Introduction

The intricate hierarchical structure of biological tissues epitomizes nature’s engineering prowess, where heterogeneous layers and structures spanning multiple length scales synergistically contribute to complex physiological functions1,2,3. This sophisticated architecture, characterized by gradient transitions in mechanical properties and microstructures, poses a significant challenge in the field of biomaterials and tissue engineering. Replicating such intricate designs in synthetic materials is particularly crucial for load-bearing tissue interfaces, exemplified by various tendon systems including the rotator cuff, Achilles tendon, and patellar tendon, where seamless transition from soft muscle to stiff bone is essential for efficient load transfer and joint stability4,5,6. Tendons play a vital mechanical role in force transmission and joint mobility, yet their limited regenerative capacity makes them prone to injury. The incidence of tendon tears varies significantly by anatomical location, with reported rates of 6%–39% for rotator cuff7, 0.05%-0.5% for Achilles tendon8, and 0.5% for patellar tendon injuries9. The incidence of rotator cuff tears (RCT) varies significantly with age and tear patterns. Current epidemiological data suggest that the prevalence ranges from 6% to 39%, with a marked increase in patients over 60 years old. Massive tears (>5 cm), in particular, present significant clinical challenges due to their complex pathoanatomy and high re-tear rates10,11. Current repair strategies predominantly rely on direct tendon-to-bone fixation through various suture techniques. However, these approaches often result in stress concentration at anchor points, leading to the notorious “cheese-wire” effect and subsequent repair failure (Fig. 1a)12. Conventional homogeneous patches, while aimed at reinforcing the repair, fail to address this issue effectively12,13. Insufficient mechanical strength leads to re-rupture, while excessive rigidity causes stress shielding, resulting in muscle atrophy beneath the patch and other complications that impair the healing process (Fig. 1b). These limitations underscore the need for biomimetic patches that can effectively replicate the complex mechanical transition between muscle, tendon, and bone, preventing re-tears while maintaining physiological function and promoting healing (Fig. 1c)12,14,15. The ideal patch should exhibit a continuous gradient of mechanical properties, with strength transitions spanning orders of magnitude to match the natural tissue interface from compliant muscle to rigid bone, while incorporating integrated sensing capabilities for post-surgical monitoring.

Fig. 1: Schematic of mechanically programmed strong hydrogel RCT patches.
figure 1

Schematic of repairing a massive RCT by a suturing, b suturing with a homogeneous patch, and c suturing with a mechanically programmed patch. d Illustration of mechanically programmed strong hydrogel RCT patches and monitoring of shoulder movements. e Structural schematic of a mechanically programmed hydrogel for RCT patches.

Full size image

To address these challenges, various biomaterial strategies have been explored. Traditional approaches, including 3D-printed polymer scaffolds, often exhibit limited cell infiltration and unconsiderable heterogeneously connection16,17,18, while electrospun fiber meshes, despite their biomimetic architecture, struggle with poor structural stability under physiological loading conditions19. Furthermore, these conventional materials typically lack the hydrated environment necessary for cellular functions and tissue regeneration. In contrast, hydrogels have emerged as promising candidates for tissue engineering applications due to their biocompatibility, tunable properties, and tissue-mimetic characteristics20,21. Recent advancements in hydrogel engineering have explored various approaches to fabricate heterogeneous structures with non-uniform spatiotemporal deformation properties. These include composite hydrogel structuring22,23, meso-patterned hydrogel networks24, 3D-printed hydrogel constructs25,26, and stimuli-responsive hydrogel systems utilizing localized crosslinking27,28, phase transition manipulation29,30, or selective physical/chemical modifications31. However, these hydrogel-based methods face significant limitations in achieving precise control over both mechanical properties and structural organization across the wide range necessary to mimic the muscle-tendon-bone interface. Traditional hydrogel fabrication approaches struggle with two key challenges: achieving high mechanical strength through simple, environmentally friendly methods, and more critically, programming continuous gradients in both mechanical properties and microstructural organization at interfaces. Current hydrogel systems often exhibit weak interfacial bonding between regions of different mechanical properties, compromising structural integrity. Moreover, many hydrogel fabrication techniques involve complex, multi-step manufacturing processes or require specialized equipment, limiting their scalability and clinical applicability. The creation of continuous gradients in mechanical properties and hierarchical structures within hydrogel systems, essential for mimicking natural tissue interfaces, remains a formidable challenge. Additionally, existing hydrogel materials struggle to achieve the high mechanical strength required to match the stiff bone end of the interface, while maintaining appropriate compliance at the soft tissue end, further limiting their effectiveness in replicating the full spectrum of tissue properties.

