Recapitulating the native tendon environment in a synthetic 3D anisotropic hydrogel as an engineered extracellular matrix

Abstract

The ability of tendons to transmit forces from muscle to bone is fundamentally attributed to the hierarchical anisotropy of the tissue. After injury, disorganized fibrotic scar tissue forms during the natural healing process, resulting in inferior mechanical properties that often lead to reinjury and limited restoration of function. Therefore, intervention is necessary to facilitate regenerative healing of the tendon. Polymeric biomaterials have historically been used to guide cell behavior, showing promise for the use of topological guidance and cell-mediated matrix remodeling as mechanisms for promoting regeneration. Here, we fabricated 3D scaffolds for tenocytes using anisotropic poly(ethylene glycol)-based hydrogels that recapitulate both the biophysical and biochemical properties of native tendon. These materials were synthesized using a two-stage polymerization strategy that includes an initial crosslinking step facilitated by thiol-Michael addition, an intermediate mechanical stretching step to align the polymer network, and a second-stage crosslinking step facilitated by a thiol-ene reaction. The application of 300% strain during the mechanical alignment of the network resulted in highly oriented materials (S = 0.38). Furthermore, a matrix metalloproteinase (MMP)-degradable peptide was incorporated into the network to facilitate cell-mediated remodeling of the scaffold. After 14 days of exposure to exogenous MMP2, a sufficient number of crosslinks were degraded for alignment to be lost (S = 0.03). When tenocytes were encapsulated in the 3D anisotropic hydrogels, they adopted the anisotropic morphology of the polymer network and deposited an extracellular matrix mainly comprised of type I collagen, indicating a pro-regenerative environment. Comparatively, isotropic materials of the same composition induced a random orientation of encapsulated tenocytes and a matrix containing primarily collagen III was deposited, indicating a fibrotic environment. Collectively, these results demonstrate the successful use of a synthetic scaffold with tunable biophysical and biochemical properties for recapitulating the native tendon environment and promoting regenerative cell behavior.

Graphical Abstract

Introduction

The primary function of tendon is the transmission of force from muscle to bone. The anisotropic organization of this tissue provides strength along its aligned axis, which is crucial for transmitting high-magnitude forces in the musculoskeletal system.^{1, 2} Unfortunately, injuries involving tearing or rupture of the tissue are common, with over 16 million occurring annually in the United States.³ These injuries are often debilitating; as a result of force transmission hindered by damage to the tissue, chronic pain and disability lead to reduced quality of life. As such, surgical intervention is frequently employed to repair the damaged tissue. However, these repaired tissues often heal without full restoration of function, and the reinjury rate is high.^4–8 For example, though over $400 million is spent on flexor tendon repairs in the United States, 40% of these repairs result in functional limitations.^{9, 10}

The poor outcomes of tendon repair are commonly attributed to the formation of fibrotic scar tissue during the healing process. When left to heal naturally, the tissue is not regenerated with the anisotropic orientation of uninjured tendon.^{11, 12} Instead, an inner aligned extracellular matrix (ECM) bridge is formed across the injury site, and the surrounding tissue consists of disorganized ECM.^{13, 14} Since force transmission relies on the mechanical properties of the anisotropic, organized tissue, the inferior mechanical properties of the disorganized scar tissue lead to poor restoration of tendon strength and range of motion.^15–18 Therefore, intervention in the healing process is necessary to promote regenerative, rather than fibrotic, healing after injury.

Numerous studies have shown that polymeric biomaterials with patterned structural anisotropy promote alignment of tenocytes when used as scaffolds to guide cell behavior. Materials such as collagen,^19–22 cellulose,^22–24 poly(caprolactone),^25–28 silk fibroin,²⁹ poly(p-dioxanone),³⁰ poly(L-lactic acid),³¹ polyurethane,³² and poly(vinyl alcohol)³³ have been used to create anisotropic scaffolds that leverage electrospun polymer fibers, nanofibrous composites, or mechanical deformation to promote the alignment of tenocytes. Notably, when these scaffolds are used for tenocyte culture, the cells exhibit regenerative behavior as characterized by the upregulation of tenogenic genes, such as tenomodulin, scleraxis, and mohawk. Furthermore, some results indicated increased production of collagen I relative to collagen III, indicating a regenerative healing process rather than a fibrotic one. While these findings provide a framework for the use of polymeric biomaterials for promoting tenogenic environments, the complex processing techniques or inclusion of additives for introducing anisotropy limit the scalability of material properties.³⁴ Furthermore, the materials used in many of these approaches have limited tunability of chemical components, hindering the ability to integrate biochemical cues alongside biophysical cues in the scaffolds.

The integration of biochemical cues is critical in the design of a polymeric scaffold as an engineered extracellular matrix that recapitulates the native tendon environment. In particular, cell-mediated degradation of the scaffold is critical for mimicking the remodeling of the extracellular matrix.³⁵ This remodeling of the ECM is essential for tendon homeostasis and repair.³⁶ As such, numerous studies have demonstrated that the incorporation of degradable bonds in the polymer network of a polymeric scaffold enables cells to remodel their environment, recapitulating the dynamic environment of a native ECM.^37–40 Of particular interest is the inclusion of matrix metalloproteinase (MMP)-degradable bonds as crosslinks within these networks.^41–46 These MMP-degradable bonds are often incorporated into a peptide sequence that is then integrated into the polymer via the reaction of the thiol groups on cysteine residues flanking either side of the MMP-degradable sequence. If sufficient degradable groups are included in the network such that a reverse gel point is achievable, the polymeric scaffold can be degraded and replaced by native ECM deposited by the cells.

