1 Introduction

Bone–tendon interface (BTI) injuries are one of the most commonly injured anatomical structures in the orthopedic field, with more than 30 million injuries occurring world-wide [1]. Most commonly affected areas include the anterior cruciate ligament (ACL) in the knee and rotator cuff of the shoulder, with 48% of ACL injuries occurring in the BTI area [2]. Due to its transitional composition of tissue and high biomechanical loads, BTI injuries often require surgical intervention. The annual incidence of ACL injuries is 30–78 per 100,000 persons [3, 4], requiring 400,000 ACL reconstructions annually worldwide [6]. While ACL reconstruction result in functional improvement and return to sports, graft retears and residual instability occur in 8–25% of patients [7] leading to revision surgeries in 17.5% of patients [8]. This is, in part, due to the defective healing in the BTI [9].

During ligament reconstructions, the tendon graft is fixed in a bone tunnel. BTI healing occurs via scar tissue formation between the tendon graft and bone. Scar tissue, which is disorganized and biomechanically inferior compared to fibrocartilage, may lead to a high failure rate and unsatisfactory clinical outcomes [10, 11]. This occurrence is caused by the lack of a gradient mineral distribution and collagen fiber disruption, which are essential to the native BTI's microstructure and metabolic activity [12]. A variety of bioinspired materials and related fabrication techniques including 3D printing [13, 14] and electrospinning [16, 17] were utilized to mimic the inherent hierarchical structures of BTI, with showing the usefulness of multiphasic scaffolds for reconstruction of the BTI [18, 19]. Nevertheless, the biggest obstacle to functional BTI healing is still obtaining a progressive change in physiological characteristics [20]. To address this issue, previous studies aimed to improve the BTI healing process via osteoinductive materials [21, 22] and stem cells [23,24,25,26,27,28,29,30]. Cell therapies targeting BTI utilizing bone marrow-derived mesenchymal stem cells (BM-MSCs), Synovium-derived mesenchymal stem cells (Sy-MSCs), and periosteal progenitor cells have been reported, resulting in some biomechanical and histologic benefits. To improve delivery of cells, Lui et al. utilized tendon-derived stem cell sheets rather than individual cells to wrap the transplanted graft, which resulted in improved healing without the use of scaffolds [32]. In terms of biomimicry, however, there is an unmet clinical need to repair BTI via fibrocartilage with a level of integration comparable to that observed in the native tissue.

The BTI, also known as the enthesis, is a highly specialized transitional tissue consisting of tendon, non-mineralized, mineralized fibrocartilage, and bone. These structural characteristics lead to gradual mechanical property changes allowing progressive transmission of mechanical loads from tendon to bone and withstanding load from various vectors [33]. We have previously developed scaffold-free cartilage constructs that resulted in not only excellent biomimicry toward cartilage, but also improved cell delivery and preservation of the transplanted construct after 6 months in a non-human primate model [34]. We aimed to engineer sy-MSC based scaffold-free fibrocartilage constructs and investigate in vivo BTI healing efficacy in a rat ACL reconstruction model. We hypothesized that the use of scaffold-free, fibrocartilage neo-tissue would promote fibrocartilage-like tissue healing in the BTI.

2 Methods

2.1 Cell isolation and culture

All experiment protocols involving animals were approved by the Institutional Animal Care and Use Committee. Three male Sprague-Dawley rats were used. The synovium of knee joints from the rats were harvested from the anterior fat pad. Tissues were minced, treated for 4 h at 37 °C with 0.2% type II collagenase (Worthington Biochemical Corp., Freehold, NJ, USA), and centrifuged at 1500 rpm for 10 min. After washing twice with complete growth medium (CGM; α-MEM containing 10% FBS, 100 U/ml penicillin G, and 100 μg/ml streptomycin), the digested tissue was passed through 100 μm nylon mesh filter (Falcon, BD Bioscience, NJ, USA) to yield single-cell suspensions. Rat synovium-derived mesenchymal stem cells (rSy-MSCs) were counted with a hemocytometer and seeded at a density of 8×103 cells/cm2. After 3 days, the medium was changed to remove non-adherent cells and incubated at 37°C with 5% CO2. rSy-MSCs were passaged at 70–80% confluence and cells at passages under 4 were used in this study.

