Cell Contractile Forces Drive Spatiotemporal Morphogenesis in 4D Bioprinted Living Constructs

Summary

Current 4D materials typically rely on external stimuli such as heat or light to accomplish changes in shape, limiting the biocompatibility of these materials. Here, a composite bioink consisting of oxidized and methacrylated alginate (OMA), methacrylated gelatin (GelMA), and gelatin microspheres is developed to accomplish free-standing 4D bioprinting of cell-laden structures driven by an internal stimulus: cell-contractile forces (CCFs). 4D changes in shape are directed by forming bilayer constructs consisting of one cell-free and one cell-laden layer. Human mesenchymal stem cells (hMSCs) are encapsulated to demonstrate the ability to simultaneously induce changes in shape and chondrogenic/osteogenic differentiation. Finally, the capability to pattern each layer of the printed constructs is exhibited to obtain complex geometric changes, including bending around two separate, non-parallel axes. Bioprinting of such 4D constructs mediated by CCFs empowers the formation of more complex constructs, contributing to a greater degree of in vitro biomimicry of biological 4D phenomena.

Graphical Abstract

eTOC blurb

A 4D bioprinting strategy utilizing intrinsic cell contractile forces (CCFs) as an endogenous stimulus is demonstrated. Tissue constructs printed from a smart composite bioink with mechanically adaptive properties undergo CCFs-driven shape morphing. When spatially encoded with precise spatial arrangement/placement of encapsulated cells, CCFs mediate multidirectional and progressively enhanced transformations, enabling the formation of complex tissue architectures. This approach provides a powerful tool for advanced tissue regeneration and morphogenesis modeling.

Introduction

Spatiotemporal geometric transformations are essential to the development and healing processes of tissues¹. During these processes, tissue layers undergo bending, folding, and buckling, facilitating the formation of complex geometries (Scheme 1a-c)^2,3. Cells interactions, both among themselves and with the extracellular matrix (ECM), propagate contractile forces that lead to the development of specialized tissue architectures characterized by intricate geometric shapes⁴. Cell contractile force (CCF), the mechanical force generated by the cell cytoskeletal machinery, is a driving mechanism in tissue morphogenesis steps of folding, invagination, and elongation and plays a crucial role in tissue functionalization and specialization^5,6.

Scheme 1. Tissue morphogenesis and 4D system design.

(a-c) Typical tissue morphogenesis in vivo. Created with https://BioRender.com. (d) Schematic depiction of the bilayer system’s composition, fabrication methodology, and morphogenesis during culturing in media.

Recapitulating these dynamic shape changes in vitro holds the promise of advancing tissue engineering to a new horizon, where flexible control over geometric complexity and biomechanics can be achieved⁷. Consequently, morphodynamical tissue engineering strategies have garnered considerable interest for their potential to produce complex, dynamic tissues that are otherwise unattainable using traditional static three-dimensional (3D) tissue engineering strategies^8-10. Four-dimensional (4D) tissue engineering, which combines 3D tissue engineering with an additional morphing component over time triggered by stimuli, has thus been developed and is rapidly evolving due to advancements in materials, instruments, and technologies^11-14.

Current strategies in 4D tissue engineering include morphing cell condensates^15-17 and, predominantly, morphing biomaterial matrices embedded with living cells^18-22. Integrating CCF with these strategies to enable a programmed morphing process may be valuable for partially replicating tissue morphogenic processes observed in vivo, thus providing effective tools for remodeling engineered tissues in vitro. However, morphing cell condensates based on CCF face challenges in scaling up tissue sizes and precisely controlling the morphing process. Recently, cell-only bioprinting technology has been developed to facilitate the fabrication of large cell constructs^23,24. When combined with a shape-morphing actuator such as a hydrogel system, 4D cell-condensate bioprinting has demonstrated effectiveness in generating large artificial tissues with complex architectures through controlled shape transformation²⁵. However, this strategy relies on hydrogel swelling to elicit shape transformation rather than CCF generated within the cell condensates. Similarly, current morphing biomaterials generally depend on external energy inputs such as light, heat, electric fields, and magnetic fields to drive shape changes^26-29.

Recent advances in 4D biofabrication indicate that manipulating CCF in conjunction with biomaterials holds significant potential for generating complex living constructs. For example, seeding cells on fibronectin-coated parylene microplates can generate effective contractile forces between cells and the substrate, driving the folding of the microplates³⁰. However, the cells seeded on microplates are presented in a monolayer form that exerts limited CCF, necessitating a specialized soft junction design between the nonbiodegradable microplates, which is challenging to apply in 3D tissue constructs. Alternatively, depositing loose cell clusters in the superficial region of soft Matrigel impregnated with collagen fibers via DNA Velcro technology has enabled generation of local contractile forces between cells and the surrounding collagen fibers, inducing local contraction of reconstituted tissues^31,32. Similarly, synthetic fibrous hydrogel assemblies composed of loose hydrogel fibers coated with cell-adhesive RGD motifs have demonstrated utility in programming shape via CCF-induced contraction³³. While embedded cells or cell clusters within reconstituted matrices can direct global contraction and subsequent spatial tissue remodeling, these systems require ultra-weak biomaterials to allow the relatively weak CCF to exert contraction functions. Consequently, engineering initial matrix shapes is highly challenging due to the inherent fragility of the biomaterials, thereby limiting the production of final complex living constructs.

Considering the challenges associated with current engineering systems and the critical role of tissue morphogenesis in tissue development, we present the development of a CCF-driven shape-morphing living system composed of uniformly embedded living cells within a robust microgel matrix. Scheme 1d illustrates this system as a bilayer system consisting of a layer of photocured oxidized methacrylate alginate (OMA) microgel and a layer of photocured OMA microgel/gelatin methacrylate (GelMA) loaded with gelatin microspheres (GMSs) and living cells. Key features that distinguish this system from previously reported ones include: (1) the microgel materials exhibit shear-thinning and rapid self-healing properties, allowing for extrusion printing to produce stable, free-standing constructs, (2) cells are evenly distributed throughout the biomaterials, capable of generating CCF collectively to induce global construct morphing efficiently, and (3) the ability to localize CCF to induce complex tissue morphing in a precise and controllable manner. Using this system, we demonstrated advanced shape programming via CCF generated within the living constructs post-printing. Additionally, we conducted proof-of-concept 4D tissue regeneration studies, successfully demonstrating the formation of cartilage-like and bone-like tissues with preprogrammed curvatures.

Results

Hydrogels printed from composite bioinks demonstrate mechanically self-softening properties

For 4D shape transformations to occur via CCF, the material system needs to have low mechanical properties for the cells to drive the geometric changes^34,35 yet durable to maintain stability after printing. To accomplish this, a system composed of OMA, GelMA, and uncrosslinked GMSs was developed. These polymers were combined to form a bioink that can be 3D printed with high fidelity while establishing a “mechanically soft” microenvironment capable of being deformed by cellular forces. To achieve printable bioinks, the OMA was processed into a jammed state which demonstrated shear-thinning and rapid self-healing^22,25. The GelMA was added to the polymer composition to enhance interactions between cells and the printed construct^36-38, and the GMSs were incorporated to make mesopores within the scaffold to: i) weaken the scaffold after they liquify, which benefits shape morphing by CCF, and ii) create pores for increasing nutrient diffusion and enabling increased cell proliferation and migration, which may in turn increase the number and distribution of cells contributing to deforming the scaffold (Figure 1a).

Figure 1. Rheology of composite OMA/GelMA bioink for 3D printing.