To address these limitations, we introduce a novel directional anneal-casting (DAC) method for creating gradient mechanical programmed strong hydrogels. This approach enables the design of patches that mechanically match the strength and gradient properties of natural interfaces, incorporating a strain sensing function for enhanced monitoring without causing any inflammation (Fig. 1d). Our method achieves biomimicry by spatially modulating the structure of polyvinyl alcohol (PVA) hydrogels, constructing oriented porous structures with varying densities and differentiated crystallinity (Fig. 1e). The primary innovation of our approach lies in its unit programmability, allowing precise control of mechanical properties over an unprecedented range. This broad spectrum of programmability stems from our ability to achieve exceptionally strong mechanical properties through the DAC method, with upper limits reaching tensile stress of 43.5 MPa, fracture strain of 275%, fracture toughness of 1150 kJ/m2, and Young’s Modulus of 105.6 MPa. Importantly, our method demonstrates versatility across various material systems, capable of programming the mechanical properties of multiple hydrogel systems without chemical additives. This programmability extends to one-, two-, and three-dimensional structures, broadening its potential applications beyond rotator cuff repair to a wide range of bio-fabrication scenarios. Furthermore, the DAC method offers several key advantages: it is simple, scalable, and does not require chemical additives, making it promising for large-scale manufacturing. The resulting hydrogels exhibit a continuous gradient of mechanical properties within a single, monolithic structure, closely mimicking the natural tissue interface. By bridging the gap between synthetic materials and complex biological structures, our approach opens new avenues for the development of next-generation biomaterials that can more faithfully replicate the intricacies of natural tissues, potentially revolutionizing various fields of tissue engineering and regenerative medicine.

Results

Design of mechanically programmed strong hydrogel RCT patches

To achieve structures with functional gradient mechanical behaviors similar to those found in natural tendon tissues, we employed directional anneal-casting (DAC) to characterize soft and hard structures within individual hydrogel units. Unlike the commonly observed isotropic dehydration of conventional hydrogels, the DAC method utilizes anisotropic dehydration behavior to create controlled structural gradients. The critical technique of the DAC strategy hinges on the application of encapsulation film for spatially selective encapsulation amid the annealing phase of the hydrogel (see the Experimental section). During the annealing process, the directional water migration created a gradient stress distribution from the center to the boundary of the hydrogel. This dehydration-induced stress drove the reorganization of polymer networks, accompanied by the evolution of cohesion energy as the polymer fraction increased during water molecule’ migration (Supplementary Fig. 1). The spatial control of water evaporation led to in-plate migration of polymers perpendicular to the direction of water migration, rather than flowing along the longitudinal axis. The DAC method achieved a hierarchical oriented structure and denser crystalline domains in the harder regions of polyvinyl alcohol (PVA) hydrogels, compared to the non-encapsulated parts (See Supplementary Information Text 1 and Supplementary Figs. 2, 3). This structural enhancement occurs through a dynamic process where the dehydration-induced stress facilitated the reorganization of polymer networks through the destruction and regeneration of crosslink sites. The mechanical enhancement is significantly higher than that of the non-encapsulated hydrogel parts, controlled by encapsulation layer thickness, annealing time, and annealing temperature, which collectively affect the gradient vapor leakage and subsequent polymer reorganization.

The structure-property relationship resulting from the DAC method allows for mechanical property tuning over a broad range, encompassing the strengths of human skin, tendons, and cartilage. Optimization of annealing temperature and time for overall control of individual hydrogel units is detailed in Supplementary Information Text 2 and Supplementary Fig. 4, 5. Under a characteristic thickness of 100 μm, optimal values were achieved at 8 hours and 100 °C, with strength enhanced times compared to untreated hydrogels. The control of these processing parameters determines the rate and extent of water molecule migration during the annealing process under encapsulation, ultimately defining the structural reconstruction and mechanical properties of the entire hydrogel structure. With this fundamental understanding of the anisotropic dehydration mechanism and optimized processing parameters, the DAC method can be further applied to design heterogeneous hydrogels.

Preparation of mechanically programmable hydrogel and mechanical properties

Unit programmability was achieved by establishing multiple encapsulation zones with varying thickness parameters within the hydrogel unit. A representative example of regional programming using the DAC method involves modulating PVA hydrogels, prepared by physical crosslinking through the freeze-thaw method (FT-PVA), using polyimide (PI) film. As illustrated in Fig. 2a, the process employed optimized DAC parameters of 100 °C and 8 h, while controlling PI film thickness to create mechanically programmed PVA hydrogels (MP-PVA hydrogel) with five distinct regions (regions 1–5 corresponding to 1300 μm, 1000 μm, 700 μm, 400 μm, and 100 μm, respectively). The variation in PI film thickness created different degrees of water vapor resistance, leading to distinct dehydration rates across regions. In regions with thinner PI films, the faster water migration created stronger dehydration-induced stress fields, which drove more intensive polymer chain reorganization and closer packing. This enhanced polymer chain mobility and reorganization resulted in higher polymer fractions and consequently increased cohesion energy. SEM images in Fig. 2b reveal the microstructural evolution across different regions. The systematic variation in pore characteristics directly results from the competition between polymer chain reorganization and water removal rates. In regions with thinner PI films, the higher dehydration stress promotes more efficient polymer chain alignment and densification, leading to increased pore density and more ordered structures. Figure 2c further elucidates this relationship, illustrating how the gradient in dehydration stress, controlled by PI film thickness, directs the evolution of hydrogel architecture. The faster water removal in regions with thinner PI films creates stronger driving forces for polymer network reorganization, facilitating a transition from a less ordered state to a highly oriented, densely packed configuration. This structural evolution, driven by the intensity of dehydration-induced reorganization, significantly enhances the mechanical strength, which increased from 9.31 MPa in region 1 to 43.5 MPa in region 5.

