Hybrid Multiphoton Lithography Scaffolds for Nanoscale Mechanobiological Assessment in Microscale Bone Models

ABSTRACT

We present a new hybrid 3D scaffold, fabricated via MultiPhoton Lithography (MPL), that integrates synthetic polymers (BisSR/CEA) with methacrylated collagen type I (Coll‐MA) and enables geometry‐dependent single‐cell confinement in most cages, while allowing cell–cell contact when the scaffold design permits multiple occupancy. We showcase that nanoscale structures with feature sizes down to hundreds of nanometers and locally tunable mechanical properties (kPa to MPa) can be achieved. Scaffold bioactivity is confirmed using 3D Single‐Molecule Localization Microscopy (SMLM). We quantified the dynamic 3D behavior of vinculin through its mechanoresponsive nanoscale localization. Notably, vinculin dynamics was independent of scaffold composition, including geometry, mechanical properties, and biodegradability. In contrast, collagen I and osteocalcin expression levels were strongly elevated in cells confined by hybrid scaffolds, indicating that scaffold bioactivity and geometry, rather than stiffness, govern stem cell fate. Our platform combines precisely tunable micro‐ and nanoscale environments mimicking extracellular matrix (ECM) features with super‐resolution imaging. This 3D tissue scaffold is component compatible with subsequent Organ‐on‐a‐Chip chamber integration and supports physiologically relevant bone tissue models.

This work presents a 3D hybrid tissue scaffold inspired by bone, incorporating collagen, fabricated using multiphoton lithography. The mechanical properties of collagen are tunable, and biodegradation allows for time‐dependent changes in geometry. The results demonstrate enhanced osteogenic differentiation of stem cells in our hybrid scaffolds. This study highlights the scaffold's potential for studying cellular dynamics and tissue engineering applications.

1. Introduction

Mimicking and understanding tissue properties and functions is crucial in tissue engineering. Valuable approaches are Organ‐on‐a‐Chip (OoC) systems, 3D‐bioprinting, and spheroid/organoid cultures for investigation of cellular development in vitro [1, 2, 3, 4, 5, 6, 7]. OoC systems incorporate microfluidic flow to simulate physiological conditions, including nutrient/waste exchange and mechanical stimulation. Cell bioprinting, on the other hand, enables precise spatial control over cell deposition to construct tissue‐like architectures. However, OoC and bioprinting often lack the resolution to allow for single‐cells analysis, limiting the ability to study cell‐cell and cell‐scaffold interactions at the micro‐/nanoscale in a controlled 3D environment.

Recent developments in tissue engineering, particularly 3D‐(bio)printed scaffolds, enable the in situ investigation and monitoring of cellular behavior in a complex environment, allowing researchers to assess the effect of the shape and topography of implants, or model diseases (e.g. Osteoporosis, Osteosarcoma) [8, 9, 10, 11, 12, 13, 14, 15, 16, 17].

The introduction of additive manufacturing has further advanced the field by enabling the fabrication of multi‐material scaffolds [18]. However, the integration of synthetic and protein‐based materials for tissue engineering remains challenging. An advanced method in additive manufacturing is two‐photon lithography introduced by Kawata in 1997 and has since been further advanced for high resolution 3D structuring of biomaterials [19]. In two‐/MultiPhoton Lithography (MPL), non‐linear excitation of a photoinitiator triggers radical formation [20], leading to monomer crosslinking and solidification. MPL enables the fabrication of structures ranging from the nano‐ to macroscale [21, 22, 23, 24], while achieving sub‐100 nm lateral feature sizes [25, 26, 27, 28, 29], with a current axial resolution of ∼200 nm [30]. In 3D‐(bio)printing, MPL outperforms established technologies like stereolithography and electrospinning with respect to resolution and geometric flexibility, both of which widely applied in tissue engineering [31, 32, 33, 34, 35, 36].

For tissue engineering, MPL has mainly been used with synthetic materials [17, 21, 37, 38, 39, 40]. The biocompatibility of 3D synthetic scaffolds has been tuned by various strategies, such as the introduction of amino acid‐based polymers [41, 42], protein‐adhesive photoresists and photochemical surface modification [10, 43, 44, 45]. The combination of nanostructuring and functionalization allows to observe cellular response on multi‐material (i.e. hybrid) scaffolds [46]. Recent developments focus on protein‐based photoresists to fabricate 3D scaffolds [47]. Collagen, the most abundant structural protein in the ExtraCellular Matrix (ECM), is a particularly promising protein for 3D‐(bio)printing [48], 2D guided cell growth [49, 50, 51], 3D patterning within hydrogels and for MPL fabrication due to its ability to support cell adhesion, migration and differentiation [52, 53].

Next to geometrical factors, affecting cell growth and attachment [54, 55, 56], mechanical properties play a crucial role in influencing cell responses [57, 58]. In particular, bone tissue presents a unique challenge due to its heterogeneous mechanical properties [59, 60, 61]. Substrate stiffness can be adjusted by altering composition or applying external stimuli, a common strategy in research for studying changes in cell behavior [62, 63, 64, 65, 66]. Mesenchymal stem cells (MSCs), stromal cells with multipotent differentiation potential in vivo and in vitro, are of high interest especially in tissue engineering, regenerative medicine and clinical applications [67, 68, 69, 70]. For example, hydrogel stiffness induces stem cell differentiation [71, 72], increases reactive oxygen species expression level in MSCs [73], and triggers phenotype changes in macrophages [74, 75, 76], altogether supporting tissue repair mechanisms. While the influence of 2D substrate stiffness on stem cell fate is well established, recent studies have demonstrated that 3D microenvironmental cues, including matrix stiffness, viscoelasticity, geometry, and degradability, critically regulate mechanotransduction and lineage commitment [77, 78, 79, 80].

Mechantransduction mechanisms allow cells to perceive ECM stiffness (i.e. vinculin‐ mitogen‐activated protein kinase 1 (MAPK1) mediated focal adhesion spot activity), while guiding the commitment toward specific differentiation lineages [77, 81, 82]. Vinculin, an actin‐binding protein, regulates and reinforces integrin binding, links the actin cytoskeleton to the ECM, as well as being a key component in the mechanotransductive vinculin‐talin complex [83, 84, 85].