(a) Schematic illustration of mesoporous matrix formation upon GMS dissolution, which weakens the hydrogel and facilitates CCF-mediated hydrogel contraction. Created with https://BioRender.com. Uncrosslinked composite bioink shear thinning properties, where (b) viscosity decreased as shear rate increases, and (c) the storage modulus was greater than the loss modulus at low shear strains and less than the loss modulus at high shear strains. (d) OMA/GelMA composite bioink exhibited self-healing properties, with the storage and loss moduli maintaining their initial values after multiple oscillations of strain with 50 mg/mL GMSs. (e) Storage modulus (G’) increased with addition of GMSs at concentrations of 25, 50, and 100 mg/mL to the uncrosslinked OMA/GelMA bionks. (f) Frequency sweep rheology demonstrating that after one day of culture, the 50 mg/mL of GMSs liquified in the construct composed of photocrosslinked OMA/GelMA hydrogel, causing the G’ to decrease dramatically. (g) Z-stack 3D confocal photomicrographs of photocrosslinked OMA/GelMA hydrogels before (i) and after (ii) GMS dissolution. Scale bar = 200 μm. Unless otherwise specified, all experiments use the composite ink consisting of OMA/GelMA mixed with GMSs at a concentration of 50 mg/mL bioink.

The mechanical properties of hydrogels formulated with composite bioinks containing varying concentrations of GMSs were evaluated prior to culture in media. The Young’s modulus exhibited a significant increase as the GMS concentration increased from 0 mg/mL to 50 mg/mL (Figure S1a). However, a further increase to 100 mg/mL resulted in a significant decrease in Young’s modulus, likely due to excessive GMS content interfering with hydrogel crosslinking, thereby compromising mechanical integrity. Additionally, scanning electron microscopy (SEM) imaging revealed that after the liquefaction of embedded GMSs, the hydrogels exhibited a highly porous structure, with porosity increasing proportionally to GMS concentration (Figure S1b). These findings indicate that hydrogel porosity can be effectively tuned by modulating the concentration of incorporated GMSs.

To be used for free-standing 3D printing, bioinks must exhibit shear-thinning and rapid self-healing properties²³. The viscosity of the composite microenvironment decreased dramatically as shear rate increased, confirming shear-thinning behavior (Figure 1b). Additionally, the storage (G’) and loss (G”) moduli of the bioink crossed over each other as shear strain increased (Figure 1c). At low shear strains (i.e., when the bioink was at rest), G’ was greater than G”, indicating that the bioink behavior was solid-like. However, at high shear strains (i.e., when the bioink was flowing through the needle), G’ was less than G”, indicating that the bioink behavior became more liquid-like. Along with these shear-thinning properties, an oscilliatory strain test revealed the self-healing quality of the bioink (Figure 1d). As strain oscillates between 1% and 100%, the moduli oscillated between the same values, indicating that the bioink is able to self-heal and consistently respond to reverses in strain application even after multiple exposures. Taken together, these shear-thinning and self-healing behaviors imparted the composite bioink with exceptional printability. However, since the purpose of this bioink is to aid in CCF-mediated scaffold shape changes, these rheological properties were rendered moot unless the crosslinked bioink was also soft enough to be deformed by cellular forces (G’ < 200 Pa)^39-41. Therefore, G’ was measured at frequencies less than 10 Hz for variations of the bioink containing different concentrations of GMSs. Consistent with the Young’s modulus results, the G’ of the bioink was observed to substantially increase (Figure 1e), whereas G” showed a moderate increase with increasing GMS concentrations (Figure S2a). This enhancement in the bioink’s mechanical properties underscores the beneficial impact of GMSs, promoting greater construct stability post-printing. However, since the GMSs were uncrosslinked and liquefied when cultured at 37 °C, the G’ of the photocrosslinked composite hydrogel in its swollen state after one day of culture at 37 °C was significantly reduced (Figure 1f), which may facilitate cellular forces to deform the hydrogel matrix. Nevertheless, the composite hydrogel maintained its mechanical integrity, as evidenced by the larger G' relative to G" (Figure S2b). The dissolution of GMSs under physiological conditions created enlarged pores within the swelling hydrogel, as evidenced by the presence of larger void spaces compared to constructs maintained at room temperature (Figure 1g, S3).

Bioink compostions, printing dimensions, UV crosslinking time, and cell densities influence construct morphing

To investigate the extent to which cellular forces can deform composite matrices of varied mechanics, cell-laden OMA/GelMA bioinks with various concentrations of GMSs were printed into 10 mm × 10 mm × 0.6 mm squares, crosslinked with UV light, and cultured for 14 days. Photographs of each sample (N = 4) were obtained each day using a dissection microscope. Representative photomicrographs for each condition at the day 1, 3, and 14 timepoints showed relative cell-mediated shrinkage of the constructs over time (Figure 2a, S4). In contrast, cell-free hydrogel constructs fabricated under the identical conditions maintained structural dimensions over 14 days (Figure S5). The shrinkage of each cell-laden hydrogel was quantified by measuring the area of each square, revealing the trend that increased GMS concentration resulted in reduced area at day 14 (Figure 2b). In other words, significant contraction of the hydrogel matrix was observed at higher concentrations of GMSs. The GMSs liquefied at the beginning of culture, leaving mesoscale pores within the printed constructs (Figure S6). The aforementioned findings were likely due to increasing concentration of micropshere producing pores reducing the mechanical resistance of the constructs to deformation by cellular forces, resulting in a noticeable increase in construct contraction. Cells in constructs in all conditions exhibited predominantly alive cells at days 1, 7, and 14 (Figure S7), demonstrating high cytocompatibility of the system. While the GMS concentration of 100 mg/mL condition conferred the greatest amount of contraction, it also posed challenges in generating reproducible, high-fidelity prints. Therefore, we determined that a GMS concentration of 50 mg/mL was the optimal concentration to ensure macroscopic changes in shape while maintaining hydrogel stability after microsphere liquefaction.

Figure 2. Shrinkage of 3D-printed cell-laden constructs.

(a) Images of OMA/GelMA composite hydrogels loaded with varying concentrations of GMSs. (b) Quantified area of the constructs over time revealed that increasing GMS concentration led to an increase in shrinkage. Scale bar = 10 mm. NIH3T3 cells; GMS concentration = 0-100 mg/mL bioink; Cell density = 100 M/mL bioink; UV crosslinking time = 30 s; Designed construct dimensions (L × W × H) = 10 mm × 10 mm × 0.6 mm. “*”, “†”, “#”, and “$” indicate statistical significance at day 14 compared to 0, 25, 50, and 100 mg/mL, respectively (p < 0.05).

Since cells play a critical role in generating CCFs within the constructs, we investigated the effect of cell density, ranging from 25 M/mL to 200 M/mL bioink, on construct morphing (Figure S8). Constructs printed with a cell density of 200 M/mL bioink exhibited mechanical instability after UV crosslinking, failing to maintain structural integrity during transfer from the printer platform to the tissue culture plate. This instability was likely due to the insufficient polymer crosslinking degree, which was inadequate to entrap such a high cell density. Consequently, this condition was not pursued further. Among the remaining cell densities, all constructs exhibited progressive contraction over time, with higher cell densities leading to more extensive contraction. This phenomenon can be attributed to the increased total CCF exerted by a greater number of cells, which induced greater hydrogel matrix contraction.