While the studies discussed thus far have independently demonstrated the importance of structural anisotropy for guiding cell orientation and network degradability for cell-mediated remodeling of scaffolds in the context of creating regenerative environments for tenocytes, to the best of our knowledge, there has been little work on combining these key material properties for promoting regenerative behavior of tenocytes. Motivated by these prior demonstrations, this work designs poly(ethylene glycol) (PEG)-based hydrogels as anisotropic 3D hydrogel scaffolds with MMP-degradable crosslinks. These materials are then utilized in an in vitro model to investigate key material design parameters that promote regenerative tenocyte behavior. These hydrogels are formed by a two-stage polymerization mechanism; in this process, thiol-Michael addition creates an initial lightly crosslinked network that is subsequently stretched to align the polymer chains. While maintaining applied strain, a secondary thiol-ene reaction is initiated to further crosslink the material and retain the anisotropic orientation of the network. We demonstrate that these scaffolds balance the key aspects of structural guidance and matrix remodeling over 14 days, which is representative of the early stages of healing.^{14, 47–50} Encapsulated primary human tenocytes adopt the anisotropic orientation of the polymer network, and this organization persists through 14 days. While these hydrogel networks lose anisotropy within 14 days when exposed to exogenous MMP, data confirm that structural guidance persists long enough for cells to elaborate sufficient native extracellular matrix (ECM) to retain anisotropy. Furthermore, the deposited ECM is primarily composed of type I collagen, as compared to ECM containing primarily type III collagen deposited by cells in isotropic hydrogels of equivalent composition, indicating that the combination of biophysical and biochemical cues provided by these scaffolds is key for establishing a pro-regenerative environment for tenocytes.

Materials and Methods

Materials:

Eight-arm 20 kDa poly(ethylene glycol) (PEG) was purchased from JenKem Technology and functionalized with norbornene end-groups (PEG-NB) as described in prior studies (>95% functionalization, Figure S1).⁵¹ Lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) was also synthesized according to published methods.⁵² Four-arm 20 kDa PEG functionalized with maleimide end-groups (PEG-mal) was purchased from JenKem Technologies. Peptides (MMP degradable: GKKCGPQGIWGQCKKG, Nondegradable control: GKKCGIQQWGGPCKKG, Cell adhesive epitope: CGRGDSG) were synthesized using a Liberty Blue 2.0 Microwave Peptide Synthesizer (CEM Corporation) and purified to >90% using a Prodigy 2.0 Peptide Purification System (CEM Corporation). Solid state synthesis was performed using Fmoc-Gly-Wang resin (CEM Corporation) and amino acids were used to build the peptide sequence using diisopropylcarbodiimide (DIC) coupling reactions in DMF with Oxyma added to assist with coupling efficiency. Peptides were cleaved from resin using a mixture of Trifluoroacetic acid, 2,2′-(Ethylenedioxy)diethanethiol, and Triisopropyl silane (Sigma Aldrich). Molecular weights of peptides were verified via matrix-assisted laser desorption ionization time of flight (MALDI-TOF, Bruker MicroFlex Smart LS) mass spectrometry, and purity was assessed using high performance liquid chromatography (HPLC) (Figure S2).

Hydrogel Synthesis:

All hydrogels were synthesized via two-stage crosslinking reactions from of stock solutions of four-arm PEG-mal (20 wt%), eight-arm PEG-NB (5 wt%), dicysteine functionalized peptides (GKKCGPQGIWGQCKKG or GKKCGIQQWGGPCKKG, 5 wt%), monocysteine-functionalized cell adhesive peptide (CGRGDSG, 1 wt%), and LAP (1 wt%), all prepared in phosphate buffered saline (PBS, Gibco). All macromer, peptide, and initiator stock solutions were combined in solution and diluted to an overall concentration of 5 wt% solids with 0.1 wt% LAP before network synthesis. For hydrogels with encapsulated tenocytes, cells were also incorporated in the initial starting material solution at a concentration of 1×10⁶ cells/mL. To ensure well-mixed starting materials, all thiol-based starting materials, cells, and additional PBS needed to dilute the final composition to 5 wt% solids were combined in one solution, and the PEG-based macromers were mixed with the photoinitiator in a second solution. Each of these two intermediate starting material mixtures was vortexed before mixing the two solutions together between two RainX-coated glass slides separated by 1 mm-thick rubber gasket spacers. After mixing these two intermediate mixtures into a single final solution, the first-stage polymerization was carried out by reaction of the four-arm 20 kDa PEG-mal with dicysteine-functionalized peptide (GKKCGPQGIWGQCKKG for degradable hydrogels, GKKCGIQQWGGPCKKG for nondegradable hydrogels) and monocysteine-functionalized peptide (CGRGDSG) via thiol-Michael addition for 20 minutes at room temperature in the dark with an overall thiol:maleimide ratio of 1.15:1, creating a crosslinked network with 5% pendant excess thiol groups. Dicysteine-functionalized peptides were used in a thiol:maleimide molar ratio of 1.05:1. The adhesive peptide, CGRGDSG, was incorporated in a thiol:maleimide molar ratio of 0.1:1. The resulting crosslinked gel was removed from the glass slide affixed to calipers by spanning the material across an initial gap of 0.3 cm and affixing the hydrogel to the jaws with tape. The caliper jaws were then expanded to stretch the hydrogel to a final strain value of 100, 200, or 300%. Unstretched (0% strain) networks were used for isotropic control experiments. While maintaining the applied strain, the second-stage polymerization reaction was initiated to further crosslink the material and retain alignment of the polymer chains. In the second-stage crosslinking reaction, the eight-arm 20 kDa PEG-NB was reacted with the excess thiols (1:1 norbornene:excess thiol) in the first-stage network via photopolymerization (365 nm, 5 mW/cm², 3 min). After photopolymerization, the hydrogel was cut from the caliper jaws, removing any material taped to the jaws that had not been stretched or exposed to light during the second-stage reaction. Final materials were placed in PBS (acellular materials) or tenocyte media (materials with encapsulated cells, 1:1 DMEM/F12 (Gibco) containing 10% fetal bovine serum (Bio-Techne) and 1% antibiotic-antimycotic (Gibco)). Hydrogels containing encapsulated cells were incubated at 37 °C and 5% CO₂ for 1, 3, 7, or 14 days. Media was exchanged with fresh tenocyte media one day after encapsulation and every 4 days thereafter.

