Subcutaneously engineered autologous extracellular matrix scaffolds with aligned microchannels for enhanced tendon regeneration

Abstract

Improved strategies for the treatment of tendon defects are required to successfully restore mechanical function and strength to the damaged tissue. This remains a scientific and clinical challenge, given the tendon’s limited innate regenerative capacity. Here, we present an engineering solution that stimulates the host cell’s remodeling abilities. We combined precision-designed templates with subcutaneous implantation to generate decellularized autologous extracellular matrix (aECM) scaffolds that had highly aligned microchannels after removal of templates and cellular components. The aECM scaffolds promoted rapid cell infiltration, favorable macrophage responses, collagen-rich extracellular matrix (ECM) synthesis, and physiological tissue remodeling in rat Achilles tendon defects. At three months post-surgery, the mechanical strength of tenocyte-populated ‘neo-tendons’ was comparable to pre-injury state tendons. Overall, we demonstrated an in vivo bioengineering strategy for improved restoration of tendon tissue, which also offers wider implications for the regeneration of other highly organized tissues.

Graphical Abstract

Introduction

Tendinopathy incidences contribute an estimated 30% of all musculoskeletal issues, 30–50% of sport-related injuries, and represent a substantial socio-economic burden to individuals, families, and healthcare systems [1, 2]. Disorganized tissue arrangement, poor inflammatory resolution, and unrecoverable mechanical damage exemplify tendon pathologies [1, 3]. Sparsely distributed resident tenocytes, limited vasculature and low innervation within the tendon tissue interior restrict their regenerative capacity [3, 4]. Consequently, current treatments aim to accelerate healing but often result in compromised mechanical properties, due to disorganized collagen deposition and scar tissue formation within the defect site [4–7]. In-turn, scarred tendons are at high risk of subsequent injuries and tears, causing significant dysfunction and patient disability [6, 7]. Surgical interventions that have utilized autografts, allografts and xenografts for tendon repair have faced complications including donor-site morbidity, chronic inflammatory, poor graft integration and tendon dysfunction [6]. Ultimately, tendon tissue regeneration without mechanical dysfunction resulting from persistent inflammation, heterotopic ossification or scar tissue formation, is preferred [6]. However, optimal regeneration strategies for tendon injury and tendinopathies still represent an unmet clinical and scientific challenge [4].

Tissue engineering offers the attractive strategy of mimicking native structural features, whilst overcoming the issues associated with traditional treatment methods [8, 9]. For highly organized musculoskeletal tissues, engineering scaffolds to help guide cell infiltration and ECM arrangement could enhance the regenerative capacity of injured tissues [10]. Topographical cues such as shape, surface roughness, size and alignment of scaffold pores or fibers have been shown to significantly influence mechanical properties and cell behavior in bone, muscle, cartilage and tendon tissue engineering studies [11]. Moreover, ideal scaffolds should encourage the recruitment of endogenous cells, should not provoke a foreign-body response, should regenerate functional “neo-tissue,” and have a degradation rate that matches the rate of new ECM formation [12]. To achieve biocompatibility, the use of autologous material is the optimal choice. However, despite the design and generation of many kinds of scaffolds, it remains a difficult challenge to engineer scaffolds that meet all aspects of the above requirements [13, 14].

In vivo tissue engineering utilizes the host’s own body to produce scaffolds for tissue regeneration [15]. Surgically accessible tissues such as the omentum, bone, peritoneal cavity and subcutaneous tissue have been used as bioreactors to engineer autologous-derived tissue constructs for experimental and clinical applications [16–20]. These tissue cavities provided ideal environments for the generation of tissue-capsules (polymers encapsulated within a fibrocellular capsule), which have improved biocompatibility and lower immunogenicity. However, tissue-capsules usually contain cells and have no designed porous structure to allow cell infiltration and remodeling of the tissue construct following implantation, which can lead to undesirable results. Hurd et al., demonstrated the feasibility of generating porous acellular matrix scaffolds that are inverse replicas of the original 3D polymer templates, using in vitro cell culture. The resultant ECM scaffolds possessed pore structure and spatial arrangement resembling the aligned tissue but had inadequate mechanical properties [21].

In light of these reports, we hypothesized that engineering a specialized polymer template for implantation into the subcutaneous pouch, could produce an autologous cell-derived, porous and mechanically robust ECM scaffold, with topologically beneficial structure for cell infiltration and orientation. Our strategy to provide a much-needed solution for improved regeneration of damaged tendons is summarized by the following steps: first, the rat’s subcutaneous pouch is used as a bioreactor to achieve cellularization and collagen-rich ECM deposition, both on and within implanted parallel-aligned poly (ε-caprolactone) (PCL) microfiber bundle templates; second, the subsequent removal of polymer template and cells results in an autologous ECM (aECM) scaffold with hollow and aligned microchannel structures. We speculated that aECM scaffolds could achieve enhanced recovery to rat Achilles tendon defects, by providing the strength benefits of autologous subcutaneous-synthesized ECM and supporting cell in-growth and alignment; without the risks associated with traditional tendon repair. To evaluate aECM scaffold performance, we assessed explants at different time points for functional, mechanical, histological and biochemical characterization. Overall, we demonstrated a novel approach to the formation of autologous ECM scaffolds for the enhancement of tissue regeneration and “neo-tendon” formation, which resulted in tissue mechanical strength and function comparable to uninjured native tendon tissue.

Materials and Methods

Materials

Poly (ε-caprolactone) (PCL) pellets (Mn = 70,000 – 90,000) were purchased from Sigma-Aldrich (St. Louis, Missouri, US). Analytical reagents including dichloromethane, chloroform, anhydrous methanol and absolute alcohol were obtained from Tianjin Chemical Reagent Company (Tianjin, China).

Animal experiments

Animal experiments were approved by the Animal Experiments Ethical Committee of Nankai University and were in accordance with the NIH Guide for Care and Use of Laboratory Animals. Sprague-Dawley rats (n = 95) (male, aged 8–10 weeks, weight range: 280–320 g) were purchased from the Laboratory Animal Center of the Academy of Military Medical Sciences (Beijing, China). A total of 20 rats were used to source the required amount of dNat scaffolds. 1 rat in aECM group and 2 rats in dNat group died from an unexplained reason. The remaining were randomly assigned into experimental groups for analysis (Table S2). Power analyses calculated the minimum rats required for histological analysis (n = 5 per group) and for walking function tests (n = 7 per group), to observe a significant effect of treatment. For all other experiments a minimum of n = 3 per group was set for obtaining numerical data. A separate power analysis was performed for the supplementary experiment of 3 days post-implantation analyses (n = 3, Fig. S4), and was not compared to other time points.

Fabrication of decellularized autologous extracellular matrix (aECM) scaffolds

The aECM scaffolds with aligned microchannels were prepared by the following process (as illustrated in Fig. 1): Firstly, a film comprising aligned PCL fibers was prepared by using a self-made, melt-spinning device. PCL pellets were placed within a 20 mL stainless steel syringe, equipped with a 17-G injection needle, which was placed above an aluminum-alloy collecting rod (distance between rod and needle, 1 cm). The PCL pellets were heated to 100 °C for two hours to achieve uniform consistency and transparent melted PCL, before commencing the melt-spin. Driven by a syringe pump, the spun fibers of PCL were extruded at a flow rate of 2 mL/h. The resultant fibrous membrane scaffolds, with oriented structure, were collected by a rotating aluminum-alloy rod (4 cm in diameter; 50 rpm rotation) for 90 minutes. Next, the PCL fiber films were rolled into a columnar structure for use as templates. The final diameter and length measurements were 2.41 ± 0.23 mm, 4 cm, respectively. The PCL template was soaked in 75% ethanol prior to subcutaneous implantation. A small incision was made along the back of rats, and the templates were implanted subcutaneously for 4 weeks, to allow sufficient host cell migration onto and within the gaps between the PCL fibers and formation of PCL-Cell-ECM composites. These composites were explanted from the implantation site, washed with phosphate buffered saline (PBS) and then dehydrated with increasing ethanol washes (40%, 60%, 70%, 80% and 100%). After dehydration, the compounds were immersed in chloroform for 4 cycles of 12 h each, to remove the PCL template and obtain Cell-ECM compounds. The compounds were rehydrated with decreasing ethanol washes (100%, 80%, 60%, 40% and sterile purified water). Finally, cells were removed using the acellular processing method of freeze-thaw cycles. Briefly, Cell-ECM compounds were washed three times with PBS, then placed in liquid nitrogen for 2 minutes, followed by thawing in 37 °C PBS for 10 minutes. The freeze-thaw cycles were repeated six times. To remove nuclear remnants, samples were incubated for 24 h in reaction buffer containing 50 U/mL deoxyribonuclease I and 1 U/mL ribonuclease A in PBS (pH = 7.4) at 37 °C with agitation, resulting in aECM scaffolds. The dNat scaffolds were prepared utilizing the same steps as aECM scaffolds, sans the polymer removal step. Prior to implantation, the aECM scaffolds were sterilized by immersing in 75% ethanol for 1 h and washed 5 times with physiological saline.