UV crosslinking time influences the hydrogel matrix rigidity by modulating the degree of crosslinking, which, in turn, affects the extent of hydrogel contraction induced by CCFs. When UV crosslinking times ranging from 15 s to 45 s were examined, a decreasing trend in rate and final contraction magnitude was observed with increasing crosslinking time, although all conditions exhibited substantial contraction by day 14 (Figure S9). Thus, UV crosslinking time serves as a crucial parameter in regulating CCF-mediated hydrogel contraction. Notably, while the 15 s crosslinking time resulted in the highest contraction, the hydrogels fabricated under this condition were very soft, making their transfer challenging. Therefore, a UV crosslinking time of 30 s was selected for subsequent experiments.

The effect of construct dimensions, including thickness and length, on contraction behavior was also assessed. Among the thicknesses investigated (0.4 mm, 0.6 mm, and 0.8 mm), thinner hydrogel constructs exhibited more rapid and pronounced contraction (Figure S10). However, construct length did not show a clear correlation with contraction behavior. Of the lengths tested (8 mm, 10 mm, and 12 mm), constructs with a length of 8 mm exhibited the greatest final contraction at 14 days, followed by those of 12 mm, while constructs of 10 mm exhibited the least contraction (Figure S11). The underlying mechanism for this observation remains unclear at this time.

Cells proliferate during hydrogel contraction

The transition of the hydrogels from a mechanically robust to a mechanically compliant state is critical for enabling CCF-induced contraction. The progressive increase in hydrogel contraction observed over the culture period also suggests a possible concurrent overall increase in the amount of CCF in each hydrogel. Cell proliferation would be likely a key driver of this amplification. To investigate this, the DNA content of printed cell-laden hydrogel constructs at different time points were quantified. DNA content exhibited a continuous increase over time, reaching a 2.1-fold increase at day 14 compared to day 1 (Figure S12), indicating substantial cell proliferation. This observation was further corroborated by live/dead staining, which revealed a progressive increase in cell density over time (Figure S13). In contrast, previous studies using nondegradable or minimally degradable hydrogels have reported minimal changes in DNA content during culture⁴². The observed increase in cell numbers can be at least partially attributed to the formation of interconnected pores resulting from the dissolution of GMSs, which enabling cell migration and proliferation.

To examine the structural evolution of the constructs over time, hematoxylin and eosin (H&E) staining was performed (Figure 3a). At day 0, intact GMSs (pink-stained regions, black arrows) were clearly visible within the construct, while cells with a rounded morphology were predominantly localized within the gaps of OMA microgels and GMSs (green arrows). By day 1, complete dissolution of the GMSs led to the formation of well-defined voids (blue arrows). At day 3, these voids began to diminish as cells infiltrated and occupied the available space. At later time points, a substantial increase in cell number was observed, with cells adopting elongated morphologies (Figure S14) and forming a highly organized, aligned structure (Figure S15).

Figure 3. Cellular changes in constructs during culture.

(a) H&E staining and (b) phalloidin staining of cell-laden hydrogel constructs at different culture time points. Black arrows indicate representative GMSs, green arrows indicate representative OMA microgels, blue arrows indicate representative pores. Scale bar = 250 μm. NIH3T3 cells; GMS concentration = 50 mg/mL bioink; Cell density = 100 M/mL bioink; UV crosslinking time = 30 s; Designed construct dimensions (L × W × H) = 10 mm × 10 mm × 0.6 mm.

Given that cellular contractile forces are driven by the actin-myosin complex^43,44, inhibition of actin polymerization markedly reduces a cell’s ability to generate force. To assess this, constructs were treated with cytochalasin D (CytoD), a well-established actin polymerization inhibitor, at a concentration of 5 μM. This concentration has been reported to disrupt cytoskeletal organization by inhibiting F-actin formation and impairing cell motility^45-47. Following CytoD treatment, cells remained rounded with no apparent alignment within the hydrogel matrix (Figure S16), underscoring the critical role of CCFs in mediating matrix remodeling and structural organization.

Phalloidin staining was employed to visualize changes in actin microfilament organization. Over the culture period, phalloidin staining demonstrated a transition from small, dispersed actin filaments at early time points to larger, clustered structures at later stages (Figure 3b), indicative of enhanced cytoskeletal organization and possibly increased cell density. These findings support the hypothesis that sustained cell proliferation and reorganization progressively amplified CCFs, thereby driving continuous hydrogel contraction and a corresponding reduction in construct size.

Lastly, confocal imaging of cryosectioned hydrogel constructs provided additional insights into temporal changes in cell distribution and matrix structure. At day 1, voids resulting from GMS dissolution were clearly visible (yellow arrows). However, these voids became less apparent by days 3 and 7, accompanied by an increased density of cells (Figure S17). These findings suggest progressive matrix remodeling, as well as active cell migration and proliferation over time, consistent with results from cell proliferation assays and H&E staining.

Bilayer design facilitates programmed shape morphing

To investigate the ability to control the direction of bending in our 4D system, a bilayer approach was then pursued. Here, the bottom layer consists only of OMA microgels while the top layer consists of the cell-laden composite hydrogel. The top layer was printed directly onto the bottom layer and the entire construct was photocrosslinked to obtain the desired bilayer structure. Cells within the top layer made physical connections with each other and the surrounding biomaterial matrix via cell adhesions. This enabled cytoskeleton-generated CCF in the cell-laden layer to cause macroscopic shrinkage of this layer. The OMA microgel-only layer resisted this contraction, causing the bilayer to bend towards the cell-laden layer direction, as shown in the cell growth medium (GM) condition (Figure 4a, S18, GM). In contrast, cell-free bilayers cultured under the same conditions demonstrated negligible shape changes over a 14-day period (Figure S19).

Figure 4. 4D shape morphing in bilayer constructs.

(a) To examine the role of CCF in the 4D process, printed constructs were cultured in cell growth media (GM) for 14 days. Constructs were also cultured in GM containing 5 μM CytoD as a negative control, and in GM containing 0.1% DMSO as a vehicle control. (b) Constructs in normal GM and GM with 0.1% DMSO conditions exhibited similar 4D bending, demonstrating that CytoD was responsible for the lack of bending seen in this condition. The average bending angles in the GM and DMSO conditions were not significantly different from each other at all timepoints (p > 0.05). (c) Photomicrographs of histological staining with H&E at day 14 showed cells in GM and DMSO conditions forming fibrous borders while cells in Cyto D condition did not. White scale bar = 5 mm. Black scale bar = 200 μm. Red scale bar = 50 μm. NIH3T3 cells; GMS concentration = 0-100 mg/mL bioink; Cell density = 100 M/mL bioink; UV crosslinking time = 30 s; Designed construct dimensions (L × W × H) = 18 mm × 4 mm × 0.6 mm.

To validate the role of CCF in driving the observed construct geometric changes, bilayer constructs were cultured in GM containing CytoD (5 μM). Constructs cultured in this media displayed no macroscopic changes in shape over the duration of culture (Figure 4b, Cyto D) and no sign of cell death at day 14 (Figure S20). Since CytoD was dissolved in dimethyl sulfoxide (DMSO) and then added to GM, an additional group of GM containing 0.1% v/v DMSO was included. To corroborate that this concentration of DMSO had no effect on bending, a vehicle control condition was established in which constructs were cultured in growth medium with 0.1% DMSO (Figure 4c). The normal GM and DMSO conditions showed no significant differences in bending angle over 14 days. The similar bending angles quantified from constructs culured GM and DMSO indicated comparable hydrogel deformations (Figure 4b). Histological analysis was performed to investigate the effects of CytoD treatment (Figure 4c). H&E staining of samples in each media condition illustrated that cells in the GM and DMSO groups appeared to have elongated cell bodies and were able to form a fibrous border along the outer surfaces of the cell-laden layers, consistent with normal behavior of fibroblasts⁴⁸. In contrast, cells in the CytoD group displayed predominantly rounded cell bodies and were unable to form a fibrous border, consistent with the inhibition mechanism of this drug^45,46. These results suggest that deformation was ascribed to the cell-mediated matrix contraction of the upper (cell-laden) layer. To the best of our knowledge, this is the first time that controllable matrix shape transformation replying soly on CCF was demonstrated in a uniformly loaded cell construct printed using a composite bioink with high free-standing stability.