2.2 Scaffold-free fibrocartilage construct fabrication

Schematic figure of experimental flow was suggested as Fig. 1. The effects of transforming growth factor beta-3 (TGF-β) and connective tissue growth factor (CTGF) on fibrocartilage construct fabrication were compared. 4×106 of rSy-MSCs were seeded at 60 mm tissue culture dish. After 24 h following cell seeding, cells were scraped with a cell lifter (Corning, New York, US) and placed 6 well plate with cell growth media. Subsequently, the medium was changed to differentiation medium (DM; high glucose DMEM supplemented with 1% insulin, transferrin, sodium selenite (ITS) + mix, 50 μg/mL ascorbic acid-2-phosphate, 100 nM dexamethasone, 1% penicillin–streptomycin, 100 μg/mL sodium pyruvate, and 40 μg/mL l-proline) with two growth factors (10 ng/mL of TGF-β and 100 ng/mL of CTGF all from Sigma-Aldrich, MO, USA) or without (control group) two growth factors. Subsequently, the constructs were cultured in DM for 2 weeks to induce differentiation in the respective growth factor groups.

Fig. 1
figure 1

Schematic overview of scaffold-free fibrocartilage engineering for bone-tendon interface healing

Full size image

For sulfated glycosaminoglycans (sGAG) content analysis, freeze-dried samples were weighed and digested in papain solution with 125 mg/mL papain with 5 mM l-cysteine, 100 mM Na2HPO4, 5 mM EDTA (Sigma-Aldrich, MO, USA) at 60 °C for 16 h as previously reported [34]. The amount of sGAG was quantified with Blyscan Assay following the manufacturer’s instructions (B1000, biocolor, Carrickfergus, UK).

For collagen content analysis, freeze-dried samples were solubilized in 0.1M HCl with 0.1% (w/v) pepsin at RT for 24 h. The amount of collagen was quantified with Sircol Collagen Assay (SCA) following the manufacturer’s instructions (S1000, biocolor, Carrickfergus, UK).

For histological analysis, samples were fixed with 10% formalin for 24 h at room temperature (RT) and then embedded in paraffin wax. Sample slides in 5 μm thickness were made and stained with safranin-O (Sigma-Aldrich, MO, USA) to examine the degree of GAG deposition. Other slides were used to immunostain collagen type I (1:100, Abcam, Cambridge, UK) and collagen type II (1:100, Abcam, Cambridge, UK) antibodies using an immunohistochemical assay kit (GBI labs, Bothell, WA, USA).

For gene expression analysis, fibrocartilage constructs were homogenized with Biomasher (OPTIMA INC, Japan) on day 14. Total RNA was extracted from cells using Trizol reagent (Invitrogen) according to the manufacturer’s instructions. Total RNA (500 ng) in a 20 μl mixed solution was reverse-transcribed into cDNA by iScript cDNA synthesis kit (Bio-Rad). RT-qPCR was performed in a CFX96 Touch Real-time PCR Detection System (Bio-Rad) using SYBR Green detection (Roche, Mannheim, Germany). The mRNA expression levels of samples were normalized to GAPDH (glyceraldehyde-3-phosphate dehydrogenase) as a housekeeping gene and calculated by using the comparative CT method. Expression of genes including collagen type I (COL1A1), collagen type II (COL2A1), and aggrecan (ACAN) were analyzed. The primers used are shown in Supplemental Table 1.

