides a new strategy for high-purity MSC acquisition and expansion. In another example, microfluidic-assisted preparation of cell-imprinted matrices enabled 14-day chondrogenic induction of MSCs in a closed environment, establishing a seamless transition from stem cell seeding to in vitro differentiation[124].

Machine learning and large AI models bring intelligence to microfluidic techniques. They enable efficient decoding of massive cellular datasets[125,126]. Take DropAI as an example. This high-throughput screening platform combines droplet microfluidics with AI and generates millions of reaction units within an hour. Consequently, it drastically improves multiplexed screening efficiency and highlights the considerable potential of AI-powered microfluidics in MSC sorting and drug combination[126]. The label-free deep learning method Bright2Nuc permits single-nucleus segmentation from bright-field images, quantitative morphological analysis, and stem cell differentiation prediction. Such abilities support automated induction and long-term dynamic tracking of 3D MSC differentiation on microfluidic chips[125]. In the future, integrating such real-time monitoring data with AI-driven feedback control systems will dynamically optimize biomaterial parameters and mechanical stimulation regimes. By doing so, the regulation of MSC fate will be more precise.

Scalable manufacturing is essential for translating microfluidic techniques into clinical use. Although various scale-up strategies have been put forward to meet Good Manufacturing Practice compliance[127], the throughput is still too low for some industrial applications[128]. In addition, microfluidic platforms are often restricted by complex fluid transportation instruments, which also drive up costs. High costs limit their practical application in real-time bedside monitoring[128]. The establishment of standardized protocols will be key for scalable manufacturing. The core task is to improve the biocompatibility and performance stability of chip materials[129]. Post-processing optimization offers a practical route. For instance, baking at 120 °C for 24 hours, UV irradiation, or ethanol immersion can enhance material biocompatibility and tailor mechanical properties to support MSC culture[130]. At the same time, closed and automated systems are also necessary for accelerating the commercial adoption of MSC therapies. For example, Danis et al[131] used Lonza Cocoon^®, an automated bioreactor system, to prepare six clinical samples. Five of them met Good Manufacturing Practice standards, each yielding approximately 2.3 × 10⁹ cells within 10 days (67 times the yield of the manually prepared group).

CONCLUSION

Stable and precise regulation of MSCs is essential for the long-term efficacy of MSC-based therapies. Microfluidic techniques offer miniaturization, precision, and multifunctional integration. These features make them comprehensive research tools to investigate how biomaterials and mechanical stress modulate MSC fate. In this review, we summarize three core applications of microfluidic techniques: Cell quality control, microenvironment engineering, and phenotypic monitoring. Specifically, we explore the acquisition of highly purified MSC subpopulations through label-free or affinity-based sorting to elucidate their heterogeneity; the recreation of 3D microenvironments using biomimetic hydrogels and mechanical loading; and the integration of optical and electrochemical sensors for dynamic phenotypic tracking. These three aspects form a continuous workflow. In addition, we highlight that integration, intelligence, and scalable manufacturing are key directions for microfluidic techniques. These advances will support their transition from the laboratory to clinical practice, as well as the commercialization of stem cell therapies.