Fig 1. Characterization of aECM and dNat scaffolds.

(A) PCL-Cell-ECM, Cell-ECM, aECM, dNat scaffold and native tendon microstructures were visualized using SEM cross- (top panels) and longitude- (bottom panels) sections. (B) Micro-CT scans of the transverse and longitudinal position in the aECM and dNat scaffolds. (C) Raman spectrometry analysis to demonstrate removal of PCL from scaffolds. (D) H&E and (E) DAPI staining. (F) DNA content of Cell-ECM composite, aECM, native tendons and dNat scaffolds. (G) Masson’s Trichrome, Sirius Red, Verhoeff’s, and Safranin O staining showing ECM composition and arrangement in aECM scaffolds, dNat scaffolds and native tendon. (H) Quantitative ECM (soluble collagen, elastin and sGAG) analyses of explanted PCL-Cell-ECM, Cell-ECM composite after polymer removal, aECM after decellularization and dNat scaffolds. Images and data are representative of n = 5 individual experiments, and data are shown as the mean ± standard error. Statistical analysis is shown as *P ≤ 0.05, **^/##P ≤ 0.01, ***^/###P ≤ 0.001. Statistics in (H), # represents Cell-ECM versus aECM; * represents aECM versus dNat. Scale bars, A, D, E, G: 100 μm; B: 500 μm.

Scaffold characterization (porosity, fiber diameter, microchannel and pore size)

Following aECM and dNat scaffold preparation, samples were freeze-dried and then scanned using a SkyScan 1276 Micro-CT (Bruker, Billerica, MA, US), at 50 kV voltage, 70 μA current, and a voxel of 5 μm. All data was 3D-reconstructed using NRecon Reconstruction Software v1.7.1.6 (Micro Photonics Inc., Allentown, PA, US) and analyzed by CTAn Software v1.17.9.0 (Blue Scientific Ltd., Cambridge, UK) and CTvox Software v3.3.0.0 (Blue Scientific Ltd.). The porosity of each scaffold was analyzed based on Micro-CT results. The average fiber diameter, microchannel and pore size of the scaffolds were visualized and measured using a FEI Quanta 200 scanning electron microscope (SEM; ThermoFisher Scientific Europe NanoPort, Eindhoven, Netherlands) at an accelerating voltage of 15kV. Five SEM images were acquired for each scaffold, and images (at least 50 fibers and 50 pores) were manually measured and analyzed using ImageJ software v1.5 (NIH, Maryland, US).

Fourier Transform Raman Spectroscopy (FT-RAMAN)

FT-RAMAN was used to detect the removal of PCL and measurements were obtained by accumulating 200 scans with a resolution of 4 cm⁻¹ in the range of 1000 – 1800 cm⁻¹ using RFS-100/S Raman spectrometer (Bruker, Massachusetts, US).

Histological Analysis

Histological analysis was performed on scaffolds and neo-tendon tissue sections. For sectioning and staining, the explants were fixed with 4% paraformaldehyde, dehydrated by 30% sucrose solution until the grafts sank to the bottom. Samples were embedded in OCT and sectioned to 6 μm in thickness and frozen, before staining with hematoxylin and eosin (H&E), Sirius Red, Masson’s Trichrome, Verhoeff-van Gieson’s (VVG; Verhoeff’s Stain), Safranin O (Beijing Zhongshan Golden Bridge Biotechnology, Beijing, China) and 4’,6-Diamidino-2-Phenylindole (DAPI; Sigma-Aldrich). Cell-ECM composites and native tendon tissue were also stained for use as reference controls for aECM and aECM-derived neo-tendon, respectively. Images were observed under a Leica DM3000 microscope (Leica Microsystems, Wetzlar, Germany). In addition, DNA content was assessed by DNeasy Blood and Tissue Kit (QIAGEN, Shanghai, China) and ECM content was quantified by Biocolor Sircol Collagen Assay, Fastin Elastin Assay and Blyscan Glycosaminoglycan Assay Kits (Biocolor; Beijing Qbioscience Technology, Beijing, China), according to manufacturer’s protocol.

Immunostaining and fluorescence microscopy

Frozen sections were rinsed twice with 150 mM PBS before incubation in 5% normal goat serum (Beijing Zhongshan Golden Bridge Biotechnology) for 30 min at room temperature. For intracellular antigen staining, 0.1% Triton-PBS was used to permeate the membrane before incubation with serum. The sections were then incubated with primary antibodies in PBS overnight at 4°C, followed by incubation with secondary antibody in PBS for 2 h at room temperature. Sections were stained with primary antibodies purchased from Abcam (Cambridge, UK): anti-CD31 antibody (1:100; ab28364) to detect and count capillaries; anti-CD68 antibody (1:100; ab31630) to observe inflammatory cells; and anti-Mannose Receptor antibody (CD206; 1:200; ab64693) and anti-iNOS antibody (1:50; ab15323) to visualize M2 and M1 macrophages, respectively. Collagen I (1:400; ab6308) and Collagen III (1:400; ab7778) were used to visualize collagen content of ECM. Other markers were stained in the same way using anti-CD146 (1:200; ab75769), anti-tenomodulin (1:200; ab203676), anti-scleraxis (1:200; ab58655), anti-α-SMA (1:200; ab7817), anti-Ki67 (1:250; ab16667) and anti-vWF (1:200; ab6994). Secondary antibodies, Alexa Fluor 488 goat anti-mouse IgG (1:200) and goat anti-rabbit IgG (1:200) were purchased from Invitrogen (Waltham, Massachusetts, US) and incubated with sections for 1 h, at room temperature and in the absence of light. Sections were incubated with serum IgG, without primary antibodies, for use as negative controls. Nuclei were counterstained with DAPI mounting solution. Slides were observed under a Zeiss Axio Imager Z1 fluorescence microscope (Oberkochen, Germany). Cell number was calculated using DAPI staining, five pictures were captured for each sample and cell number was averaged. The spreading area (S) and perimeter (L) of the nuclei at each time points were also calculated using ImageJ software v1.5. The nuclear shape index (NSI) was calculated using the following equation:

$$\text{NSI}=\left(\frac{4\pi S}{L^2}\right)$$

Nuclei with a linear and elongated morphology had an NSI approaching 0. Whilst nuclei with nearly circular shape have an NSI close to 1. Images were observed and captured using a Leica TCS SP5 laser scanning confocal microscope (Leica, Germany).

