(Ultra) Structural Details of Cells and Matrix during Early Cartilage Regeneration after Chondrocyte Implantation in the Equine Knee Joint

Objective: This pilot study of cartilage defect treatment was designed to establish the horse as a suitable animal model for MACT (Matrix-Associated Autologous Chrondrocyte Transplantation) transplants and to analyse the morphological aspects of repair tissue development and differentiation. Design: Hyaluronan-based and collagen-based biodegradable scaffolds were seeded with autologous chondrocytes and implanted into large (1.5×2 cm) defects in the trochlear ridge of the distal femur of three horses. A non-treated, empty defect was used as control. Three months after surgery, samples of the defect area were investigated using bright field and polarized light microscopy, immunohistochemistry and electron microscopy. Results: In MACT-treated lesions with good defect filling, the repair tissue integrated well into the defect and showed features of differentiation in transition to native cartilage; the matrix was partially masked by proteoglycans, strongly stained for collagen type II, and the fibres had the typical vertical arrangement of articular cartilage. In areas with less intense staining, collagen type II formed a network around the cells. Calcified cartilage was partially decalcified and osteoclasts as well as osteoblasts reorganised the subchondral bone. The control defect was almost empty. Conclusion: This study showed that the horse is a suitable animal model for MACT. Differentiation starts early after transplantation in the periphery of the defect. Collagen type II precedes proteoglycan deposition and undergoes a kind of vertical self-arrangement which is hypothesized to develop from a fibre network around the chondrocytes. Journal of Tissue Science & Engineering J o u r n a l o f T iss ue S cience &ngine e r i n g


Introduction
In the treatment of traumatic articular cartilage defects, clinical science presently focuses on long-lasting tissue engineering approaches such as "Matrix-associated Autologous Chondrocyte Transplantation" (MACT). During this procedure the patient's own (autologous) chondrocytes are expanded in monolayer cultures, seeded on a 3D carrier (scaffold) and reinserted into a defect [1].
Despite generally favourable clinical outcomes, a certain amount of transplant failures have been observed, the mechanisms of which are not fully understood. Therefore, in vivo investigations in an appropriate animal model have been suggested to help clarify open questions in the postoperative development of the transplants and the Repair Tissue (RT) [2].
The horse is generally considered to be a good animal model for cartilage defect treatment because joint dimension as well as cartilage morphology and physiology are similar to that of humans and because horses suffer from the same degenerative and traumatic injuries [3][4][5]. A detailed characterisation of equine chondrocytes on two structurally very different MACT scaffolds has recently been published [6] ( Figure   1). The study showed that the cells reacted well and in a similar way to human chondrocytes on these materials, and that the distribution of the chondrocytes strongly depended on the material density. One aim of the present study was to demonstrate that soft MACT scaffolds can indeed be applied to horses in vivo using the same surgical procedure as in humans, without postoperative complications (e.g. reaction to the scaffold) and that the transplants are able to give rise to RT. Furthermore, we used the animal model to analyse the morphology of the early (three month) transplant-derived RT development as has previously been done with other cartilage regeneration methods in the horse [7][8][9]. In particular, we focused on matrix architecture. This is of special interest because matrix architecture and development have been studied mainly in Native Cartilage (NC) and during ontogenesis, but little is known about the process in the context of regenerating articular cartilage tissue [10,11].

Animal model
Four healthy yearling mares that did not meet breeding criteria were obtained from the Haflinger Stud Meura, Thüringen, Germany. During the entire study the animals were kept and treated at the horse clinic of the Research Centre of Medical Technology and Biotechnology (fzmb GmbH) in Bad Langensalza, Thüringen, Germany, under the surveillance of veterinarians. The study was verified and approved by the local ethical board (Thüringer Landesamt für Lebensmittelsicherheit und Verbraucherschutz Nr. 14-03/03).