2.3 Collagenase treatment on tendon

A sheath known as the paratenon may deter BTI integration and are usually dissected off during ACL autograft reconstructions [35]. In addition to this idea, collagenase treatment was utilized to improve the attachment of rSy-MSCs on the surface of the flexor digitorum profundus (FDP) tendon graft. The volume of collagenase was optimized to induce slight digestion of the tendon surface which resulted in 0.5 μl/mm2. Briefly, a total of 18 (3 grafts for each group), 2 mm × 5 mm × 1 mm FDP grafts were harvested from 9 rats. The effects of collagenase treatment were evaluated according to treatment time; 0 , 5, 15, 30, 60 min. The group without any treatment was named as the control group. A mechanical abrasion group was also analyzed, where we applied mechanical abrasion by rubbing the tendon with sandpaper (220 grit) three times to the tendon surface for comparison with collagenase treatment. After treatment, the degree of integration with tendon and cells was evaluated by the histological observation and mechanical test.

Sirius red (SR) staining was used to clearly distinguish collagen fibers structure of tendon. Briefly, sample were incubated for 1 h in a 0.1% Sirius Red solution diluted in picric acid. After rinsing in 0.5% Glacial acetic acid, the samples were dehydrated and mounted using Permount Mounting Medium (Fisher Scientific). All PSR imaging was performed on an axioscope 5 microscope (Zeiss) with linear polarizers (PSR-POL).

For cell adhesion analysis, 5 × 106 cells were labeled with PKH26 Red Fluorescent Cell Linker Mini Kit For General Cell Membrane Labeling (Sigma-Aldrich, MO, USA). Subsequently, cells were seeded on the tendon surface (n = 3). Following treatment, cells remaining on the cell culture plate that did not attach to the tendon surface were detached by from the culture plate using 0.05% trypsin (Gibco BRL) and counted with a hemocytometer. Finally, the percentage of cell adhesion was calculated by ratio of cell number counted from hemocytometer to initial seeding number of cells. Three independent experiments were repeated for the reliability.

Tensile test was performed by connecting the FDP with a length of 5 mm between two custom-made zigs. The pulling test was applied to measure the longitudinal axis strength, and this was carried out using a ZWICK/Roell Z005 testing machine with a 5 N loading cell (ZWICK/Roell, Ulm, Germany). The tests were performed with a 1 mm/min pulling rate for each sample. Loading values measured with a sensitivity of 0.1% by the device's load cell were entered into the software, and force values (F) were obtained; the tests were terminated after the detection of a force decrease and rupture of the sample. The Young's elastic modulus is typically calculated by analyzing the linear portion of the stress–strain curve, indicating the elastic resistance to compression.

2.4 in vitro bone/scaffold-free fibrocartilage construct/tendon integration

The effects of fibrocartilage differentiation and collagenase treatment were tested in an in vitro BTI reconstruction model. FDP tendons and bone tissue from the femur were obtained from pig carcasses (Finish pig, Yorkshire, female, >1 year, free of knee osteoarthritis) within 1 h of sacrifice. For bone preparation, cancellous bone was cut in to 6 mm diameter discs with 2 mm height. A cylinder-shaped defect (3 mm diameter) was created in the center using a 3 mm biopsy punch. For tendon tissue preparation, tendon samples were cut into 3 mm diameter, 2 mm height disc shapes perpendicular to the tendon fiber orientation. Overall, five groups were tested (n= 6 per group); (1) Empty (2) Col-D− (3) Col+D− (4) Col-D+ (5) Col+D+. Col+ groups were subjected to collagenase treatment of the tendon surface (15 min, 1% collagenase, 0.5 μl/mm2) before inserting the tendon disc into bone tissue. D+ groups were subjected to insertion of fibrocartilaginously differentiated scaffold-free construct using two growth factors, while D-groups were subjected to insertion of scaffold-free constructs cultured without growth factor treatment. After assembling the tendon tissues within the bone tissues, the in vitro BTI reconstruction constructs were cultured in high glucose DMEM supplemented with 1% ITS for 4 weeks. Afterwards, the histological and biomechanical evaluation was performed with these experimental groups.

For integration failure stress analysis, mechanical test using a push-out test was done as previously described [23]. While recording load, a specialized cylindrical-shaped indenter with 4-mm-diameter coupled with Universal testing machine pushed the inner tendon part with the rate of 1 mm/min. The load at failure and interface area were divided to compute failure stress (integration strength). TestXpert II was used for both the initial preparation of the tests and the post-test analysis of the data.