Western blot analysis

CD68, CD206, iNOS, Tenomodulin (Tnmd), Scleraxis (Scx), Collagen I and Collagen III were quantitatively assessed by Western blot. At different time points after implantation, half of each explanted graft was cut into pieces, and the total protein was extracted using Tissue Protein Extraction kit (Thermo Fisher Scientific, Waltham, Massachusetts, US) in the presence of protease inhibitors (Beijing Solarbio Science & Technology, Beijing, China). Samples were homogenized, and lysates centrifuged at 12,000 rpm for 10 min at 4 °C. The supernatant was collected, and protein concentrations were measured by BCA assay (Beyotime Biotechnology, Shanghai, China). Samples were mixed with SDS-PAGE loading buffer and boiled for 5 minutes in a water-bath. Protein samples were separated by SDS-PAGE using 4–12% polyacrylamide gels. The separated proteins were transferred onto Immobilon-P Transfer membranes (Merck Millipore, Darmstadt, Germany) and blocked for 2 hours with 5% nonfat dry milk, before incubation with the appropriate primary antibody including rabbit anti-tenomodulin (1:500), rabbit anti-scleraxis (1:1000), rabbit anti-Collagen I (1:1000) and rabbit anti-Collagen III (1:1000), (Abcam), or mouse anti β-actin antibody (1:5000; Cell Signaling Technology, Danvers, Massachusetts, US). The membranes were then incubated with HRP-conjugated goat anti mouse IgG (H+L) antibody (1:1000) or HRP-conjugated goat anti rabbit IgG (H+L) antibody (1:1000) (Beyotime Biotechnology) and washed with Tris-buffered saline containing 0.1% Tween20 (Sigma-Aldrich). Protein bands were detected by adding Chemiluminescent HRP Substrate (Merck Millipore). Relative densitometries were calculated using ImageJ software v1.5 (NIH).

In vivo implantation.

The Sprague-Dawley rats were anesthetized with chloral hydrate (300 mg/kg) by an intraperitoneal injection. 6 mm long incisions of approximately 2mm width and depth, were performed at the Achilles tendon of the left leg of the rats. The aECM grafts were sutured in an end-to-end way using interrupted 6–0 monofilament nylon sutures (Yuan Hong, Shanghai, China). The wound was closed with 3–0 monofilament nylon sutures. The Achilles Functional Index (AFI) was tested at predetermined time points (−2 day, 1, 2, 3, 4, 8 and 12 weeks). Animals were sacrificed by chloral hydrate overdose and grafts were explanted for histological, transmitting electronic microscope (TEM), biomechanical and Western blot analysis.

Measurement of Achilles Functional Index (AFI)

The Achilles tendon functional test was performed according to Murrell’s method [22]. A walkway lined with white paper, 10 cm × 60 cm, was used for functional testing. The left hind paws of the rats were coated using a red inkpad and then rats could walk along the walkway (Fig. S3A). The pawprints were scanned and relevant parameters, such as print length, print width (distance between first and fifth toe) and intermediate toe width (distance between second and fourth toe) were measured (Fig. S3). The print length factor (PLF), toe spread factor (TSF) and intermediate toe factor (ITF) were then calculated using ImageJ software v1.5 (NIH). Pawprints were collected before the operation at day −2, and on post-operative weeks 1, 2, 3, 4, 8 and 12. Achilles tendon functional index (AFI) was determined in accordance with the formula:

$$\text{AFI}=74\left(\text{PLF}\right)+161\left(\text{TSF}\right)+48\left(\text{ITF}\right)-5$$

Mechanical Tests

Mechanical properties of the aECM scaffolds, before and after implantation, were assessed by an Instron 5865 tensile-testing machine (Norwood, Massachusetts, US) using 100 N load cells. Normal Achilles tendon tissue was used as a control. The cross-sectional areas of Achilles tendon were measured using Vernier calipers. The ends of the Achilles tendon that connect to bone and muscle were fixed in place using custom-made clamps. The load was imposed along the direction of the aligned microchannels of the aECM scaffolds. After applying a preload of 0.01 N, each scaffold was pulled at a rate of 5 mm/min until rupture. Maximum stress and strain at rupture were recorded. The elastic modulus representing elasticity was calculated according to the stress-strain curve in the elastic region. The structural properties of the aECM scaffolds are represented by Maximum stress and Elastic modulus (MPa).

Transmission electron microscopy (TEM)

Collagen fiber diameter and alignment were assessed by TEM. Explants were fixed by standard procedures for TEM observation. A total of 400 collagen fibrils were used to calculate the average diameter of the collagen fibrils.

Enzyme-linked Immunosorbent Assay (ELISA)

Inflammation-related cytokines were measured using the AimPlex Basic Kit for Premixed Panels and a diluent kit following the provided assay procedures (Quantobio Technology).

Statistical analysis

All quantitative results were obtained from at least three samples, for data from larger experimental groups, sample size is indicated in figure legends. GraphPad Prism Software v5.0 (San Diego, California, US) was used for statistical analysis. For testing across multiple sample groups, multiple comparisons were performed using a one-way ANOVA and Tukey’s post-hoc analysis. For testing across multiple groups with more than one variable, a two-way ANOVA with Bonferroni multiple comparison analysis was used. For data with two independent experimental groups, an unpaired two-tailed Student’s t test was used to determine significance. Significance was accepted at a P value below 0.05. Data displayed on graphs are expressed as the mean ± standard error of the mean (SEM).

Results

Scaffolds generation and characterization

According to our design criteria, melt-spun PCL microfibers with an average diameter of 86.31 ± 18.56 μm were bound together to form fiber bundles for use as templates. The average final macroscopic diameter of the templates was 2.41 ± 0.23 mm, approaching the approximate size of native rat tendon (2.35 ± 0.44 mm) (Table S1). The templates were then implanted subcutaneously into the rat for 4 weeks. It was noted that at no point during the implantation time was redness, swelling or signs of rupture evident among any of the implant sites, owing to PCL’s inert reactivity. Upon evaluation, the gaps among the aligned microfibers were filled by dense host tissue, as observed by SEM from transverse and longitudinal cross-sections (Fig. 1A). Following the removal of PCL templates, aligned microchannels were present (Cell-ECM). After cell removal (aECM), the ECM architecture and spatial orientation were maintained, without collapse or deformation (Fig. 1A). Micro-CT scanning of different cutting positions further confirmed the presence of uniform and orientated microchannel structures within aECM scaffolds, whereas no obvious pore structure was observed across the decellularized native rat tendons (dNat) (Fig. 1B; Movie S1 and Movie S2). Based on these Micro-CT analysis, the porosity of aECM scaffolds was significantly higher than that of dNat scaffolds (77.38 ± 3.35% vs. 10.44 ± 1.18%). No PCL remnants could be detected within aECM scaffolds following polymer leaching, as identified using Raman spectrometry (indicated on PCL chart as vC=0, δCH₂, ωCH₂, vC-C; Fig. 1C). The aECM scaffold possessed uniformly orientated microchannels with an average pore size of 94.41 ± 16.22 μm (Table S1). In contrast, the dNat scaffolds had few and scattered small pores of 6.44 ± 2.05 μm (Table S1) and showed an overall dense ECM structure in cross and longitudinal section (Fig. 1A). Accordingly, the porosity of the dNat scaffolds was significantly lower than that of the aECM scaffolds (Table S1). H&E staining demonstrated clear, equally spaced, and substantial ECM deposition, which was retained following polymer leaching and decellularization (Fig. 1D). Furthermore, DAPI staining of nuclei showed that a large quantity of host cells occupied the gaps among the aligned fibers (Fig. 1E). Further analysis revealed that these cells were predominantly α-smooth muscle actin (α-SMA)-positive cells (Fig. S1A) mixed with a small number of CD68 cells distributed around the template fibers (Fig. S1B). Also, the immunostaining results suggested that a number of capillaries were distributed throughout the templates (von Willebrand factor, vWF⁺ cells, Fig. S1C; CD31⁺ vessel-like structures, Fig. S1D). The effectiveness of cell removal from the aECM and dNat scaffolds was evidenced by H&E and DAPI staining (Fig. 1D, E). Indeed, there were significant decreases in DNA content in aECM and dNat, following decellularization of Cell-ECM and native tendon, respectively (Fig. 1F). Residual levels of DNA in aECM and dNat were detected at 31.56 ± 0.01 and 28.00 ± 0.01 ng/mg, respectively (Table S1), lower than the minimal criteria for acellular products (50 ng/mg) [23]. In addition, histological staining revealed that aECM scaffolds were mainly composed of collagen (Masson’s, Sirius Red), elastin (Verhoeff’s) and sulfated glycosaminoglycans (sGAG; Safranin O); these were also present within, but distributed differently, throughout the dNat scaffolds (Fig. 1G). The content of each component decreased after decellularization, as expected (Cell-ECM to aECM; Fig. 1H), but not after the initial polymer removal step (PCL-Cell-ECM to Cell-ECM); suggesting that polymer templates could be used in a range of versatile ways, without the polymer removal process compromising the content of the ECM. The collagen content in aECM scaffolds was lower than that of dNat scaffolds, whereas the content of elastin and sGAG was higher than in dNat scaffolds. These results can be explained by the cell types responsible for producing the scaffolds. The aECM scaffolds were synthesized by subcutaneous cells, likely fibroblasts, whereas the dNat scaffolds are tenocyte produced matrices. Finally, we tested the mechanical strength of aECM scaffolds (Table S1). The suture strength was 3.34 ± 0.96 N, which is slightly less than 6–0 surgical suture (4.24 ± 0.41N). However, these values still meet the United States and British Pharmacopeia tensile strength requirements of a suture material [24] and thus, aECM scaffolds were deemed acceptable for use in the attachment of tissues. Additionally, the elastic modulus and the maximum stress of aECM scaffolds were 3.95 ± 1.51 MPa and 3.58 ± 0.41 MPa, respectively.