Most prior studies have relied on bulk or multicellular 3D systems [78, 79, 80], limiting mechanistic insight into single‐cell behavior. Platforms that enable dynamical remodeling of the truly 3D scaffolds with nanoscale control over geometry, stiffness, and bioactivity are therefore essential to quantify single‐cell mechanosensing and direct mesenchymal stem cell differentiation [86, 87], with cell spreading and morphology critically influencing osteogenic commitment. While 2D micropatterning allows precise control over cell shape and adhesion, it fails to mimic the 3D bone microenvironment, where scaffolds must balance structural support, mechanical properties, biodegradability, and biocompatibility [88]. Current in vitro models often lack the microscale spatial organization needed for fine‐scale mechanosensing and spatially controlled differentiation. Emerging single‐cell platforms, including microfluidic and encapsulation systems [89, 90, 91], offer rigorous control of individual cell microenvironments, yet their applicability in studying mechanosensing and differentiation remains limited.

We report, for the first time, 3D hybrid scaffolds integrating collagen type I and biocompatible polymers, fabricated via MPL with a bone‐mimicking geometry allowing local dynamic remodeling and biodegradation. Unlike bulk 3D scaffolds or standard bioprinting approaches, which lack nanoscale resolution and dynamic remodeling, these scaffolds offer complete spatial control over cellular microenvironments (even down to a single‐cell confinement), allowing confinement and, when combined with high‐resolution/nanoscopy, direct monitoring of cell–scaffold interactions while minimizing measurement artefacts like autofluorescence. We characterized the nanoscale mechanical properties and feature sizes of collagen‐based nanostructures (Coll‐MA lateral features: 243 ± 34 nm; Young's moduli: 49 ± 32 kPa to 81 ± 10 MPa) using AFM and two‐color 3D SMLM. At the single‐cell level, SMLM and confocal imaging enabled analysis of vinculin clusters to quantify 3D mechanosensing and focal adhesion organization, as well as high‐resolution visualization of time‐dependent degradation of the nanoscale Coll‐MA structures.

The methacrylated type I collagen was produced by modifying the free amine groups, primarily the ɛ‐amines of lysine residues and the α‐amines at the N‐termini. In our material, approximately 20% of the total lysine residues were methacrylated (according to the provider).

We show that scaffold remodeling drives cell spreading, ECM deposition, and enhances osteogenic marker expression at single‐cell resolution, revealing that both local geometry and bioactivity guide stem cell differentiation. MPL‐enabled tunable micro‐ and nanoscale scaffolds provide comprehensive control over stiffness and architecture. Moreover, our innovation paves the way for investigations of cell–ECM interactions in physiologically relevant 3D microenvironments in unprecedented detail. Finally, our scaffolds will be applicable as a versatile, modular platform for Bone‐on‐a‐Chip and Organ‐on‐a‐Chip tissue engineering, with the option to easily integrate more complex scaffold architectures in the future.

2. Results

2.1. 3D Hybrid Scaffold Designed for Single Cell Confinement and Observation

Imitating complex tissue properties is key to understanding cellular development in vitro. First, we focused on designing a 3D scaffold to support cell growth within a controlled 3D microenvironment, where material composition, geometry, and mechanical properties influence cell differentiation. The scaffold was specifically designed to enable the investigation of cell‐matrix interactions at the single‐cell level in this 3D context. We tailored a stadium‐like structure (500 × 200 × 30 µm³, Figure 1a) inspired by bone structures with pore sizes (∼400µm²) supporting osteoblast differentiation, and high mechanical stability using the commercial monomer Ormocomp [21, 92]. The side walls of the stadium‐like structure were designed with a tilt of 70° for an increased opening surface for easier cell seeding. The stadium‐like structure segregates cells within the scaffold, minimizes direct cell‐cell interactions, and prevents cell migration in and out of the scaffold (see Figure S1). The structure provided structural support for the more delicate microenvironments, which were printed in a second step inside of the stadium (scanning electron microscopy (SEM) insets in Figure 1a, for stadium‐like structure with cell cages see Figure S2). We printed symmetrical cell cages with dimensions of 45 × 45 × 24 µm³. The cell cages were fabricated 6 µm lower than the stadium‐like structure to ensure cell segregation (stadium‐like structure height: 30 µm, cell cage height: 24 µm). Hard cell cages were comprised of biocompatible synthetic polymer bars in the x‐ and y‐ direction, while hybrid cell cages were a combination of synthetic polymer bars in the x‐direction and methacrylated collagen (Coll‐MA) bars in the y‐direction (Figure 1b).

FIGURE 1.

3D scaffold design for single‐cell analysis. (a) Schematic drawing of the stadium‐like structure, designed to separate cells growing around the scaffolds from cells growing within the scaffolds. Insets show SEM images of hard cell cages (BisSR/CEA) and hybrid cell cages (BisSR/CEA and Coll‐MA). In the hybrid cell cages, the hard material supports BisSR/CEA supports the soft biofunctional methacrylated collagen type 1 bars (Coll‐MA) incorporated in hybrid cell cages. (b) Schematic drawing of hybrid cell cages composed of 5 × 5 × 24 µm³ pillars and 45 × 1 × 3 µm³ bars consisting of BisSR/CEA and Coll‐MA, respectively. Pillars are separated by 1 µm and start 3 µm above the substrate to ensure cell‐cell contacts between neighboring cells within the cell cages. The schematic 3D view of the cages with an additional inset highlights an SEM image of the bars separated by a gap. (c) White light image (left) of MSC growing in cell cages 6 days post‐seeding and fluorescence image of cell nuclei stained with DAPI. Mostly single cells were housed within the scaffold (n_cages = 40 cages/scaffold (n = 3); ∼70% of cages with cells, ∼90% of cages have only 1 cell). The lower confocal fluorescence image shows the actin cytoskeleton labeled with phalloidin Alexa647Plus (red) and vinculin labeled with anti‐vinculin IgG Alexa488 (green). Schematics are not drawn to scale.

