Time series assessment of biocompatibility of decellularized bovine tendon in rat anterior cruciate ligament reconstruction model up to 52 weeks

There is an essential demand for the development of biocompatible graft substitute for knee anterior cruciate ligament reconstruction (ACLR). We aimed to investigate biocompatibility of decellularized bovine-derived tendon using rat knee ACLR model. The Sprague–Dawley rats with 12 weeks old were used. At weeks 1, 2, 4, 8, 16, 26, and 52 (each period, n = 6) after ACLR, rats using decellularized bovine tendon (group D, n = 42) and rats using autologous exor digitorum longus tendon as a graft (group A, n = 42) were assessed for bone mineral density (BMD) of the peritibial bone tunnel, histology, and immunohistology. BMD increased over time in both groups until week 16 and then maintained without signicant group differences. Initially, cellularity in group D was lower than in group A, but by weeks 4–8, both groups were comparable to native anterior cruciate ligament and were maintained until week 52. Initially, group A had more M1 macrophages, which represent inammation, while group D had more M2 macrophages, which represent tissue regeneration. However, at most periods, the counts of M1 and M2 macrophages were comparable in both groups. This study revealed excellent biocompatibility of the decellularized tendon in the cross-species model.


Introduction
Anterior cruciate ligament (ACL) injury is one of the most common and troublesome injuries of the knee 1 .
Due to its poor self-healing ability, the majority of patients who want to return to sports must undergo ACL reconstruction (ACLR) to restore knee stability 2 . In the United States, it has been reported that 200,000 ACL injuries are expected to occur annually, and the direct cost of ACLR was estimated as $3 billion per year 3 . ACLR has used three types of grafts: autograft 4 (using each patient's tissue), allograft 5 (using cadaveric tissue), and arti cial synthetic ligament 6 . Autograft is the gold standard graft of ACLR because the clinical outcomes of ACLR using arti cial ligament were relatively poor 7 and allograft has its speci c problem such as disease transmission 8 and tissue damage by gamma irradiation for sterilization 9 . However, there are still unresolved issues with autograft. To begin with, donor site morbidity, such as anterior knee pain and patella fracture caused by harvesting bone patella tendon-bone graft 10,11 , and skin sensory change over the lower leg caused by harvesting medial hamstrings tendon 12 . Second, the graft's volume is insu cient. This issue may arise not only in multiple ligament reconstruction or revision ACLR but also in primary ACLR. ACLR using hamstring grafts < 8 mm in diameter has been reported to have a higher failure rate 13,14 . Furthermore, if the patient is short in stature, the volume of the hamstring may be smaller 15 , resulting in a graft < 8 mm in diameter 16 . Third, harvesting autografts from the patient's body would be time-consuming and labor-intensive if done during surgery.
Decellularized tissues from animal sources are attractive because the decellularized tissues have the biological tissue matrix unique to target tissues. The purpose of this study is to investigate the biocompatibility of the decellularized bovine-derived xenograft tendon in a rat ACLR model.