Cell infiltration and macrophage response to scaffolds.

The aECM and dNat scaffolds were implanted subcutaneously in rats for 2 weeks to evaluate cell infiltration and macrophage responses. H&E and DAPI staining revealed that all the microchannels of aECM scaffolds were occupied by host cells, whereas cells were mainly concentrated on the margin of dNat scaffolds (Fig. S2A). Accordingly, this suggested the number of cells within aECM scaffolds were significantly higher than that in dNat scaffolds, and cell counts verified this (Fig. S2B). Immunofluorescent staining revealed that CD68, CD206 and iNOS positive cells were identifiable both within and around the aECM scaffold, whereas they were mainly dispersed on the marginal areas of dNat scaffolds (Fig. S2C). More staining for CD206 and comparatively less staining for iNOS, was observed in the aECM group (Fig. S2C). Furthermore, Western blot analyses indicated that whole tissue protein expression of CD68 showed no significant differences between the scaffold groups, however CD206 expression in aECM scaffolds was higher, whereas iNOS expression was lower than that of dNat scaffolds (Fig. S2D). Therefore, a higher ratio of total tissue M2/M1 macrophages was calculated within the aECM group (approximately 1.5:1) compared with the dNat group (approximately 0.6:1) (Fig. S2E). These data suggested that the key feature of aECM scaffolds, the aligned microchannels, were the likely reason for enabled macrophage cell infiltration and regulation of M1 to M2 polarization and phenotypic switch; whereas dNat scaffolds, which attenuated macrophage infiltration to the interior, failed to reciprocate the polarization response.

Functional and mechanical restoration of tendon defects following implantation.

Functional performance and biomechanical properties are both important criterion for evaluation of Achilles tendon regeneration. Rat pawprints were collected at different time points, before and after implantation (Fig. S3A), and used to collate the necessary parameters for Achilles Functional Index (AFI) assessment (Fig. S3B). The pawprints were narrower compared to those taken prior to tendon resection (−2 days), both in aECM and dNat group at 1 and 2 weeks, displaying a trend towards recovery. The pawprints of the aECM group exhibited noticeable recovery from 2 weeks onwards, whereas those from the dNat group did not (Fig. 2A). The AFI was calculated as −6.88 ± 10.26 for uninjured tendon and this decelerated sharply in both groups, as calculated at week 1 (Fig. 2B). Tendon function of the aECM group showed significant recovery at 3 weeks and displayed continuous improvement until returning to an AFI values at 12 weeks, which was comparable to the pre-injury state values. The overall AFI score of aECM group remained consistently higher than that of the dNat group, with significant differences observed at weeks 2, 3 and 4. Next, we tested the mechanical strength of the remodeled scaffolds. Notably, the stress-strain curve of the neo-tendon at 12 weeks exhibited a clear toe region, similar to native tendons (Fig. 2C). The elastic modulus (Fig. 2D) and the maximum stress (Fig. 2E) were significantly increased at 4 weeks, when compared to the original pre-implantation aECM scaffolds. Both elastic modulus and maximum stress of neo-tendon displayed no significant differences to native tendons, at 12 weeks. No significant differences were observed for the elastic modulus of the dNat group, however the maximum stress of dNat scaffolds was significantly lower than that of both native tendon and neo-tendon at 12 weeks. The results shown here suggest that despite the initial low stress-strain and elasticity thresholds of aECM scaffolds, they were more successful at rapid remodeling into strengthened structures than dNat scaffolds. This rapid change in mechanical strength was able to restore walking function and AFI to a state comparable to the mechanical functionality observed in pre-injured models.

Fig. 2. Functional and mechanical testing of aECM and dNat scaffold implants.

(A) The pawprints of aECM and dNat groups prior to surgery (−2 day) and post-implantation at 1, 2, 3, 4, 8 and 12 weeks. (B) Functional testing of the Achilles tendon over 12 weeks was performed to evaluate recovery of motion. Values indicate the mean AFI (y-axis) with standard deviation error bars (n = 7). (C) The representative stress-strain curve of different samples including aECM scaffolds, the remodeled aECM scaffolds at 4 and 12 weeks, native tendon (Nat) and remodeled dNat scaffolds at 12 weeks (dNat 12w) (n = 3). The (D) elastic modulus and (E) maximum stress of the different groups (n = 3 for each). Graph data are displayed as the mean ± standard error. Statistical analysis is shown as *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ns = no significance.

Tendon remodeling.