Transplant generation
Chondrocytes were harvested from cartilage chips (1-2 g) taken from the talocrural joints of all horses except for the control animal. All the following steps were performed for each biopsy/horse separately as described in detail in Nürnberger et al. [6]. Chondrocytes were enzymatically isolated, seeded in culture flasks and passaged when confluent. In passage two, chondrocytes were transferred to two structurally very different biodegradable carrier materials:1) the loose hyaluronan web (HYAFF ® ; Fidia, FAB, Italy) and 2) the dense collagen type I/III fleece (BioGide ® , Geistlich Pharma, Switzerland) at a concentration of 10 6 cells/cm 2 and cultivated for a further three weeks.

Surgery
Three horses were treated with transplants eight weeks after the biopsy. Anterolateral arthrotomy of the left knee joint was performed with the horses in dorsal recumbency. A 1.5×2 cm defect was set in the trochlear ridge of the distal femur without damaging the Subchondral Plate (SP; consists of calcified cartilage plus subchondral bone). Two horses were treated with one transplant each, either the hyaluronan web or the collagen fleece. The third animal was treated with both transplants (one of each transplant type) in a double sandwich-like layer with the hyaluronan web facing the SP and the collagen fleece on top. Fibrin glue (Beriplast ® P, Combi-Set 3 ml, Fa. Aventis Behring, Marburg) was applied at the interface of transplants to the host cartilage. The stability of the graft in the defect was tested by passive movement of the leg over the entire range of motion. In the control animal, the same defect was created but not treated with transplants. All surgeries were carried out without complications. Horses were kept for two weeks in boxes, then exercised (walked) under controlled conditions for 30 days and thereafter kept in paddocks. Postoperative follow-up care was also without complications.

Sampling
Three months after implantation, the horses were euthanized and the defect area prepared for taking overview images and further processing. Tissue sections were taken from the interface regions and the centre of the defect. Sections were fixed with 2.5% glutaraldehyde in 0.05 M sodium cacodylate (pH 7.3), rinsed, photographed and postfixed with 0.5% osmium tetroxide containing 1% potassium ferrocyanide. Then they were dehydrated in a gradient series of alcohol and embedded in Agar-100 resin via propylene oxide. Semithin sections (1 µm) and ultrathin sections (70 nm) were sequentially cut. The former were either stained with toluidine blue or deplastified and used for immunohistochemical reaction against a monoclonal collagen type II antibody (Neomarkers, CA, USA) and visualized with diaminobenzidine tetrahydrochloride and haematoxylin counterstaining. Ultrathin sections were stained with uranyl acetate and lead citrate. Alternatively, some tissue sections were fixed with 7.5% formaldehyde, dehydrated, blocked in paraffin and sectioned for immunohistochemical reaction as described above. Light microscopic sections were examined under a Nikon Eclipse E800 and Olympus BX41 with polarization filter for collagen fibre orientation. Ultrastructural investigations were done at a Zeiss 902 (Oberkochen, DE) and FEI Morgagni electron microscope (Eindhofen, NL).

Gross anatomy
In the macroscopic view, the surfaces of the three transplanted defects were totally covered by RT (Figure 2A and F). The control defect contained little tissue which only sparsely covered the CC at some sites ( Figure 2K). In two defects treated with a single transplant layer, RT reached the level of NC and strongly resembled the whitish NC in appearance (Figure 2A-2D). In contrast, the defect treated with a double layer transplant was subtotally filled ( Figure 2F-2I) suggesting that the upper layer was pulled out of the defect in the course of the postoperative period. In this case, the RT in the defect was somewhat translucent and appeared slightly reddish in top view because of the subjacent bone ( Figure 2F). In all samples the centre of the defect was less dense and less highly filled with RT (about ½ to ¾ of the thickness) than the periphery of the defect. In the control defect, fringy patches of translucent fibrous tissue were loosely attached to the SP ( Figure 2L).