In-situ Rheology:

The formation of 300 μm-thick hydrogels via two-stage polymerization was monitored by in situ oscillatory shear rheology at 1% strain and 1 rad/s on a DHR-3 rheometer equipped with a quartz photo-curing plate and an 8 mm parallel plate geometry (TA Instruments). The starting material solutions were prepared as described for hydrogel synthesis and deposited between the rheometer plates. The first-stage crosslinking step was monitored via oscillatory shear measurement for 20 minutes in the dark, followed by 3 minutes of light exposure (365 nm, 5 mW/cm²).

Tensile Measurement of Elastic Modulus:

Tensile tests were conducted on an RSA G2 DMA (TA Instruments) at a strain rate of 20%/min. Anisotropic hydrogels were cut into strips of 1.5 mm width and clamped with the direction of network alignment either parallel or perpendicular to the direction of strain. Elastic modulus was calculated from the linear region of the stress-strain curve (5–15%).

Wide-Angle X-ray Scattering (WAXS):

Static 2D wide-angle x-ray diffraction patterns were collected on a SAXS/WAXS system housed within the BioPACIFIC facilities at the University of California, Santa Barbara. Acellular hydrogels were dried before WAXS characterization to improve signal. Diffraction patterns were acquired on samples using 9 keV beam energy and 60 s exposures. Data reduction and order parameter calculations were performed in Igor using the Nika analytical package and through a custom script in MATLAB, respectively. Azimuthal intensity integration was performed over the entirety of the anisotropic lobes in the diffraction pattern, with intensity averaged over a 10 pixel width (0.075 CCD pixel size). Background subtraction was performed using a diffraction pattern collected with no sample between the beam and the detector. The Hermans orientation parameter (S) was calculated in Matlab using equations 1 and 2 where ϕ is the azimuthal angle and I(ϕ) is the angle-dependent scattering intensity from the WAXS patterns⁵³:

$$〈cos2\varphi 〉=\frac{{\int }_0^{\pi }I(\varphi )co{s}^2\varphi sin\varphi d\varphi }{{\int }_0^{\pi }I(\varphi )sin\varphi d\varphi }$$ (1)

$$S=\frac{3〈\text{cos}2\varphi 〉-1}{2}$$ (2)

MMP Hydrogel Degradation Experiments:

Acellular hydrogels (isotropic and anisotropic) of both degradable and nondegradable compositions were synthesized by procedures described above and placed in solution with 10 nM human MMP2 recombinant protein (PeproTech) in buffer (50 mM Tricine (Thermo Scientific), 50 mM NaCl (Sigma Aldrich), 10 mm CaCl₂ (Sigma Aldrich), 50 μM ZnCl₂ (Thermo Scientific), and 0.05 wt% Brij35 (Thermo Scientific) in diH₂O, pH 7.4). MMP2 solution was discarded and replaced every 48 hours until hydrogels were removed from the solution and dried after 14 days.

Collagenase Hydrogel Degradation Experiments:

Acellular degradable hydrogels were synthesized by procedures described above and placed in either PBS or tenocyte media (1:1 DMEM/F12 containing 10% fetal bovine serum and 1% antibiotic-antimycotic). The hydrogels were incubated (37 °C, 5% CO₂) for either 1 or 3 days before being placed in solution with 1000 U/mL collagenase type II (Gibco) dissolved in tenocyte media or PBS (collagenase solution type was kept consistent with the type of solution used for incubation of each hydrogel, i.e. PBS or media). Hydrogels were incubated in the collagenase solution for 24 hours, with visual observation after the initial 30 minutes, 1 hour, and 2 hours to determine if gels had degraded. After 24 hours, media or PBS was removed to confirm the presence or absence of nondegraded gels.

Primary Human Tenocyte Isolation and Culture:

Tenocytes were isolated from whole human quadricep tendon tissue via digest in collagenase type I solution. Tendon tissue was obtained under University of Oregon Institutional Review Board approval (IRB Protocol STUDY00000501) 0.15 wt% collagenase type I (Gibco) was dissolved in tenocyte culture media (1:1 DMEM/F12 (Gibco) containing 10% fetal bovine serum (Bio-Techne) and 1% antibiotic-antimycotic (Gibco)). A volume of approximately 1 cm³ of tendon tissue was added to 1 mL collagenase solution with 1 mL fresh tenocyte media and incubated overnight at 37 °C and 5% CO₂. After incubation, the solution was strained through a 100 μm cell strainer. The filtrate was collected in a conical tube and centrifuged at 2000 rpm for 5 minutes at 20 °C. The supernatant was aspirated from the cell pellet, and the cells were resuspended in fresh PBS and centrifuged again at 2000 rpm for 5 minutes at 20 °C. Supernatant was aspirated, and cells were resuspended in 10 mL fresh tenocyte media. The resulting cell suspension was placed in a 10 cm² tissue culture flask and incubated at 37 °C and 5% CO_2. The media was exchanged with fresh tenocyte media after the initial 48 hours and every 4 days thereafter.