Extensive and rapid infiltration of correct cell types are crucial for constructive tendon remodeling. We examined the histology, cellular influx, and the cell phenotypes within aECM implants. Both H&E and DAPI staining demonstrated copious migration of host cells into the interior of aECM scaffolds 3 days after implantation (Fig. S4A, B). In contrast, host cells predominantly distributed around the exterior of the dNat scaffolds, indicating that a low number of cells had migrated into the interior of the dNat scaffolds (Fig. S4A, B). Accordingly, the number of cells within the aECM scaffolds were significantly higher than that of dNat scaffolds (Fig. S4C). The gross appearance of the aECM scaffolds changed drastically throughout the 12 weeks of implantation time (Fig. 3A). Upon implantation, the transparent aECM scaffolds immediately reddened and at 2 weeks, there was moderate adhesion between the tendinous tissue and host’s tissue, which were both surrounded by abundant vasculature. At 4 weeks, the remodeled aECM scaffolds began to whiten and showed less adhesion. By week 12, the regenerated tendon or “neo-tendon” was compact, white in color and exhibited a smooth surface comparable to native tendon (Native) (Fig. 3A). Neo-tendons were virtually devoid of prominent vasculature and showed no tissue adhesion. At 2- and 4-weeks post-implantation, the neo-tissue formed from the aECM scaffolds had an enlarged appearance, as evidenced by the increase in tendon cross-sectional area (Fig. 3B). The average cross-sectional area of neo-tendon tissue had reduced by 12 weeks, to a size nearer to native tendon. In contrast, dNat scaffolds were visibly prominent at 2 weeks (Fig. S5A) and could still be observed at 12 weeks post-implantation (Fig. S5B), despite tendon-like tissue formation beneath the scaffold implantation site, indicating a different mode of tissue reparation. As such, the cross-sectional areas of tissue sections taken from the dNat group were significantly larger than the neo-tendon formed from the aECM scaffold group (Fig. 3B). Furthermore, H&E staining at 2 weeks revealed that the large quantity of cells had infiltrated into the aECM scaffolds and the extensive ECM was further condensed and assembled into an ordered ECM throughout the implantation time (Fig. 3C). Whereas in the dNat scaffold group, H&E staining of tissues sections revealed that tendon-like tissue had formed beneath the scaffolds (Fig. S5A) and contained large and apparent areas of defects at 12 weeks (Fig. S5B). Cell nuclei were visualized by DAPI staining, showing a large quantity of cells that occupied the aECM scaffolds had become more elongated over the 12 weeks, to a similar shape seen in native tendon (Fig. 3D). The Nuclear Shape Index (NSI) confirmed the DAPI observations, showing a decrease over time as the nuclei changed to more spindle-like (Fig. 3E), and this was dissimilar to nuclei observed within the dNat group. The quantity of overall cellular DNA content also gradually decreased, indicating the cellularity of neo-tendon lessened over time and approached to that of native tendon at 12 weeks (Fig. 3F). Immunostaining (Fig. 3G) revealed that cell proliferation during the remodeling process was highest at 2 weeks, as detected by Ki67 staining, whereas its expression was nearly undetectable at 4 and 12 weeks. CD31 staining suggested that capillaries were distributed within the aECM scaffolds throughout the remodeling process. The number of these capillary-like structures significantly decreased after 4 weeks and showed no significant differences in numbers at 12 weeks, when compared to native tendon (Fig. 3H). CD146-positive cells, a marker of tendon stem cells with clonogenic and multilineage differentiation abilities, were widely distributed around the implanted aECM scaffolds at 2 weeks. The expression of CD146 decreased over time and was comparable to the levels seen in native tendons by week 12. Scleraxis (Scx) is a transcriptional regulator first expressed in tendon progenitor cells, and its expression was widely distributed within the aECM scaffolds at 2 weeks. Scx expression had a sharp decrease and was difficult to detect at 4 and 12 weeks. Tenomodulin (Tnmd) is a marker of mature tenocytes, its expression first decreased (4 weeks) and subsequently displayed a marginal increase (12 weeks) to an expression level approximating to that observed in native tendons. Assessment of CD146, Tnmd and CD31 within the interior of the dNat scaffolds (Fig. S5C) and within the tendon-like tissue that formed beneath the dNat scaffold implants (Fig. S5D) yielded marked differences in staining. In all instances, staining was poor within the dNat scaffold interior, suggesting limited cellularization. In the tendon-like tissue that formed beneath the scaffold, distinct areas of CD146 staining were comparable to the neo-tendon formed from aECM implants, however there was less Tnmd, indicating that despite the presence of tendinogenic progenitor cells there was limited differentiation into mature tenocytes. The lack of CD31 staining suggested that there was minimal vascularization throughout the tissue. Western blot analyses comparing aECM groups to native tendon for Scx and Tnmd total protein expression (Fig. S6A) were in-line with the immunostaining, both Scx and Tnmd had highest expression levels at 2 weeks. Scx decreased over the 12 weeks to show no significant differences to protein levels detected in native tendons (Fig. S6B). Tnmd expression dipped at 4 weeks, but recovered by 12 weeks, showing no significant differences compared to native tendon (Fig. S6C). These data indicated that aECM implants were infiltrated in an ordered manner by appropriate cell types, which were altered over the course of the implantation time to achieve a cellular content that closely resembled native tendon tissue.

Fig. 3. Remodeling of aECM scaffolds into neo-tendon.

(A) Representative images of the gross morphology of the remodeled aECM scaffolds at 2, 4 and 12 weeks and the native tendon (Native). White arrows indicate the suture sites. (B) Measurements of the tissue’s cross-sectional area (n = 5). (C) H&E staining showing the scaffolds remodeled into neo-tendon throughout the 12 weeks implantation time. (D) DAPI staining of cell nuclei to show quantity and changes to nuclear morphology. (E) Nuclear shape index (NSI) of aligned cell nuclei and (F) DNA content were determined. (G) Immunostaining for phenotypic markers Ki67, CD31, CD146, Scx and Tnmd, and counterstained for nuclei by DAPI (blue). (H) The change in capillary number throughout the measured duration of scaffold remodeling. All images are representative of n = 5 individual experiments and graph data is presented as the mean ± standard error. Scale bars: (A), 1 mm; (C, G), 100 μm; (D), 50 μm. Statistical analysis is shown as *P ≤ 0.05, **P ≤ 0.01, ns = no significance.

ECM deposition and reorganization.

We next assessed the impact of host remodeling on the aECM scaffolds by investigating the newly synthesized ECM arrangement, composition and similarity to native tendon, using histological and immunofluorescent staining, at the earlier 2 and 4-week time points (Fig. S7A) and at 12 weeks (Fig. 4A). Sirius Red staining showed that the fibrillar collagen deposition increased, became more longitudinally oriented and tightly packed over the course of the 12 weeks implantation time. Verhoeff’s and Safranin O staining revealed that the elastin and sGAG content were well distributed throughout the remodeled aECM scaffolds during the 12 weeks of implantation time. Immunostaining showed collagen type I was localized throughout the remodeled aECM scaffolds, whereas collagen type III had low expression and was primarily cell-associated at 12 weeks (Fig. 4A). There were no observable differences between the collagen arrangement in aECM-derived neo-tendon and native tendon; both had long collagen type I and short collagen type III bundles, which were aligned in a more cell-orientated manner. Tissue ultrastructure was investigated using transmission electron microscope (TEM). With prolonged implantation time, the fibrous tissue increased in alignment and density, as observed in longitudinal (Fig. S7A, Fig. 4A) and cross-sections (Fig. S7B), respectively. These ECM components, including collagen I and III, elastin and sGAG were also distributed in the tendon-like tissue that grew below the dNat scaffolds at the 12-week time-point, however, the collagen fibers had irregular arrangement (Fig. 4A). Moreover, the average diameter of collagen fibrils steadily increased in the aECM group and were significantly thicker than those found in the dNat group at 12 weeks, although both scaffolds had a smaller fibril diameter than the native tendon (Fig. 4B). Quantitative data showed that the collagen content gradually increased (Fig. 4C), whilst the content of elastin (Fig. 4D) and sGAG (Fig. 4E) decreased progressively with longer aECM implantation time. Initial differences in content were expected, as aECM scaffolds were likely derived from fibroblast cells during subcutaneous implantation. Despite this, over the course of the 12 weeks, the aECM group demonstrated a remarkable trend that approached the ECM content detected within native tendons, and no obvious differences were seen between native tendon and aECM-derived neo-tendon at 12 weeks. However, at 12 weeks, the total soluble collagen content in the dNat groups was significantly lower, the elastin content was marginally higher, and the sGAG content was significantly higher than that of both the neo-tendon and native tendon. Western blot analyses indicated that within the aECM group, the total tissue protein expression of collagen I gradually increased, whereas collagen III showed a declining trend throughout the remodeling process (Fig. S7C). To ensure that the collagen deposited was not associated with fibrocartilaginous tissue formation, we stained tissue sections for mineralization using Von Kossa’s stain and for collagen II deposition (Fig. 4F, G). The aECM-derived neo-tendon was negative for mineralization, calcification and collagen type II (Fig. 4F). However, a large proportion of the tendon-like tissue formed beneath the dNat scaffolds exhibited tissue calcification, and these areas stained positive for collagen type II (Fig. 4G). These data suggested that the collagen content seen within the dNat experimental group was collagen type II. A limited amount of calcium deposits were expected in native tendon tissue, owing to general wear and microtears with poor repair due to tendon hypocellularity and reduced blood supply [25]. We assessed native tendon sections and could not detect mineralization, calcification (Fig. S8A) or collagen II (Fig. S8B). This could be explained by the young age of the animals utilized in this study (approximately 20 weeks), thus it would not be uncommon to visualize isolated cases of calcification in older and larger animal tendon tissue or implants. Taken together, these data suggested that the large amounts of cartilaginous tissue that formed in the dNat group were causative of the attenuated and delayed remodeling activity after implantation. Overall, these data offer an explanation wherein the improved mechanical and functional properties observed were due to the rapid remodeling of aECM scaffolds over the 12 weeks implantation period; suggesting the ECM content and structure would be similar to that of mature native tendon tissue ECM, following a prolonged implantation time.

