Defects in articular cartilage are classified as full or partial-thickness, according to whether or not they penetrate the marrow spaces of subchondral bone7,38,40,58,59. Partial-thickness defects are analogous to the clefts and fissures that are seen in the early stages of osteoarthrosis in humans; these fibrillated lesions grow larger and deeper during the course of the disease but never repair spontaneously. Full-thickness defects heal transiently but imperfectly1,44,59.
It has been suggested that partial-thickness defects do not heal because they are walled off from marrow and have no access to the macrophages, endothelial cells, and mesenchymal cells that reside therein5,12,22,39,59. The observations that are described in this report indicate that this is not the case. Partial-thickness defects do not need access to cells in marrow to undergo repair. Under appropriate conditions, which will be described, mesenchymal cells can be induced to migrate from the synovial membrane across the articular surface into the defect, where they proliferate and fill its cavity.
The rationale behind our experiments can be outlined as follows. Articular cartilage contains dermatan sulfate and other proteoglycans that confer antiadhesive properties on the surface of the defect34,53,55. Decorin and biglycan, for example, are known to inhibit adhesion of cells to macromolecules, such as fibronectin, in the extracellular matrix34,55. Removal of these molecules from the surface of the defect by controlled enzymatic digestion would thus be expected to overcome this problem. However, this treatment alone was insufficient to elicit proliferation of cells throughout the defect. The implication of this finding is that cells cannot grow into empty space but need a matrix or scaffolding on which to grow. The defect was therefore filled with a biodegradable matrix (fibrin), which became sparsely populated with cells.
Cell density was enhanced considerably by incorporating a mitogenic growth factor into the fibrin clot. This procedure promoted the recruitment of mesenchymal cells from the synovial membrane and subsynovial area into the cavity of the defect, where they proliferated and formed a fibrous connective tissue. We describe the histological features of the repair process that was elicited with the use of this treatment protocol.
*One or more of the authors has received or will receive benefits for personal or professional use from a commercial party related directly or indirectly to the subject of this article. Funds were received in total or partial support of the research or clinical study presented in this article. The funding sources were the AO/ASIF Foundation, Bern, Switzerland, and Orthogene, Sausalito, California (E. B. H.), and National Institutes of Health Grant R01 AR 34614 (L. C. R.).
†M. E. Muller Institute for Biomechanics, University of Bern, Murtenstrasse 35, P.O. Box 30, CH-3010 Bern, Switzerland.
‡Orthopaedic Research Laboratories, Montefiore Hospital, 111 East 210th Street, Bronx, New York 10467.
Materials
Fibrinogen was isolated and purified from rabbit plasma by affinity chromatography with use of Gly-Pro-Arg-Pro-Lys peptides covalently bound to a TSK Fractogel (Merck, Darmstadt, Germany) column. Chondroitinase ABC and epidermal growth factor were purchased from Sigma Chemical (St. Louis, Missouri): human recombinant growth hormone, from Novo Nordisk (Copenhagen, Denmark); and thrombin, from Kabi (Stockholm, Sweden). Lyophilized transforming growth factor-ß1, derived from porcine platelets and containing bovine serum albumin (one microgram of transforming growth factor-ß1 per fifty micrograms of bovine serum albumin), was supplied by R and D Systems (Minneapolis, Minnesota). Basic fibroblast growth factor and insulin-like growth factor-1 were purchased from GibcoBRL (Gaithersburg, Maryland), and epidermal growth factor was obtained from Sigma Chemical.
Operative Procedure and Treatment of the Defect
In the adult synovial joint, hyaline articular cartilage forms a thin layer, which is 200 to 300 micrometers deep in the rabbit and 700 to 900 micrometers deep in the minipig. This layer is anchored to the subchondral bone by a zone of calcified cartilage. The creation of a precisely defined defect restricted to the substance of cartilage (that is, a partial-thickness defect) requires custom-built instrumentation (Fig. 1). The defects that were created in the articular cartilage from mature rabbits were one millimeter wide and 0.2 millimeter (patellar groove) to 0.25 millimeter (medial femoral condyle) deep. The defects in the mature minipigs were 0.5 millimeter wide and 0.6 millimeter deep. The length of the defect was controlled by the surgeon and was four to six millimeters in the rabbits and seven to nine millimeters in the minipigs.
Skeletally mature New Zealand White rabbits (at least eight months old)41 were anesthetized with use of an intramuscular injection of ketamine hydrochloride (twenty-five milligrams per kilogram of body weight) and xylazine. Sedation was maintained by intravenous administration of ketamine and nitrous oxide. The knee joint was exposed, and a defect of the defined size and shape was created with use of a specially designed instrument (Rolf Haenggi-Engineering, Grenchen, Switzerland) (Fig. 1).