Once the cells reached 80% confluence, the media was aspirated, and the cells were rinsed with PBS before being treated with 3 mL of 0.5% Trypsin-EDTA and incubated for 3 minutes. 7 mL fresh tenocyte media was then added, and the cell suspension was centrifuged at 2000 rpm for 5 minutes at 20 °C. The supernatant was aspirated, and cells were resuspended in freezing media (1:1 DMEM/F12 containing 20% fetal bovine serum, 10% dimethyl sulfoxide, and 1% antibiotic-antimycotic) at a concentration of 200,000 cells/mL. 1 mL aliquots of cell suspension were then cooled to −80 °C at a rate of 1 °C/min in a Corning CoolCell LX Cell Freezing Vial Container before transferring to storage in liquid nitrogen. All cells used in these experiments were frozen once after isolation and used in experiments on passage 2. Isolated cells were characterized by qPCR to verify the presence of key tenogenic genes (TNMD, SCX, MKX, COL1A1, COL3A1). Isolated cells quantitatively expressed all five genes when normalized to GAPDH (Table S1), verifying that the cells used for hydrogel experiments were tenocytes. To prepare samples for qPCR analysis, cells were rinsed with PBS and placed in TRK lysis buffer (Omega Bio-tek). RNA was extracted using an E.Z.N.A. extraction kit (Omega Bio-Tek) and reverse-transcribed using a qScript cDNA Supermix kit (Quanta Bio). Both extraction and reverse transcription were conducted according to the manufacturer’s instructions. Quantitative PCR was conducted on a QuantStudio 5 real-time PCR system (Thermo Fisher Scientific) using PerfeCTa SYBR Green FastMix (Quanta Bio). Genes tested and corresponding primer sequences were sourced from prior studies^54–58 and are listed in Table S1. qPCR experiments were run with an initial hold at 95 °C for 3 minutes, followed by 40 cycles of denaturation at 95 °C for 10 second and annealing at 58 °C for 30 seconds.

To prepare tenocytes for encapsulation in hydrogels, 1 mL of cell suspension was thawed and added to a 75 cm² treated polystyrene tissue culture flask with 9 mL fresh tenocyte media (1:1 DMEM/F12, 10% fetal bovine serum, 1% antibiotic-antimycotic) and incubated at 37 °C and 5% CO₂. Media was exchanged with fresh tenocyte media after the initial 48 hours and every 4 days thereafter. Once the cells reached 80% confluence, the media was aspirated, and the cells were rinsed with PBS before being treated with 3 mL of 0.5% Trypsin-EDTA and incubated for 3 minutes. 7 mL fresh tenocyte media was then added, and the cell suspension was centrifuged at 2000 rpm for 5 minutes at 20 °C. Cells were counted, and the suspension was divided into aliquots containing the appropriate number of cells for encapsulation in a single hydrogel at 1 × 10⁶ cells/mL. The aliquots were centrifuged at 2000 rpm for 5 minutes at 20 °C. The supernatant was then aspirated, and the cell pellet was resuspended in the starting mixture of thiol-functionalized components before being mixed into the final hydrogel solution.

Fluorescent staining:

Tenocytes were fixed and stained using rhodamine phalloidin reagent (Abcam) and DAPI (Sigma Aldrich) for visualization of the actin cytoskeleton and nucleus, respectively, while encapsulated within the hydrogels. All steps were performed at room temperature and hydrogels were immersed in 1.5 mL of each solution unless noted otherwise. Hydrogels containing encapsulated cells were first removed from tenocyte media and placed in fresh PBS for 5 minutes. Then, hydrogels were immersed in 4% paraformaldehyde (Thermo Scientific) for 15 minutes, followed by 5 minutes in PBS. A solution containing 1% bovine serum albumin (Thermo Scientific) and 0.1% Triton X-100 (Sigma-Aldrich) was then prepared in PBS, and the hydrogels were immersed in this solution for 10 minutes. The hydrogels were then subjected to another 5-minute wash in fresh PBS before being immersed in a solution containing 1 μg/mL DAPI and 1 μL/mL rhodamine phalloidin reagent stock solution (99.9% DMSO, 0.1% phalloidin-tetramethylrhodamine) in PBS. After removing the hydrogels from the DAPI/rhodamine phalloidin solution, they were immersed in fresh PBS for 5 minutes. The PBS was then exchanged for fresh PBS and the hydrogels were stored at 4 °C for 24 hours. The PBS was then exchanged again for fresh PBS before imaging.

Separate samples containing encapsulated tenocytes were used for characterization of cell viability. After 14 days in culture, hydrogels with encapsulated tenocytes were subjected to two five-minute washes at room temperature in 1.5 mL fresh PBS. After PBS washing, the hydrogels were immersed in 1 mL of a solution containing 2 μM calcein and 4 μM ethidium homodimer-1 in PBS (diluted from the starting components of the LIVE/DEAD Viability/Cytotoxicity Kit for mammalian cells, Invitrogen) and incubated at 37 °C for 30 minutes. Hydrogels were then immersed in 1.5 mL fresh PBS for 5 minutes. The PBS was then exchanged for 1.5 mL fresh PBS before imaging.

Cell Imaging:

Fluorescent imaging of cells encapsulated within hydrogels was performed on multiple microscopes, including a Zeiss LSM 900 confocal microscope with plan-apochromat 10x NA 0.45 objective and Leica Thunder Imager with 10x objective. All images were taken at least 500 μM away from the outer edges of gels and at least 200 μM from the well plate to exclude cells that may have been impacted by contact-guidance effects of edges.

Characterization of Cell Orientation:

The actin cytoskeleton was used for quantifying cell orientation within hydrogels. Fluorescent images were rotated to align the direction of hydrogel network alignment at 0°. Z-stacks of images spanning 800 μm through the hydrogel thickness were stacked via orthogonal projection for orientation quantification. Orientation analysis was performed using ImageJ. Images were converted to 32 bit and inverted with a gaussian blur of 2.0 and the “find edges” function applied. OrientationJ horizontal alignment analysis was used with σ = 5 pixels, which is the approximate width of a single tenocyte in pixels. Counts from the OrientationJ were converted to frequencies and frequency plots were averaged across replicates for each sample type, with data presented as averages ± standard deviation Full width half maximum (FWHM) was calculated from the averaged frequency plots for each sample type.

Viability Quantification:

Fluorescent images of cells stained with calcein/ethidium homodimer-1 were used for quantifying viability in ImageJ. Images were converted to 8-bit, thresholded, and segmented. The number of cells was quantified with the Analyze Particles tool. Percent viability was calculated as the number of cells in the live cell channel relative to total number of cells in the live and dead channels.