2.5 Bone-to-tendon healing in vivo efficacy

A total of 56 SD rats (27 weeks, 350g, Male) were used in this study. Rats were allocated in to ‘ACL recon’ group (n= 28), and ‘ACL recon+FC’ group (n= 28). Time points of 2 and 4 weeks were analyzed, with 14 animals allocated at each time point. The following is the brief description of the surgical procedure. The rats were anesthetized by 3% isoflurane with 2 L/min oxygen. Each rat was injected with dexmedetomidine (0.2 mg/Kg, IP) and ketamine (80 mg/Kg, IP). Oxygen (2 L/min) was delivered via a nose cone during the surgical procedure. Buprenorphine (0.1 mg/kg, SC) and meloxicam (1 mg/Kg, SC) were given after the induction. The left knee of the animals was shaved and prepped in an aseptic manner. A lateral parapatellar incision was made. After patellar dislocation, the native ACL was removed. Bone tunnels were created with a 1mm diameter drill in the femur and tibia. FDP tendons were sutured at the ends and the suture was passed through the bone tunnels. The graft was seated at the native ACL site and fixed with 1.5 mm stainless steel cortical screws at the femur and tibia tunnel. For the ‘ACL recon+FC’ group, the FDP tendon graft was treated with collagenase as previously described at the proximal and distal ends of the tendon graft that come in contact with the bone tunnels. During graft passage, scaffold-free FC constructs were inserted in the bone tunnels in the surface opposite of the interference screws. Layer by layer closure of the skin was done. Postoperative analsesia was provided with meloxicam (0.1 mg/Kg, SC, qd for 4 days). The animals were single housed and fed ad libitum.

After euthanasia, 7 knees of each group and time point were analyzed for histology and biomechanical analysis, respectively. For biomechanical testing, the interferences screws were removed. All joint tissues were dissected off the knee joint excluding the ACL graft. The femur and tibial ends were placed on jigs in a biomechanical testing machine (ZWICK/Roell Z005, Ulm, Germany) directed along the longitudinal axis of the graft. After preloading of 1 N per specimen, ultimate failure load was calculated at a speed of 0.25 mm/s. Stiffness of constructs were calculated using TestXpert II.

2.6 Statistical analysis

Statistical analysis was conducted with Graph Pad Prism 8 (GraphPad Software, La Jolla, CA, USA) using one-way analysis of variance (ANOVA) with Tukey’s multiple comparison test as described in each assay. The data are expressed as means ± SD. The statistical significance was determined as *p < 0.05, **p < 0.01, and ***p < 0.001.

3 Results

3.1 Scaffold-free fibrocartilage construct fabrication in vitro

Safranin-O staining was strongly stained in the TGF-β group, but weakly stained in the control and CTGF group. CTGF+TGF-β group showed areas of strong safranin-O staining on the outer regions of the construct and weakly stained area in the center (Fig. 2A). Immunohistochemistry staining for collagen types I and II showed that the TGF-β group had more type II collagen, while the CTGF group had more type I collagen. CTGF+TGF-β group showed expression of both collagen types (Fig. 2A). Biochemical quantification results showed the most sGAG expression in the TGF-β group, followed by CTGF+TGF-β group and CTGF group (Fig. 2B). Additionally, collagen contents were significantly higher in the CTGF group, but there was no significant difference between CTGF+TGF-β and other groups (Fig. 2B, Control: 9.987 ± 3.84 μg/mg, CTGF: 22.28± 6.97 μg/mg, CTGF+TGF-β: 16.99 ± 1.14 μg/mg, TGF-β: 19.68± 2.28 μg/mg). Moreover, the gene expression profile showed higher expression of COL1A1 with CTGF exposure, whereas chondrogenic marker, COL2A1 and ACAN expression were significantly increased upon TGF-β treatment (Fig. 2C). The CTGF + TGF-β group showed intermediate COL1A1, COL2A1 and ACAN expressions compared to CTGF and TGF-β treated groups. In all, histologic, biochemical and gene expression results showed elevated fibrocartilage features in the CTGF + TGF-β group.