Distribution and characteristics of the hyaline-like repair tissue (RT)
Well-filled defects, treated with one layer of transplant (cell-seeded hyaluronan web or collagen type I/III fleece) showed hyaline-like characteristics in the periphery of the defect. In sections, the tissue stained strongly with toluidine blue and for collagen type II, was homogeneous and contained spherical chondrocytes ( Figure 2C-2E). Locally this tissue formed up to two-thirds of the RT-height ( Figure 2C and 2D). The hyaline-like matrix was dense, was almost as homogeneous as in NC and merged with it completely, without any obvious edge at the interface ( Figure 3A). The fibres, however, were not entirely masked by proteoglycans, making their alignment discernible. The fibres had  the cartilage-specific vertical orientation stretching from the calcified plate towards the cartilage surface, but were slightly wavy ( Figure 3A and B). Sometimes fibres crossed each other, transverse to the general orientation and appeared to form a network around the chondrocytes ( Figure 3C and D). An especially fine network was also present around chondrocytes in matrix with weak collagen type II staining and no vertical fibre strands ( Figure 3E). At the upper end of the collagen type II positive area, the stained fibres extended into the dedifferentiated region of disorderly arranged, unstained fibres, indicating collagen type I ( Figure 3F and G). The control, in comparison, shows the different appearance of collagen type II fibres from fringy cartilage edges stretching into fibrous RT. In these locations, bunches of parallel fibres extended locally from the very dense and homogeneous NC edge ( Figure 3H).
Both in the hyaline and the fibrous RT, occasionally dense and thick fibre strands were embedded in the Extracellular Matrix (ECM) ( Figure  4A and B). The morphology and dimension of this material resembled the fibre strands of the collagen type I/III fleece as observed in in vitro investigations ( Figure 4C). In the horse study, those structures were found only in the defect treated with this specific transplant type, indicating that they were transplant residuals.
Cell density in the hyaline-like RT was generally higher than in the NC but all cells were clearly spaced apart from each other, with a separation of about one to several cell diameters ( Figure 2C, 3A and Figure 5A). Only in double or multichondron-like chondrocyte groups were cells even closer, but a rim of matrix still prevented intercellular contact ( Figure 5G). Most chondrocytes embedded in the hyaline-like matrix had the typical characteristics of differentiated chondrocytes, albeit not as well expressed as in NC. Cell shape was spherical ( Figure  5F-H) or somewhat wavy and dented ( Figure 5E). The nuclei were typically located peripherally in the cells and were spherical or ovoid. Some of the chondrocytes contained glycogen, lipid droplets and small bundles of intermediary filaments around the nucleus ( Figure 4A and 5G). A pericellular matrix had sometimes developed ( Figure 5G).

Fibrous Repair Tissue (RT)
With increasing distance to the hyaline-like RT, both matrix as well as cellular characteristics of differentiation became less expressed towards the centre and the surface of the defect. The matrix stained progressively less intensively with toluidine blue, which corresponds to a decrease in proteoglycans ( Figure 2C and 5A). Due to the reduced masking, the matrix appeared more fibrous. The collagen fibres lost their vertical orientation and formed disorderly arranged collagen bundles that did not stain for collagen type II ( Figure 2D, 3G, 5A, 6A). With the decrease in the hyaline character of the matrix, the cells formed more cell extensions ( Figure 5D). In completely fibrous areas, cells were polygonal with long slender cell projections, which were frequently      oriented alongside the collagen bundles ( Figure 5C). Adapted to the cell shape, the nuclei had an elongated, curved or irregular outline. The intracellular features such as intermediary filament-bundles around the nuclei and lipid droplets became rare. Instead, the endoplasmic reticulum was especially expressed and sometimes dilated ( Figure 6B); the number of mitochondria increased and abundant small vesicles (caveolae) were formed at the cell membrane ( Figure 6C). Strands of microfilaments (actin) stretched especially into the cell projections, anchored locally to the cell membrane and formed focal junctions to the ECM ( Figure 6C). Sometimes individual collagen fibrils penetrated the cell in deep invaginations of the membrane ( Figure 6C). The matrix surrounding the cells was generally fibrous but sometimes a very thin layer of granular material was present ( Figure 6B).
In deeper regions of fibrous RT, vascular-like structures were present, especially in the centre of the defect (Figure 6A and D). In the area around those structures, large cells containing numerous large vesicles with dense granular content were identified as mast cells.
The same morphological characteristics were found in the fibrous RT of the double layer transplanted defect ( Figure 2H-J) and also in the tissue of the control ( Figure 2M-O). In the latter, though, the ECM was less dense and more irregular; the cells were more ramified with many cell processes, exhibited an especially dark cytoplasm, and often showed a partially extended ER with bright lumen ( Figure 2O).