To facilitate cell‐cell contact and communication (i.e. paracrine signaling), which are both crucial for cell growth and development [93], the cell cages were designed with 1 µm gaps between cage bars with an additional 3 µm gap to the substrate. Furthermore, this ensures a genuine 3D architecture in which cells interact with all surrounding walls, a feature not achievable with simpler planar designs. Unlike open micropatterns, our cage geometry prevents multi‐layering and lateral interactions. The 1 µm gaps, while structurally defining compartments, maintain functional isolation. Absence of gaps completely suppresses development, validating the design's balance of confinement and minimal structural support. Combining biocompatible Bisphenol A‐Glycidyl MethAcrylate (Bis‐GMA), ethoxylated bisphenol A dimethacrylate (SR348C) (BisSR) and protein‐binding 2‐CarboxyEthyl Acrylate (CEA) in a weight ratio of 1:3.5:0.1 allows for coating of the hard cell cages with collagen prior to cell seeding [10, 94]. We utilized the biodegradability of Coll‐MA to fabricate a 3D hybrid scaffolds with a progressively changing geometry to investigate cellular development. The cell cages provided the required support for cell attachment and expansion in 3D for 21 days of cell culture. The consistent enclosure of mostly single cells (n_cages = 40 cages/scaffold (n = 3); ∼70% of cages with cells, ∼90% of cages have only 1 cell) within the scaffold was visualized by cell nuclei staining using 4',6‐DiAmidino‐2‐PhenylIndol (DAPI). The actin cytoskeleton was labeled with phalloidin Alexa647Plus as an indicator for cell morphology/spreading and cytoskeleton organization (red in the lower panel in Figure 1c). The focal adhesion‐associated protein vinculin, involved in mechanotransduction, has been visualized using anti‐vinculin IgG Alexa488 to assess cell‐matrix adhesion (green in the lower panel in Figure 1c). We determined the best‐suited grid size for cell spreading and confinement via visualization of actin cytoskeleton in tested grid sizes: 30, 35, and 45 µm (see Figure S3) [55, 58]. Initially, 30 – 40% of cell cages were filled during seeding. Cell cages with the 3 µm gap housed cells in all cages 6 days post‐seeding (Figure 1c), while for cell cages with no gap to the substrate it took up to 10 days for the same occupancy. Additional exclusion of the 1 µm gaps between bars resulted in impeded cell expansion (see Figure S3, 30 µm grid size). We communicate a 3D hybrid scaffold enabling reliable cell confinement and multi‐color single‐molecule imaging to monitor cell spreading and cell–scaffold interactions over time. High‐resolution 3D structuring confines autofluorescent regions and heterogeneous refractive indices to minimal volumes, mitigating their impact and enabling reliable super‐resolution imaging.

2.2. Functionality and Mechanical Properties of Collagen‐Based Structures

While the development of cells in 3D environments is a growing field of research [95, 96, 97], cell behavior within biodegradation‐driven scaffolds remains poorly understood. To address this, we also optimized material composition on top of scaffold geometry, focusing on mechanical properties and the biofunctionality of the structures We quantified the mechanical properties of BisSR/CEA and Coll‐MA‐based structures and developed a strategy to match the variations of the Young's modulus (YM) observed in bone and other tissues (ranging from kPa to GPa) [59, 77]. To achieve this, we utilized Poly(Ethylene Glycol) DiAcrylate (PEGDA) as a crosslinking agent to precisely modulate elasticity [47]. Starting out, we fabricated suspended lines, as previously outlined by Buchroithner et al. [21] to achieve 3D features minimizing surface effects, between Ormocomp bars for mechanical characterization with AFM.

We used nanoindentation to characterize the YM of the components of the hybrid scaffold and to assess the modulation of YM of Coll‐MA with the addition of PEGDA. As a reference, BisSR/CEA exhibited a YM of 81 ± 10 MPa (STD; median = 72 MPa). With the addition of 5 wt% PEGDA to Coll‐MA, the YM increased by a factor of ∼7 compared to pure Coll‐MA lines (Coll‐MA/PEGDA: YM  =  368 ± 150 kPa (median = 359 kPa), Coll‐MA: YM  =  49 ± 32 kPa (median = 38 kPa)) (Figure 2a). The high standard deviation in YM in Coll‐MA based structures can be attributed to the heterogeneity in the material distribution after printing in comparison with BisSR/CEA (see right panel in Figure 2c; Figure S15). Additionally, we determined a compression of the gel‐like shell of Coll‐MA structures by a factor of ∼4, as the height profile of the structure was significantly influenced by the applied force (see Figure S4).

FIGURE 2.

Multi‐modal characterization of collagen‐based MPL‐structures. (a) BisSR/CEA YM was measured to be 81 ± 10 MPa, 368 ± 150 kPa for Coll‐MA/PEGDA, and 49 ± 32 kPa for pure Coll‐MA. (b) Fluorescence image of suspended Coll‐MA lines. The Coll‐MA has been labeled with anti‐collagen type I IgG Alexa647. SMLM was used to detect single emitters showing a homogenous distribution of collagen (estimated 49.4 ± 11.2 emitters µm⁻¹ along the line, N_samples   =  9) and proved the functionality of collagen after printing. The yellow inset shows a super‐resolved image of the lines, with a line width (i.e. full width at half maximum) of 243 ± 34 nm. (c) AFM height and nanoindentation images of BisSR/CEA, Coll‐MA/PEGDA and Coll‐MA. The table shows the Root Mean Square (RMS) roughness of BisSR/CEA, Coll‐MA and Coll‐MA/PEGDA along the grey dashed lines. The RMS roughness was measured at 2.84 nm for BisSR/CEA, 7.69 nm for Coll‐MA/PEGDA and 50.89 nm for Coll‐MA. The Coll‐MA line exhibits filaments spreading from the core of the line. Scale bar represents: 0.5 µm for Coll‐MA/PEGDA and 2 µm for BisSR/CEA and Coll‐MA.

The bioactivity of scaffolds in tissue engineering is as crucial as their mechanical properties. To evaluate the biofunctionality of the Coll‐MA lines after printing, we used anti‐collagen type 1 IgG Alexa647 for immunolabeling of the lines (Figure 2b). Homogeneous spreading of bound antibodies along the written lines was observed via SMLM, with an estimated antibody density of 49.4 ± 11.2 emitters µm⁻¹ (N_samples   =  9; median = 49.6) (yellow insets in Figure 2b). The single emitters were localized with a lateral position accuracy of 44 ± 12 nm (17.32 signal‐to‐noise ratio). Additionally, we estimated the width of lines (at full width at half maximum) to be 243 ± 34 nm, proving sub‐diffraction limited feature sizes for Coll‐MA based structures for the first time. (for comparison to results from the AFM measurements see Figure S4). Despite the homogenous spreading of antibodies, we found that the 1D roughness quantified by Root Mean Square (RMS) values of pure Coll‐MA lines differs from Coll‐MA/PEGDA lines by a factor of ∼6h (see Figure 2c).