There were five groups of rabbits. In group I (control; twelve animals), the defect was left untreated. In group II (twelve animals), proteoglycans were removed from the surface of the defect with chondroitinase ABC at a concentration of one unit per milliliter of phosphate-buffered saline solution; this concentration was chosen on the basis of data obtained from preliminary experiments. The defect was blotted dry and then filled with chondroitinase ABC with use of a Hamilton syringe connected to a fine Teflon (polytetrafluoroethylene) tube. The solution was maintained in contact with the defect for five minutes before removal (by drying), and the joint was then thoroughly irrigated with physiological saline solution. Histological and electron microscopic studies were carried out to determine the depth to which proteoglycans had been removed from the surface of the defect. In two of the twelve animals from this group, trypsin (2.5 per cent in Tris-buffered saline solution) was tested as an alternative to chondroitinase ABC.
In group III (twelve animals), one of five growth factors was applied topically to the defect after it had been treated with chondroitinase ABC. Transforming growth factor-ß1 (six nanograms per milliliter of phosphate-buffered saline solution) was used in four defects; insulin-like growth factor-1 (fifty nanograms per milliliter of phosphate-buffered saline solution), in two defects; basic fibroblast growth factor (ten nanograms per milliliter of phosphate-buffered saline solution), in two defects; epidermal growth factor (twenty nanograms per milliliter of phosphate-buffered saline solution), in two defects; and growth hormone (sixty nanograms per milliliter of phosphate-buffered saline solution), in two defects.
The defects in group IV (eleven animals) were filled with a fibrin clot after treatment with chondroitinase ABC. The fibrin matrix was introduced as a solution (one milligram of fibrinogen per milliliter of phosphate-buffered saline solution), with thrombin (100 units per milliliter of phosphate-buffered saline solution, pH 7.4) added shortly before application, with use of a sterile Hamilton syringe connected to a fine Teflon tube. The defects were filled to the point of bulging above the edges, so that after retraction of the clot an approximate leveling was achieved. The clot formed within two to four minutes.
In group V (thirty-four animals), the defect was treated with chondroitinase ABC and filled with a fibrinogen solution containing one of the growth factors. Transforming growth factor-ß1 (six nanograms per milliliter of fibrinogen) was used in twenty-two defects; insulin-like growth factor-1 (fifty nanograms per milliliter of fibrinogen), in six defects; basic fibroblast growth factor (ten nanograms per milliliter of fibrinogen), in two defects; epidermal growth factor (twenty nanograms per milliliter of fibrinogen), in two defects; and growth hormone (sixty nanograms per milliliter of fibrinogen), in two defects. In two of the defects in which transforming growth factor-ß1 was used, a gelatin matrix (Gelfoam; Upjohn, Kalamazoo, Michigan) was substituted for fibrin as a control against possible fibrin-specific effects.
Six adult (two to four-year-old) Yucatan minipigs were treated according to the protocols outlined for groups I, IV, and V, with two animals in each group. In the two minipigs treated according to the protocol for group V, only transforming growth factor-ß1 (at a concentration of six nanograms per milliliter of phosphate-buffered saline solution) was tested.
Rabbits and minipigs were allowed free cage activity during the entire recovery period.
Processing and Analysis of Tissue
At various times after the treatment of the defect, the rabbits were given a lethal dose of Pentothal (thiopental); for the minipigs, a lethal dose of potassium chloride was infused during anesthesia. The hindlimb was amputated immediately, and the knee joint was freed of all soft tissue. The distal part of the femur was removed, and those that were to be studied by light microscopy only were trimmed with a diamond band saw (Exakt Medical Instruments, Oklahoma City, Oklahoma); they were chemically fixed as a whole by complete immersion of the suspended joint16 for one hour in 2.5 per cent glutaraldehyde solution (containing 0.1-molar sodium cacodylate buffer [pH 7.4] and 2.5 per cent cetylpyridinium chloride27). Each femur was then cut, perpendicular to the long axis of the defect, into 1.2-millimeter slices with use of a diamond saw (Leco, Warrendale, Pennsylvania). Dehydration of the tissue was prevented by continuous application of buffered saline solution. The specimens were then fixed for an additional three to four hours. Blocks of tissue were dehydrated in ethanol and embedded in methylmethacrylate. Each block was glued onto a plane and polished Plexiglas object holder and was milled on a Polycut E (Reichert-Jung, Heidelberg, Germany) to a thickness of approximately eighty to 150 micrometers. Surface-staining was effected by using MacNeal tetrachrome, toluidine blue O, and basic fuchsin54.