Collagen Staining and Imaging:

Collagen deposited by tenocytes within the hydrogel scaffolds was characterized using picrosirius red staining. Anisotropic and isotropic degradable hydrogels with encapsulated tenocytes were rinsed with PBS and fixed with 4% paraformaldehyde as described for fluorescent staining experiments on day 14 post-encapsulation. After fixing, samples were immersed for 24 hours in a solution of 1 w/v% poly(vinyl alcohol) (MW 31,000–50,000, Sigma Aldrich) dissolved in PBS. Samples were then embedded in O.C.T. compound (Fisherbrand) and frozen at −80 °C. Frozen samples were cut into 20 μm sections at −20 °C using a cryostat (Leica CM3050 S) and mounted on glass slides. A Polysciences Picrosirius Red Stain Kit was used according to the manufacturer’s protocol to stain the 20 μm sections. After staining, the sections were dried and imaged using a Zeiss Axio Imager A2 microscope with an EC Epiplan-Neofluar 10x/NA 0.25 Pol M27 objective. One image was taken in brightfield polarized transmission mode with the sample between crossed polarizers. A second image was taken of each sample in the same configuration with polarizers removed. The area of collagen I (red) and collagen III (green) regions of the sample were quantified by color threshold and area quantification in ImageJ, and the anisotropy score for each sample was quantified using FibrilTool in ImageJ.

Statistical Analysis:

All averages are presented as average ± standard deviation. OriginPro was used for statistical analysis. Paired t-tests were used to compare elastic moduli in the directions parallel and perpendicular to the direction of alignment. p < 0.05 was considered significant. All cell experiments were performed with 3 biological replicates and 3 technical replicates.

Results and Discussion

Poly(ethylene glycol) (PEG)-based hydrogels that mimic the native tendon environment were designed as engineered extracellular matrices (eECM) for tenocytes to promote alignment and regenerative behavior of the encapsulated cells (Figure 1). These 3D scaffolds integrated both biophysical and biochemical cues by combining anisotropic polymer network topology for structural guidance with enzymatically degradable crosslinks for cell-mediated matrix remodeling. These hydrogels were synthesized by a two-stage polymerization method. In the first stage, a lightly crosslinked network was formed via thiol-Michael addition (Figure 1a–b). The lightly crosslinked polymer network was then uniaxially strained to align the polymer chains (Figure 1c). While maintaining the applied strain, the material was further crosslinked by a photoinitiated thiol-ene reaction to retain the anisotropic orientation of the polymer chains in the freestanding gel once the applied strain was removed (Figure 1c–d). To facilitate this sequence of reactions, a combination of 4-arm 20 kDa maleimide-functionalized PEG, 8-arm 20 kDa PEG, dicysteine-functionalized peptide (GKKCGPQGIWGQCKKG), and monocysteine-functionalized peptide (CGRGDSG) was combined in solution at an overall concentration of 5 wt% (Figure 1f). Lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) was also included in the solution as a photoinitiator at 0.1 wt%. When using these materials as 3D scaffolds for tenocytes, the cells were also mixed in this initial solution. However, initial experiments were performed on acellular hydrogels to characterize material properties and verify both anisotropy and degradability.

Figure 1.

Anisotropic hydrogels were synthesized as 3D scaffolds for encapsulated tenocytes by a process in which (a) all starting materials were mixed in solution before (a to b) a lightly crosslinked network was formed via thiol-Michael addition, followed by (c) mechanical stretching of the hydrogel and (c to d) a secondary photoinitiated thiol-ene crosslinking step to retain applied strain and orientation of the network. The anisotropic polymer network provides a template for (e) alignment of tenocytes in a pro-regenerative environment. (f) Shows representative starting materials.

Overall, the ratio of thiol to maleimide functional groups in the starting materials was held at 1.15:1 to create a network structure that contained pendant thiols after the Michael addition reaction between thiols and 4-arm PEG-maleimide. The thiol groups were introduced into the network with a combination of monocysteine- (0.1:1 thiol:maleimide) and dicysteine-functionalized (1.05:1) peptides. A dicysteine-functionalized peptide (GKKCGPQGIWGQCKKG) facilitated the crosslinking of multi-arm PEG macromers and introduced a platform for cell-mediated network remodeling.⁴¹ A monocysteine peptide (CGRGDS) was included as an adhesive epitope necessary for in vitro cell culture experiments.⁵⁹ The thiol-Michael reaction was carried out for 20 minutes, resulting in a crosslinked gel with a storage modulus of 650 Pa measured using in-situ shear rheology (Figure 2a,i, plotted with loss modulus in Figure S3). The resulting pendant thiol groups were then reacted with the 8-arm PEG-norbornene, which was added in a 1:1 ratio with the excess thiols, to form additional crosslinks in the photoinitiated thiol-ene step (5 mW/cm², 3 min). This second thiol-ene crosslinking step resulted in further stiffening of the material, indicated by the increase in storage modulus to 1400 Pa, measured via in-situ shear rheology (Figure 2a, ii). Notably, our prior work has demonstrated that thioether bonds formed by thiol-Michael addition of thiols and maleimides are susceptible to radical-mediated degradation, so it is likely that some of the first-stage crosslinks are degraded during the thiol-ene reaction and act as sacrificial crosslinks, leading to a greater proportion of second-stage crosslinks contributing to material properties and reinforcing the anisotropic orientation of the network.^{60, 61}

Figure 2.

(a) The two-stage polymerization process used for synthesizing hydrogels was monitored via in situ shear rheology (1% strain, 1 rad/s). The first stage of crosslinking proceeded via thiol–Michael addition between maleimide- and thiol-functionalized precursors (purple shading, schematic a, i), followed by second-stage crosslinking facilitated by a photoinitiated thiol–ene reaction between norbornene-functionalized precursors and pendant thiols in the first-stage network (orange shading, schematic a, ii). (b) Uniaxial mechanical stretching was applied between the two crosslinking stages to induce anisotropy in the hydrogel network (schematic b, i). Wide-angle x-ray scattering (WAXS) was used to evaluate network alignment in hydrogels stretched to (b, ii) 0%, (b, iii) 100%, (b, iv) 200%, and (b, v) 300% strain during fabrication. Arrows indicate the direction of polymer alignment and inset S values represent Hermans orientation parameters calculated from each diffraction pattern. All hydrogels were synthesized with MMP-degradable crosslinks. (c) Hydrogels subjected to a 300% applied strain during the mechanical alignment step of fabrication were characterized by uniaxial tensile testing (20%/min) in directions parallel (pink) and perpendicular (gray) to the alignment. Elastic moduli (E) were calculated from the linear region of the stress–strain curve (5–15% strain); n = 4.