To show the effect of a biodegradation‐driven change in geometry on cell expansion during 21 days of cell culture, we tested the biodegradation of Coll‐MA in hybrid cell cages. The difference in RMS roughness did not influence the time scale of cells degrading the Coll‐MA and Coll‐MA/PEGDA structures (for results 11 days post‐seeding see Figure S5a). Once the cells degraded the collagen (typically 6 days post‐seeding), further expansion throughout the scaffold was observed (see Figure S5b). To prove that Coll‐MA is indeed degradable, the enzymatic degradation of nanostructures without cell was also performed (see Figure S17). While no significant difference for Coll‐MA and Coll‐MA/PEGDA in biodegradation was observed over a period of 11 days, the surface roughness variations may influence early cell adhesion and degradation, affecting the cellular response. In contrast, cells within the hard cell cages experience lateral spreading restrictions for the entire cultivation period, permitting only axial growth. We demonstrate the use of Coll‐MA to fabricate 3D scaffolds with tunable mechanical properties with Young's modulus covering three orders of magnitude (∼50 kPa to >80 MPa) and biodegradation‐driven geometry changes, effectively mimicking various tissue types and enabling dynamic remodeling that guides mechanosensitive MSC spreading, ECM deposition, and osteogenic differentiation.

Furthermore, the ability to match tissue‐specific elasticity, using PEGDA as a cross‐linking agent in a Coll‐MA based scaffold, highlights the potential for the use in scaffolds to guide mechanosensitive MSC differentiation.

2.3. Vinculin Distribution as an Indicator for Cell‐Matrix Interactions

Next, we focused on the comparison of cell‐matrix interactions in hard collagen‐coated BisSR/CEA and hybrid BisSR/CEA in combination with Coll‐MA cell cages. To investigate the effect of scaffold stiffness and geometry on cell‐matrix interactions, we examined the actin cytoskeleton and the vinculin distributions in MSC. For that, we performed confocal and SMLM imaging of stained MSC 6 and 11 days post‐seeding (for confocal images see Figure 3a; for SMLM images see Figure S6). Our results on hard‐/hybrid cell cages verified proper cell adhesion, as the actin filaments associated and co‐localized with the vinculin clusters formed at the substrate 6 days post‐seeding. Furthermore, we used SMLM data of vinculin for quantitative analysis of mechanosensing at the single molecule level. Vinculin was stained using fluorescently labeled (Alexa647 or Alexa488) anti‐vinculin IgG. Figure 3b depicts exemplary images of the vinculin distribution in cells within hybrid cell cages 6 days post‐seeding (right panel) and within hard cell cages 11 days post‐seeding (left panel). The single IgG conjugated Alexa647 emitters in samples 6 days post‐seeding were localized with an average position accuracy of 45 ± 9 nm lateral and 109 ± 22 nm axial (signal‐to‐noise ratio (SNR) 13.62) and the single IgG conjugated Alexa488 emitters in the samples 11 days post‐seeding with an average position accuracy of 49 ± 9 nm lateral and 83 ± 15 nm axial (SNR 17.02), respectively [98]. As shown in Figure 3c, the vinculin spatial distributions have been arranged into clusters [99], showing the estimated molecular organization of vinculin within a single cell (Figure S7 for exemplary clusters). We observed that, for some cells at an early stage, the z‐position of cluster centroids (z_pos ) indicates the presence of two distinct and spatially separated populations, shown as yellow and blue dots in Figure 3d, respectively. The upper population primarily comprises protein clusters involved in cell‐scaffold interactions, while the lower one contributes to cell adhesion to the glass substrate (Figure 3d). 11 days post‐seeding, the upper population redistributes toward the top of the cell (for population filtering see Figure S8). As shown in Figure 3e, the average z_pos for all clusters was at 153 ±99 nm above the substrate for 6 days post‐seeding and 510 ± 154 nm for 11 days post‐seeding (for further data of z_pos see Figure S9). No correlation between the scaffold version (i.e. hard/hybrid cell cages) and z_pos was observed at this early developmental stage. The 2CALM‐based cluster (SMLM data) comparison performed at the single‐molecule level [100], revealed that all clusters are similar in density distribution, regardless of cell cage composition and z_pos (for 6 and 11 days post‐seeding). Overall, a 73 ± 19% (median = 75%) similarity of cluster density distribution between all analyzed samples has been determined (Figure 3f). An additional single‐molecule cluster analysis comparing the upper‐population (6 days post‐seeding) and the overall vinculin population (11 days post‐seeding), revealed no significant differences (see Figure S10). Our results showed gradual increments of z_pos , suggesting a similar trend in upper cell layers. However, methodological limitations (such as low SNRs and z‐range restrictions) hindered the acquisition of SMLM data regarding the vinculin distribution above 1 µm of the substrate. Additional non‐quantitative confocal imaging showed a reduced vinculin population within the cell redirected from the substrate toward the top of the scaffold, along the growth direction (see Figure S11). Moreover, confocal images revealed that cells, forming the uppermost cell layer, exhibit a vinculin distribution directed toward the cells beneath them, instead of direct interaction with the scaffold (see Figure S12).

FIGURE 3.

Vinculin cluster formation in 3D scaffolds over time. (a) White light image and the corresponding confocal image of vinculin (green) and the actin cytoskeleton (red) of MSC in hybrid cell cages 6 days post‐seeding, showing a healthy cell state within the cell cages [101]. (b) 3D dSTORM images showing the distributions of vinculin in MSC cultivated in cell cages for 6 days (left) and 11 days (right). Vinculin was labeled with anti‐vinculin IgG Alexa647 in samples 6 days post‐seeding and anti‐vinculin IgG Alexa488 in samples 11 days post‐seeding, respectively. The focus was set 1 µm above the substrate. (c) The image shows vinculin clusters identified using DBSCAN [99, 100] clustering algorithms (with at least 5 nearest neighbors) in hybrid (6 days post‐seeding, left) and hard (11 days post‐seeding, right) cell cages [99]. The localized emitter signals are clustered with a radius of r  =  230 nm (left) and 100 nm (right). (d) A distinct vinculin cluster sub‐population above the substrate indicates the axial migration of vinculin clusters in MSC 6 days post‐seeding over time. (e) The average z_pos was 153 ± 99 nm in 6 days post‐seeding samples, while in 11 days post‐seeding samples, z_pos was 510 ± 154 nm, indicating a gradual redistribution (N_samples   =  9). (f) A matrix showing the similarity (in %) of vinculin cluster density distribution for cells 6 days post‐seeding and 11 days post‐seeding analyzed using 2CALM [100]. The results indicate an overall 73 ± 19% similarity in vinculin density distributions (i.e. cluster density and curvature), regardless of the cell cage composition. The plot was generated for DBSCAN cluster radii ranging from 10 to 1000 nm in 30 nm increments.