Results
Gross appearance of the femur-graft-tibia complex at each time point

BMD of peritibial tunnel
Micro-CT images of slices parallel to the tibial tunnel at each time point in each group are shown ( Fig. 2a). There was no signi cant enlargement of the tibial tunnel in either group over time. In each group, the BMD of the peritibial tunnel increased over time up to week 8 and was plateau after week 16 ( Fig. 2-b).
There was no signi cant difference between the groups at each time point. Supplemental videos of the BMD at weeks 4 (Supplementary Video S1) and 52 (Supplementary Video S2) in group D showed that there was no enlargement in tunnel diameter at week 52, and the BMD increased at week 52 than at week 4 at the interface between graft and bone.
Cellularity in the intra-articular graft and intratibial tunnel graft at each period and group N (native ACL, day 0). At week 1, the cellularity of group D was lower than of group A (p = 0.01), and both groups were lower than of group N (p = 0.007, p = 0.01, respectively). At week 2, there was no longer a signi cant difference between groups A and D due to increased cellularity in group D. Nonetheless, both groups D and A were lower than group N (p = 0.007 and p = 0.02, respectively). At week 4, there was no signi cant difference between groups D and N due to increased cellularity in group D. On the other hand, the cellularity was signi cantly lower in group A than in group N (p = 0.007). At week 8, because the cellularity of both groups A and D increased further, no signi cant difference was shown between the three groups. After week 16, the cellularity of the three groups became comparable and was maintained thereafter. Fig. 3-b depicts cell in ltration into the intra-articular graft at various time points in representative cases from groups D and A. At week 1, there was a small number of in ltrated cells in group D. At week 2, cell in ltration increased in group D. At week 4, cell in ltration further increased in group D. Cell in ltration increased further in both groups by week 8. Furthermore, cells with oval or rod-shaped nuclei were present in the graft at week 4. Following that, the proportion of cells in the graft with oval or rod-shaped nuclei increased even more, and they were aligned longitudinally in parallel with regularly oriented bers. Fig. 3-c shows the box plots showing the count of cells into the intratibial tunnel graft of group D, group A, and group N. At week 1, the count of cells in group D was lower than in group A (p = 0.04), and both groups A and D were lower than in group N (p < 0.0001, p < 0.0001, respectively). At week 2, the cell count of both group D and group A increased, and no signi cant difference was found between the groups D and A. The number of cells in both groups, however, was signi cantly lower than in group N (p = 0.01, p = 0.0008, respectively). After week 4, there was no signi cant difference between the three groups, and this was maintained until week 52. Fig. 3-d depicts cell in ltration into the intratibial tunnel graft at different time points in representative cases of group D and group A. At week 1, there was a small number of in ltrated cells in group D. At week 2, cell in ltration increased in both groups. At week 4, cell in ltration further increased in both groups and the cellularity appeared to be comparable. At week 8, cells with oval or rod-shaped nuclei appeared, which were then aligned longitudinally and oriented in parallel with collagen bers in group D. After week 26, almost all cells were rod-shaped or oval nuclei and aligned longitudinally in parallel with regularly oriented bers in groups D and A.
Count of M1 and M2 macrophages in decellularized bovine and autologous tendons in the intra-articular and intratibial bone tunnel sites.
The count of M1 macrophages of groups D and A in intra-articular grafts at each time point is shown ( Fig. 4-a). The number of M1 macrophages in group D was signi cantly lower than in group A at week 1 (p = 0.03). Week 2 was signi cantly higher than week 1 for both groups (p = 0.04, p = 0.04, respectively). Group A tended to be higher than group D, but the difference was not signi cant (p = 0.08). At weeks 4 and 8, no signi cant difference was present between the groups. At week 16, group D was signi cantly lower than that group A (p = 0.002). At weeks 26 and 52, no signi cant difference was present between the groups. After the mid-term, a non-signi cant downward trend in the count of M1 macrophages was observed in both groups.
The count of M1 macrophage in the intratibial tunnel graft of both groups at each time point is shown ( Fig. 4-b). At week 1, no signi cant difference was found between the groups D and A. In comparison to week 1, group A increased numerically but with no statistical signi cance in week 2, whereas group D showed a signi cant increase (p=0.003). At weeks 2 and 4, there was no signi cant difference between the groups. Group D was signi cantly lower than group A at week 8 (p = 0.046). After week 16, there was no signi cant difference between the groups. Group A decreased numerically with no statistical signi cance at week 52 compared to week 26, while group D decreased with statistical signi cance (p = 0.02).
The count of M2 macrophage of both groups in the intra-articular-graft at each time point is shown ( Fig.   4-c). At weeks 1, 2, there was no signi cant difference between the groups. At week 4, both groups increased signi cantly compared to week 2 (p = 0.01, p = 0.01, respectively). Furthermore, group D was signi cantly higher than group A (p = 0.002). After week 8, there was no signi cant difference between the groups.
At each time point, the count of M2 macrophages of both groups in the intratibial tunnel graft is shown ( Fig. 4-d). At week 1, there was no signi cant difference between the groups. At week 2, both groups increased signi cantly compared to week 1 (p = 0.01, 0.007, respectively), with no signi cant difference between the groups. At week 8, group D tended to be numerically higher than group A with no statistical signi cance (p = 0.08). After week 8, the count of M2 was maintained and comparable between the groups.