The femora that were to be analyzed by both light and electron microscopy were fixed in 2 per cent glutaraldehyde solution containing 0.1-molar sodium cacodylate buffer (pH 7.4) and 0.7 per cent ruthenium hexaammine trichloride for four to five hours and were postfixed in 1 per cent osmium tetroxide (containing 0.1-molar sodium cacodylate buffer [pH 7.4] and 0.7 per cent ruthenium hexaammine trichloride) for three to four hours. Thick (approximately 1.5-millimeter) sections were produced with use of a diamond sawing machine (Leco). The sections were dehydrated in a graded series of increasing concentrations of ethanol, beginning at 70 per cent (volume per volume), and embedded in Epon 81225,26. Blocks of tissue, one millimeter by one millimeter, were 2cut from the embedded thick sections; from these blocks, semithin (one-micrometer) and ultrathin (forty to sixty-nanometer) sections were cut with use of an ultramicrotome (Reichert Ultracut E; Leica, Canada, Ontario, Canada). The semithin sections were stained with toluidine blue O, and the ultrathin sections were stained with uranyl acetate and lead citrate25.
Light microscopy was carried out with use of a Vanox AH-2 microscope (Olympus, Zurich, Switzerland), and electron microscopy was performed with use of a Hitachi 7100-B electron microscope (Tokyo, Japan).
Partial-thickness defects in articular cartilage do not heal spontaneously (Figs. 2, A; 3, A; and 3, B)39,43,56,58. The reasons for this phenomenon are not well understood, although a number of explanations have been suggested12,21,38,56,59. The hypothesis that has been expressed most frequently is that, because there are no blood vessels in mature articular cartilage, cells from perivascular mesenchymal pools4,8,39 cannot enter this area. Another proposal has been that articular cartilage does not have access to stem cells in marrow, which have a high potential for inducing repair, because articular cartilage is walled off from the subchondral bone marrow by calcified tissue35,64. Indeed, the lack of a source of cells for repair is usually the reason given for the absence of healing, even though no experimental evidence has supported this contention, to our knowledge.
The purpose of the current study was to examine the factors that prevent healing in partial-thickness defects—that is, to ascertain whether the lack of healing is attributable to the absence of a potential source of cells or whether some such source of cells exists that can be activated. We sought to determine whether cells (of unknown origin), once induced to migrate into the defect, would proliferate and completely fill its volume or whether it would be necessary to furnish a space-filling biodegradable matrix as a scaffolding for ingrowth of cells.
It is important to remember that in this investigation a defect in the cartilage was defined as a structural defect from which tissue was missing or had been removed. Other types of injury or repair models, in which the resident chondrocyte population and collagenous network are left intact and cartilage is simply depleted of matrix proteoglycan molecules without any structural deficiency, were not studied48,65.
Our original rationale for treating the surfaces of cartilage defects with a proteoglycan-degrading enzyme and removing these molecules from the surface was twofold. First, we expected that repair cells might adhere better to the surface of the defect after the collagen network had been exposed by the removal of the proteoglycans. Proteoglycan molecules have been shown to inhibit adhesion of cells in vitro6,34,53,55, whereas collagen fibrils and matrix proteins, such as fibronectin, could promote this process by exposure of their R-G-D sequences15,30,33,51. Moreover, exposure of additional matrix proteins13,33,45,60,67 to cells within the cavity of the defect could facilitate repair reactions. Second, we expected that the structural integration of repair tissue with the surrounding host cartilage matrix (that is, the absence of clefts) would be improved by this treatment step.
Our experiments showed that spontaneous adhesion of cells was sporadic and patchy along the surface of untreated defects (group I) and that this response was notably enhanced after the digestion of proteoglycans (group II) (Figs. 2, A and B, 4, B and B). These findings appear to support our hypothesis that cartilage matrix has an intrinsic antiadhesive property. Apart from the scattered formation of chondrocyte clusters along the edges of the defect (Fig. 4, C and D), which may represent an abortive type of cartilage repair4,46, no other repair response was observed in the host tissue; chondrocytes did not migrate from the clusters into the defect.
Thus, treatment of the surface of the defect with chondroitinase ABC increased the coverage by mesenchymal cells but did not result in their outgrowth to fill the entire volume of the defect with repair tissue. The question was raised whether the local application of a growth factor with chemotactic or mitogenic properties, or both, might induce such a response. A series of growth factors was therefore tested, and the degree of coverage by cells was markedly enhanced, most notably by transforming growth factor-ß1 (Fig. 5). However, although bilayers and multilayers of cells were laid down along the surfaces of defects, their volume was still not filled completely.