Concurrent CCF-driven 4D morphogenesis with chondrogenesis

Coordinating cell differentiation and new cartilage tissue formation with cell-mediated 4D construct shape transformations could advance biomimetic tissue engineering⁴⁹. To explore this, we investigated the effects of chondrogenic differentiation of hMSCs within the cell-laden layers and extent of 4D geometric changes in the bilayered constructs.

To accomplish this, hMSCs at a density of 1 × 10⁸ cells/mL bioink were suspended in the composite bioink and bilayer constructs were printed as designed in Scheme 1d. Constructs were cultured in one of two media: (1) normal GM, or (2) chondrogenic pellet medium (CPM), a serum-free medium which includes TGF-β1. Construct bending angles were measured from photomicrographs of constructs in each condition over 21 days of culture (Figure 5a, S21). The bending angles of constructs increased progressively during culture, with constructs in CPM exhibiting significantly greater (p < 0.05) bending than those in GM on days 3, 5, 7 and 14 (Figure 5b). Previous studies have shown that the scaffolds seeded with chondrocytes exhibited more pronounced contraction compared to those seeded with MSCs^50,51. Therefore, the hydrogels encapsulating hMSCs could experience enhanced deformation when exposed to chondrogenic differentiation conditions. By day 21, constructs in both GM and CPM displayed comparable bending angles, indicating that exposure to chondrogenic factors induced a greater rate of bending in hMSC-laden constructs but did not increase the maximum bending angle achieved by day 21.

Figure 5. Chondrogenesis in 4D bilayer constructs.

hMSCs were printed in bilayer constructs and cultured in normal growth media (GM) or chondrogenic pellet medium (CPM) for 21 days. (a) Macroscopic images of constructs cultured over 21 days. (b) Quantification of bending angle over time (N = 4). (c) Quantification of GAG production and GAG/DNA levels for both GM and CPM conditions (N = 4). Constructs cultured in CPM produced significantly more GAG (p < 0.05). (d) Histological staining on day 21 qualitatively confirmed substantial GAG production in constructs cultured in CPM (N = 2). Scale bar in (a) = 5 mm, scale bar in (d) = 0.2 mm. GMS concentration = 50 mg/mL bioink; Cell density = 100 M/mL bioink; UV crosslinking time = 30 s; Designed construct dimensions (L × W × H) = 18 mm × 4 mm × 0.6 mm.

During chondrogenic differentiation, hMSCs secrete a matrix rich in negatively charged polysaccharides known as glycosaminoglycans (GAGs), which contribute to the unique mechanical properties of cartilage⁵². Therefore, the presence of GAGs is a useful measure to evaulate the extent of chondrogenesis in the bilayer constructs cultured in GM and CPM media conditions. Biochemical analysis revealed that constructs cultured in CPM had significantly higher (p < 0.05) GAG content normalized to DNA content, consistent with an increase in chondrogenesis (Figure 5c). Additionally, histological analysis was performed to visually corroborate the biochemical results (Figure 5d). H&E staining revealed that constructs cultured in CPM formed a more dense and compact tissue early in culture compared to the those cultured in GM. This suggests that chondrogenic differentiation of hMSCs resulted in a higher degree of condensation and, consequently, generated a higher local cell density and enhanced cell-cell interactions than those in GM, leading to a more pronounced bending observed visually. Staining for Safranin O and Fast Green revealed the presence of negatively charged GAGs and collagen, respectively. In both CPM- and GM-cultured constructs, much of the collagen staining was localized to the surface of cell-laden layer directly in contact with media (inner side, as indicated by the yellow arrows), consistent with previous results where cells from cell aggregates in direct contact to the media often produce a fibrous layer (outer side) during chondrogenic differentiation^53-55. While the OMA matrix in the construct can also be nonspecifically stained by Safranin O, CPM-cultured constructs exhibited more intense Safranin O staining within the inner bounds of the cell-laden layer, which supports the increase in normalized GAG levels measured in the biochemical analysis. To further substantiate this observation, constructs were additionally stained with Alcian blue at a pH of 0.2, which selectively stains strongly sulphated proteoglycans from cell-produced GAG without staining the OMA microgels or GelMA. The lack of intense blue staining in the GM-cultured constructs showed that the majority of Safranin O staining is due to the presence of OMA and not cell-produced GAG. Additionally, the intense blue staining of the CPM-cultured constructs revealed that much of the Safranin O staining is indeed cell-produced GAG. These results demonstrate that cells subject to chondrogenic differentiation culture conditions exhibited amplified and accelerated initial construct morphing. Additionally, the encapsulated cells were able to produce a cartilage-like matrix. Therefore, this system exhibits immense potential for CCF-based 4D chondrogenic tissue regeneration.

Concurrent CCF-driven 4D morphogenesis with osteogenesis

The role of CCFs in 4D morphogenic tissue engineering in this system was further investigated by culturing cell-laden bilayer constructs in osteogenic medium (OM) for 28 days. Constructs maintained in normal GM served as controls. Similar to the chondrogenic condition, the constructs exhibited progressive increases in bending angles during culture (Figure 6a, S22), eventually forming well-defined spiral configurations (Figure 6b) with high structural stability (Video S1, S2). Constructs cultured in OM showed a more rapid and pronounced deformation compared to those in GM, with significantly greater bending (p < 0.05) observed on days 1, 3, and 14 (Figure 6c). However, by day 14 and onwards, bending angles between the two groups became comparable. To assess bone-like tissue formation, biochemical analyses were conducted to quantify osteogenic markers alkaline phosphatase (ALP) activity and calcium ion (Ca²⁺) deposition (Figure 6d, 6e, S23). Constructs cultured in OM for 28 days exhibited significantly higher levels of ALP and Ca²⁺ compared to the GM group, indicating enhanced osteogenic differentiation and bone-like matrix formation. Alizarin red staining revealed pronounced mineral deposition, a key marker of osteogenic differentiation, in the OM constructs (Figure 6f), indicating strong matrix mineralization⁵⁶. Therefore, these findings support the robust potential of CCF-based 4D printing strategies to guide osteogenic tissue morphogenesis and regeneration.

Figure 6. Osteogenesis in 4D bilayer constructs.

hMSCs were printed in bilayer constructs and cultured in normal growth media (GM) or osteogenic medium (OM) for 28 days. (a) Macroscopic images of constructs in GM and OM for 28 days. (b) Macroscopic images of GM and OM constructs in PBS at day 28. (c) Quantification of bending angle over time (N = 3). (d, e) Quantification of (d) ALP/DNA and (e) Ca²⁺/DNA levels after 28 days of culture (N = 3). (f) Alizarin red staining of sectioned GM and OM constructs at day 28. Scale bar in (a) = 10 mm, scale bar in (b) = 3 mm, scale bar in (f) = 0.2 mm. GMS concentration = 50 mg/mL bioink; Cell density = 50 M/mL bioink; UV crosslinking time = 30 s; Designed construct dimensions (L × W × H) = 18 mm × 4 mm × 0.6 mm.