Fig. 2
figure 2

in vitro characterization of fibrocartilage constructs according to two growth factors addition. (A) Histologic (B) biochemical characterization of fibrocartilage constructs. (C) Gene expressions were quantified by qRT-PCR analysis. Differences among the groups were determined by one-way ANOVA with Turkey’s multiple comparisons test (*p < 0.05, **p < 0.01 and ***p < 0.001). Scale bar = 100 μm

Full size image

3.2 Improvement of integration properties in tendon via collagenase treatment

Sirius red staining showed maintenance of collagen arrangement at tendon until 15 min collagenase treatment time (Fig. 3A). More than 30 min of collagenase treatment destroyed fibrillar crimp structure. At 60 min, dense fiber bundles were completely disorganized and collagen fluorescence decreased considerably. As shown Fig. 3A and B, cell attachment on tendon, on the other hand, increased according to collagenase treatment time (Control: 35.16 ± 3.9%, Abrasion: 51.52 ± 7.51%, 5 min: 56.46 ± 5.57%, 15 min: 67.23 ± 2.3%, 30 min: 75.08 ± 2.77%, and 60 min: 80.85 ± 2.1%). In collagenase treatment groups for >30 min, cells penetrated the tendon surface and adhered to the stroma (Fig. 3A). Tensile modulus was not considerably affected until collagenase treatment of 15 min, however, it reduced dramatically at >30 min (Fig. 3C, Control: 134.02 ± 38.57 Kpa, Abrasion: 124.7 ± 40.45 Kpa, 5 min: 120.93 ± 42.97 Kpa, 15 min: 120.93 ± 42.97 Kpa, 30 min: 75.58 ± 6.75 Kpa, and 60 min: 18.23 ± 8.65 Kpa). After these results, collagenase was applied to tendon for 15 min to improve cellular adhesion without altering tensile strength.

Fig. 3
figure 3

Optimization of collagenase treatment time for enhanced integration of rSy-msc to tendon. (A) Histological analysis of integration with rSy-MSCs and tendon and (B) quantitative analysis of PKH-labeled rSy-MSCs attachment to tendon and (C) biomechanical evaluation of tendon after treatment of collagenase. Differences among the groups were determined by one-way ANOVA with Turkey’s multiple comparisons test (*p < 0.05, **p < 0.01 and ***p < 0.001). Scale bar = 100 μm

Full size image

3.3 Integration analysis of scaffold-free fibrocartilage construct in vitro

Safranin-O stained tissue section revealed the presence of positive staining regardless of collagenase treatment in the group with fibrocartilage differentiation (Fig. 4A). Additionally, collagenase treatment to tendon showed enhanced integration of bone-to-scaffold-free fibrocartilage construct as well as the integration of construct-to-tendon (Fig. 4Ae–Ah). In contrast, detachment of scaffold-free fibrocartilage construct-to-tendon interface was observed in the group without collagenase treatment (Fig. 4Aa–Ad). The integration failure stress was significantly highest in the Col+D+ group, but there was no significant difference between other groups except empty group (Fig. 4B, Col+D+: 22.494 ± 13.74 Kpa, Empty: 0.46 ± 0.13 Kpa, Col-D−: 5.72 ± 1.76 Kpa, Col+D−: 7.41 ± 2.67 Kpa, and Col-D+: 7.93 ± 2.91 Kpa). Consequently, fibrocartilage differentiation and collagenase treatment mimicked native bone-to-tendon interface and promoted bone-to-scaffold-free fibrocartilage construct-to-tendon interface integration.