Discussion
The most important nding of this study was that decellularized bovine-derived tendons were repopulated with recipient rat cells and the grafts were retained for 52 weeks, as were autologous grafts in the rat ACLR model.
First, the BMD of the peritibial tunnel in groups D and A was comparable at each time point, increased up to 16 weeks, and then stabilized. One of the causes of early graft failure is the failure of osteointegration into the graft 17 . The success of the healing phase of osteointegration into the graft coincided with an improvement in load-to-failure 18 . Therefore, tendon-bone healing is one of the key points in the success of ACLR 19 . Tendon-bone healing starts with the in ltration of loose, unorganized brovascular tissue into the gap between the tendon and bone tunnel, followed by the expression of osteoinductive factors like bone morphogenetic proteins (BMPs) 20,21 . After four weeks, the bone tunnel wall surrounding the tendon graft thickens even more 19,20 . Later, Sharpey-like bers, considered to be the earliest sign of osteointegration into the tendon [22][23][24] appear and connect the bone and tendon. It has been reported that allograft tendons may have inferior remodeling in bone tunnels compared to autograft in ACLR in humans 25 . However, our study revealed that the BMD of decellularized bovine tendon xenograft and rat autograft at peritibial tunnels was comparable over time. The decellularized bovine tendon demonstrated autograft-like bone-tendon healing.
The cellularity of the graft of intra-articular and intratibial regions became comparable in both groups after weeks 4, 8 respectively, equivalent to native ACL. It is important to note that the cellularity of group D did not exceed that of group A at weeks 1 and 2. Because if the decellularization was inadequate, there may be many in ammation-derived cells in ltrating in the xenograft in the early stage, and the cellularity of the decellularized xenograft might exceed that of the autologous graft [26][27][28] . The fact that cellularity in both intra-articular and intratibial tunnel grafts in group A was lower at weeks 1 and 2 than in group N could be attributed to graft necrosis 18,29-32 . Kondo et al. 32 reported that two weeks after ACLR in sheep, the core of an autologous semitendinosus graft was nearly acellular. They concluded that the graft necrosis occurred at week 2, followed by revascularization and cellular in ltration over time, resulting in graft regeneration. The cellularity in group A was lowest at week 1, but it was not acellular. In terms of tissue regeneration, rats are most likely ahead of sheep 33 . As a result, the autologous tendon's cellularity may be lowest before week 1 or between weeks 1 and 2. Furthermore, in the sheep model, the orientation of cells with rod-shaped or oval nuclei began to appear in the grafts at week 12 32 , whereas in the current study, the orientation of cells began to appear at week 8. It is also important to note that the cellularity in group D was maintained after reaching the equivalent level to that in group N. If the number of cells in group D increased far exceeding that in group N, the decellularized graft might become cancerous [34][35][36] . Therefore, the decellularized bovine tendon grafts were considered safely integrated as a new ACL in rats.
M1 macrophages have been reported to be activated by type I cytokines such as interferon-γ (IFN-γ) and tumor necrosis factor-α (TNF-α), and have anti-proliferative, cytotoxic, and pro-in ammatory activities [37][38][39][40][41][42][43] . In contrast, M2 macrophages have been reported to suppress the production of pro-in ammatory cytokines and have angiogenic and tissue repair properties [42][43][44][45] . During the healing process, these macrophages can plastically change their polarity in response to local stimuli 42,46,47 . Brown et al. 48 compared four groups of rats that received four types of tissue transplanted into abdominal wall defects: cellular autograft, acellular allograft, cellular xenograft, and acellular xenograft. They observed that the acellular tissue transplant group showed a constructive M2-type remodeling response, while the cellular tissue transplant group showed an M1-dominant strong in ammatory response, resulting in connective tissue deposits and scarring, and indicated that transplantation of tissue containing cellular components, even if they are autologous, causes an M1-type in ammatory response. Previous studies in tendon injuries reported similar results to the current study, with a large increase in the count of M1 macrophages in the rst two weeks 49,50 , implying that M1 macrophages were involved in acute in ammation 49,51 .