CCF patterning drives formation of complex geometries

Next, it was investigated whether the direction of contraction can be controlled by patterning the cell-laden layer to induce complex 4D events within growth medium (Figure 7a). For example, printing parallel lines of cell-laden hydrogel on a rectangular hydrogel layer resulted in the formation of a cylindrical tube-like structure. Similarly, printing parallel cell-laden lines diagonally across a rectangular hydrogel layer resulted in the formation of a helical structure. Using 3D printing to create these structures also allowed for the generation of 4D constructs that change shape along multiple axes simultaneously. To illustrate this, a 4-armed bilayer “gripper” shape was printed. Each arm was observed to bend upward and inward, similar to how one’s fingers bend to grip an object in one’s palm. Additionally, a bilayer rectangle was printed with the cell-laden layer on top and adjoined to a second bilayer rectangle printed cell-laden layer first. Accordingly, bending was observed in opposite centrosymmetric directions around same axis, resulting in an “S”-shaped structure. These results indicate that cellular forces were sufficient to drive spatiotemporal changes within this biopolymer microenvironment, and that these changes persisted over 14 days of culture. Importantly, the complex final structures were generated and controlled by the precise patterning of the printed hydrogel and cell-laden layers.

Figure 7. Complex patterning of the cell-laden and cell-free hydrogel layers.

(a) Complex patterning of the cell-laden layer. Schematics of the printed constructs followed by photomicrographs at days 0, 3 and 14. (b) Complex patterning of the cell-free hydrogel layer. Rectangular hydrogel patterns were printed either vertically or horizontally onto a cell-laden hydrogel rectangle. The direction of the hydrogel pattern influenced the bending direction of the constructs. (c) Complex patterning of both the cell-laden and cell-free hydrogel layers. Layers were printed sequentially according to the schematic, resulting in bending around two separate, non-parallel axes. Scale bars = 5 mm. NIH 3T3 cells, GMS concentration = 50 mg/mL bioink; Cell density = 100 M/mL bioink; UV crosslinking time = 30 s. Designed construct dimensions: (a) Bars: hydrogel-only layer = 20 mm × 15 mm × 0.2 mm, cell-laden bar = 15 mm × 1.5 mm × 0.4 mm, 1 mm spacing; Diagonal Bars: hydrogel-only layer = 15 mm × 5 mm × 0.2 mm, inclined cell-laden bar (45°) = 5 mm × 2 mm × 0.4 mm, 1mm spacing; Gripper: hydrogel-only layer = 20 mm × 5 mm × 0.2 mm, cell-laden layer = 20 mm × 5 mm × 0.4 mm; “S”: hydrogel-only layer (half segment) = 10 mm × 4 mm × 0.2 mm, cell-laden layer (half segment) = 10 mm × 4 mm × 0.4 mm; (b) Vertical Bars: hydrogel-only bar = 9 mm × 1.5 mm × 0.2 mm, 2 mm spacing, cell-laden layer = 9 mm × 12 mm × 0.4 mm; Horizontal Bars: hydrogel-only bar = 12 mm × 1.5 mm × 0.2 mm, 1 mm spacing, cell-laden bars = 12 mm × 9 mm × 0.4 mm; (c) T-shape: Arm: hydrogel-only layer = 12 mm × 4 mm × 0.2 mm, cell-laden layer : 12 mm × 4 mm × 0.4 mm, Base: hydrogel-only layer = 12 mm × 4 mm × 0.2 mm, cell-laden layer = 12 mm × 4 mm × 0.4 mm.

Up to this point, directional bending has been controlled solely by patterning the cell-laden layer. Therefore, we investigated whether the direction of 4D bending could be influenced by patterning the cell-free layer. To demonstrate this, the composite cell-laden bioink was printed into 9 mm × 12 mm × 0.6 mm rectangles. A patterned cell-free hydrogel layer consisting of either horizontal or vertical bars was then printed onto the composite cell-laden layer. The cell-free hydrogel bars were observed to induce bending of the constructs around an axis perpendicular to the direction of the bars (Figure 7b). Differences in initial printed geometries (e.g., rotated 90 degrees) contributed to differences in shape change over the duration of the 14-day culture period.

In addition to controlling the direction of contraction by separately patterning the cell-free and cell-laden layers, multiple bending axes can be incorporated into a single construct by patterning both layers simultaneously. To demonstrate this, T-shaped bilayer structures were printed in a step-by-step protocol. Figure 7c presents the schematic of printing, where the base of the T was first formed by printing a rectangular cell-free hydrogel layer, followed by a matching cell-laden layer directly on top. Next, the arm of the T was formed by printing a cell-free hydrogel layer directly perpendicular to the base and subsequently printing a cell-laden layer adjacent to the perpendicular cell-free hydrogel layer. This initial geometry caused the arm of the T to bend around the y-axis, while the base of the T exhibited bending around the x-axis. Multi-axial bending around two non-parallel axes has not previously been demonstrated using generated CCF in cell-laden biomaterials.

Discussion

Controlling shape transformation in 4D reconstituted tissues has garnered significant attention in tissue regeneration. This capability is critical as it allows engineered tissues to recapitulate essential morphogenic processes that occur in vivo, providing a powerful tool to mimic natural tissue development⁵⁷. The programming and control of shape morphing in engineered living constructs are fundamental for advancing tissue engineering. Fortunately, emerging 4D tissue bioprinting⁵⁸, which produces shape-transformable objects, offers a promising solution to push this frontier forward.

However, 4D tissue bioprinting faces critical challenges due to the demanding requirements for cytocompatible materials for cell embedding, long-term culture, and cytocompatible stimuli that permit morphing under physiological conditions. Consequently, recent developments have focused on identifying suitable biomaterials. Typically, engineered constructs based on these biomaterials undergo shape morphing in response to external stimuli such as light, magnetic fields, and solvents. CCF plays a crucial role in cell differentiation, migration, and growth during development, tissue remodeling and healing. It also plays a vital role in morphogenesis by driving tissue bending, stretching, alignment, and repositioning⁵⁹. These forces are generated by actomyosin networks, where myosin motors slide along actin filaments and transmit mechanical tension across the plasma membrane to the ECM via integrin-mediated adhesions⁴⁴. Thus, developing a printable system capable of promoting controllable construct shape transformation on a large scale in response to internal CCF is highly significant, as it enables the creation of structurally dynamic tissues. However, the utilization of inherent CCF generated by cells within artificial tissues has been scarcely reported.

Current systems for this purpose are based on morphing cell condensates and ultrasoft biomaterials, which are limited to microscale size, local morphing, or in-plane volume changes. Microscale cell condensates, such as organoids, exhibit strong cell-cell and cell-matrix interactions, generating substantial CCFs that facilitate rapid contraction and are advantageous for studying morphogenesis^60,61. However, directly scaling these models to large tissues without incorporating foreign materials for controlled macroscopic shape changes has not yet been demonstrated. While embedding living cells within ultrasoft biomaterials such as collagen enables construct contraction due to CCFs^34,62, printing such soft biomaterials into stable, freestanding tissue constructs is difficult. Furthermore, such constructs typically undergo linear shrinkage rather than out-of-plane morphing. The development of bioinks with sufficient mechanical stability for extrusion printing (or direct ink writing) enables the fabrication of multiscale cell-laden constructs with cost-effectiveness and shape-shifting capabilities, offering promising solutions for biomedical applications^63,64. However, this remains challenging, as mechanically robust hydrogels tend to resist deformation under weak CCFs. With this in mind, in this work we developed a system with bioink mechanical properties robust enough for printing stand-alone constructs while allowing for transformation by weak CCFs generated by uniformly distributed cells throughout the matrix. This was achieved by developing printable jammed OMA microgel bioinks and incorporating GelMA to enhance cell adhesion and interactions. Although direct visualization of cell-matrix interactions at the subcellular level remains technically challenging, these interactions are primarily mediated by cell adhesion peptide motifs, such as the RGD sequence, present in GelMA⁶⁵. These motifs are recognized by integrin receptors on the cell surface, which enable physical coupling to the ECM and can activate downstream mechanotransduction pathways⁶⁶.