Surface of the repair tissue
In all defects (treated with one or two transplant layers), the surface of the RT was generally straight, smooth and well delimited from the synovial space by ECM. Although the RT was fibrous in the upper region of the defect, the outermost layer (about 20 µm) showed local similarities to the NC surface. Elliptic cells were surrounded by a pericellular matrix and embedded in a partially masked collagen matrix ( Figure 7A). Both cells and collagen fibres were oriented parallel to the surface. A fine layer of electron-dense, granular matrix of about 80 nm covered the superficial margin over large distances and seemed to lubricate the surface ( Figure 7A and B). At other sites, the cell morphology was more polygonal or fibroblast-like and embedded between more or less masked collagen strands ( Figure 5A and 7B). The control resembled the latter situation but the matrix and the cells were more irregular, bore numerous large and small cell processes and sometimes contained an extended endoplasmic reticulum ( Figure 7C).

Calcified Cartilage (CC) and Subchondral Bone (SB)
During defect preparation, Calcified Cartilage (CC) was not removed but only superficially scratched. Consequently, the tide mark forming the upper margin of the CC should have been partially interrupted. Three months after transplantation, the CC and parts of the tide mark were still present in most sites but showed a different degree of degradation.
Completely unchanged CC was found at sites where the tide mark was intact and some NC remained ( Figure 7D). In regions where the tide mark was missing, the CC had an irregular upper outline and a slightly inhomogeneous consistency. This probably resulted from endogenous decalcification and degradation of proteoglycans during the three-month regeneration of the defect ( Figure 7D and E). Especially in the centre of the defect, the trabeculae of the subchondral bone also showed signs of degradation and were very thin or missing. At those sites the adjacent tissue (RT or endogenous decalcified cartilage) directly flanked the bone marrow cavities ( Figure 7E). In some cases the RT close to the fenestrations was strongly positive for collagen type II ( Figure 7F). The situations described above did not form a continuous transition from the periphery to the centre, but were recorded locally along the SP. Nevertheless, the CC and the RT had an intense contact with each other and a continuous transition at all sites.
In the Subchondral Bone (SB), osteoclasts advanced the medullar space towards the CC ( Figure 7G and H). We also recorded osteoblasts, which deposited new osteoid ( Figure 7I and J). Generally, the bone level was very irregular and either above or below the original level ( Figure  2C). Figure 8: Hypothesis on the development and maturation of the collagen type II (coll II) arrangement in the RT. A: Initially, Chondrocytes (chcy) secrete collagen type II into their direct surroundings. B: The fibrils continuously form a fibre network (illustration corresponds to Figure 3E). C: With increasing collagen type II secretion, vertical fibre strands arrange and make up the bulk of the ECM fibres (illustration corresponds to Figure 3A). The previous network (of stage B) remains at least in parts but becomes quantitatively insignificant.

J Tissue Sci Eng
Cartilage and Bone Tissue Engineering ISSN:2157-7552 JTSE an open access journal The present study was designed to verify the horse as a suitable animal model for MACT transplants using soft scaffold materials as is done in humans. During this pilot study we asked the additional question of how tissue develops from the very different transplant materials at the early postoperative time point of three months.