After confirming the successful synthesis of a two-stage network with distinct crosslinking steps, the amount of applied strain required to invoke anisotropic orientation of the polymer network was investigated. Following the thiol-Michael crosslinking step, hydrogels were strained uniaxially to 100%, 200%, or 300% (Figure 2b, i) before photoinitiation of the thiol-ene reaction to further crosslink the material. Hermans orientation parameters were used to compare the degree of anisotropy of the polymer chains within each of the resulting networks relative to the amount of strain applied during network synthesis.⁶² The value of Hermans orientation parameter ranges from zero to one, with higher values indicating higher degrees of alignment. Order parameters were calculated from 2D wide angle x-ray scattering (WAXS) diffraction patterns (Figure 2b, ii–v). These results showed that order parameter increased as the magnitude of applied strain increased, with 300% strain resulting in the most highly oriented network (S = 0.38, Figure 2b, v). Notably, the samples prepared with 0% strain between crosslinking steps (Figure 2b, ii) and networks synthesized using only the first-stage Michael addition (Figure S4) reaction both exhibited order parameters of approximately 0, emphasizing the importance of applying strain between crosslinking steps to invoke anisotropy in these hydrogels. Furthermore, samples could not be reliably strained beyond 300% without tearing or slipping from their anchor points during stretching (Figure S5). Therefore, since the material synthesized with 300% strain exhibited an order parameter characteristic of a highly oriented polymer network,⁶³ all subsequently studied anisotropic hydrogels were synthesized with 300% applied strain. While, to the best of our knowledge, no prior studies have quantified anisotropy of the collagen matrix in native tendon tissue using metrics that can be directly correlated to the order parameter of the oriented polymer network, numerous prior studies have quantitatively confirmed by other methods suitable for biological tissues that healthy tissue possesses strong anisotropy.^64–71 Therefore, the order parameter of 0.38 was deemed suitable for recapitulating the anisotropic environment of native tendon since this value is within the range of previously characterized highly oriented polymer networks.⁶³ After alignment and second-stage crosslinking, the mechanical properties of the resulting anisotropic hydrogels were characterized to corroborate WAXS results and confirm that the anisotropy was inherent to the polymer network. We found that the hydrogel was significantly stiffer (p < 0.001) when strained in the direction parallel to the direction of mechanical stretching during network synthesis (E = 5.9 ± 0.05 kPa) as compared to being strained in the direction perpendicular to mechanical stretching (E = 3.6 ± 0.5 kPa) (Figure 2c, full curves Figure S6). The directional differences in stiffness observed indicate that the polymer chains had been elongated in the direction of mechanical stretching during synthesis, confirming the inherent anisotropy in the hydrogel networks.

In addition to providing anisotropic structural guidance for tenocytes within a synthetic hydrogel scaffold to promote regenerative behavior, it is also important to consider the cells’ ability to remodel the hydrogel and replace the synthetic matrix with their own native extracellular matrix (ECM), since matrix remodeling is a key component of native tendon homeostasis and repair.³⁶ The dicysteine peptide incorporated within the anisotropic hydrogels (GKKCGPQGIWGQCKKG) enables remodeling via matrix metalloproteinase (MMP)-mediated degradation. To study degradability, the anisotropic hydrogels were exposed to exogenous MMP2 for a total of 14 days, which is representative of the critical timeframe for tendon healing.^{13, 14, 47, 50} As a control experiment, we synthesized a nondegradable analog of the anisotropic hydrogels using a scrambled version of the dicysteine-functionalized peptide sequence (GKKCGIQQWGGPCKKG). Mechanical properties and orientation parameters of these nondegradable hydrogels were similar to those of the degradable hydrogels (Figure S7, S8), confirming similar network structure between the compositions. Since the anisotropy of the network is retained by the elastic free energy of the crosslinks, it is expected that the anisotropy will be lost and the order parameter will decrease as the crosslinks degrade. Therefore, each material was again characterized by Hermans orientation parameter from WAXS diffraction patterns upon synthesis and after 14 days of MMP exposure (Figure 3). Notably, these experiments were conducted on acellular hydrogel samples, enabling these results to isolate enzymatic degradation behavior of the polymer network without convolution of structural support from cells or elaborated ECM. As expected, both the degradable (S = 0.38, Figure 3a) and nondegradable (S = 0.41, Figure 3c) hydrogels exhibited orientation parameters indicative of anisotropic networks after initial material synthesis. After 14 days of MMP exposure, the degradable network lost nearly all anisotropy (S = 0.03, Figure 3b), whereas the nondegradable hydrogel retained its anisotropy (S = 0.38, Figure 3d), confirming that the enzymatically degradable network facilitates cell-mediated degradation over the 14-day timespan. Interestingly, although sufficient crosslinks were degraded for the hydrogels to return to an isotropic state, the hydrogels did not fully degrade upon digestion with collagenase II, indicating that some non-degradable crosslinks had formed. Interestingly, this non-degradable behavior was only observed when the gel was subjected to digestion with collagenase II after being incubated for 3 days; after 1 day of incubation, 30 minutes of collagenase II digestion resulted in complete degradation of the material. Based on these results, it is possible that some non-degradable crosslinks are formed between unreacted maleimide or norbornene groups over time. Since these are both multifunctional macromers with functionality greater than 4, it would be expected that less than 25% conversion of functional groups would result in a crosslinked network,⁷² and therefore prevent complete degradation of the material. Despite this incomplete degradation, these experiments demonstrated that the number of enzymatically degradable bonds in the network was sufficient for partial degradation and loss of anisotropy. Therefore, these materials are still suitable for facilitating cell-mediated remodeling, as observed in native tendon.