Importantly, we devised sacrificial GMSs to transition the resulting mechanically robust printed and photocrosslinked hydrogel scaffold to a mechanically soft state during culture while maintaining structural integrity. Strategies for fabricating porous hydrogels primarily include freeze-drying, liquid-liquid phase separation (LLPS), and 3D bioprinting. Freeze-drying generates pores through sublimation of ice crystals but offers limited architectural control and is unsuitable for live cell encapsulation due to harsh processing conditions⁶⁷. LLPS enables the formation of interconnected voids by modulating the miscibility of immiscible liquids⁶⁸, offering tunable and rapid fabrication with controllable pore architectures⁶⁹. However, LLPS-based hydrogels often exhibit structural instability and limited fidelity in replicating complex tissue geometries, which can compromise long-term biological performance. 3D bioprinting allows precise control over scaffold geometry and has been employed to define porosity via printed patterns⁷⁰. More recently, porous hydrogels have been created by assembling shear-thinning microgels into interconnected networks, with tunable pore size governed by microgel dimensions and packing density⁷¹. In addition, photopolymerization-induced phase separation, which triggers phase demixing during light-mediated crosslinking, enables spatiotemporal control over pore formation in the presence of living cells, supporting tissue morphogenesis for 3D cell culture and biofabrication^72,73. The development of sacrificial microgel-based bioinks offers a promising and biologically compatible strategy for 3D printing porous hydrogels⁷⁴. In our system, GMSs serve as sacrificial porogens that dissolve under physiological conditions, enabling adjustable porosity and mechanical properties simply by varying GMS concentrations (Figure 1, S1), while retaining full geometric control through extrusion-based bioprinting.

To control spatial shape change, an inert thin OMA hydrogel layer was used as a CCF resistant layer (Figure 4). As a result, the bilayer composed of an OMA layer and a cell-laden composite hydrogel layer exhibited predefined deformation. With this system, we generated complex shape transformations by carefully arranging the materials’ spatial positions with the aid of extrusion printing without the need of a support structure or slurry bath (Figure 6). Additionally, when subjected to a differentiation environment, cartilaginous and bone-like tissues were generated while promoting CCF-mediated tissue reshaping (Figure 5). Thus, this system, demonstrating CCF dynamism, marks a significant step towards mimicking in vivo tissue development, signifies a major advancement in 4D technology, and has the potential to profoundly impact 4D tissue engineering by enabling engineered tissues to adapt to the geometries of damaged host tissues while providing essential biomechanical cues (e.g., tensile/compressive strain) that guide tissue differentiation and growth, in addition to enhancing structural complexity. It is also possible that incorporating various cell types, such as smooth muscle cells and endothelial cells, into a layered design while leveraging programmed shape morphing and scalability could facilitate the engineering of complex, hierarchically organized tissues such as vasculatures. Furthermore, by demonstrating the ability to generate complex curvatures, this system can effectively emulate some intricate tissue morphogenesis observed during native tissue development.

Future research could leverage this system as an advanced tool for normal and diseased tissue modeling by incorporating cell-directed tissue dynamics in addition to the bulk properties typically controlled in traditional 3D models. Additionally, this platform has potential applications in the development of soft biohyrid robotics⁷⁵ that utilize living cell actuation, offering advantages such as enhanced energy efficiency and autonomous motion. Another promising direction for future work includes the design of self-driven cargo or drug delivery systems that operate independently of external energy input. However, the current system also faces challenges, including: (1) low construct resolution due to the inherent limitations of extrusion printing; (2) multimodal shape transformations demonstrated in a single construct are still not complex enough to mimic many intricate shape changes occurring in vivo. Future developments must address these limitations. The rapid progress of this emerging technology is expected to have a profound impact on regenerative medicine, biomedical engineering, tissue engineering, and beyond.

In summary, a novel bioink composed of OMA microgels, GelMA, GMSs, and living cells has been developed for 4D bioprinting of living constructs that undergo controllable spatiotemporal geometric changes mediated by cell-generated forces, i.e., CCF. By forming bilayered constructs, we demonstrated that the direction of 4D bending can be controlled. Multiple degrees and directions of deformation were observed within a single construct, showing the capacity for complex shape transformations. hMSCs can be encapsulated in the living constructs at high cell density, with the supplementation of chondrogenic/osteogenic signals leading to an increased rate of 4D bending while simultaneously guiding cells down the chondrogenic/osteogenic lineages. Additionally, each layer of the bilayered constructs can be spatially arranged with the aid of 3D printing to generate complex initial and final structures. Both the cell-free and cell-laden layers can be patterned to provide new shape transformations in culture, illustrating the system’s versatility. Our system provides the first report of CCF-mediated 4D bioprinted living constructs that demonstrate highly stable, complex shape transformation in a preprogrammed and controlled fashion.

Methods

Cell culture

NIH 3T3 (ATCC) were cultured in low-glucose Dulbecco’s modified eagle medium (DMEM, Sigma, cat#D5523-50L) supplemented with 10% fetal bovine serum (FBS, Sigma, cat#18N103) and 1% penicillin/streptomycin (P/S, Gibco, cat#15140122). Upon reaching 90% confluence, cells were trypsinized, counted, and pelleted at 100 million cells per vial to be combined with 1 mL of composite bioink. After printing, constructs were cultured in high-glucose DMEM (Sigma, cat# D5648-10x1L) supplemented with 10% FBS and 1% P/S.

OMA microgel synthesis

Alginate was modified with 2% oxidation and 30% methacrylation according to previously published protocols⁷⁶. Briefly, 1% sodium alginate (10 g, Protanal LF 20/40, FMC Biopolymer) solution was dissolved in ultrapure deionized water (diH₂O, 900 mL) by stirring overnight at room temperature (RT). 216 mg of sodium periodate was dissolved in 100 mL of diH₂O, mixed with the alginate solution to achieve 2% theoretical alginate oxidation and reacted in the dark at RT for 24 hrs under stirring. 2-morpholinoethanesulfonic acid (MES, 19.52 g, Sigma) and sodium chloride (NaCl,17.53 g, Sigma) were then dissolved in the oxidized alginate solution and the pH was adjusted to 6.5 using 4 N sodium hydroxide (NaOH, Sigma). N-hydroxysuccinimide (NHS, 1.77 g, Sigma) and 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide hydrochloride (EDC, 5.84 g, Sigma) were dissolved into the mixture. AEMA (2.54 g, Polysciences) was then slowly added to the solution to achieve a theoretical methacrylation level of 30%. The reaction was conducted at RT for 24 hrs in the dark. The reacted OMA solution then was poured into excess acetone to precipitate the OMA. The precipitate was dried in a fume hood and subsequently dissolved in diH₂O at a 1% w/v concentration. The OMA solution was dialyzed for purification using a dialysis membrane (MWCO 3500, Spectrum Laboratories Inc.) for 3 days. The dialyzed OMA solution was collected and treated with activated charcoal (5 g/L, 50-200 mesh, Fisher) for 30 min. The solution was further purified and sterilized by filtering through a 0.22 μm pore membrane and then lyophilized. The ¹H NMR characterization of the OMA is shown in Figure S24. The actual methacrylation degree was determined to be 9.9% through ¹H NMR analysis, as described in literature⁷⁷.

