Rabbit xenogeneic transplantation model for evaluating human chondrocyte sheets used in articular cartilage repair

Abstract Research on cartilage regeneration has developed novel sources for human chondrocytes and new regenerative therapies, but appropriate animal models for translational research are needed. Although rabbit models are frequently used in such studies, the availability of immunocompromised rabbits is limited. Here, we investigated the usefulness of an immunosuppressed rabbit model to evaluate directly the efficacy of human chondrocyte sheets through xenogeneic transplantation. Human chondrocyte sheets were transplanted into knee osteochondral defects in Japanese white rabbits administered with immunosuppressant tacrolimus at a dosage of 0.8 or 1.6 mg/kg/day for 4 weeks. Histological evaluation at 4 weeks after transplantation in rabbits administered 1.6 mg/kg/day showed successful engraftment of human chondrocytes and cartilage regeneration involving a mixture of hyaline cartilage and fibrocartilage. No human chondrocytes were detected in rabbits administered 0.8 mg/kg/day, although regeneration of hyaline cartilage was confirmed. Histological evaluation at 12 weeks after transplantation (i.e., 8 weeks after termination of immunosuppression) showed strong immune rejection of human chondrocytes, which indicated that, even after engraftment, articular cartilage is not particularly immune privileged in xenogeneic transplantation. Our results suggest that Japanese white rabbits administered tacrolimus at 1.6 mg/kg/day and evaluated at 4 weeks may be useful as a preclinical model for the direct evaluation of human cell‐based therapies.

of the elderly population mean that more people are living longer with disability and a reduced quality of life (Ondrésik et al., 2017).
Using this scaffoldless technology, we have reported the usefulness of layered chondrocyte sheets (Kokubo et al., 2016) in the treatment of partial-thickness defects in rabbits (Kaneshiro et al., 2006) and full-thickness defects in rats (Takaku et al., 2014); rabbits (Ito et al., 2012); and minipigs (Ebihara et al., 2012). In a clinical study from 2011 to 2014, we treated eight patients with autologous chondrocyte sheets and confirmed the safety and effectiveness of this method in treating patients with OA (Sato, Yamato, Hamahashi, Okano, & Mochida, 2014).
However, autologous transplantation requires two surgeries, and the cell source is often limited. Thus, we are currently investigating the possibility of creating allogeneic chondrocyte sheets. For cartilage regeneration, both differentiated (Ham et al., 2015) and undifferentiated mesenchymal stem cells (Richardson et al., 2016) from various sources such as bone marrow, adipose tissue, and synovial tissue have been investigated as potential sources. Hyaline cartilaginous tissue generated from embryonic stem cells (Oldershaw et al., 2010) and, more recently, from induced pluripotent stem cells (Yamashita et al., 2015) shows great potential. To address the feasibility and traceability issues, we have been investigating the use of chondrocytes obtained from young polydactyly patients (Nasu, Takayama, & Umezawa, 2016) in preparation for a first-in-human clinical study for allogeneic chondrocyte sheets that has been approved by the Ministry of Health, Labour and Welfare of Japan.
The main objective of this study was to establish a preclinical animal model that would allow the direct evaluation of human chondrocyte sheets for translational research purposes. Rabbits are phylogenetically and anatomically more closely related to humans than rodents (Schnupf & Sansonetti, 2012), and the size of defects that can be created allows for the transplantation of a single chondrocyte sheet without overpopulating the defect. Human adult chondrocyte sheets were fabricated from surgical remains of total knee arthroplasty (TKA) and transplanted into osteochondral defects in Japanese white (JW) rabbits immunosuppressed by tacrolimus, a potent immunosuppressant used in organ transplantation (Ikebe et al., 1996;Kino et al., 1987). We hypothesized that a xenogeneic transplantation rabbit model would be a cost-effective method for the direct evaluation of the in vivo efficacy of human chondrocyte sheets.