Resident chondrocytes along the surface of the defect again showed no signs of migration into the defect; they exhibited only scattered proliferation within cartilage tissue near the surface of the defect.
Topical application of growth factors led to an interesting observation. Inspection of serial sections from defects that had been treated with transforming growth factor-ß1 or basic fibroblast growth factor (with or without a fibrin clot) revealed a continuous layer of mesenchymal cells extending from the synovial membrane (and subsynovial space) across the superficial tangential zone of normal articular cartilage into the defect, clearly indicating that the synovial tissues served as the source of repair cells (Figs. 7 and 9, B and C). This argument is corroborated by the morphological similarity between cells populating the site of the defect (mesenchymal-like and macrophage-like cells) and those in the synovial membrane (types A and B synoviocytes and subsynovial mesenchymal cells)23,28. That this streaming phenomenon was observed after the use of transforming growth factor-ß1 and basic fibroblast growth factor but only rarely after the use of insulin-like growth factor-1, epidermal growth factor, and growth hormone2,19,37,47,69,61 probably stems from the potent chemotactic effects of the former two growth factors2,3,19,50,52,62,63. The responsiveness of these synovial cells to external stimuli (such as growth factors) has been reported in a number of disease states, including rheumatoid arthritis10,18. Their high potential for differentiation also has been demonstrated experimentally14,31,44, and their capacity to differentiate into chondrocytes in vitro has been documented recently29. On the basis of these findings in the literature, our data demonstrating that cells derived from the synovial membrane may be stimulated to migrate into defects in cartilage and to act as a source of cells for repair are not surprising.
As filling of the cavity of the defect with mesenchymal cells could not be induced by the application of growth factors alone, we introduced a fibrin clot to furnish a scaffolding for cellular ingrowth. Fibrin was chosen because of its biodegradability (necessary for final replacement by cartilage repair tissue) and ease of handling. In the presence of fibrin alone, the entire cavity became populated with cells but only sparsely so (Fig. 6); inclusion of any of the five growth factors potentiated this response considerably, such that the matrix became densely populated (Figs. 7, 8, and 9). A stream of migrating cells from the synovial membrane to the defect was again observed frequently after treatment with transforming growth factor-ß1 or basic fibroblast growth factor. At five weeks (and at seven weeks in the minipigs), the fibrin matrix had been completely replaced by a loose connective tissue (Figs. 7, 8, and 9), which consisted predominantly of collagen fibrils, indicating that the mesenchymal cells were metabolically active and possessed the capacity to remodel the extracellular space11,20,36. However, this type of fibrous connective tissue persisted, even at forty-eight weeks after treatment of the defect, with no sign of cartilage formation. These findings point to the importance of a matrix or scaffold within which mesenchymal cells can migrate and to which they can adhere. The surface of the defect was still treated with chondroitinase ABC before deposition of the fibrin matrix, as this step is important for its attachment and for adequate integration between repair and host tissue (unpublished data).
Cartilage did not form in any of the experiments in which a growth factor was included in the fibrin matrix. It was somewhat surprising that this did not occur after treatment with transforming growth factor-ß1, as this growth factor has been described as being cartilage-inducing9,42,57. One of the reasons for this may be that the concentration applied (six nanograms per milliliter) was too low, but this seems unlikely, as doses greater than five to ten nanograms per milliliter have been clearly shown to inhibit chemotactic or mitogenic effects, or both52,63,66, effects that are of paramount importance to achieve filling of a defect with cells in this healing process. Moreover, in a preliminary experiment, we found that the intra-articular application of this factor at greater concentrations (more than 100 nanograms per milliliter) led to effusion and swelling of the joint. This problem has been identified by other authors17. Another factor that may play a critical role is the biomechanical environment, as certain kinds of controlled motion of the joint during the healing phase of full-thickness defects have been shown to improve the outcome32. Our experimental animals did, however, have free cage activity.
Although this treatment protocol did not result in the production of cartilage, our systematic investigation revealed that the failure of partial-thickness defects to heal is not entirely due to the lack of access to mesenchymal cells in perivascular tissue or marrow. The identification of synovial cells as a potential source of cells to induce healing of articular cartilage is a very promising finding. It is hoped that further research in this area will lead to the identification of the intrinsic limitations of these cells to differentiate and, thus, yield a rational and biological basis for developing a treatment protocol for complete healing of superficial defects.
NOTE: The authors are grateful to H.-U. Staubli, M. Schafer, L. Gorgievski, and K. Jurgensen for their assistance during the operations. They are also indebted to D. Mettler and his ESI-operation team of the Inselspital, Bern.