Similarly to Brown's study 48 , M1 macrophage-dominated in ammation was observed in group A, particularly in the early stages, until the cells in the graft disappeared even if they were autologous cells.
Group D had fewer M1 macrophages than group A in early periods. Therefore, group D may have fewer M1 macrophages in intra-articular graft than group A at week 16.
In a rat's Achilles tendon repair model, Sugg et al. 49 reported that the concentration of M2 macrophages became signi cant at 28 days. Similarly, Kawamura et al. 52 found that M2 macrophages reached maximum accumulation by day 28 in a study of bone to autologous tendons healing using the rat ACLR model. In the current study, the number of M2 macrophages was highest at week 4 in both groups for intra-articular grafts and intratibial tunnel grafts, as previously reported. Several studies have reported that increase of M2 macrophage was associated with promotion of tendon healing 53,54 . Barboni et al. 53 reported that transplantation of human amniotic epithelial cells into injured Achilles tendons of sheep increased the expression of M2-related genes and promoted angiogenesis and extracellular matrix (ECM) remodeling. Gelberman et al. found that exor tendon repair in dogs with adipose-derived mesenchymal stromal cells (ASCs) and recombinant bone morphogenetic protein-12 resulted in increased levels of M2 macrophages, decreased levels of in ammation, and increased levels of proteins involved in ECM production 54 . In this study, signi cantly more M2 macrophages appeared in group D in the intra-articular grafts at week 4, and numerically more M2 macrophages appeared in the intratibial tunnel grafts at week 8. Throughout the study, the number of M2 macrophages of group D both in intra-articular and intratibial tunnel grafts was never lower than that of group A. This data indicated that the decellularized tendon might be a better graft than the autologous tendon.
This study has several limitations. First, the biomechanical properties of explanted decellularized tendons and autologous tendons were not tested, because of extensively small size of the rat knee. Indeed, to obtain reliable data in tensile tests, it is necessary to develop chucks for the femur and tibia so that the tissue can be placed horizontally with the tensile direction. We are currently conducting ACLR using a bovine-derived decellularized tendon in sheep. Therefore, biomechanical remodeling properties of the decellularized tendon will be evaluated in the experiments.
Secondly, it is unclear whether the observation period of 52 weeks in rats is su cient to evaluate thorough in vivo remodeling processes. The remodeling period of the ACLR has been reported to be 12-tendon graft 55 . Fifty-two weeks exceeds the remodeling period of the human bone-patellar tendon graft but falls short of the hamstring tendon remodeling period. However, remodeling is much slower in human grafts than in animals 56 . Therefore, 52 weeks in rats is considered su cient to evaluate in vivo remodeling processes.
This study revealed that bovine tendon-derived decellularized grafts sterilized with ethylene oxide gas after freeze dry had the excellent time series biocompatibility in the rat ACLR model. The BMD of the peritibial tunnel increased over time, equivalent to that of the autologous tendon. Cell in ltration occurred over time, equivalent to that of the autologous tendon. Moreover, the number of M2 macrophages, which are responsible for tissue repair, tended to be higher at 4-8 weeks after ACLR in the decellularized xenograft tendon than in the autologous tendon. These ndings encouraged us to further study the potential of the decellularized tendon in a cross-species large-animal model.