Fig. 4. Analysis of ECM deposition, quantification and calcification.

(A) Histological analysis panel, including Sirius Red (collagen, red), Verhoeff’s (elastin, black) and Safranin O (sGAG, red) staining, immunostaining for collagen I and III, and TEM of collagen fiber ultrastructure. (B) Average diameter of collagen fibrils across the different groups. Quantitative analysis of the (C) total soluble collagen, (D) elastin, and (E) sGAG content within the experimental groups. Von Kossa’s stain for mineralization (dark purple/black), immunofluorescent staining for Collagen II of the remodeled (F) aECM and (G) dNat scaffolds at 12 weeks, areas outlined by red dashed lines indicate the remaining dNat scaffold. All images are representative of n = 3 individual experiments and graph data is presented as the mean ± standard error. Scale bars, (A), histology panels: 100 μm; TEM images: 500 nm; (F, G), Von Kossa’s staining whole sections, 500μm; magnified and immunostaining, 100 μm. Statistical analysis is shown as *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ns = no significance.

Early macrophage infiltration and inflammatory resolution.

Immunofluorescence staining showed that at an early time-point of 3 days, there was predominantly CD206-positive macrophages in aECM scaffolds, whereas iNOS-positive staining was evident in large quantities around the dNat scaffolds (Fig. S9). Double staining for CD68/CD206 and CD68/iNOS indicated that macrophages were still detectable within aECM scaffolds at 2 weeks and showed a decrease at 4 weeks (Fig. 5A). At 12 weeks, the inflammatory response appeared to have been resolved, as the numbers of CD68, CD206 and iNOS positive cells within the neo-tendons were comparable to native tendon’s resting tissue state (Fig. 5B). In the case of dNat scaffolds, M1 macrophage cells were distributed around the remaining dNat material, as confirmed by elevated CD68 and iNOS expression level at 12 weeks. Additionally, we assessed the dynamic changes in inflammation-related cytokine expression throughout the remodeling process. Monocyte chemoattractant protein-1 (MCP-1), a key mediator of proinflammatory response, decreased in aECM scaffolds at 2 and 4 weeks, and showed no obvious differences with native tendon expression at 12 weeks (Fig. 5C). However, MCP-1 expression in dNat scaffolds was significantly higher than neo-tendon and native tendon at 12 weeks. Proinflammatory cytokines including interleukins (IL)-1β (Fig. 5D), IL-6 (Fig. 5E), and tumor necrosis factor α (TNF-α) (Fig. 5F) were also analyzed. Their expression decreased to the level of native tendons by the 12-week time-point. In contrast, the expression of IL-1β and IL-6 in dNat scaffolds at 12 weeks was significantly higher than neo-tendon. TNF-α on the other hand, showed no significant difference. Anti-inflammatory cytokines IL-4 (Fig. 5G) and IL-10 (Fig. 5H) expression decreased over the course of 2 weeks to 12 weeks, towards the levels produced by native tendon. The production levels of these cytokines in dNat scaffolds remained high at 12 weeks. Stromal cell-derived factor-1 (SDF-1) plays an important role in stem cell recruitment, proliferation and differentiation. Its expression in aECM scaffolds displayed a steady downwards trajectory during the remodeling process, and at 12 weeks, SDF-1 production levels were indistinguishable from native tendon (Fig. 5I). SDF-1 expression in dNat scaffolds was slightly higher than that of neo-tendon and native tendon but exhibited no significant difference. Interferon (IFN)-γ expression remained low and showed a slight downwards throughout the remodeling process, and its expression levels showed no obvious differences between aECM scaffolds and native tendon, which were significantly lower than that of dNat scaffolds at 12 weeks (Fig. 5J). Taken together, these data demonstrated the effectiveness of the aECM scaffolds in the rapid resolution of the inflammatory response, in contrast to the persistent inflammatory cytokine expression induced by dNat scaffolds. Regulation of the inflammatory response is an important feature of engineered scaffolds and suggests that aECM scaffolds had a reduced risk of rejection or provoking foreign body responses, thus preventing chronic inflammation.

Fig. 5. Inflammatory response during remodeling.

(A) Double staining showed the distribution of the CD68 positive pan-macrophage (red) and CD206-positive M2 type macrophages (green), or iNOS-positive M1 type macrophages (green). (B) The number of CD68, CD206 and iNOS-positive cells was calculated. The dynamic changes of inflammation-related cytokine expression were detected by ELISA, including MCP-1 (C), IL-1β (D), IL-6 (E), TNF-α (F), IL-4 (G), IL-10 (H), SDF-1 (I), and IFN-γ (J). All images and data are representative of n = 5 individual experiments and graph data is presented as the mean ± standard error. Scale bars, 100 μm. Statistical analysis is shown as *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ns = no significance.

Discussion

A key challenge in tissue engineering is the design and fabrication of 3D scaffolds mimicking the specialized anisotropic architecture and composition of native tissues, with the capability to restore mechanical function to the injured tissue [26, 27]. Our investigation aimed to achieve this in tendons, a typically hard-to-repair connective tissue, by providing specialized scaffold implants with the capacity to promote the host cells’ tissue remodeling potential. In the present study, we developed sacrificial polymer templates for subcutaneous implantation, which supported widespread cellularization, ECM deposition and vascularization. Polymer leaching and decellularization of the explanted templates resulted in autologously generated porous aECM scaffolds with aligned microchannels to accommodate a regenerative tendon microenvironment. Within 3 days, aECM scaffolds displayed extensive macrophage infiltration. At 2 weeks, there was M2 macrophage repolarization, marked proliferative tendon cell in-growth and angiogenesis. Between 4–12 weeks, there was remission of the inflammatory response, decreased vascularization, and the scaffold was remodeled into an organized and collagen-rich ECM. At 12 weeks, neo-tendon had the morphology, ECM composition and cellularity similar to native rat Achilles tendon (Fig. 6A, B). In comparison, tendon-like tissue formed below the dNat scaffolds, indicating a different mode of remodeling throughout the 12 weeks of implant time (Fig 6A, Fig. S10). Additionally, disorganized tissue regeneration, poor host-scaffold integration and substantial calcification were occurred in tendon-like tissue part. The dNat scaffolds also resulted in chronic inflammation within the 12 weeks of implant time (Fig. 6 A, C).

Fig. 6. The remodeling process of aECM and dNat scaffolds.

(A) H&E staining of whole defect site sections showing the aECM scaffold remodeled into neo-tendon throughout the implantation time and at 12 weeks for the dNat group. Native tendon is shown alongside for comparison. Schematic representations of tendon remodeling process of (B) aECM and (C) dNat scaffolds. Scale bar, 500 μm.

We made the interesting observation that both vWF⁺ and CD31⁺ cells were present within the PCL-Cell-ECM constructs, following subcutaneous explant. This suggested that pre-formed microvasculature structures may have been present following polymer leaching and decellularization, and it could be speculated that these may have supported early CD31 endothelial cell infiltration and capillary functionality following implantation within tendons [28]. The biocompatible aECM and microchannels supported host cell infiltration, the major source was likely to be from the tendon tissue flanking the graft site, in addition to cells from the surrounding tissue. The formation of long, aligned microchannels within the scaffold allowed the host’s cells to penetrate the interior of the scaffold and occupy most of the internal space, as early as 3 days post-implantation. Incorporation of aligned microchannels allowed for simultaneous intrinsic and extrinsic scaffold remodeling by host cells, in contrast to the slow responses seen in tendinopathies and as observed within the dNat group. Within the dNat scaffolds, the host’s cells could not penetrate the interior of the scaffold and were restricted to remodeling the dense ECM in order to progress through the scaffold, thus only extrinsic remodeling could take place. This observation could also explain why tendinopathies typically have limited regenerative capabilities with pathological healing outcomes [29]. Therefore, the incorporation of aligned microchannels into decellularized native tissues, or clinically available acellular allogenic and xenogeneic tendon scaffolds, through physical alteration may be sufficient to overcome the shortfalls of the dense and difficult to infiltrate native tissue ECMs; providing the matrix maintains shape without collapsing. Indeed, it would be interesting to investigate this within large tissue constructs wherein such alterations have more probable feasibility.