Using 3D SMLM localization data of vinculin, we performed cluster analysis to extract quantitative parameters of focal adhesion organization. This analysis provided two key insights: (i) determination of the absolute nanoscale dimensions of vinculin clusters, allowing investigation of their role in mechanosensing and their association with the actin cytoskeleton at adhesion sites, and (ii) comparative evaluation of clusters in hard and hybrid cell cages at 6 and 11 days post‐seeding, which revealed no significant difference in cluster size or organization between the two scaffold types. Notably, the axial position z_pos of vinculin clusters increased over time, indicating dynamic vertical redistribution during culture. This vertical relocation likely reflects scaffold remodeling processes that influence how cells sense and interact with their environment, thereby linking focal adhesion dynamics to downstream MSC spreading, ECM deposition, and osteogenic differentiation.

2.4. Effect of 3D Scaffolds on Osteogenic Differentiation of MSC

Herein, we address the role of mechanosensing as a response to ECM cues, which are crucial for the development and differentiation of stem cells. Osteogenic differentiation is well characterized and depends on a precisely controlled environment, including medium composition, spatial constraints, and ECM stiffness [35, 72, 81, 92]. However, there is currently a lack of studies and tools, tracking the underlying mechanisms at the single‐cell level. Building on our previous results, where we showcased the redistribution of vinculin clusters over time (see chapter 2.3), our aim is now to demonstrate the potential of the 3D scaffold for studying osteogenesis and its link to mechanotransduction. We visualized the formation of vinculin clusters to determine whether vinculin organization influences MSC differentiation and to further investigate the potential impact on osteogenesis. Therefore, we tested the effect of our scaffolds on the expression of osteogenic markers: ALkaline Phosphatase (ALP), an early osteogenesis marker; collagen type I, an ECM protein, involved in the early stage of bone formation; osteocalcin, a late‐stage marker indicating bone maturation. First, we seeded MSC onto 3D hard cell cages and differentiated them using osteogenic medium. Interestingly, already after 14 days of osteogenic treatment, we saw an increased ALP activity within the cell cages compared to the cells growing in a 2D environment and to cells without osteogenic treatment (See Figure S13). Second, to compare the results of osteogenic differentiation with our previous findings on cell‐matrix interactions within 3D hybrid scaffolds over time, we performed immunostaining of MSC after 21 days under proliferative and osteogenic conditions. Vinculin was stained as outlined in chapter 2.3. Collagen type 1 was labeled using either anti‐collagen type 1 IgG Fluor CoraLite Plus488, or anti‐collagen type 1 IgG Alexa488. Osteocalcin was visualized using primary IgG rabbit anti‐osteocalcin and secondary IgG goat anti‐rabbit Alexa647. Confocal imaging of vinculin localization in cells treated with expansion medium and osteogenic medium for 21 days revealed no difference regardless of treatment (see Figure S12). However, cells treated with an osteogenic medium exhibited strong expression of both markers (Figure 4a). In contrast, cells treated with expansion medium expressed less collagen type 1 and osteocalcin expression was absent (see Figure S14). Müller et al. [102] demonstrated that stem cell repopulation is most effective in single‐cell compartments in the short term (seven days), highlighting spatial restrictions as a simple yet powerful tool for directing stem cell fate. Our results showed decreased cell spreading in the hard cell cages (70% of cell cages occupied) under osteogenic conditions compared to cells under expansion conditions and only a subset of the cells (30% of the filled cell cages) expressed both markers (Figure 4b). Yet, our observation of osteogenic marker expression at the border of the 3D scaffold consisting solely of hard cell cages, suggests that spatial constraints influence stem cell differentiation, aligning with Müller et al.’s findings. The combination of hard cell cages (yellow inset in Figure 4) and hybrid cell cages (red inset in Figure 4) allowed us to study the impact of a biodegradation‐driven change of geometry on the cell spreading within the same scaffold. In the hybrid cell cages, after biodegradation of the Coll‐MA structures, cell expansion along the axis of the hard bars of the scaffold was observed (Figure 4c). Collagen type I expression was increased by approximately a factor of two in the hybrid cell cages, with an average intensity of 54.61 ± 6.83 a.u./pixel (median = 56.23), compared to 23.54 ± 6.36 a.u./pixel (median = 23.54) in the hard cell cages. This increase is further supported by the normalized pixel‐intensity distributions shown in Supplementary Figure S16, derived from 20 hybrid‐ and 16 hard‐scaffold collagen‐fluorescence images (osteodifferentiated MSCs). The distributions show a clear shift toward higher pixel intensities on the hybrid scaffolds, consistent with the increased collagen production. A Kolmogorov–Smirnov test (p = 0) confirmed that the two distributions differ significantly. The observed increase in collagen type I signal highlights scaffold‐dependent osteogenic differentiation. Measurements were collected from multiple regions across independent scaffold areas, confirming the consistency of these patterns.

FIGURE 4.

Osteogenic differentiation of MSC in 3D hybrid scaffolds. The first line shows the white light images. The second line shows the corresponding confocal images. The scale bars hold true for both lines within the same column. ECM was labeled with anti‐collagen type 1 IgG Fluor CoraLite Plus488, or anti‐collagen type 1 IgG Alexa488 (green), Osteocalcin, a late marker for osteogenic differentiation was labeled with primary IgG rabbit anti‐osteocalcin and secondary IgG goat anti‐rabbit Alexa647 (red), and the cell nuclei were stained with DAPI (blue). Autofluorescence of the stadium‐like structure is also shown in blue. (a) shows a 3D scaffold with hard and hybrid cell cages. MSC have been in culture and differentiated in osteogenic medium for 21 days. The expansion of MSC and the distribution of collagen type 1 and osteocalcin was scaffold‐dependent. Hybrid cell cages showed spreading of the cells throughout the scaffold (red inset), while expansion in hard cell cages was impeded (yellow inset). In the scaffold part with no cell cages, cells were not able to adhere for the 21‐day period (purple inset). (b) A magnified image of the hard cell cages shows limited cell activity, referring to restricted cell spreading and actin cytoskeleton organization along the scaffold axis, resulting in reduced ECM deposition compared to hybrid cages. Additionally, cells remained spatially constrained with no observable mobility within the hard cages. Only a subset of the cells (30% of cells) within the cell cages formed a collagen type 1 matrix and lower osteocalcin expression was observed compared to MSC in hybrid cell cages. (c) In the hybrid cell cages, the cells biodegraded the Coll‐MA bars, allowing the cells to expand along the remaining hard part of the cell cage. The hard BisSR/CEA structure supported cell spreading in the z‐direction. Distinct clusters of osteocalcin have been detected throughout the hybrid cell cages, indicating osteogenic differentiation. (d) Confocal image of collagen type 1 in the hard cell cages. The impeded differentiation of the MSC resulted in expression of collagen pre‐dominantly near the substrate (e) Confocal image of collagen type 1 in the hybrid cell cages. After biodegrading the collagen, the cells were able to expand across the whole scaffolds resulting in collagen expression across the entire scaffold height. Z‐axis in (xz)‐ and (yz)‐scans scaled for better readability, scale bar 20 µm.