Horse model
Despite the similarity of joint dimension as well as cartilage thickness, morphology, physiology and pathology in humans and the horse there are, of course, differences in the anatomy and the demands put on the knee of the two species [4][5][6][7][8][9][10][11][12][13]. Although the horse has a higher body weight, the load on the cartilage is spread over 4 legs (60% on the forelegs) rather than two and loading on the cartilage appears comparable between the two species. However, it was necessary to test whether the cells and the soft carrier materials can endure the loading conditions in the equine joint [14]. In addition, in previous studies, human as well as equine chondrocytes showed very different cell distribution on the scaffolds depending on the material density [6]. Therefore, we also aimed to test if the transplants are able to give rise to a defect-filling RT. Three months after implantation, all transplanttreated defects were filled with RT while the control defect was almost empty. This indicates that the transplants remained in the defect and a tissue was able to develop from both the loose and dense transplant type. Nonetheless, the defect filled with a double-layer transplant was less filled with RT than those treated with a single-layer transplant only. This suggests that the upper transplant was pulled out of the defect during the postoperative period. Only the lower transplant, the soft hyaluronan web, remained compressed by the upper one and no longer completely filled the defect after the upper layer was lost. These results show that the soft transplants withstand the mechanical loading conditions in the equine joint, but still are at risk of being displaced if they are subjected to shear force on the joint surface. The surgery and the postoperative follow-up care were uncomplicated, showing that the surgical procedure can also be considered suitable for horses from the technical and ethical points of view.

Peripheral differentiation of the repair tissue (RT)
In the horse cartilage, differentiation of the RT (collagen type II and homogeneous toluidine blue O staining) clearly appeared in the periphery adjacent to the NC and in the depth of the defect. This is a common feature in cartilage regeneration and according to the present observations, it can be related to either mechanical reasons or the influence of NC; at the interface to the NC, mechanical loading is probably more intense than in the middle, where the RT is somewhat collapsed (only ½ to ¾ the height of the RT in the periphery) [15][16][17]. Dynamic mechanical load is one of the key stimuli of differentiation, in vitro as well as in vivo, and influences matrix composition, collagen structure and dimension [18][19][20][21][22].
Apart from the mechanical loading conditions, the proximity of the differentiated RT to the NC makes it probable that the latter influenced the development of tissue in the defect. As known from experimental studies, differentiated chondrocytes may have a "teaching effect" on other cells such as dedifferentiated chondrocytes or stem cells (MSCs, ASCs); the trigger proceeds via cytokines or direct cell contact [23][24][25][26]. It is possible that the chondrocytes in the bordering NC released cytokines which triggered the implanted chondrocytes in the defect to differentiate.

Organisation of the hyaline-like matrix
As mentioned above, the matrix of the peripheral RT stained strongly for collagen type II, which is the most common criterion for cartilage differentiation [17,27]. Another important but frequently disregarded aspect of the functionality of articular cartilage is the matrix architecture. Typical for articular cartilage is collagen arranged in vertical arcades that consist of leaf-like or tube-like substructures [28][29][30]. Vertical arrangement of the collagen type II fibres was also observed in the differentiated part of the RT. The arrangement apparently developed from a collagen type II-network, which was formed directly around the cells ( Figure 8A). This network becomes tangential and finally vertical with increasing collagen type II fibre deposition. It seems likely that the collagen type II-network around the chondrocytes remains, at least temporarily, but becomes overlaid by an increasing amount of vertically arranged fibres ( Figure 8B and C). Tangential fibrils crossing the main fibre direction are also known from NC and were previously identified using scanning and transmission electron microscopy [31]. However, a development of the collagen architecture from network-like to vertical has, to our knowledge, never been described before, neither in regeneration nor developmental studies. Rather, ontogenetic cartilage maturation is described as replacement of a horizontally oriented collagen (parallel to the joint surface) by a vertical one [32,33]. Why such a process as proposed in Figure 8 has not been described in earlier studies could be due to the following reasons: 1) The network is too fine to be determined with normal paraffin histology (here we used deplastified 1 µ-resin sections) and too large to be assessed in (70 nm-thick) ultrathin sections using transmission electron microscopy, and/or 2) a fine fibre network is difficult to distinguish inside pre-existing tissue such as that present during the matrix maturation of postembryonic development. The present study provides first insights into a possible matrix organisation in developing RT. Nonetheless, further studies are required to collect more information about this process. One interesting aspect is a possible difference in matrix deposition and maturation depending on whether collagen type II replaces a pre-existing matrix or is deposited in an empty space.
Despite the large correspondence of the differentiated RT with mature cartilage, the organisation of RT still differs from the mature tissue in terms of high cellularity, lower matrix density, lower proteoglycan content as well as the lack or rare appearance of a pericellular and territorial matrix. All those features are, however, typical for immature articular cartilage and support the suggestion that this tissue is in the process of maturation [34][35][36][37]. This further means that collagen type II is the earliest morphologically distinctive sign of RT-differentiation, preceding proteoglycan deposition and matrix compartmentalisation.