Fig. 2: The illustration of the DAC strategy for a mechanically programmed strong hydrogel.
figure 2

a Schematic of the fabrication of mechanically programmed strong PVA hydrogel involving the DAC strategy. b SEM images of PVA hydrogels with different PI encapsulation thicknesses. c Illustration of mechanical and structural evolution of hydrogel in the programmed DAC strategy. d Illustration of the DAC process leading to the optimization of structure and crystallinity for PVA hydrogel.

Full size image

Figure 2d illustrates how the DAC method modulates both the crystallinity and microstructure of PVA hydrogels. The process transforms PVA hydrogels from a low-crystallinity, disordered state to a highly crystalline, directionally aligned structure through several key steps. Initially, during freeze-thaw treatment, the formation of ice crystals within the hydrogel facilitates physical crosslinking between PVA chains, though the PVA molecular chains remain randomly oriented at this stage32. Subsequently, the directional migration of water molecules during the DAC process guides the PVA chains into an aligned configuration. This alignment is accompanied by enhanced crystallinity due to the annealing effect, resulting in the formation of more robust fibrous structures. Finally, the salting-out process, driven by the Hofmeister effect33,34, further enriches the crystalline domains within the hydrogel, leading to a more stable structural configuration. As the process progressed from 2 to 8 hours, the PVA hydrogels evolved to exhibit a refined 3D hierarchical porous structure, with the extent of refinement varying according to PI film thickness. As shown in Supplementary Fig. 6 and Supplementary Fig. 7, regions with thinner PI films, particularly region 5 (100 μm), experienced higher effective annealing temperatures during initial DAC, leading to increased crosslinking density and enhanced crystallinity. The oriented migration of free water drove this evolution by compressing pore walls and optimizing pore size distribution, culminating in a stable, robust oriented porous structure at 8 h (Supplementary Fig. 8). The resulting structure in region 5 is characterized by more compact anisotropic microchannels and enhanced local concentration of molecular chains, significantly improving resistance to delamination during stretching. In contrast, regions with thicker PI films may experience structural collapse and poor orientation, potentially disrupting crystal domains. Consequently, region 5 demonstrates more abundant crystalline domains compared to region 1 (1300 μm PI) and exhibits higher crystallinity than other regions with multilayer PI film (Supplementary Fig. 7), resulting in superior mechanical strength.

The structural variations induced by encapsulation thickness across regions within a single MP-PVA hydrogel result in a gradient change of soft and hard properties. Seamless gradient thickness DAC facilitates the formation of continuous mechanical regions (Supplementary Information Text 3 and Supplementary Fig. 9). This continuous encapsulation yields a mechanical gradation of MP-PVA hydrogel across five regions, as evidenced under a uniform mechanical loading of 30% strain (Fig. 3a). Notably, all regions are rehydrated to equilibrium for subsequent mechanical testing with swelling ratio is shown in Supplementary Fig. 10. Under these loading conditions, compliant regions exhibit larger strain relative to stiffer regions, with strain gradually decreasing from the softest to the hardest area. Correspondingly, the simulated stress profile of the regionally programmed MP-PVA corroborates these programmed mechanical characteristics, demonstrating a progressive increase in stress from the softest to the hardest region (Fig. 3b). This gradual transition in mechanical properties from soft to hard materials effectively mitigates stress concentration in the integrated material. Additionally, the control discontinuous PI encapsulation regions to realize selective mechanical programming of DAC methods (Supplementary Information Text 4 and Supplementary Fig. 11).

Fig. 3: Mechanical performance of the MP-PVA hydrogels.
figure 3

a Optical photograph of stretching the MP-PVA hydrogel at 30% strain with the different regions marked with different colors. b The simulated stress profile of regionally programmed MP-PVA at 30% strain. c Tensile stress-strain curves of MP-PVA hydrogels with different PI encapsulation thicknesses. d Mechanical gradient at the tendon-bone interface. e Optical photograph of stretching the RCT patch prepared by MP-PVA hydrogel at 100% strain (the soft and hard regions are marked using different colors). f Consecutive 500 cycles loading-unloading curves of the RCT patch prepared MP-PVA hydrogel at 40% strain for durability test at a frequency of 0.2 Hz. g SEM image of the interface between the 100 μm PI encapsulation region and the 400 μm PI encapsulation region. h Suture retention test of the MP-PVA hydrogel patch attached to the sutures at 25 N for 500 cycles.

Full size image

Analysis of the programmed MP-PVA hydrogel revealed a significant gradient in mechanical properties that closely mimics the diverse biomechanical characteristics of natural tissues. This gradient transitions smoothly from the softest (region 1) to the hardest (region 5) area, replicating the gradual change observed in complex structures with gradient performance changes, such as RCT tendons (Fig. 3c, Supplementary Fig. 12). The DAC strategy enables precise programming of mechanical properties across a wide range, with Young’s Modulus spanning from 5.68 to 105.6 MPa, ultimate stress from 9.31 to 43.5 MPa, and toughness up to 95.7 MJ m−3. Remarkably, these values align closely with those of natural RCT tendons, which typically exhibit a Young’s Modulus of 10–170 MPa, ultimate stress of 1–16 MPa, and toughness of 3.5 MJ m−3 35. The fracture toughness of different regions is quantitatively compared in Supplementary Fig. 13, with regions having thinner PI films showing superior performance (up to 1150 kJ/m2 in region 5) due to enhanced polymer chain alignment and densification. This demonstrates a notable programmable biomimetic mechanism, where the comprehensive biomimetic gradient not only replicates the mechanical properties of natural tissues but also preserves the crucial interfacial relationships between different regions. Importantly, the crystallinity and mechanical properties tests in Supplementary Figs. 14, 15 reveal consistent performance between inner and outer regions within each programmed zone, with no statistically significant differences. This remarkable uniformity stems from the uniform crystal nucleation and growth kinetics enabled by our precisely controlled annealing conditions, where the temperature field and duration were optimized to ensure homogeneous polymer chain reorganization throughout each region.