| MATERIALS AND METHODS
The animal experiments were approved by the Institutional Animal Chondrocytes and synovial cells were enzymatically isolated, and chondrocyte sheets made from cells obtained from TKA (TKA sheets) were prepared separately from each donor using the coculture method as previously described (Kokubo et al., 2016). Briefly, cartilage tissue and synovial tissue were separately minced and digested with 5 mg/ml collagenase Type I (Worthington Biochemical Corp., Lakewood, NJ, USA) in Dulbecco's modified Eagle's medium-Nutrient Mixture F-12 (DMEM/F12; Gibco, Waltham, MA, USA) supplemented with 20% fetal bovine serum (FBS; Ausgenex, Molendinar, Australia) and 1% antibiotic-antimycotic solution (AA; Gibco). The cells were filtered through a 100-μm cell strainer (Becton, Dickinson, and Company [BD], Franklin Lakes, NJ, USA) and washed in phosphate-buffered saline (PBS; Gibco). Primary chondrocytes were stored at −80°C in Cellbanker 1 cryopreservation medium (Zenoaq, Fukushima, Japan), and synovial cells were cultured to Passage 1 and then stored similarly.
To fabricate TKA sheets, synovial cells at Passage 1 were first plated as feeder cells in six-well dishes (BD) at a density of 1 × 10 4 cells per cm 2 with 3 ml of medium. After 1 hr, temperature-responsive culture inserts (Cellseed, Tokyo, Japan) were placed in the wells, and primary chondrocytes were seeded without direct contact to synovial cells at a density of 5 × 10 4 cells per cm 2 with 2 ml of medium. The medium was changed every 3 or 4 days thereafter with the addition of 100 μg/ml ascorbic acid (Wako Pure Chemical Industries, Osaka, Japan). At 2 weeks, three sheets of only chondrocytes were layered using a support membrane of polyvinylidene difluoride, and the layered sheets were cultured for another 7 days.
To count the cells, fabricated TKA sheets were digested enzymatically and stained with trypan blue. For histological analysis, TKA sheets were fixed in 4% paraformaldehyde in phosphate buffer and embedded in optimal cutting temperature compound (Sakura Finetek Japan, Tokyo, Japan). Twenty-micrometre-thick sections were stained To measure the concentrations of humoral factors produced by TKA sheets, fabricated sheets were placed in DMEM/F12 supplemented with 1% FBS and 1% AA, and the culture supernatants were collected after 72 hr, centrifuged at 15,885 g for 5 min to remove debris, and stored at −80°C. Commercial enzyme-linked immunosorbent assays were used to quantify the concentrations of transforming growth factor-β1 (R&D Systems) and melanoma-inhibitory activity (MIA; Roche, Basel, Switzerland).

| Measurement of blood tacrolimus concentration in JW rabbits
All JW rabbits were purchased from Tokyo Laboratory Animals Science Co. (Tokyo, Japan). The blood concentration of tacrolimus was monitored in three female JW rabbits (average weight = 3.0 kg) independent from the transplantation experiment. Tacrolimus (Astellas Pharma, Tokyo, Japan) was administered daily for 14 days intramuscularly at a dosage of 1.6 mg/kg/day. Blood samples from the ear were collected into EDTA 2 K tubes (Tokuyama Sekisui Co., Yamaguchi, Japan); frozen at −30°C; and sent to SRL Inc. (Tokyo, Japan) for analysis. Tacrolimus concentration in blood was measured using an electrochemiluminescence immunoassay on a Cobas e 411 immunoassay analyser (Hitachi High Technologies Co., Tokyo, Japan) with a minimum detection level of 0.5 ng/ml.