Calcified cartilage (CC) and subchondral bone (SB)
The role of CC in defect healing in full-thickness defects in in vivo studies is controversial. In the horse, CC was reported to hamper adhesion of RT after microfracture [38]. In dogs, either good or variable adhesion to CC was found after different implantation methods (cells, scaffolds, periost) without SB perforation [17,39]. In the present study, the CC was not removed but superficially scratched during surgery. Three months after implantation, the CC was locally decalcified and the transition of the RT to the CC continuous. Likewise, a stress-line was prevented and the interface between the CC and the RT was strengthened. The present observations reveal that partial endogenous degradation of CC may contribute to a good bonding of RT to the SP.
Locally, the SB displayed fenestration in the third postoperative month, connecting the defect area to bone marrow. Since no bleeding was observed during surgery, micro-perforation during defect preparation is unlikely. Instead, the perforations apparently develop postoperatively through intrinsic mechanisms such as the decalcification of the CC and the resorption of the SB by osteoclasts found in the trabecular system. In previous studies, postoperative bone remodelling has been attributed to endogenous cells reacting to the degradation products of the scaffold or to implanted cells [16,39]. In a further scaffold-implantation study, untreated control defects also revealed subchondral remodelling, indicating that remodelling was not related to the implanted material [40]. The detailed histological and ultrastructural findings of this study also showed no indication that SB-remodelling is related to the implanted material (e.g. foreign body reaction or inflammation). Rather, the SB remodelling may well be related to a reduced loading in those particular regions. Previous immobilisation studies support this suggestion because they also reported an association between SB-remodelling and mechanical demand [41,42]. Osteoblasts found in some areas at the base of the SP suggest counteracting bone reconstruction, which probably explains the alteration of bone surface levels found in this and other studies [7,43].

Clinical relevance & limitations
The detailed morphological analyses presented in this study could give insights into the clinical relevant topics of cartilage repair, tissue differentiation and integration. Complete defect filling seems to support repair tissue differentiation indicating that transplants should be at equal level as the defect and stiff enough to keep the shape under the demand of movement. In this study defects differentiation started at the periphery and was probably positively influenced by vital chondrocytes in the adjacent native cartilage. This implicates that defect preparation for removal of degraded cartilage is important, but has to be done with care to prevent further cell damage. The study further contributes to the controversial discussion on the function of calcified cartilage in repair tissue integration demonstrating that good bonding is possible even with calcified cartilage. Furthermore, morphological details of matrix characteristic allowed describing a sequence of matrix deposition which may possibly be specific to the regeneration process. This raised the question whether the sequence of matrix maturation is essential for cartilage tissue differentiation, and whether this process could be supported by adequate biomaterial architecture.
The study is limited by the low sample size in this pilot study. Therefore no statistically relevant difference could be observed between transplant types. The focus of this study was the in vivo development of the transplants in the course of the first months after implantation. However data on the long term outcome may contribute further relevant information in the future.

Conclusion
Morphological investigations revealed that differentiation starts at the periphery of the defects and that collagen type II is the first morphologically detectable ECM component. We propose a sequence of collagen type II fibre deposition from network-like to vertical. Remodelling of CC was found to lead to good bonding of RT to the SP. The complication-free surgery and post-operative follow-up care, good defect filling and initial differentiation after three months confirm that the horse is an appropriate animal model. Long-term experiments with a higher number of defects treated with MACT soft scaffolds would be helpful to obtain further information about defect development and differences between scaffold types.