The scaffolds we designed for this study were intended to induce the host cell’s innate regenerative abilities and enable in situ tendon repair [12, 30]. Until now, a plethora of strategies for the repair of tendon and other tissues opted for the use of cell-seeded scaffolds [4, 5]. Investigations using tenocyte-seeded scaffolds, such as poly-HBHx (3-hydroxybutyrate-co-3-hydroxyhexanoate) combined with collagen hydrogels and decellularized tendon matrix, demonstrated tenogenic ECM and tissue formation with improved biomechanical properties [31–33]. Tendon stem-cell progenitors has also shown great promise [5, 34]. However, isolation of cells directly from tendon tissues inevitably leads to donor site morbidity, therefore reducing the practicality of harvesting cells from uninjured tendons [5]. The use of autologous adipocyte-derived stem cells (ADSCs) and mesenchymal stem cells (MSCs) have also been proposed for tendon repair strategies. For example, ADSC-seeded composite scaffolds composed of polyglycolic acid and polylactic acid fibers were used for the reparation of rabbit Achilles tendon defect [35]. The neo-tendon that formed showed similar histological structure to native tendon, with functional recovery and partial mechanical restoration. Despite the documented success in controlling stem cell differentiation, extensive research is still required to delineate exact methodology required to definitively promote stem cell differentiation into tenogenic lineages and to account for variables such as donor age and stem cell survivability [5, 36]. Instead, we showed that our aECM scaffolds encouraged the endogenous cells to populate and remodel the scaffold. Consequently, we demonstrated that our cell-free scaffolds could be used to effectively regenerate tissue without the added uncertainty that can come with exogenous stem cell co-delivery. We were able to visualize the proliferation marker Ki67 alongside high expression of CD146, Scx and Tnmd, within the aECM scaffolds at 2 weeks post-implantation. It can be speculated that these early infiltrating and highly proliferative cells could be tendon progenitor cells. Tendon progenitor cells mature into tenoblasts and then non-proliferative tenocytes, and this could explain the sharp decreases in Ki67, CD146, and Scx observed at 4 weeks. Staining for Tnmd at 12 weeks, indicated that most of all observable cells with elongated nuclei were positive for Tnmd, suggesting that tenocytes were the most abundant cell type present in the neo-tendon. Compared to cell-seeding approaches, our strategy not only avoided the complications associated with seeded cell manipulation, but also could effectively regenerate tendon, comprised of the correct cell type, ECM composition and arrangement, in addition to restoring of mechanical integrity to pre-injury state.

An increasing number of studies are demonstrating biomaterial-based scaffold immunoregulation can promote a positive remodeling response, through acute induction of macrophage accumulation and phenotype polarity, dependent on ECM scaffold topography and arrangement [37, 38]. There is a limited amount of information available regarding macrophage polarization in direct response to autologous ECM scaffold implants during tendon repair. More often, macrophage phenotype is assessed to predict the outcome of scaffold remodeling, suggesting that how the ECM itself is structured may have more precedence over the source of ECM scaffold [39]. In our study, both subcutaneous and tendon in situ implantation of aECM scaffolds accumulated macrophages with higher marker expression consistent with M2 phenotype switch. Indeed, previous research has shown that biophysical cues provided by an aligned ECM structure modulated macrophage phenotype, and the incorporation of microchannels improved constructive tissue remodeling [39, 40]. Furthermore, consistent with these prior studies, degradation products from autologous aECM scaffolds were shown to contribute to the promotion of constructive M2 macrophage polarization, which in-turn facilitated migration and maturation of progenitor cells [41]. In contrast, macrophage polarization as induced by subcutaneous implantation of the dNat scaffolds, was predominantly associated with the M1 phenotype. Following tendon implantation, these results were reciprocated, host cells could not migrate into and intrinsically remodel the dNat scaffold, instead cartilaginous tendon-like tissue formed below the scaffolds. Moreover, the accumulation M1 macrophages on the external surface of the dNat scaffolds persisted until the 12-week time-point, and this would likely continue until the remnants of the dNat scaffold could be completely degraded. These data indicated that, without suitable pore structure within scaffolds, the host’s immune cells remained unregulated, causing an undesirable persistent inflammatory response accompanied by disorganized ECM remodeling and calcification. Furthermore, within our investigation, the acute inflammatory response was decreased at 4 weeks and was resolved by 12 weeks in aECM regenerated tendon, whereas the dNat scaffolds showed significantly higher expression of inflammatory factor and thus, chronic inflammation. The linear relationship between time and inflammatory resolution, tending towards native tendon levels at the later phase of remodeling, suggested that the formation of aECM scaffold-derived neo-tendon was physiological, rather than the pathological remodeling observed in the dNat group [2, 42].

The aECM scaffolds we generated as part of this investigation had the remarkable capability to induce the host’s remodeling response. The results shown here suggest that the scaffold’s structural and compositional optimization were the reason for the promotion of host cell remodeling and tendon regeneration. We suggest that three features of our scaffolds were essential. First, biocompatibility owing to their autologous nature, which resulted in a minimal induction of an inflammatory reaction with rapid resolve. The degradation products of aECM scaffolds may have actively contributed to constructive remodeling, for example, as observed in healthy wound healing and granulation tissue remodeling [41, 43]. Second, the aligned microchannels throughout the length of the scaffold that were favorable for cell access and deep scaffold infiltration, induced macrophage polarization and allowed for simultaneous intrinsic and extrinsic remodeling. Microchannels also facilitated the orientated distribution of cells and newly synthesized ECM, which gradually replaced the original aECM scaffolds. The microchannel features resulted in the preferential remodeling time and ECM structure that was suited to support tendon functionality. Finally, the suitable mechanical properties of our grafts were beneficial for suturing to surrounding tendon tissue and supported mechanical loading and the functional use of the tendon tissue, as evidenced by the improvement of the rat’s walking ability throughout the regeneration of the tissue. Mechanical loading may have also supported and initiated the rapid cellular infiltration and alignment during the early stages of regeneration (e.g. activation of tenocyte progenitors), which were crucial for the later stages of tissue remodeling [44]. Whereas these factors were inferred through the results shown in this study, we also speculated that additional properties of a subcutaneously generated ECM scaffold may have played a role in immunomodulation and tissue regeneration. In addition to the structural support the ECM provides, pleiotropic effects can be conveyed to cells by mechanical cues, direct stimulation by ECM components or through the provision of access to soluble growth factors and enzymes trapped within the matrix. The dynamic modulation of cellular activity by ECM has been shown to be prominent during tissue development, morphogenesis and remodeling [45, 46]. In our investigation, aECM was derived from active ECM production and deposition by fibroblasts in the subcutaneous. We propose that this ECM may have also contained a significant amount of fibroblast-synthesized bioactive factors, that when used as tendon replacement implants, provoked a stronger remodeling response from the infiltrating macrophages, tendon progenitor cells, tenocytes and capillary-forming endothelial cells. From our study, it is evident that the ECM is intricately involved in promoting the remodeling response. The in-depth assessment of the aECM scaffolds proteomic content and mechanisms that these components could play on tendon regeneration, in particular the induction and source of CD146⁺ cells, warrants closer examination in future investigations.