Figure 3.

The loss of alignment of anisotropic hydrogels synthesized with MMP-degradable crosslinks was characterized by 2D WAXS diffraction patterns collected (a) before and (b) after 14 days of MMP2 exposure. Comparatively, the diffraction patterns for anisotropic hydrogels synthesized with non-degradable crosslinks (c) before and (d) after 14 days of MMP2 treatment exhibited retention of alignment. Hermans orientation parameters (S) were used to quantify the extent of order in each sample. Hydrogels were subjected to a 300% applied strain during the mechanical alignment step of fabrication.

While the results presented thus far demonstrate that the synthesized hydrogels possess inherent network anisotropy and enzymatic degradability, using these materials to control tenocytes during healing requires a balance of these properties. Initially, structural guidance from the anisotropic network is necessary to align the cells in a single direction; therefore, the network must not degrade or lose anisotropy too quickly. Once cells align and begin to deposit their own ECM, the structural guidance provided by the polymer network will become less critical, as the native ECM maintains anisotropic orientation and continues to provide structural guidance. After the cells begin depositing native ECM, they leverage its anisotropy for structural guidance rather than the synthetic hydrogel. As such, we next sought to determine whether the balance of structural guidance and degradability in these anisotropic, MMP-degradable hydrogels effectively facilitates the in vitro alignment of encapsulated tenocytes. Primary human tenocytes were mixed with the hydrogel starting materials in solution to facilitate encapsulation in the hydrogel (1×10⁶ cells/mL) during the first-stage crosslinking step of network synthesis, as depicted in Figure 1. After aligning the hydrogel and facilitating second-stage crosslinking using the same procedure described for acellular experiments, the tenocytes were cultured within the anisotropic hydrogels for 14 days, which is representative of the early stages of tendon healing that are critical for establishing a regenerative environment After 1, 3, 7, and 14 days in culture, the actin cytoskeleton was used to characterize the alignment of cells relative to the direction of hydrogel alignment. Cell orientation was analyzed in 800 μm-thick regions of the hydrogels, where cells remained evenly distributed over 14 days (Figure S9). On day 1 post-encapsulation, although the cells had not yet begun to spread substantially throughout the matrix, some cells encapsulated in the anisotropic material assumed an oblate morphology aligned with the network (Figure 4b, i). This alignment was characterized by the corresponding orientation frequency plot (Figure 4c, i) with a full width half maximum (FWHM) of 57°. On day 3, there was minimal cell spreading, which, as observed in fluorescent images (Figure 4a, ii), appeared to align with the direction of hydrogel alignment in the anisotropic samples. The corresponding orientation frequency plots for day 3 showed a stronger correlation with the alignment, as evidenced by a decrease in FWHM to 51° (Figure 4c, ii). On day 7, cells had spread throughout the anisotropic matrix and elongated in the direction of the network alignment based on visual observation (Figure 4a, iii). The increasing prominence in cell alignment was confirmed by further sharpening of the orientation frequency plot and a decrease in FWHM to 48° (Figure 4c,iii). This alignment persisted through day 14, when the cells appeared to have formed a network throughout the hydrogel matrix (Figure 4a, iv). Interestingly, although the network visually appears to correlate strongly with the direction of polymer network alignment, the analysis of orientation frequency plots on day 14 yielded an FWHM of 53° (Figure 4c, iv), indicating a weaker correlation with alignment direction than on day 7. This increase was attributed to the formation of a cell/ECM network throughout the hydrogel, with cells branching to form connections across the thickness of the material, accounting for the slight deviation from collective anisotropic orientation. Since exogenous MMP degradation experiments in the absence of cells indicated that hydrogels degrade and lose anisotropy by 14 days, the retention of anisotropy in cell orientation through 14 days confirms that sufficient native ECM has been deposited to maintain structural guidance for the cells once the synthetic scaffold has been degraded. Without the native ECM retaining the anisotropic structure, it would be expected that cells would lose orientation in parallel with the hydrogels as was seen in acellular experiments. Therefore, in the case when cells are present, the guidance of cellular orientation is initially dominated by the hydrogel scaffolds, but guidance has transitioned to be dominated by the oriented, native ECM after 14 days of encapsulation.

Figure 4.

Primary human tenocytes were encapsulated in (a) anisotropic and (b) isotropic hydrogels. Anisotropic hydrogels were subjected to a 300% applied strain during the mechanical alignment step of fabrication, and all materials were synthesized with MMP-degradable crosslinks. The actin cytoskeleton of cells (phalloidin) was used to characterize their orientation via fluorescence microscopy for each sample type on days (i) 1, (ii) 3, (iii) 7, and (iv) 14 post-encapsulation. Scale bars = 500 μm. Representative summed projections consist of 17 images spanning 800 μm thickness. Arrows in (a) represent the direction of polymer network alignment within the anisotropic hydrogels. Collective cell orientation within each hydrogel was quantified by (c) frequency distributions calculated using the corresponding anisotropic (purple) and isotropic (gray) samples at each time point. n = 6–9 with 3 biological replicates. Orientation was quantified on the summed projection spanning 800 μm thickness and 2.5 mm × 2.5 mm area. Data is represented as means ± SD.

The importance of network anisotropy in guiding cell orientation is highlighted by characterizing cells in analogous isotropic, MMP-degradable samples at the same time points: 1, 3, 7, and 14 days. On day 1, cells remain spherical (Figure 4b,i). On days 3, 7, and 14, the tenocytes elongate and spread throughout the matrix to a similar extent as in the isotropic samples, but with no preferential orientation (Figure 4b,i–iv). Frequency plots confirm these observations, with all four time points resulting in a relatively even distribution of orientations (Figure 4c). These results confirm that although the hydrogels undergo MMP-mediated degradation, the polymer network alignment provides structural support long enough for cells to adopt and maintain the prescribed anisotropic orientation as they deposit their own ECM. Furthermore, we confirmed that matrix degradability is critical for cells to spread and adopt any elongated morphology, regardless of network orientation, by demonstrating that cells remained spherical throughout 14 days when encapsulated in hydrogels synthesized with a non-degradable peptide as the dithiol component of the network (Figure S10). Notably, cells remained >95% viable after 14 days in culture, regardless of whether they were encapsulated in degradable or nondegradable hydrogels (Figure S11), indicating that degradation was necessary for cell spreading and the adoption of anisotropic morphology, but not for survival.