Dry OMA was dissolved in MilliQ water to form a 2% w/v solution. The solution was then added dropwise to a beaker of 0.2 M calcium chloride under vigorous stirring and allowed to ionically crosslink for 4 hrs. Crosslinked OMA was collected and placed in a blender (Oster) with 100 mL of 70% ethanol. OMA was blended for two min before adding 50 mL of 70% ethanol, and then blending was continued for 2 min. OMA microgels and ethanol were collected into 50 mL conical tubes, centrifuged at 4200 rpm (Sorvall ST40R, thermofisher) for 5 min at 20 °C, and stored at 4 °C for future use.

GelMA synthesis

GelMA was synthesized according to previously established protocols^78,79. Briefly, 10 g of gelatin (type A, Sigma Aldrich) was dissolved in 100 mL of PBS (pH 7.4) and heated to 50 °C. Then 10 mL of methacrylic anhydride was added into the 10% gelatin solution and reacted for 1 hr at 50 °C and then stirred overnight at RT. GelMA was precipitated with acetone, purified via dialysis at 50 °C for 7 days with a MWCO 12-14k membrane (Spectrum Laboratories Inc.), sterilized via a 0.22 mm pore filter, and then lyophilized. The ¹H NMR characterization of the GelMA is shown in Figure S25. The actual methacrylation degree was determined to be 84% through ¹H NMR analysis, as described in literature⁷⁷.

GMS synthesis

To synthesize GMSs, 10 g of gelatin type A (Sigma-Aldrich, cat#G2000-500G) was dissolved in 90 mL of diH₂O (11% w/v) by heating at 60 °C. The gelatin solution was then added dropwise into 500 mL of preheated olive oil (45 °C) under vigorous stirring at a rate of 10 mL/min. After 10 min, the temperature was reduced to 15 °C while maintaining stirring. After 30 min, 200 mL of chilled acetone was added to the emulsion, and the mixture was stirred for 5 min. The gelatin microparticles were then collected by filtration and washed five times with acetone to remove any residual olive oil. Finally, the microspheres were dried in a fume hood overnight. The as-synthesized GMSs had an average size of 67 ± 34 μm, as shown in Figure S26.

Cell-free bioink and cell-laden composite bioink preparation

The cell-free bioink is composed solely of OMA microgels, which were prepared as follows: OMA microgels were reconstituted through three washes of 0.05% photoinitiator (2-hydroxy-4’-(2-hydroxyethoxy)-2- methylpropiophenone, Sigma, PI)-containing MilliQ water and two washes of low-glucose DMEM with 0.05% PI. To prepare the cell-laden composite bioink, lyophilized GelMA was weighed and dissolved directly in the above reconstituted OMA microgels (3% w/v GelMA). Lyophilized GMSs were weighed and rehydrated for 15 min in low-glucose DMEM with 0.05% PI at a ratio of 15 μL/mg. The desired volume of OMA/GelMA mixture was measured and added to the rehydrated GMSs. One mL of this solution was added to a pellet of 100 million cells to form the cell-laden bioink.

Young’s modulus measurement

Disc-shaped hydrogel samples (d = 8 mm, h = 1.0 mm) containing varying concentrations of GMSs (0, 25, 50, and 100 mg/mL) were prepared as described above. Uniaxial, unconfined compression up to 60% strain at a constant strain rate of 0.8%/s at room temperature was applied. Mechanical testing was conducted using a 225 lb actuator (TestResources, MN, USA) equipped with a 5 N load cell. The Young’s modulus of each sample, under the specified sample dimensions and testing conditions, was determined from the initial linear region of the stress-strain curve (0–10% strain) (N = 3).

SEM analysis

Cell-free bioinks containing varying concentrations of GMSs (0, 25, 50, and 100 mg/mL) were photocrosslinked at an intensity of 12 mW/cm² for 30 s to form hydrogels with a thickness of 1.0 mm. The hydrogels were then sectioned using a blade to expose the cross-sectional surface and incubated at 37 °C for 4 h to liquefy the GMSs. Subsequently, the samples were rapidly frozen in liquid nitrogen for 5 min, lyophilized for 24 h, and sputter-coated with a 10 nm layer of gold for SEM imaging. SEM was performed using a JSM-IT500HR (JEOL Ltd., Tokyo, Japan) at an accelerating voltage of 5 kV with a secondary electron detector (SED).

Porosity characterization

Cell-free composite hydrogels containing 50 mg/mL GMSs were prepared as described above and incubated in a solution of high molecular weight FITC-dextran (Mw 2,000,000 Da, Sigma-Aldrich Inc., Cat#FD2000S-100MG, 5 mg/mL in Milli-Q water) for 30 min. To evaluate porosity before and after GMS dissolution, hydrogels were cultured at either room temperature (GMS-undissolved) or 37 °C (GMS-dissolved). 3D z-stack images were acquired using a Zeiss LSM 710 BIG confocal microscope (Oberkochen, Germany), with 134 z-slices captured for the GMS-undissolved samples and 225 z-slices for the GMS-dissolved samples. Image processing and analysis were performed using Fiji software (NIH, MD, USA).

Histological cell orientation analysis

Histological cell orientation analysis was performed using Fiji software equipped with the built-in Directionality plugin. H&E-stained images in Figure 3a were first converted to 8-bit grayscale to enhance structural clarity, followed by contrast normalization and histogram equalization to improve feature detection. To reduce background noise and facilitate edge identification, a Gaussian blur (sigma = 2.0) was applied prior to analysis. The Directionality plugin was then executed using the Fourier components method. The angles were measured relative to a baseline orientation corresponding to the horizontal axis (0°). The resulting orientation histogram quantified the relative frequency of alignment angles across a −90° to +90° range. Orientation data, including dominant angle and distribution profile, were extracted and exported for further analysis.

Single layer printing

A 3 mL syringe with a 22 gauge needle was loaded with cell-laden composite bioink and placed in a Cellink BIOX3D printer (Celllink, San Diego, CA). Square constructs measuring 10 mm × 10 mm × 0.6 mm were printed from a custom-made STL file using a print speed of 4 mm/s, an extrusion rate of 1.2 μL/s, and an infill density of 60%. Printed constructs were crosslinked with UV light at an intensity of 12 mW/cm² for 30 s and subsequently transferred to 6 well plates containing 8 mL of media. To minimize disturbance to the constructs during culture, half the media was removed and replenished every day. Constructs were imaged daily using a dissection microscope and a ditital camera.

Bilayer printing

To accomplish bilayer printing, two syringes were loaded into the 3D printer: the first contained only reconstituted OMA microgels (cell-free bioink); the second contained the cell-laden composite bioink. For bilayer rectangles, a 25 mm × 25 mm × 0.2 mm square was printed with the cell-free bioink using a 4 mm/s print speed, 1.2 μL/s extrusion rate, and 80% infill density. Three cell-laden rectangles measuring 18 mm × 4 mm × 0.6 mm were then printed using the cell-laden composite bioinks directly on top of the cell-free square at parameters of 4 mm/s, an extrusion rate of 1.2 μL/s, and an infill density of 60%, spaced 4 mm apart. After crosslinking constructs with UV light at an intensity of 12 mW/cm² for 30 s, the cell-free layer was cut with a razor blade to match the geometry of the cell-laden rectangles, forming bilayer rectangles (N = 4).