By controlling the encapsulation regions through selective application, it is also possible to achieve distinct mechanical properties in adjacent areas. This approach enables the realization of biomimetic tendon-bone bonding with a mechanical gradient (Fig. 3d, Supplementary Fig. 3). A hard-soft-hard DAC mode utilizing 100 μm and 400 μm PI thickness is implemented to design the MP-PVA hydrogel, which can be defined as an RCT patch (Fig. 3e). This configuration mimics the rotator cuffs’ tendon-bone junction, a critical area in shoulder biomechanics. The programmed rigidity offered by the 100 μm PI regions provides strain isolation capability, while regions with 400 μm PI can bear more strain. A comprehensive investigation into the reversibility, reusability, and stability of the DAC-PVA hydrogel through multiple loading-unloading tests is illustrated in Fig. 3f and Supplementary Fig. 16. Furthermore, the MP-PVA hydrogel patch demonstrates fracture energy of ~890 kJ m−2 and fatigue threshold of ~1125 J m−2, matching the characteristics of natural tendons36. To evaluate resilience under dynamic conditions, additional testing at physiologically relevant conditions (Supplementary Fig. 17) validated the hydrogel’s cyclic loading stability at higher strain levels (100%) and frequencies (0.5 and 2 Hz). The material maintained mechanical integrity after 300 cycles, with characteristic hysteresis attributed to reversible hydrogen bond dynamics serving as an effective energy dissipation mechanism. The evolution of the microstructure at the joint of the soft and hard interfaces, showing continuous and stable connections, indicates that the DAC strategy effectively modulates the mechanical properties of the hydrogel units (Fig. 3g). Shear resistance testing (Supplementary Fig. 18) revealed a shear modulus of 24 MPa and fracture toughness of 1094 kJ/m2 for the interface, exceeding the fracture energy of natural load-bearing tissues36. The gradient interface design enhances swelling resistance (Supplementary Table 1) with smooth transitions between regions rather than abrupt changes (Fig. 3a). Mechanical performance analysis (Supplementary Fig. 19) showed that the integrated patch response reflects the synergistic behavior of all regions, with fracture consistently initiating in the more porous region 4 rather than at interfaces. Thus, it permits sufficient interfacial hydration to maintain structural continuity and stable interface connections. In the context of tendon repair, the integrated interface connection provides strength and toughness to prevent breakage during stretching. Given that rotator cuff repairs frequently fail due to tendon cut-through of sutures, suture retention tests were performed to assess the ability of the tendon-like RCT patch to sustain surgical repair (Fig. 3h, Supplementary Fig. 20). No significant damage is observed in either the hard or soft regions during the suture retention test (at 25 N for 500 cycles), indicating the potential for successful suturing and fixation in vivo.

Furthermore, the DAC mechanical programming technique demonstrates universality across other hydrogel systems (alginate and polyacrylamide) and encapsulation materials, with an order of magnitude improvement in tensile strength underscoring the broad potential of the proposed DAC technology (Supplementary Information Text 5 and Supplementary Fig. 21). This versatility is complemented by structural and mechanical stability across varying environmental conditions. SEM characterization (Supplementary Fig. 22) shows that while the pore size slightly increases at higher water contents due to molecular chain movement during solvation equilibrium, the oriented porous structure maintains its characteristic morphology across different hydration states (~30%, ~60%, and ~90%). More importantly, MP-PVA hydrogels demonstrate consistent mechanical performance regardless of water content (Supplementary Fig. 23), maintaining fracture stress of approximately 43 MPa even at high water content (~90%). This stability, attributed to the robust hierarchical structure created by our directional anneal-casting strategy, is particularly advantageous for biological applications where hydration levels may fluctuate.