Finally, to demonstrate that the anisotropic, enzymatically degradable hydrogels promoted regenerative behavior instead of fibrotic behavior of the tenocytes, we characterized the collagen deposited by encapsulated cells after 14 days. Hydrogels were sectioned and stained with picrosirius red to distinguish between collagen I (red staining in polarized images), which is representative of a regenerative environment, and collagen III, which indicates a fibrotic environment. The cells encapsulated in the anisotropic hydrogels deposited a matrix primarily composed of collagen I (Figure 5a, c). Comparatively, cells encapsulated in isotropic hydrogels deposited a matrix that included a combination of both collagen I and collagen III (Figure 5b, d), indicating fibrotic characteristics in this matrix and confirming that the anisotropic network morphology of the hydrogel scaffold is critical to invoking a regenerative environment. When comparing the area of collagen I (red) to collagen III (green) in the polarized images of the deposited ECM in each condition, the anisotropic sample (Figure 5c) exhibited a collagen I to collagen III ratio of 71:1, while the isotropic sample (Figure 5d) exhibited a ratio of 4:1. These results quantitatively confirm that the ECM deposited by cells in the anisotropic hydrogels more closely resembled native tendon versus scar tissue as compared to the ECM deposited by cells in the isotropic hydrogel. Furthermore, enhanced overall anisotropy of the deposited collagen structure deposited within hydrogels was confirmed quantitatively using FibrilTool in ImageJ.73 The anisotropy scores, which range from zero to one, of the collagen network deposited in the anisotropic and isotropic samples were 0.11 and 0.08, respectively. These results confirm that the anisotropic hydrogel scaffold promoted deposition of native ECM with greater anisotropy, similar to native native tendon.

Figure 5.

Tenocytes encapsulated in the 3D hydrogel scaffolds deposited an ECM that was abundant in collagen, as visualized by picrosirius red staining of 20 μm sections of (a) anisotropic and (b) isotropic samples on day 14 post-encapsulation. Collagen deposition within (c) anisotropic hydrogels was composed primarily of collagen I (red), while collagen deposited in (d) isotropic hydrogels also contained a substantial fraction of collagen III (green) when the samples were visualized between crossed polarizers. Scale bars: 200 μm. Arrows indicate the direction of crossed polarizers.

Overall, these studies demonstrate that integration of balanced structural guidance from anisotropic network topology and matrix degradability from enzymatically degradable crosslinks in a synthetic hydrogel provides a regenerative environment for tenocytes on a timeline that is commensurate with the early stages of tendon healing. The implementation of a two-stage polymerization strategy for synthesizing these hydrogels enables encapsulation of cells in a 3D matrix that is representative of the native tissue. These studies serve as an in vitro model demonstrating that the balance of intrinsic structural anisotropy and cell-mediated degradation in this class of synthetic scaffolds positively impact the regenerative behavior of tenocytes. The balance of structural guidance and enzymatic degradability is important because, as demonstrated by these results, directional cell spreading is initially guided by the oriented polymer network acting as a template for the tenocytes. While some enzymatic degradability is necessary in this initial stage for cell spreading to occur, the degradation must be slow enough to support the oriented cells while native ECM is deposited. After sufficient ECM has been deposited, the hydrogels scaffold is no longer required, as the tenocytes are then supported by the anisotropic native ECM. While the materials investigated here possess some limitations in translatability, such as a fraction of permanent crosslinks prohibiting full degradation of the hydrogel, this work serves as a foundational in vitro study that elucidates important considerations for material design when implementing anisotropic hydrogels as scaffolds for promoting regenerative tendon healing. Permanent crosslinks may influence long-term matrix remodeling as native ECM and surrounding tissue integrate with the synthetic scaffold. However, it remains unclear whether incomplete degradation would occur in vivo or, if so, whether it would negatively affect healing. Given the low cell density and poor vascularization of native tendon, some hydrogel may integrate benignly into the regenerated tissue. Nevertheless, additional in vivo studies are necessary to determine whether residual hydrogel could affect biomechanical properties or long-term tendon homeostasis. Based on these results, future work will focus on further innovation of these materials to translate these materials to clinical applications for tendon healing.

Conclusions

Anisotropic PEG-based hydrogels with integrated biochemical cues were synthesized via a two-stage polymerization process, where the inherent anisotropy of the polymer network was invoked through mechanical stretching between crosslinking steps. The resulting materials were susceptible to cell-mediated degradation due to the inclusion of MMP-degradable peptide sequences in the crosslinked network, exhibiting a loss of alignment after 14 days when exposed to exogenous MMP2. When tenocytes were encapsulated in the 3D anisotropic hydrogels, they adopted the anisotropic morphology of the polymer network and deposited an extracellular matrix that contained mostly type I collagen, indicating a pro-regenerative environment. Comparatively, isotropic materials of the same composition induced random orientation of encapsulated tenocytes, and deposition of a matrix containing primarily collagen III was observed, indicating a fibrotic environment. Collectively, these results demonstrate the successful use of a synthetic scaffold with tunable biophysical and biochemical properties for recapitulating the native tendon environment, thereby opening opportunities for promoting regenerative healing of tendons after injury.

Supplementary Material

Supporting Information. NMR spectra, MALDI spectra, rheology data, tensile measurement data, WAXS diffraction patterns, cell distribution images, cell encapsulation images, PCR primer information, tenocyte gene expression data