Ethics on Animal Experiments
This study was approved by the Animal Research Ethics Committee of Waseda University (approved number: 2015-A021). All animal experiments complied with the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines and were carried out following the National Institutes of Health guide for the care and use of laboratory animals (NIH publications No. 8023, revised 1978).

Preparation of decellularized tendon
Bovine tendons were decellularized using the following procedure 57 . Bovine tendon was incorporated into the pulsatile circulation pf physiological saline solution with 1wt% of sodium deoxycholate (Sigma), under a mean ow and pressure of 5L/min and 80 mmHg, respectively. A microwave was irradiated simultaneously. The temperature of the circulating solution was maintained below 37 degrees Celsius, not to induce thermal denaturation to the tissues. The tissues were then treated with Benzonase Ò nuclease (Merck). The decellularized tissues (Fig. 5-a) were freeze-dried, and then sterilized with ethylene oxide. The tissues were rehydrated with physiological saline solution before use as grafts (Fig. 5-b). The residual amount of DNA in the decellularized tendon were assessed using Quant-iT™ PicoGreen Ò ds-DNA Assay Kit (Invitrogen). The decellularized tissues had a DNA residue of less than 1 ng/mg dry weight, which is signi cantly less than the value of 50 ng/mg recognized as a safety threshold for decellularized tissue DNA residue 58 .

Experimental animal and evaluation period
Male Sprague-Dawley rats (12 weeks old, 350 ± 50 g) were used for ACLR of the right knee. The rats were divided into two groups: group D in which decellularized bovine tendon-derived tissue was used as a graft for ACLR (n = 42), and group A in which the autologous exor digitorum longus tendon of the right hind limb was used as a graft for ACLR (n = 42). Each group was evaluated at weeks 1,2,4,8,16,26, and 52 after ACLR with six animals in each period. The bone mineral density (BMD) of the right per-tibial tunnel and the cellularity of the native ACL of the left knee were assessed in six animals with a 0-day implantation period (group N), for a total of 90 rats. ACLR in rat ACLR in rats was performed based on the methods of previous studies 52,59 . General anesthesia was induced by intraperitoneal injection of 60 mg/kg of sodium pentobarbital (Kyoritsu Seiyaku, Tokyo, Japan). If insu cient, additional doses of 6.5 mg each were administered, not to overdose. As a preventative measure, 4 mg/kg of cefazolin sodium (LTL pharma, Tokyo, Japan) was injected intraperitoneally. In group D, decellularized bovine tendon were trimmed in a diameter of 1.5 mm and a length of 15 mm, with the longitudinal diameter parallel to the ber direction. In group A, a longitudinal incision was made on the medial side of the ipsilateral lower leg, from which the exor digitorum longus tendon was harvested at a size of approximately 1.5 mm in diameter and 15 mm in length. In each group, a 3-0 silk thread (Akiyama medical, Tokyo, Japan) was sutured at both ends of the graft. Following that, the procedure was the same in both groups. The ACL was resected in its natural state. A 1.8-mm diameter Kirschner-wire was used to create the femoral tunnel from the footprint of the native ACL to the lateral femoral epicondyle. The tibial tunnel was created in the same way. The graft was inserted into the tunnels, and silk thread was secured to the periosteum around each tunnel aperture, being careful not to loosen the graft (Fig. 5-c,d). The joint capsule and skin were sutured using 5-0 nylon thread (Alfresa Pharma Corporation, Osaka, Japan). Immediately after surgery, the rats were transferred to a warm cage and allowed to move freely upon awakening from anesthesia. All rats did not develop any complications during the observation period.

Sample harvest
Six rats were euthanized with an overdose of sodium pentobarbital via intraperitoneal injection at the time speci ed above. The meniscus and other ligaments were carefully removed from the femur-grafttibia complex. The samples were xed in 10% formalin solution at 4°C after rinsing the blood thoroughly with saline.
Measurement of BMD of the peritibial tunnel using micro-CT CT images of the femur-graft-tibia complex were taken using a 3-D measuring X-ray CT scanner (TDM 1300-IS, Yamato Scienti c Co., Ltd, Tokyo, Japan) (set voltage: 45 kV, set current: 236,0 µA, lter: 0.1 mm brass plate, acquisition time: 10 min) to evaluate the overtime peritunnel bone reconstruction at the interface between the graft and the tibial tunnel. Under the same conditions, phantoms of known density were imaged, and calibration curves were obtained. Based on the calibration curve, 3D BON (RATOC system engineering Co., Ltd, Tokyo, Japan) was used to analyze the BMD. In a slice along the long axis of the tibial tunnel, the BMD was measured at the top, middle, and bottom of a region 2 mm distal to the tibial epiphyseal plate (Fig. 6).