Demonstration of the potential application for the RCT patch

The MP-PVA patch demonstrates significant advances in cellular, tissue, and clinical validations due to its biomimetic properties and performance advantages. Utilizing PVA, a traditionally biocompatible material, and employing a pure physical fabrication method without chemical additives, the patch exhibits excellent biocompatibility. Cell Counting Kit-8 (CCK-8) assays and live/dead staining of rat myoblast cells (L6) confirmed high cell viability (Supplementary Fig. 24), providing a solid foundation for in vivo applications. Besides, Supplementary Fig. 25 demonstrates the microstructure of MP-PVA hydrogel at different scales for different PI encapsulation thicknesses. The surface structure consists of oriented fibers with nanoscale interconnections, providing micro- and nanoscale roughness that facilitates cell adhesion (Supplementary Fig. 26). Cellular inflammation analysis (Supplementary Fig. 27) confirmed that the MP-PVA hydrogel did not trigger inflammatory responses, with IL-1β and IL-6 expression levels comparable to control groups. This reorganization would better align the content with the logical flow from mechanical properties to biological performance. In vivo evaluation using typical rabbit massive RCT models further demonstrated the MP-PVA patch’s potential for clinical reconstruction of ruptured rotator cuffs (Fig. 4a). The study compares three groups: suture-only repair, homogeneous patch (H-patch, prepared by DAC method with 100 μm PI) with suture, and MP-PVA patch with suture. The surgical process involved stitching the broken rotator cuff stump to the humerus head using medical suture, then covering the suture site with either the MP-PVA or H-patch (Fig. 4b). Assessment at 2nd and 4th week post-surgery revealed superior outcomes for the MP-PVA patch group, with force-displacement curves indicating enhanced therapeutic effects compared to the H-patch (Fig. 4c). By week 4, repaired RCT with MP-PVA patch showed obviously improved mechanical strength comparing with H-patch or without patches (Fig. 4d). Meanwhile, histological analysis using H&E and Masson trichrome staining reveals orderly arranged collagen fibers and minimal inflammatory cell infiltration in the MP-PVA group (Fig. 4e). Quantitative analysis shows significantly lower expression of inflammatory marker IL-6 in the MP-PVA group (0.66 ± 0.20) compared to the H-PVA group (1.00 ± 0.16), suggesting enhanced tissue integration (Fig. 4f, together with same verification via TNMD and COL3A1 shown in Supplementary Fig. 28). Additionally, the MP-PVA patch exhibits a strain-dependent electrical response that correlates with shoulder joint movements (Supplementary Fig. 29a). Within the common range of motion (50%), it demonstrates good linearity in resistance response with a gauge factor of 1. Meanwhile, the MP-PVA hydrogel patch shows real-time, identifiable resistance signal responses during shoulder-specific rehabilitation exercises (Fig. 4g, Supplementary Fig. 29b, and Supplementary Movie 1). This property suggests potential applications in monitoring joint kinematics during the healing process, though further research is needed to establish its clinical relevance and reliability.

Fig. 4: Application demonstration of MP-PVA hydrogel as RCT patch.
figure 4

a Repair of massive RCT using RCT patch in a rabbit model. b Photographs showing the procedure of massive RCT reconstruction surgery using suturing, H-PVA patch, and MP-PVA patch, respectively. c Photographs evaluating rotator tissue reconstruction repaired using suturing, MP-PVA patch, and H-PVA patch, respectively. d Representative force-displacement curves for biomechanical tests of RCT repairs using MP-PVA hydrogel patch, homogeneous patch, and without the patch in 4 weeks. e H&E and Masson staining of RCT patch treated by MP-PVA patch after 4 weeks (regions B, I and T corresponding to bone, interface, and tendon, respectively). f Evaluation of the inflammatory response implanted patch in rabbits after 28 days (**P value < 0.01). g Monitoring arm stretching forward and bending back of shoulder joint activity characteristics in vivo experiments.

Full size image

The scalability of DAC technology

The DAC strategy represents a universal, physical, and facile hydrogel preparation method that enhances mechanical properties while enabling continuous mechanical programming of different regions. To demonstrate its industrial potential, we proposed a scalable continuous production system for manufacturing large-scale 2D mechanically programmed hydrogel films with hierarchical anisotropic aligned microporous structures (Fig. 5a). In this system, printing technology is employed to gradient-wrap the encapsulation material on a continuous hydrogel film, followed by mechanical encapsulation and in-situ annealing for directional performance enhancement. After removing the encapsulation layer, a two-dimensional hydrogel film with a gradient strength can be collected. This industrial approach offers significant advantages: customizable mechanical properties, large-scale production capabilities, high throughput for efficient mass production, and versatility across various hydrogel materials and encapsulation layers. Notably, the absence of additional reagents simplifies industrial procedures and minimizes environmental impacts from chemical discharges, while also benefiting in vivo applications.

Fig. 5: Application development for DAC strategy.
figure 5

a Schematic of the continuous fabrication of a large-area 2D mechanically programmed hydrogel patch. b continuous fabrication of mechanically programmed hydrogel fiber. c Schematic diagram of the hierarchical structure of the stomach wall. d 3D mechanical programming diagram of directional anneal-casting. e SEM image showing cross-section microstructure of region A (0 μm PI) and region B (100 μm PI) in 3D hydrogel with different pore sizes.

Full size image

The versatility of the DAC strategy extends beyond two-dimensional applications to one-dimensional and three-dimensional hydrogel materials, enabling the mimicry of hierarchical structures found in natural organs such as the mucous layer and muscle layer (See Supplementary Information Text 6). For one-dimensional programming, the DAC strategy modulates directional water migration in hydrogel fibers to achieve gradient mechanical properties along their length (Fig. 5b). In the three-dimensional context, controlled directional water migration in bulk cylindrical hydrogels demonstrates the strategy’s effectiveness for spatial mechanical programming (Fig. 5c). A representative example of 3D mechanical programming involved applying the DAC strategy to bulk cylindrical hydrogels by encapsulating 100 μm PI in their lower portion, allowing precise control of longitudinal compressive properties (Fig. 5d). This process generates an aligned anisotropic porous structure in the encapsulated region B, while region A remains uniform and non-porous without PI encapsulation (Fig. 5e). The oriented microporous structure in region B provides superior compressive performance compared to the disordered microstructure in region A (Supplementary Fig. 31), with region B showing a 16.4-fold increase in compressive modulus. Importantly, as shown in Supplementary Information Text 5, the DAC strategy has been proven effective not only with single-network PVA hydrogels but also with more complex systems such as double-network PVA/sodium alginate ionic hydrogels. This versatility in both dimensional scalability and material compatibility provides a potential platform for developing artificial tissues and organs with biomimetic mechanical properties.