| Transplantation of TKA sheets
Thirty female JW rabbits (average weight = 3.0 kg) were used in the transplantation experiment. The animals were housed one animal per cage and were given daily standard chow and access to water ad libitum. Before surgery, rabbits were randomly assigned by weight to one of five groups: A (defect only, 4 weeks, 1.6 mg/kg/day of tacrolimus); B (TKA sheet, 4 weeks, 0.8 mg/kg/day); C (TKA sheet, 4 weeks, 1.6 mg/kg/day); D (defect only, 12 weeks, 1.6 mg/kg/day); and E (TKA sheet, 12 weeks, 1.6 mg/kg/day). TKA sheets fabricated from each of the donors were equally allocated to each transplantation group.
Tacrolimus was administered intramuscularly daily for 10 days starting 2 days before transplantation and then every other day until 4 weeks after surgery. The intramuscular injections alternated between the right and left hind legs.
For surgery and transplantation, the rabbits were anaesthetized with 2 L/min nitrous oxide, 1 L/min oxygen, and 2.5-3.0% isoflurane (Pfizer, New York City, NY, USA). A medial parapatellar incision was made to the right knee, and the patella was dislocated to access the patellar groove of the femur. A 5-mm biopsy punch (Kai Industries, Gifu, Japan) was used as a marking guide, and a 5-mm drill was used to create an osteochondral defect (diameter = 5 mm; depth = 3 mm). Slight bleeding from the subchondral bone was confirmed, and physiological saline (Nipro, Osaka, Japan) was used to clean the defect and prevent thermal damage. For transplantation groups B, C, and E, one TKA sheet was transplanted into each defect without suturing. After restoration of the patella, the quadriceps femoris muscle and tendon were sutured to prevent dislocation.

| Monitoring of biochemical markers in blood
Blood monitoring was performed weekly for selected rabbits (n = 3 for each group) in defect group A and transplantation group C from Day 0 (before surgery) to Day 28 (before euthanasia). Blood samples were collected from the ear, placed in EDTA 2 K tubes and BD Vacutainer SST II Advance tubes, and frozen at −30°C. Samples were sent to Fujifilm Monolith Co. (Tokyo, Japan) for analysis. Abnormalities in blood chemistry were monitored, especially to detect changes in kidney and liver function.

| Pain evaluation
The Linton Incapacitance Tester (Linton Instrumentation, Diss, Norfolk, England) was used to evaluate the degree of pain, inflammation, or discomfort, as previously reported (Ito et al., 2012). Measurements were made before surgery, on Days 1,4,7,10,14,17,21,24,and 28 for the first 4 weeks, and on Days 35,42,49,56,70, and 84 for the following 8 weeks. The average damaged limb weight distribution ratio (%) of the hind limbs was calculated from 10 repeated measurements for each animal and averaged for all groups as follows.

| Statistical analysis
Numerical results are expressed as mean and standard deviation unless otherwise noted. ICRS scores are expressed as mean and standard error of the mean. Repeated measures analysis of variance was used to analyse measurements from the monitoring of biochemical makers in blood. Analysis of variance was used to analyse ICRS scores, and Tukey's honest significance test was used for post hoc analysis. The weight distribution ratios were compared with values before surgery using the paired t test.

| Properties of TKA sheets
An average TKA sheet contained 1.6 ± 0.2 × 10 6 cells and had a thickness of 50.0 ± 6.5 μm. The sheets were layered and manipulated using a polyvinylidene difluoride support membrane, which was removed upon transplantation (Figure 1a

| Blood tacrolimus concentration in JW rabbits
The blood tacrolimus concentration (ng/ml) in three JW rabbits admin-

| Xenogeneic transplantation of TKA sheets in immunosuppressed JW rabbits
The surgeries were uneventful, and the TKA sheets fully covered the defect areas. Loss of appetite and diarrhoea were observed after surgery, and a subsequent decrease in body weight was observed; largest decrease in average weight for each group was as follows: 0.16 kg at Day 14 in Group A; 0.32 kg in Group B at Day 21; 0.09 kg in Group C at Day 21; 0.24 kg in Group D at Day 10; and 0.13 kg in Group E at Day 4. Adverse events were detected near the end of the 4 weeks in two rabbits from Group C and in four rabbits from Group B. Selfinflicted wounds to the end of the hind limbs were observed, but no abnormalities to the surgical areas were detected. Adverse events were detected in two rabbits from Group E. Self-inflicted wounds to the end of the hind limbs were observed in one rabbit, and swelling in the surgical knee joint was detected in the other. In all rabbits, muscle stiffness and muscle loss were observed in areas where tacrolimus had been administered.