Fig. 4
figure 4

Integration analysis of scaffold-free fibrocartilage construct in vitro. (A) The representative histologic images stained using safranin-O at integrated region as low magnification image (left) and high magnification images (right, a–h) (B) Quantification of failure stress after 4 weeks differentiation. Differences among the groups were determined by one-way ANOVA with Turkey’s multiple comparisons test (*p < 0.05, **p < 0.01 and ***p < 0.001 compared to the Col + D + group). B; Bone, T; Tendon, FC; Fibrocartilage construct. Scale bar = 100 μm

Full size image

3.4 Bone-to-tendon healing efficacy in vivo

At week 2, ACL Recon group showed fibrous scar at BTI. In contrast, the representative hematoxylin and eosin (H&E) stained image of ACL Recon+FC group indicated hypercelluar tissue mass attached directly bone and tendon (Fig. 5A). After 4 weeks, both groups showed enhanced integration between bone and tendon graft compared to 2 weeks. Particularly, the formation of hypercellular, fibrocartilage-like tissue was observed at the ACL Recon+ FC group (red arrow). In addition, safranin-O stained images showed increased sGAG deposition in the ACL Recon+FC group with the group compared to ACL Recon group (red arrowhead). The results of biomechanical test indicated the enhanced mechanical properties of both groups at 4 weeks operation compared to 2 weeks in line with histologic results (Fig. 5B). All knees failed at the BTI of the femoral tunnel during the biomechanical examination. In particular, the estimated stiffness of ACL Recon+FC group was increased by 88% (p= 0.0102) when compared to the ACL Recon group at 4 weeks operation. In all, histologic and biomechanical results showed improved integration in interface as well as fibrocartilaginous tissue formation at the group of ACL Recon+FC group.

Fig. 5
figure 5

Bone tendon healing efficacy in vivo. (A) The representative histological images stained with H&E and safranin-O at bone and tendon integrated region. The black rectangle indicates the region of interest for safranin-O stained images. Red arrow indicated fibrocartilage-like tissue at the ACL Recon + FC group, whereas arrowhead indicated increased sGAG deposition in the ACL Recon + FC group with the group compared to ACL Recon group (B) the biomechanical evaluation for measuring load to failure and stiffness by using universal testing machine. ST; Scar tissue, B; Bone, TG; Tendon graft, FC; Fibrocartilage cell construct, and S; Screw. Scale bar = 100 μm

Full size image

4 Discussion

The goal of this study was to engineer sy-MSC based scaffold-free fibrocartilage constructs and investigate its efficacy in an in vivo rat ACL reconstruction model. Treatment of both CTGF and TGF-β resulted in fibrocartilage differentiation of the scaffold-free constructs. Collagenase treatment on the tendon surface resulted in enhanced cell-to-tendon graft integration. Fibrocartilage constructs, together with tendon collagenase treatment resulted in BTI healing similar to native BTI and increased biomechanical strength compared to the current tendon reconstruction group both in vitro and in vivo. In all, scaffold-free fibrocartilage engineering together with tendon collagenase treatment offers a potential biomimetic solution to improve current tendon/ligament reconstructions.

Fibrocartilage, by definition, is a composite connective tissue with both fibrous and chondrogenic properties. A number of growth factor combinations utilizing BMP-2 [36], BMP-7 [37, 38], IGF-1 [39, 40], CTGF [41], and TGF-β3 [42] have been used to induce fibrochondrogenic differentiation in vitro and in vivo. Lee et al. [43] described a sequential treatment protocol of CTGF followed by TGF-β3 treatment on MSCs to induce fibrochondrogenic differentiation in a scaffold-based model for meniscus regeneration. We have utilized a combination of CTGF and TGF-β condition media for fibrochondrogenic differentiation and have found both fibrous and cartilaginous properties within 2 weeks of in vitro culture in the respective scaffold-free constructs (Fig. 2). CTGF directs fibroblast differentiation in MSCs [44] while TGF-β1 induces MSC condensation and chondrocyte differentiation. Fibrochondrogenic differentiation protocols may be influenced by MSC type, growth factor combinations and dosage, duration of differentiation, and culture conditions.