Discussion

In conclusion, this work presents a green, universal, and facile directional anneal-casting (DAC) strategy for preparing strong and mechanically programmable biomimetic hydrogels. The simplicity of the DAC approach in spatio-selective programming enables seamless integration of hydrogel modules with varying tensile strengths and moduli within a single hydrogel unit. Through a single-step process, DAC constructs differentiated directional porous structures in distinct regions of the hydrogel, allowing for significant enhancement and localized, precise adjustment of mechanical properties. A key feature of this method is the maintenance of good interface connection characteristics between adjacent regions with stiffness gradients. This capability has enabled the design of a biomimetic patch that effectively simulates the tendon-bone interface strength, offering a promising alternative to traditional suturing paradigms and potentially reducing the risk of tendon re-tearing after rotator cuff repair. The versatility of the DAC strategy extends beyond two-dimensional applications, demonstrating potential for mechanical property programming in one-dimensional and three-dimensional hydrogel materials. This scalability allows for the creation of complex, biomimetic mechanical gradients that closely mimic the hierarchical structures found in natural tissues and organs. The absence of additional reagents in the DAC process enhances biocompatibility for in vivo applications while demonstrating potential for industrial scale-up. A continuous production model has been developed, showcasing the feasibility of large-scale manufacturing of mechanically programmed hydrogels with customizable properties. Therefore, the DAC strategy opens new avenues for comprehensive tissue biomimicry, providing a powerful tool for advancing the fields of tissue engineering and regenerative medicine.

Methods

Materials

Polyvinyl alcohol (PVA-1799, 98–99% hydrolyzed, Aladdin), sodium citrate (Na3C6H5O7. nH2O, 99%), acrylamide (AM, 99%), N, N’-methylenebisacrylamide (MBAA, 99%), ammonium persulphate (APS, 98%) and sodium citrate (Na3C6H5O7. nH2O, 99%) were procured from Macklin Chemical Reagent Co., Ltd. Additionally, Polyimide (PI) was purchased from Dow Corning, USA. All these reagents were used without any additional treatment. Deionized (DI) water, serving as a solvent, was consistently utilized in all preparation processes.

Preparation of MP-PVA hydrogel using mechanically programmable DAC process

Solutions of PVA (polyvinyl alcohol) at a concentration of 20 wt% are fabricated through dissolving PVA powders in DI water. This process involves intense stirring and high temperature at 90 °C. Following this, the solution is degassed through sonication, resulting in a clear, homogeneous mixture. This mixture is then carefully poured into Teflon molds. The next step involves freezing the mixture at −20 °C for 8 hours, following a thawing process at room temperature for 3 hours in an ambient air environment. To achieve the final FT-PVA (freeze-thaw PVA) hydrogels, this freeze-thawing cycle is repeated three times.

For MP-PVA hydrogel preparation, firstly, the two ends of the FT-PVA hydrogel prepared by the freeze-thaw method are fixed by the homemade fixture. PI of different thicknesses is encapsulated on the regional surface of the hydrogel to regulate the directional migration of water. Firstly, the PI films with various thicknesses (i.e., 100 μm, 400 μm, 700 μm, 1000 μm, 1300 μm) are encased completely and tightly on the hydrogel except for both ends fixed at the fixture (constructing a pathway for water molecules to move out). Thereafter, the sample together with the holding fixture is dried in the oven at different annealing temperatures (i.e., 50 °C, 75 °C, 100 °C, 125 °C) and times (i.e., 4 h, 6 h, 8 h, 10 h). The hydrogel obtained by the DAC method with fixed PI thickness was named H PVA hydrogel. Subsequently, the annealed hydrogel sample is recovered, and the surface-cured PI film is peeled off. The hydrogel is then swelled to equilibrium in a 1.5 M sodium salt solution to obtain the target hydrogel.

Preparation of non-directional anneal-casting PVA (nDAC-PVA) hydrogel

The PVA hydrogel is fixed to a homemade fixture. Then, the hydrogel is transferred to the oven along with the homemade fixture. Subsequently, the PVA hydrogel is isotropically dried at 100 °C for 8 h, and the dried PVA hydrogel is immersed in a 1.5 M sodium salt solution, achieving equilibrium to obtain the nDAC-PVA hydrogel.

Preparation of PAM, MP-PAM, alginate, MP-alginate hydrogel

PAM hydrogel is synthesized using a previously reported chemical-crosslinking method, utilizing acrylamide as the monomer. The process begins by dissolving 14 g of AM monomer in 100 mL of DI water. Subsequently, MBAA powders and thermal initiator APS are added into DI water to obtain 0.1 M and 0.15 M solutions, respectively. For every 1 mL AM solution, 1.5 μL APS solution, and 5 μL MBAA solution are added. This mixture is then transferred to a mold and cured at 60 °C to obtain the PAM hydrogel. Curing time is two hours. The PAM hydrogel is further processed using the mechanically programmable DAC method at 60 °C for 4 hours to create MP-PAM hydrogel.