| Monitoring of biochemical markers in blood
Blood monitoring was performed weekly for selected rabbits (n = 3) in defect group A and transplantation group C (Table 1)  and 14 in both groups, but these decreases may have reflected the muscle damage caused by the surgery and intramuscular administration. No significant differences were detected between the two groups for the measured biochemical markers.

| Pain evaluation
The weight distribution ratio was used as a measure of pain and was followed for 4 weeks ( Figure 3a) and 12 weeks (Figure 3b). This ratio recovered to the value before surgery by Day 21 in transplantation groups B (p = 0.972) and C (p = 0.214) but did not return fully to the   A modified version of the ICRS grading system was used to evaluate cartilage repair (Figure 6). At 4 weeks, the scores were significantly higher in transplantation groups B (30.4 ± 2.8, p = 0.020) and C (31.0 ± 2.2, p = 0.014) than in defect group A (20.1 ± 2.0). At 12 weeks, the scores did not differ significantly (p = 0.07) between transplantation group E (18.2 ± 2.8) and defect group D (25.8 ± 1.6).

| DISCUSSION
Hyaline cartilage regeneration using chondrocyte sheets may provide an effective and long-term treatment for OA. To ensure the safety and efficacy of this treatment using different cell sources, preclinical models are needed for evaluating human chondrocyte sheets directly.
Such models will also be critical for evaluating differences associated with donor age, gender, health status, and other factors yet to be identified. In this study, we have shown the usefulness of a rabbit xenoge- were observed on both hind legs, but the results of the weight distribution ratios were comparable with those reported in our previous study (Ito et al., 2012). viously (Hamahashi et al., 2015). The paracrine effect was also reported to be the major mode of action of cell sheet treatment of ischaemic cardiomyopathy in a porcine xenogeneic transplantation model (Kawamura et al., 2012;Kawamura et al., 2015). The rejection of transplanted cells may occur in parallel with the paracrine effect and may result in regeneration of hyaline cartilage by activated host cells even when no donor cells remain.
We also examined whether this model could be used to evaluate the remodelling of articular cartilage over the long term. We hypothesized that, after successful engraftment and matrix production, immunosuppression may be unnecessary. However, histological evaluation at 12 weeks after transplantation (i.e., 8 weeks after termination of immunosuppression) showed that immune rejection had occurred.
The articular cartilage has long been considered a relatively immune- A key limitation of our study is that tacrolimus has been shown to reduce OA-like responses and to protect cartilage matrix integrity in vitro and in vivo (Siebelt et al., 2014 Another limitation is that tacrolimus administration was accompanied by adverse events such as weight loss and self-inflicted wounds. Self-inflicted wounds and muscle loss increased the variability in the weight distribution ratio. Blood monitoring did not indicate kidney or liver failure, but these adverse events limited tacrolimus administration to 4 weeks in this study and would limit its use in longer studies. Differences in tacrolimus toxicity between rabbit species must also be considered in order to translate our results to other rabbit species. Severe tacrolimus toxicity was reported in the Dutch-Belted rabbit, and a much lower dosage of 0.08 mg/kg/day has been suggested as feasible (Giessler, Gades, Friedrich, & Bishop, 2007).
JW rabbits can tolerate 1.6 mg/kg/day, as first described by Ikebe et al. (1996) in bone xenogeneic transplantation, but the optimal concentrations need to be determined in further studies.   Group D: defect only, 12 weeks, 1.6 mg/kg/day; and (e) Group E: TKA sheet, 12 weeks, 1.6 mg/kg/day. At 4 weeks, ICRS scores were significantly higher for groups B ( * p = 0.020) and C ( * p = 0.014) than for Group A. At 12 weeks, the scores did not differ significantly between Groups D and E (p = 0.07)