Histology
After micro-CT imaging, samples were decalci ed with 9% formic acid for 4-6 weeks and then embedded in para n. 5-µm-thick longitudinal para n sections parallel to the direction of the tibial tunnel and intraarticular graft, and which parallel to the femoral tunnel were prepared 59 and stained with hematoxylineosin (HE). Fluorescence microscopy (BZ-8100, KEYENCE, Osaka, Japan) was used to observe the whole image (magni cation ×2) and local view (magni cation ×200). The graft between the apertures of the femoral and tibial tunnels was de ned as the "intra-articular graft." The term "intratibial tunnel graft" refers to a graft that enters the tibial tunnel deeper than the tibial tunnel aperture. To assess the cellularity of the graft, the cells inside the intratibial tunnel graft and the intra-articular graft were counted.
Immunohistochemical study for macrophage in ltration into the graft Immunohistochemical staining was performed to investigate the phenotype 48 of macrophages (M1, M2) that accumulate in the grafts. Rabbit anti-CCR7 (Cell Applications, San Diego, CA) was used for M1 macrophage staining, and mouse anti-rat CD163 (Serotec) was used for M2 macrophage staining. The staining was done with reference to the previous study 48 . The macrophages of each phenotype in ltrating the intratibial tunnel graft and in the intra-articular graft were counted.

Cell count
The entire tissue was observed on a slide using a uorescence microscope (Fig. 7-a). Both in the core of the intra-articular graft approximately equidistant from the tibial and femoral tunnel apertures ( Fig. 7-b) and in the core of the intratibial tunnel graft from the tibial tunnel aperture distally (Fig. 7-c), cells in the grafts were counted. The number of cells were counted using Image-J software 60 in six 200-µm 2 square regions, respectively. Then, the average number in the six squares was used. The number of cells in the native ACL of the left knee was counted (n = 6) using the same protocol. In immunohistochemical staining, the number of M1 and M2 macrophages were counted using the same protocol ( Fig. 7-d,e,f).

Statistical Analysis
Box-and-whisker plots were used to display the distribution of each data. The Shapiro-Wilk W test was used to test the normality of continuous variables. Depending on whether the distribution is normal or not, the student t-test or Mann-Whitney U test was used to compare the mean differences between the two groups. Depending on whether the distribution is normal or not, the Kruskal-Wallis test or one-way analysis of variance was used to compare the means of the three groups. When there was a signi cant difference in the three groups, the student t-test or the Mann-Whitney U test was used to evaluate the difference in the means of each group as a post hoc analysis. Holm's multiple comparison method was used for the p-value correction. P < 0.05 was considered statistically signi cant. JMP Pro version 13.2.1 (SAS, Cary, NC, USA) was used for statistical analyses.

Declarations Data availability
All data generated or analyzed during this study are included in this published article. Figure 1 The gross appearance of right knee in representative cases in each time point of group D and group A.

Figures
The black dashed line shows the outline of the graft. Group D, decellularized bovine tendon graft; group A, autologous tendon graft    Anterior cruciate ligament reconstruction in rats using the decellularized bovine tendon and autologous tendon grafts. a: Hematoxylin-eosin staining of the decellularized bovine tendon. No cell components can be seen and the extracellular matrix and ber arrangement are preserved. b: The gross appearance of the decellularized bovine tendon after rehydration with physiological saline solution. c: Schematic diagram of the completion of anterior cruciate ligament reconstruction in a rat. d: Gross appearance of anterior cruciate ligament reconstruction in right knee of the rat at completion. F, femur; T, tibia; G, graft; S, 3-0 silk thread Figure 6 Measurement of bone mineral density at the interface between the graft and tibial tunnel. The de nition of the measurement area of bone mineral density. The CT image shows slices along the long axis of the tibial tunnel. The BMD was measured at the top, middle, and bottom of a region 2 mm distal to the tibial epiphyseal plate. The major and minor tunnel diameters were measured at three cross-sections, namely, upper (0), middle (1), and lower (2) ends of the region 2 mm distal to the growth cartilage (red square) in the femur-graft-tibia complex. The data of day 0 after anterior cruciate ligament reconstruction were used