In the preparation of alginate hydrogel, sodium alginate salt is combined with a calcium sulfate (CaSO4) slurry dispersed in DI water to form a 2 wt% sodium alginate mixture, as per previous reports1,2. This mixture is immediately molded and undergoes gelation in a mold, followed by storage in a 5 °C cooler for one day to complete the ionic crosslinking and form sodium alginate hydrogel. The sodium alginate hydrogel is then subjected to the DAC method at 60 °C for 4 hours, resulting in DAC-alginate hydrogel.

For the double-network hydrogel comprising acrylamide and alginate, the weight ratio between acrylamide and alginate is maintained at 3:2, and the mixture is dissolved in water. The MBAA crosslinker is added at 0.003% of the AM weight, while the physical crosslinker CaSO4 is 0.13% of the alginate weight. For each 1 mL of the precursor mixture, 1.5 μL of the APS initiator solution is added. The precursor is then cured at 60 °C for 2 hours to obtain the alginate/PAM hydrogel. Finally, MP-alginate/PAM hydrogel is prepared by applying the mechanical programming DAC method, involving 4 hours of annealing at 60 °C.

Toughness and fracture toughness test

The fracture behavior of hydrogels is tested using notched and unnotched rectangular samples. These samples have dimensions ranging from 10–20 mm in width, 1–2 mm in thickness, and 40–50 mm in length. Tested in pairs, they are used to evaluate the toughness and fracture energy. The hydrogel is subjected to a uniaxial tensile test to acquire the strain-stress curves. The stretch speed is 10 mm min-1. The area under the stress-strain curve was calculated to determine the toughness of the unnotched hydrogel. The formula for this calculation is as follows:

$$\mathrm{Toughness}={\int }_{0}^{\mathrm{\varepsilon b}}\sigma \,d{\rm{\varepsilon }}$$
(1)

where σ and εb are the tensile strength and corresponding fracture strain, respectively.

The notched samples, each featuring a 5 mm notch in the middle of the edge, are uniaxially stretched at the same speed. According to strain-stress curves, the strain (εc) is obtained, which refers to the strain at crack propagation. The corresponding unnotched samples are also tested to reach εc. The fracture toughness is then calculated by multiplying the area under the stress curve of the unnotched samples by the initial clamp distance (H). The formula for this calculation is as follows:

$$\mathrm{Fracture\; toughness}=H{\int }_{0}^{\mathrm{\varepsilon c}}\sigma \,d{\rm{\varepsilon }}$$
(2)

Pure shear experiment

The height H0 = 15 mm, thickness T = 2 mm, and the width L0 = 60 mm in the undeformed state correspond to the sample dimension parallel and perpendicular to the loading direction, respectively. For characterizations of shear resistance, unnotched samples were monotonously loaded until rupture. The shear modulus μ was calculated by the linear fitting of the slope in the linear region of the nominal stress–strain curve. μ is one quarter of the slope for the pure shear test. The notched samples, each featuring a 15 mm notch in the middle of the edge, are uniaxially stretched at the same speed. According to strain-stress curves, where the crack started to propagate was denoted as εc. The corresponding unnotched samples are also tested to reach εc. The fracture toughness is then calculated by multiplying the area under the stress curve of the unnotched samples by the initial clamp distance (H). The fracture toughness is calculated according to formula (2).

Water content measurement

The water content of the hydrogel samples is tested by recording weight loss after thoroughly drying. Any residual water on the hydrogel surface is removed using dust-free paper. The weight before (mw) and after (md) drying are recorded, and then water content is calculated using the formula: (mw − md)/mw×100%.

Crystallinity content measurement

In DSC (Differential Scanning Calorimetry) measurement, the sample is placed in a Tzero pan and heated from 50 °C to 250 °C in a nitrogen atmosphere. The heating rate is 20 °C min−1 with a flow rate of 30 mL min−1. From 200 °C to 250 °C, the heat flow curve typically displays a narrow peak, corresponding to the melting of crystalline domains. The integration of this endothermic transition from 200 °C to 250 °C yields the enthalpy of the melting of the crystalline domains per unit mass of the dry samples, denoted as Hcrystalline. The total mass of the freeze-dried sample, m, is measured. The mass of the crystalline domains, mcrystalline, is determined using the formula mcrystalline = m×Hcrystalline /H0crystalline, in which H0crystalline = 138.6 J g−1 is the enthalpy of the fusion of 100 wt% crystalline PVA. The crystallinity of dry hydrogel, Xdry, is calculated as Xdry = mcrystalline /m. The water content of the hydrogel, fwater, is obtained by comparing the weight change after freeze-drying. The crystallinity in the swollen state, Xswellon, is then calculated using the formula Xswellon = Xdry (1-fwater).