To print constructs with complex cell-laden geometires, similar protocols were followed. A large cell-free layer was printed, followed by a patterned cell-laden layer. If necessary, the cell-free layer geometry was cut to match the cell-laden layer. Constructs with dual layers were constructed by printing a rectangle with dimensions of 12 mm × 9 mm × 0.6 mm. The second layer was then printed directly on top of the first layer using the cell-laden composite bioink. With the cell-free bioinkprinted first with using the cell-free bioink-ladenbioink

CytoD treatment

To further illustrate the role of cell-generated forces in the observed shape changes, the effect of CytoD (Invitrogen, cat#PHZ1063), a known inhibitor of actin polymerization, on bilayered constructs composed of a cell-free and a cell-laden layer was studied. CytoD was weighed and dissolved in sterile DMSO at a concentration of 5 mM and stored at 4 °C. Bilayer rectangles were printed and cultured in GM supplemented with 0.1% v/v CytoD, resulting in a final CytoD concentration of 5 μM. Half of the media was replaced every day, with fresh CytoD added at 0.1% v/v each time. To ensure that any change in construct behavior resulted from CytoD treatment and not DMSO, four bilayer rectangles were cultured in GM with 0.1% v/v DMSO. These media conditions were also compared to bilayer rectangles cultured in normal culture media. A total of four samples were prepared for each test unless otherwise specified (N = 4).

Chondrogenesis/osteogenesis in 4D printed constructs

Human mesenchymal stem cells (hMSC) were harvested according to previously established protocols⁸⁰. For this study, hMSCs passage 3 for chondrogenesis were cultured in low-glucose DMEM supplemented with pre-screened 10% FBS, 1% P/S, and 10 ng/mL fibroblast growth factor-2 (R&D, cat# 233-FB-MTO), while hMSCs passage 3 for osteogenesis were cultured in high-glucose DMEM supplemented with 10% FBS + 1% P/S. Upon reaching 80% confluence, cells were trypsinized, counted, and pelleted at 100 million cells (chondrogenesis) or 50 million cells (osteogenesis) per vial to be combined with 1 mL of composite bioink. After printing with passage 4 cells, constructs for chondrogenesis were cultured in high-glucose DMEM (Sigma, cat#D5648-10X1L) supplemented with 10% ITS+ (Fisher, cat#CB40352), 1% non essential amino acids (NEAA, Gibco cat#11140050), 1% P/S, 100 mM sodium pyruvate (Fisher, cat#SH3023901), 100 nM dexamethosone (Sigma, cat#D4902-100mg), L-ascorbic acid phosphate (Wako USA, cat#013-12061), and 10 ng/mL TGF-β1 (Peprotech, cat#100-21-10UG), and constructs for osteogenesis were cultured in high-glucose DMEM, 10% FBS, 1% P/S, 10 mM β-glycerophosphate (Sigma, cat#35675-100GM), 50 μM ascorbic acid (Sigma, cat# A8960), 100 nM dexamethasone (Sigma, cat#D4902-100mg), and 100 ng/mL BMP-2 (Genescript, cat# Z02913). Constructs were collected for biochemical analysis (N = 3) and histological analysis (N = 2) at 3 weeks (chondrogenesis) or 4 weeks (osteogenesis).

Histology

After culture duration was complete, samples were fixed overnight in 10% neutral buffered formalin at room temperature (N = 2). Samples were then dehydrated through one-hour incubations in 70% ethanol, 95% ethanol, 100% ethanol (Fisher, cat#1SF1C163), 1:1 volume ratio of ethanol and xylene, and two separate 100% xylenes (Fisher, cat#x3s-4). Samples were then submerged in molten paraffin (Epredia, cat#8336) for at least 24 hrs before being embedded in paraffin blocks. 5-μm sections were obtained from blocks using a Leica RM2255 microtome (Leica Biosystems, Nussloch, Germany) and mounted on microscope slides. Mayer S Hematoxylin (Fisher, cat#TA125MH) and Eosin-phloxine solution, alcoholic (Electron Microscopy services, cat#26763-03) (H&E) sections were stained with hematoxylin for 2 min followed by counterstaining with eosin for 30 s. SafO (Acros Organics, cat#477-73-6) and Fast Green (Fisher, cat#F99-10) sections were stained with 0.1% w/v SafO for 5 min followed by counterstaining with 0.05% w/v Fast Green for 1 min. Alcian blue (Fisher, cat#92-31-9) and nuclear fast red (Polysciences, cat#09773) sections were stained with alcian blue (pH = 0.2) for 30 min followed by counterstaining with nuclear fast red for 5 min.

Biochemical assays

DNA and GAG values were quantified according to previously described methods⁸⁰. Briefly, GAG values were quantified by measuring the absorbance of DMMB-bound samples at 595 nm using a plate reader (N = 3). DNA values were similarly quantified by measuring fluorescence intensity of PicoGreen-bound samples at an excitation of 480 nm and emission of 520 nm (N = 3).

ALP and calcium levels were quantified according to previously described methods²¹. Briefly, Briefly, ALP activity was assessed by measuring the absorbance of pNPP-treated samples (37 °C, 30 min) at 405 nm using a plate reader (N = 3). Calcium content was quantified using a calcium assay kit (Pointe Scientific, Cat# C7503-480), with absorbance measured at 570 nm (N = 3).

Bending angle quantification

Dissection microscope images were used to measure bending angle as described in literature²¹. As shown in Figure S27, a circle with crosshairs was superimposed on each image using Microsoft Powerpoint. The dimensions of the circle were modified to fit the arc of the construct in the image. The image and circle were then copied into ImageJ and the angle tool was used to measure the angle between one end of the construct, the intersection of the crosshairs, and the other end of the construct. Using this method, a construct that has bent into a half circle was measured as 180 degrees, while a construct whose ends are touching were measured as 360 degrees.

Live/dead staining

Printed constructs were stained with fluorescein diacetate (Sigma, cat# F7378) and propidium iodide (Sigma, cat# P4170-25MG) to visualize live and dead cells, respectively. These stains were incubated with printed constructs for 5 min, after which all media was removed, and samples were immediately imaged using a Nikon Eclipse TE300 fluorescence microscope (Nikon, Tokyo, Japan) equipped with a AmScope MU1403 camera (AmScope, Irvine, California). N = 3.

Statistics

All graphs are reported as mean ± standard deviation (±SD). For data with more than two groups, significance was determined using one-way ANOVA with a post-hoc Tukey HSD test. For data with only two groups, significance was determined using the Student’s t-test. p ≤ 0.05 was considered significant unless otherwise specified.

Supplementary Material

1

Video S1. Robust spiral configuration of the tissue construct cultured in GM for 28 days (4x speed).

2

Video S2. Robust spiral configuration of the tissue construct cultured in OM for 28 days (4x speed).

Document S1. Figures S1–S27.

Progress and Potential.

Current 4D bioprinting of tissues primarily relies on external physical or chemical stimuli for shape transformation, often overlooking the critical role of intrinsic cell contractile forces (CCFs), which drive tissue growth and remodeling. While CCFs offer a biologically compatible mechanism for shape change, the relatively small forces they generate limit their ability to induce substantial morphing in previously reported printed constructs. Herein, we have developed a composite bioink that enables the fabrication of tissue constructs with mechanically adaptive properties. These printed, freestanding constructs transition from a mechanically robust to a soft state, allowing CCFs to induce significant and progressively enhanced hydrogel contraction. This strategy provides a powerful approach for programming shape morphing in engineered tissues, facilitating the formation of complex architectures and modeling morphogenesis. This system holds great promises for advancing tissue engineering and regenerative medicine.

Highlights.

1. A composite bioink was developed for high-fidelity printing of cell-rich constructs.

2. The printed constructs underwent programmed morphing driven by CCFs.

3. Complex tissue architectures were formed through preprogrammed shape transformations.

4. The system demonstrated the integration of 4D morphogenesis with tissue regeneration.