Analyzing the cell expansion in the z‐direction, we found that cells in the hard cell cages expressed collagen type 1 predominantly near the substrate (Figure 4d). In hybrid cell cages, we observed collagen type 1 and osteocalcin expression over the entire scaffold height (Figure 4e). Our previous results showed time‐ and spatial independence in vinculin clustering of MSC (see chapter 2.3). However, osteogenic markers in hard and hybrid cell cages revealed distinct patterns, suggesting a strong dependence on scaffold properties. Despite the importance of ECM stiffness and the mechanotransduction of cells during osteogenesis [81], the lack of a clear correlation between the vinculin distribution and osteogenic markers highlights the complexity of osteogenesis [103, 104, 105].

We observed a correlation in the shape and spreading of cells in hybrid scaffolds, and increased marker expression was determined. Our results demonstrate that the combination of hard (as supporting structure) and biodegradable soft materials (guiding cell expansion) facilitate guided cell expansion over time, thus enabling the fabrication of 3D hybrid scaffolds for tissue growth in any desired 3D geometry.

3. Conclusion

We present a modular 3D tissue scaffold for Organ‐on‐a‐Chip applications, enabling single‐cell and single‐molecule resolution studies of cell–matrix interactions during expansion and osteogenic differentiation. The platform links mechanosensing to cell differentiation outcomes in physiologically relevant 3D microenvironments.

We demonstrate this connection with an exemplary analysis of vinculin clusters in a confined adhesion scenario using super‐resolution microscopy. We report the gradual migration of vinculin clusters in MSCs after 6 days in confinement. Notably, these findings were achieved using single‐molecule sensitive techniques, offering a more detailed understanding of vinculin dynamics and cellular behavior compared to previous studies. In contrast to findings by Lo et al. [106], reporting a strong association between increased vinculin clustering and accelerated osteogenesis in single cells, our results revealed only a weak correlation between vinculin clustering and the expression of osteogenic markers.

This decoupling distinguishes 3D confinement from 2D models, where vinculin clustering correlates with osteogenic fate. While our results do not show a clear stiffness‐dependent effect, previous studies suggest that such differentiation can proceed via vinculin‐independent routes (e.g., talin unfolding or YAP/TAZ signaling) [106] which was not tested in current work. The similarity of clusters observed in both hard and hybrid microenvironments may be due a generalized adaptive response in addition to a stiffness‐specific mechanotransduction, indicating a predominant influence from the rigid glass substrate.

A systematic analysis of additional focal adhesion components such as talin and paxillin would be a promising next step to address whether vinculin‐independent mechanisms trigger osteogenic differentiation in our 3D scaffold system. However, such an extended molecular characterization is beyond the scope of the present work, which focuses on vinculin‐based readouts. Future studies targeting talin, paxillin and related proteins could help disentangle scaffold‐wall versus glass‐substrate contributions and clarify how different adhesion complexes cooperatively or differentially regulate MSC fate in confined 3D environments [107].

From our analysis, vinculin z‐positions shift upward earlier in hybrid scaffolds (+28% vs. glass at day 6), indicating that wall stiffness influences focal adhesion organization despite basal glass substrate dominance. This is consistent with reports on systems with stiff underlying surfaces, where early rigid contact biases initial adhesion polarity [108]. While fully soft or suspended architectures would provide deeper mechanobiological insight, our 3D tissue scaffold benefits from mimicking the lacunae‐canaliculi‐mineralized bone interface. By day 11, cells migrate into the scaffold volume with upper layers losing glass contact, as shown by the progressive z‐position shift and confocal imaging, reflecting transition from substrate, to scaffold/ECM‐mediated mechanosensing [108].

This highlights the complexity of osteogenesis, which is intricately influenced by mechanical and biochemical cues, such as ECM composition, cell density, scaffold architecture, and timing [109].

Coll‐MA and BisSR/CEA polymers provide tunable stiffness (∼50 kPa to >80 MPa) and mechanical as well as biochemical cues guiding osteogenesis. Multi‐modal imaging (AFM, SMLM, confocal) characterized nanoscale features and confirmed time‐dependent scaffold remodeling, enabling dynamic geometry changes that direct cell behavior. We find that the RMS roughness of the scaffolds did not significantly affect the time scale of cellular degradation of Coll‐MA and Coll‐MA/PEGDA structures, with collagen typically being degraded around six days post‐seeding, after which cells continued to expand throughout the scaffold. Nonetheless, the increased roughness observed in collagen‐based barriers compared to other polymeric or hybrid structures indicates that free collagen fibers may protrude from the scaffold surface. While this feature does not markedly alter overall biodegradability, it could influence early cell adhesion and spreading, mechanotransduction [110, 111, 112], and the capacitive properties of the scaffold through interactions with collagen‐binding proteins [113], potentially affecting cell behavior during the initial stages of scaffold colonization. These findings demonstrate that moderate methacrylation preserves the biological functionality and cell‐mediated bioresorbability of collagen I, supporting the suitability of Col‐MA scaffolds for applications requiring controlled remodeling and tissue integration.

This versatile modular system supports controlled 3D osteogenesis and is compatible with both Bone‐on‐a‐Chip and Organ‐on‐a‐Chip tissue engineering platforms. Moreover, our hybrid scaffold provides a framework to decouple biochemical and mechanical cues by temporally controlling cell confinement and expansion, enabling single‐cell and single‐molecule studies of dynamic microenvironmental regulation of stem cell fate. Unlike conventional 3D bulk or multicellular systems [77, 78, 79, 80] our modular protein‐based scaffold demonstrates that local geometry and bioactivity, rather than overall scaffold stiffness, drive enhanced osteogenic marker expression at the single‐cell level. MSCs in hybrid scaffolds exhibit stronger osteogenic marker expression, highlighting the potential of such scaffolds for controlled, microenvironment‐guided osteogenesis. While our platform does not yet emulate the hierarchical complexity of bone structures as compared to osteocyte lacunae or canaliculi, it offers the potential to do so and uniquely supports quantitative, time‐resolved mechanobiology studies, providing a basis for future integration of co‐cultures, bioactive cues, and complex architectures compatible with Organ‐on‐a‐Chip and disease modeling.