Evaluation of mechanical properties in combination with ECM composition gave the strongest evidence for tendon remodeling. Achilles tendons are mainly composed of highly aligned collagenous fibers arranged in a longitudinal way that parallel to the mechanical axis and therefore confer high tensile strength. Conversely, poor arrangement of collagen fibrils in repaired tendons often leads to incomplete mechanical recovery and risk of re-rupture [1, 47]. Moreover, non-collagenous macromolecules are also essential, for instance, the viscoelastic properties of elastin and presence of proteoglycans enables tendons to withstand compressive and tensile forces [47]. Although the initial mechanical strength of our aECM was less that of native tendon, the mechanical strength rapidly and significantly increased after implantation; and when remodeled into neo-tendon by 12 weeks, showed no differences when compared to native tendon. We did not observe any breakage or ruptures in our aECM scaffolds, suggesting that scaffolds could withstand dynamic loading strain during early remodeling. The data presented in this study suggested that the rate of ECM and neo-tissue regeneration was fast enough to compensate for the mechanical loss caused by material degradation. This inference was further supported by the time-lapse evaluation of the structure and composition of the synthesized ECM. The collagen content of the original aECM implants increased throughout the remodeling process, whereas the sGAG and elastin content decreased. In addition, the density, diameter and degree of orientation of collagen fibers increased steadily. All components analyzed tended toward similar levels detected in native tendon. Moreover, previous studies have verified that increases in collagen type III were associated with the formation of scar tissue, which can result in further mechanical dysfunction and increased risk of re-rupture [2, 48]. Notably, we found that the expression level of type I collagen increased, while type III collagen decreased gradually throughout the remodeling process. The high collagen content detected in the dNat group was identified as collagen I and large amounts of collagen II. Taken together with the higher sGAG content, this may help explain the calcification within the tendon-like tissue, which formed below the dNat scaffolds. Large areas of calcification and hardened tissue would lead to abnormal mechanics and chance of re-rupture [49]. Our aECM scaffolds were able to overcome this risk, in addition to demonstrating enhanced and healthy mechanical and functional restoration, from scaffold-only implants.

The use of the subcutaneous space as a bioreactor for autologous aECM generation is a preferential technique for scaffold fabrication for tissue regeneration [50, 51]. Autologous scaffolds can effectively mediate the immune response and eliminate risk from foreign body reactions. Generation of autologous tendon ECM scaffolds could also be achieved in vitro, yet this will increase preparation time and would usually rely on single cell-synthesized ECM; for example, from autologous tenocytes taken from the patient (collateral donor site co-morbidity risk), to generate the ECM among PCL sacrificial template fibers within bioreactors, which is also a resource and time expensive method with added risk of infection across long-term culture [35]. Subcutaneous implantation circumvents the issues with time, reducing production time from 10 to 6 weeks for autologous ECM-only scaffold generation [35, 52]. In addition, subcutaneous scaffold generation enables the generation of an ECM containing multiple facets synthesized by multiple cell types, including the identified SMA-positive cells (activated fibroblasts; main ECM scaffold base), vWF/CD31-positive cells (potentially pre-vascularised scaffold) and macrophages (actively remodeling, account towards acute inflammation resolution during subcutaneous implantation, which we speculate affords scaffold topography that can enhance the elimination of M1 faster, in preference of M2 phenotypes). Overall, we suggest that the approach used in this study generates pro-regenerative autologous aECM scaffolds, as opposed to the mature or densely cross-linked ECM by cultured tenocytes, which may not have favorable reactions with the hosts’ immune system (tenocyte-only remodeled topography) and may have lower potential for rapid angiogenesis (lack of vascular-like structures within the scaffold to facilitate endothelial cell influx). For these reasons, we believe that subcutaneous implantation for the generation of autologous ECM is a worthwhile option to apply in tissue regeneration.

As with any new technology, translation from laboratory to clinic presents scientific and regulatory challenges not only to address safety and efficacy, but also to elucidate the cellular and molecular mechanisms influenced by structured aECM scaffolds [26, 53]. A more comprehensive investigation is necessary to define the specific roles of the various kinds of cells from the tendons, as well as the surrounding tissues and the mutual influence of each on the other. It would also be of interest to examine the parameters of frequent mechanical stimuli on beneficial cell response and tendon remodeling. Here, we used tendon resection to replace a section of tendon tissue with the generated scaffolds, to mimic the commonly used open surgical debridement procedure performed in patients with tendinopathies [26]. This procedure aims to remove the inflamed and damaged tendon tissue, although it is noted that it would be of interest to assess how our scaffolds perform following a prolonged period of inflammation, as seen in clinical tendinopathy cases, to further assess inflammation’s impact on tendon tissue regeneration and the immunoregulatory potential of our scaffolds [54]. In terms of clinical accessibility, despite subcutaneous implantation being a simple surgical procedure and a safe method, there is inevitably a delay in treatment for the time it takes for the template to be covered by autologous ECM (4 weeks) and the subsequent scaffold processing (approximately 2 weeks); although we believe that this is an acceptable compromise given the effectiveness in regenerative capabilities that were shown in this investigation. The approach outlined here is straightforward and has the potential to be applied to larger tendon defects in larger animals. Furthermore, our approach could be utilized to generate scaffolds for numerous other aligned and specialized tissue types; easily customizable polymer templates can be used to generate a broad range of ECM scaffolds that go beyond adjustment of pore or microchannel size (aligned, concentrically aligned, non-linear, meshes, etc). Subsequent polymer leaching of these aECM scaffold would thus yield in unique and catered structures ideally suited for the patient’s needs. Based on previous work, ECM products can be stored long-term without mechanical strength decrease, highlighting the versatility in scaffold application for clinical procedures [52, 55, 56].

Perhaps a key limitation that does involve significant changes would be the loss of bioactivity from any factors that remain trapped within the ECM structures of scaffold. In order to circumvent this, removal of advanced templates with mild leaching method and appropriate cryo-preservation and storage measures would require investigation in future studies. Optimization across different animal models would be required to not only to perfect scaffold generation time, microchannel size, and means of enhancing initial mechanical strength, but to also assess safety and efficacy before consideration for clinical implementation [54]. Furthermore, we have utilized our aECM scaffolds in an autologous setting, it would be worth investigating whether the scaffolds could support inflammatory resolution and positive remodeling in both allogenic and xenogeneic situations, therefore reducing procedure time by offering ‘off-the-shelf’ scaffolds for implantation.

Conclusions

In summary, we have shown that aECM scaffolds with aligned microchannels supported tendon remodeling and achieved enhanced restoration of functional and mechanical properties to Achilles tendon defects in rat models. Our proposed strategy for scaffold engineering was able to overcome the common complications faced by multiple tissue engineering approaches that contribute to the difficulty of clinical translation. Regeneration of tendon, a tissue with high mechanical strength-orientated roles, could be achieved by rapid remodeling of engineered autologous scaffolds - containing the key feature of aligned microchannels - by the host’s endogenous macrophage and tendon cells, without the need of additional cell seeding or bioactive factor modifications. Furthermore, we showed that defects to a tissue that bears high mechanical loading strength could be achieved with using precision designed scaffolds, despite their initial low mechanical strength. Thus, it lowered the requirements of initial matching mechanical strength for tissue engineering scaffolds. More broadly, this in vivo engineering approach could be applied to the repair of other tissues with aligned architecture. Such tissues include muscle, cardiac, vasculature, esophageal, urethral and nerve, among others. Overall, the present study provides an innovative concept that combines multiple elements of traditional tissue engineering techniques and delivers new routes for regenerative medicine and materials science, towards translation and a clinically viable treatment.

Supplementary Material

1

Movie S1. Micro CT scanning demonstrated the 3D structure of aECM scaffolds.

2

Movie S2. Micro CT scanning demonstrated the 3D structure of dNat scaffolds.

3

Fig. S1. Cellularization of template scaffolds explanted from the subcutaneous pouch.

4

Fig. S2. Subcutaneous cellularization and macrophage response to aECM and dNat scaffolds.

5

Fig. S3. Functional evaluation.

6

Fig. S4. Analysis of implanted scaffold cellularization.

7

Fig. S5. Remodeling of dNat scaffolds.

8

Fig. S6. The expression of Scx and Tnmd in aECM.

9

Fig. S7. Additional assessment of ECM deposition and arrangement.

10

Fig. S8. Von Kossa’s and Collagen II staining for native tendon.

11

Fig. S9. Early macrophage infiltration in scaffolds.

12

Fig. S10. The remodeling process following dNat scaffold implantation.