A sheath known as the paratenon covers most tendons which allow smooth gliding of the tendon via proteoglycan rich matrix [45]. These sheaths may deter BTI integration and are usually dissected off during ACL autograft reconstructions [35]. Previous studies have reported the effects of enzyme treatment on connective tissue surfaces such as collagenase [46, 47] and hyaluronidase [48] to improve integration. We observed that collagenase treatment time correlated with cell adhesion while inversely correlating to biomechanical strength of the graft. Collagenase treatment resulted in cells and surrounding tissue to interdigitate with the graft tissue, resembling native BTI (Fig. 3). Collagenase treatment, however, should be optimized to the graft being used and clinical scenarios to maximize the integration effect while minimizing biomechanical strength loss of the graft.

Regarding the rat ACL reconstruction model, we have utilized an interference screw to fix the graft to the bone tunnel, following methods from previously reported studies [49, 50]. Interference screws are commonly used in human reconstructions to provide firm fixation and compression of the graft to the bone tunnel. We have utilized interference screws in our rat model to reduce motion of the graft within the bone tunnel, considering the soft cancellous bone of rodents compared to humans. Furthermore, interference screws increased the reproducibility of the in vivo results by preventing migration and potential loss of the implanted construct. Histology and biomechanical analyses of both experiment groups were equally carried out after removal of the interference screws.

The mode of action of our constructs can be explained by biomimicry, which was achieved by fibrocartilaginous scaffold-free constructs and collagenase treatment of the tendon grafts. Fibrocartilage constructs mimicked the native fibrocartilage layers situated between the tendon and bone, while collagenase treatment resulted in interdigitating integration between the tendon graft and fibrocartilage constructs. The effect of these two factors was shown in the in vitro BTI model, where fibrocartilage constructs together with collagenase treatment resulted in the highest biomechanical strength in the BTI (Figs. 4B and 5B). Previous strategies attempted to improve BTI healing through enhanced osteogenesis, in situ tissue formation via scaffolds, or increased MSCs in the BTI. Several studies utilized growth factor delivery via biomaterials such as bone morphogenetic protein (BMP) [27, 38] and fibroblast growth factor (FGF) [51, 52], or insertion of nano-scaffolds [53, 54]. Some studies also tested the efficacy of sy-MSCs in BTI healing. Ouyang et al. observed higher percentage of oblique fiber formation after sy-MSC treatment in BTI but could not detect fibrocartilage formation until the week 4 of transplantation [25]. Wrapping the tendon graft with tendon-derived stem cell sheets also enhanced mineralized tissue formation inside bone tunnel thus facilitating early graft healing in a previous study [32]. Our study adds on such previous strategies in that we have attempted to engineer the BTI rather than deliver certain cells or growth factors. The use of scaffold-free constructs in turn may have a beneficial role in BTI engineering in terms of non-uniform application and prevention of cell loss [24].

This study has the following limitations. The rat ACL model used in this study does not reflect the human knee environment after major ligament reconstructions of the knee. Furthermore, longer observation time periods may be necessary to fully validate the fate of the implanted scaffold-free construct in the BTI. Guo et al. reported that the scaffold-free construct preserved cells from acute inflammation and ECM of the construct facilitated integration of host cells and neo-epicardial tissue [55]. While some researcher revealed the presence of MSC up to week 8 at the graft-to-bone tunnel interface in rat model [30], other reported the disappearance of transplanted MSC within 4 weeks [25]. Therefore, further research is required to identify the fate of cells after implantation by using cell tracking. Furthermore, the in vivo data could be improved with further biological and microstructural characterizations of the implanted construct. Collagen microstructure and fiber orientation, as well as mineral density of the bone tunnel are important factors in BTI healing. Further investigations concerning such factors are required in a large animal model to further investigate the biomimicry potential of the constructs and potential areas for improvement.

Taken together, the implantation of scaffold-free fibrocartilage constructs at the BTI achieved considerable biomimicry, via fibrocartilage formation and collagenase treatment of the tendon graft. Future research directed to mimic the native BTI may further enhance clinical results of current tendon/ligament reconstruction.