Biocompatibility experiments

Rat myoblast cells(L6) were cultured in high glucose Dulbecco’s modified Eagle’s medium (Solarbio) (Containing 10% fetal bovine serum, 100 IU/mL penicillin, 100 μg/mL streptomycin) in a humidified incubator (37 °C, 5% CO2). And cells in the logarithmic growth phase were selected for experimental research. PBS was used as a blank control group to detect the cell adhesion, proliferation, and morphology of the sensor. One mL of myoblast cells at a concentration of 2 × 103 cells mL−1 were seeded onto the sensor. Myoblast cell were stained with fluorescein diacetate (FDA) for 5 min at room temperature after being cultured for 1, 3, 5, and 7 days. After being washed with serum-free medium, the cell adhesion was observed immediately by an inverted fluorescence microscope. The Cell Counting Kit-8 kit (CCK-8) was used to evaluate the proliferation activity of cells quantitatively. The cells were seeded at a density of 1500 per well on the 96-well plate. After 1, 3, 5, and 7 days, the absorbance value of each well at a wavelength of 450 nm was recorded by a microplate reader.

Cell adhesion and cell inflammation assessment

Mesenchymal stem cells (MSCs) were cultured in a medium with 10% fetal bovine serum at 37 °C and 5% CO2. When the fusion degree was 80% ~ 90%, cell passage was carried out, and we selected the cells in the logarithmic growth stage for experimental study. Concerning cell viability staining, once the cells reached a density of about 80%, the original culture medium was discarded and rinsed with PBS. Introduce the Calcein AM/PI assay kit (Beyotime, Shanghai, China) to each well, then incubate in a 37 °C cell culture incubator in the dark for 30 min, and observe under an inverted microscope (Olympus, Japan). The live cells are marked with green fluorescence from Calcein AM, whereas the dead cells are marked with red fluorescence from PI. The cell adhesion of the MP-PVA hydrogel was detected by inoculating MSCs onto the surface of the DAC MP-PVA hydrogel. After 3 days of culture, they were stained with Calcein AM at 37 °C for 30 min. Cell adhesion was observed using an inverted fluorescence microscope after washing with PBS. Total RNA was isolated from the control and hydrogel-seeded cells using RNAiso TM Plus (Takara, Tokyo, Japan) and reverse-transcribed using Evo M-MLV Kit (ACCURATE BIOLOGY AG, Changsha, Hunan). The relative level of RNA was detected using the LightCycler-480 system (Roche Diagnostics GmbH, Mannheim, Germany) with SYBR Green SupTaq HS (ACCURATE BIOLOGY AG) for qRT-PCR (Quantitative reverse-transcription polymerase chain reaction).

Animal surgery and demonstration in vivo

This investigation was approved by the Institutional Animal Care and Use Committee and carried out according to the Shandong University Animal Experimentation Regulations (permission number DWLL-2023-070). Eighteen adult male New Zealand rabbits were used in this study. Their age was 10 weeks, and their mean weight was 2.5 kg. This study adhered to the 3 R principles (Replacement, Reduction, Refinement). Anesthesia was induced via marginal ear vein injection with 20% urethane (4 mg/kg), and analgesia was maintained using flurbiprofen axetil (50 mg/mL). The skin prepared, and a 3-cm median incision was made at the greater tuberosity of the humerus, and the deltoid muscle was bluntly separated until the subscapularis muscle was exposed. A blade was used to transversely break off the subscapularis muscle tendon stop point, and the remaining tendon tissue on the greater tubercle was cleaned. The subscapular tendon (5 × 5 mm) was resected from the greater tuberosity to create rotator cuff defects. One 1 mm–diameter tunnels were drilled using an electric drill from the middle site of the subscapular tendon enthesis to the lateral area of the greater tuberosity. And tendon edge was sutured directly to the humeral head using a nylon suture. The defect was subsequently sutured with MP-PVA hydrogel patches and H-PVA hydrogel patches. After irrigation, the surgical incision was closed. The blank control group was sutured without patches. All rabbits were allowed to move freely within their cages after the surgery. And 400,000 U of penicillin was administered into the muscle to prevent infection three days post-surgery. Euthanasia in this experiment was performed via CO2 inhalation overdose.

Histological analysis

On the day of sacrifice at week 4, the samples of each group were fixed with 4% neutral formaldehyde buffer solution, followed by decalcification in a 10% EDTA solution for 8 weeks. Subsequently, the specimens underwent dehydration and were embedded in paraffin. The resulting 10-μm sections were stained using standard hematoxylin and eosin (H&E) staining or Masson’s trichrome procedures.

Inflammation response

The surrounding tissue was quickly separated, total RNA is extracted using the Trizol, and the RNA was reverse-transcribed into cDNA using the reverse transcriptase kit. IL6, TNMD, and COL3A1-specific primers designed by the manufacturer are used, including GAPDH controls. The PCR reaction system is constructed, and the real-time quantitative PCR instrument is used for determination. The reverse transcription products are amplified by PCR. The data are expressed as relative Ct value, and the relative expression is calculated by 2−ΔΔCt method.

The sequence of primers required for PCR is as follows:

IL6: forward,5’- ACTGGCGGAAGTCAATCTGC-3’

reverse,5’-GAACTCCATCAGCCCCGAAG-3’

TNMD: forward, 5’-CAAAATGAGCAGTGGGTGGTC-3’

reverse,5’-TCCTCACTTGCTTGTCTGGC-3’

COL3A1: forward, 5’-AACCCGAACCGTGCCAAATA-3’

reverse,5’-CAACAGTGCGGGGAGTAGTT-3’

GAPDH: forward,5’-TGGTGAAGGTCGGAGTGAAC-3’

reverse,5’-GCCGTGGGTGGAATCATACT-3’