This study introduces: (1) BisSR/CEA‐Coll‐MA hybrid scaffolds with degradable confinement; (2) single‐cell cages with 1 µm gaps for functional isolation; and (3) 3D SMLM quantification of vinculin dynamics. To our knowledge, this represents the first single‐molecule analysis of focal adhesions in MPL‐printed, degradable 3D scaffolds.

In closing, 3D structuring of bioreactive, protein‐based materials via MPL offers unprecedented control over scaffold geometry, chemistry, and mechanics. The resultant structures are a versatile platform to study single‐cell mechanosensing and enable time‐resolved studies in dynamically (re‐)malleable microenvironments exhibiting variable degrees of anisotropy. Future developments may include co‐culture capabilities, incorporate additional bioactive cues, and integrate dynamic microfluidic systems to engineer complex tissue interfaces, such as the bone–cartilage junction. Most importantly, our platform enables the quantitative investigation of confinement‐driven cell differentiation, concomitantly offering new avenues for mechanobiology and serving as a scaffold component for future Organ‐on‐a‐Chip applications.

4. Experimental Section/Methods

4.1. Materials

Ormocomp (Micro Resist Technology) was used as monomer for the stadium‐like structure. For the hard cell cages and the hard part of the hybrid cell cages, a mixture of the monomers of Bisphenol A‐Glycidyl MethAcrylate (Bis‐GMA, Esschem Europe), ethoxylated bisphenol A dimethacrylate (SR348C, SARTOMER) and 2‐CarboxyEthyl Acrylate (CEA, Sigma Aldrich) in a weight ratio of 1:3.5:0.1 were used (BisSR/CEA). 1 wt% of Irgacure 819 (Sigma Aldrich) was added as photoinitator.

For the soft part of the hybrid cell cages, methacrylated collagen (Coll‐MA, Sigma Aldrich) was dissolved at a concentration of 6 mg mL⁻¹ in 20 mm acetic acid over night at 4°C. 0.25 wt% Riboflavin (Vitamin B₂) (Sigma Aldrich) was added prior to printing as photoinitiator. 5 wt% of Poly(Ethylene Gycol) DiAcrylate (PEGDA, M_n  =  575, Sigma Aldrich) was added as a crosslinking agent for Coll‐MA/PEGDA structures.

4.2. Multi‐Photon Lithography (MPL)

MPL was performed with a customized lithography system (Workshop Of Photonics (WOP)) using an ultra‐short pulsed laser at 515 nm (CARBIDE, 1 MHz repetition rate, >290 fs pulse duration, Light Conversion) and a 3‐axis stage (AEROTECH Nanopositioner) for sample movement, as described previously [21]. 3D MPL scaffolds were prepared using an air objective lens (50x, 0.42 NA, Mitutoyo). The stadium‐like structure and the supporting bars for the suspended lines were fabricated with an excitation intensity of 0.067 TW cm^{− 2} and a writing speed of 1 mm s⁻¹, followed by rinsing with 100% acetone (Roth). BisSR/CEA structures were printed as follows: pillars were written with an excitation intensity of 0.038 TW cm^{− 2} at 1 mm s⁻¹ writing speed while bars were generated via double illumination of each voxel with an excitation intensity of 0.059 TW cm^{− 2} at the same writing speed. BisSR/CEA structures were developed by rinsing with 100% ethanol (Roth). Coll‐MA/PEGDA structures were fabricated using double illumination of each voxel with an excitation intensity of 0.084 TW cm^{− 2} and 0.1 mm s⁻¹ writing speed. Similarly, Coll‐MA structures were produced with a double illumination of each voxel with an excitation intensity of 0.091 TW cm^{− 2} and a writing speed of 0.1 mm s⁻¹. Coll‐MA based structures were developed by gentle washing with 0.5 mm acetic acid (Sigma Aldrich) and stored in ultra‐pure water until further use.

4.3. MSC from Wharton's Jelly Culture

MSC were thawed and expanded in MSC expansion medium (StemMACS, Milteny Biotec) supplemented with MSC‐CytoMix (Milteny Biotec) and 1% penicillin‐streptomycin (P/S). MSC were passaged using 0.25% Trypsin Versene (VWR) and centrifuged at 220xg for 5 min. The cells were expanded twice before seeding on 3D scaffolds.

4.4. MSC Seeding on 3D Scaffolds

Prior to cell seeding, the samples were sterilized with UV‐light for 30 min, washed 3x times with sterile phosphate buffered saline without Ca/Mg (PBS, Roth). 3D scaffolds with only hard cell cages were coated with 50 µg mL⁻¹ collagen type 1 from rat tail (Ibidi) in 17.5 mm acetic acid for 15 min prior to seeding. 10 000 cells were seeded on the scaffolds in MSC expansion medium and kept at 37°C and 5% CO₂ for 6 to 11 days, exchanging the medium at least every 2 days.

4.5. Osteogenic Differentiation of MSC

For osteogenic differentiation, MSC were seeded at a density of 4 000 cells per scaffold in MSC expansion medium (StemMACS, Milteny Biotec) and incubated overnight. Next day, medium was exchanged to osteogenic differentiation medium DMEM‐low glucose (Thermo Fisher) containing 10% fetal calf serum (FCS, Sigma Aldrich), 2 mm L‐glutamine (Thermo Fisher), 100 U mL⁻¹ P/S (Thermo Fisher), 10 nm dexamethasone (Sigma Aldrich), 150 µm ascorbat‐2‐phosphate (Sigma‐Aldrich), 10 mm β‐glycerophosphate (StemCell Technologies) and 10 nm dihydroxy‐vitamin D3 (Sigma‐Aldrich) or control medium consisting of DMEM‐low glucose, L‐glutamine with 10% FCS and 100 U mL⁻¹ P/S. Medium was changed every 3–4 days.

4.6. Immunostaining

MSC were washed with PBS at 37°C 3 times, fixed using 4% paraformaldehyde (Arcos Organics) in PBS at 37°C for 20 min, permeabilized in 0.5% Triton X‐100 in PBS and blocked in 1% albumin from chicken egg white (ACE, Sigma‐Aldrich) for 1 h. The actin cytoskeleton was labeled with 6.6 nm phalloidin Alexa647plus (Thermo Fisher) in ACE for 20 min. Vinculin was labelled either with 1 µg mL⁻¹ anti‐vinculin IgG Alexa488 (Abcam) or 1 µg mL⁻¹ anti‐vinculin IgG Alexa647 (Abcam) in ACE for 1 h. Collagen type 1 was labelled with 5 µg mL⁻¹ anti‐collagen type 1 IgG Fluor CoraLite Plus 488 (Fisher Scientific) or 1 µg mL⁻¹ anti‐collagen type 1 IgG Alexa647 (Abcam) in ACE for 1 h. Osteocalcin was labelled with 10 µg mL⁻¹ primary IgG rabbit anti‐osteocalcin (Abcam) in 1% ACE overnight at 4°C and 2 µg mL⁻¹ secondary IgG goat anti‐rabbit Alexa647 (Fisher Scientific) in ACE for 1 h. The cell's nuclei were stained with 2 µg mL⁻¹ DAPI (Thermo Fisher) in PBS for 15 min.

4.7. Fluorescence Microscopy

Fluorescence images were acquired using a modified Olympus IX81. Precise sample positioning was done using a XYZ piezo stage with a nanometer precision (Physik Instrumente GmbH) in combination with a mechanical stage with a 1 × 1 cm range (JPK Instruments) for coarse movement. MSC were illuminated through a 60x magnification objective lens (1.4 NA, Olympus) and an additional 1.6x magnification tube lens for a final magnification of 96. For 3D imaging, a cylindrical lens (f  =  1 000 mm, Thorlabs) was used in the optical detection pathway of the microscope. A 642 nm solid‐state diode pumped laser (Toptica Photonics), a 488 nm solid‐state pumped diode laser (Toptica Photonics), and a 405 nm diode laser (Insaneware) were used as light sources. A dichroic filter (ZT405/488/561/640rpc nm, Chroma) and an emission filter (446/523/600/677 nm BrightLine quad‐band band‐pass filter, Semrock) were used. An additional emission filter (ET 700/75 M, Chroma Technology GmbH) was used for imaging of phalloidin Alexa647plus, anti‐vinculin IgG Alexa647 and an additional emission filter (ET 525/50 m, Chroma Technology GmbH) was used for imaging of anti‐vinculin IgG Alexa488. Fluorescence signals were collected with an Andor iXonEM+ 897 (back‐illuminated) EMCCD camera (16 µm pixel size).

For 3D imaging with nanometer accuracy, direct STochastic Optical Reconstruction Microscopy (dSTORM) with a cylindrical lens in the optical detection path was implored. Calibration of the axial‐dependent deformation of the point spread function was done by using a sample of homogeneously distributed TetraSpeck beads (0.1 µm, Invitrogen) on a glass cover slip. 200 images were taken at a defined step size of 10 nm along the axial axis. 10 000 images were acquired per imaging sequence with 20 ms illumination time and a sampling rate of 25 images s⁻¹. The single‐molecule signals were analyzed and visualized using a custom written software [98].

Confocal imaging was performed with a Leica Stellaris microscope using either an air objective lens (20x, 0.75 NA), or a glycerin immersion objective lens (63x, 1.30 NA). The 3D stacks were recorded by illuminating the sample with 400 nm, 491 nm, 640 nm laser light. Images were analyzed with Fiji/ImageJ V2.

4.8. Cluster Analysis

For comparison of the 3D vinculin localization data, clustering‐based analysis using 2CALM was performed [100]. The 2‐sample Comparative Analysis of 3D Localization Microscopy Data platform (2CALM) is an analysis pipeline that organizes localization microscopy data into protein clusters of varying dimensions and calculates statistical parameters using various numerical methods. Localization microscopy images were treated as 3D point clouds, and a comparative analysis required extracting features that represent their geometric structure without compromising accuracy. To achieve this, point clouds (vinculin localizations) were clustered within a constant radius. For each cluster, the relative density (compared to the overall density of the sample) and curvature were calculated. These distributions were then pairwise compared using the Kolmogorov‐Smirnov and Wilcoxon tests, and the p‐values were determined and aggregated. This process generates a curve of p‐values for each radius in the given sequence (from 5 to 1000 nm, with a step size of 30 nm). Based on this p‐value curve, the level of similarity was determined. The platform enabled the assessment of protein (i.e. vinculin) distributions in two biological samples, identifying nanoscale differences through parameters such as cluster density and curvature. Its automated classification system integrated multiplexing and multi‐level statistical techniques into a unified parameter, facilitating sample comparison and similarity testing. We employed DBSCAN‐based clustering to assess the relationship between density and curvature as a function of cluster size for both samples [99]. The DBSCAN clustering radius was automatically calculated for each sample as the average minimum distance between sample points, multiplied by a factor of 2. This method also enabled the exclusion of noise from unspecifically bound fluorescent antibodies. Furthermore, the cluster centroids were extracted to calculate the relative distance of the clusters above the substrate.

4.9. Atomic Force Microscopy (AFM)

The AFM was mounted on top of an inverted epi‐fluorescence microscope (AxioObserver, Zeiss) allowing simultaneous optical and AFM imaging. AFM measurements were performed with an JPK Nano Wizard 4 using QI (Quantitative Imaging) mode. In this mode simultaneous topography and mechanical parameters of the surface we measured by recording in each pixel full force‐distance curve. To extract Young's Modulus maps the recorded force‐distance curves were fitted with Hertz/Sneddon model with following parameters: Tip shape—Paraboloid; Tip radius—20 nm; Poisson ratio – 0.5. The fitting was done with indentation depth ∼ 10% from the overall line's height, in order to avoid the contribution of the stiff glass slide (See figure S15 for example of force‐distance curve (from line presented in Figure 2) fitted with Hertz model). The YM of BisSR/CEA and compression experiments on Coll‐MA were measured with a PPP NCHR cantilever (Nanosensors, nominal spring constant = 42 N m‐1) and the YM of Coll‐MA and Coll‐MA/PEGDA was measured with a PPP FMR cantilever (Nanosensors, nominal spring constant = 2.8 N m‐1). The exact cantilever spring constant was determined ad hoc before each measurement via contact‐based calibration method (JPK). All fitting and image processing were done with the JPKSPM Data processing software.

Funding

The work was supported by the Austrian Science Fund (FWF) (Grant No. 10.55776/P31827). The financial support by the Austrian Federal Ministry of Labour and Economy, the National Foundation for Research, Technology and Development and the Christian Doppler Research Association is also gratefully acknowledged.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: smll73501‐sup‐0001‐SuppMat.docx.

Data Availability Statement

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