Articular cartilage is a highly specialized and uniquely structured biomaterial that forms the smooth, gliding surface of the diarthrodial joints. It consists of an extracellular matrix that is synthesized by the sparsely distributed resident cells, i.e., the chondrocytes. Osteoarthritic (OA) cartilage degeneration is largely a process of destruction and failure of this extracellular matrix which serves as the functional component of connective tissues. However, matrix turnover, including anabolic regeneration, is solely dependent on the chondrocytes, which, in contrast to the “dead” matrix, are the active players within the tissue. Thus, anabolic activity, phenotypic alterations, and, finally, survival of the chondrocytes are essential for the maintenance of proper articular cartilage. Therefore, failure of the cartilage matrix always implicates a failure of the involved cells.
Adult human articular cartilage is avascular. Thus, there is no external cell supply to compensate for cell loss caused by necrosis, apoptosis, or other cellular mechanisms. Also, adult articular cartilage does not have any sort of germinal cell layer, as occurs in epithelia. Therefore, adult articular cartilage is thought to be a postmitotic tissue with virtually no cell turnover, with resident cells encased within the impermeable extracellular cartilage matrix.
OA, the most common disabling human condition in the Western world, is not a single disease entity, but rather, represents a disease group with different underlying pathophysiologic mechanisms. The typical feature of lacunar emptying in OA cartilage (1–3), as well as the presence of cellular debris–like material apparent at the ultrastructural level in OA cartilage, have led to the long-standing assumption that cell death is a central feature in the degeneration of OA cartilage (2, 4, 5). More recently, apoptotic cell death has become a focus of interest and has been suggested to be an important event in OA cartilage degeneration (6–8).
Modes of cell death
Strictly speaking, necrosis encompasses all forms of programmed and nonprogrammed cell death, but commonly, the term necrosis is used for nonprogrammed cell death (formerly referred to as oncosis), whereas the term apoptosis is used for “programmed” cell death.
Apoptosis was originally described as a specific form of cell death, involving a strict schedule of cellular events, some of which are morphologically identifiable (9, 10). This phenomenon plays an important role in physiologic cell removal, particularly during fetal development. This is also true for the removal of hypertrophic chondrocytes in the fetal growth plate cartilage, as will be discussed later. Apoptosis represents a complex sequential process (for review, see ref. 11) involving an effector, a degradation, and a clearing phase.
The effector phase is characterized by an increase in intracellular free Ca++, activation of endonucleases, tissue transglutaminase, and cellular proteases (especially the caspases), as well as mitochondrial dysregulation. In addition, the expression of several genes, most importantly, c-myc and members of the bcl-2 family, as well as p53, is initiated or activated. Collectively, these factors lead to a “point of no return” in which the apoptotic destiny of the cells is fixed. In the degradation phase, extensive degradation of the nucleic acids and proteins occurs. Endonucleases and tissue transglutaminase are further activated, and the cytoskeleton is reorganized. The clearing phase is characterized by the clearance of the apoptotic cells by macrophages, a feature not relevant in adult articular cartilage because of its lack of macrophages.
One leading criterion of apoptotic cell death is intranucleosomal fragmentation of DNA by Ca++/Mg++-dependent endonucleases, which appears as a DNA ladder in gel electrophoresis. Another characteristic feature is the redistribution of phosphatidylserine residues from the inner to the outer plasma membrane. However, none of the constitutive elements of apoptotic cell death are specific, nor are they found in every apoptotic process (12, 13) and certainly not at all stages of the process.
Nonapoptotic programmed cell death due to other cell death mechanisms, with somewhat different cellular features, might also occur. Recently, a process named paraptosis, which might better describe the events seen in OA cartilage, was suggested (14, 15). Paraptosis is characterized by cytoplasmic vacuolization without nuclear fragmentation. Similar to apoptosis, it seems to be an active process requiring transcription and de novo protein synthesis. Paraptosis can be initiated by insulin-like growth factor type 1 (IGF-1) receptor. Intriguingly, both cytoplasmic vacuolization and the expression of the IGF-1 receptor are known to occur in adult (OA) articular chondrocytes (16, 17). The basic features of necrosis, apoptosis, and paraptosis are summarized in Table 1.
Table 1. Major cellular features of apoptosis, necrosis, and the recently described paraptosis*
Apoptotic cell death can, in many cases, be identified at the histomorphologic level by the appearance of cells with the classic fragmented nuclei, the so-called “apoptotic bodies.” However, this approach is laborious and only reflects the late phases of this multistep event. Thus, several different techniques have been developed in order to identify apoptotic cell death at earlier stages. For example, in situ end-labeling (ISEL) or TUNEL technologies allow detection of DNA strand breaks within tissues. Also, DNA-laddering of genomic DNA isolated from the cells is a rather reliable criterion. Electron microscopy provides a more detailed ultrastructural characterization of apoptotic cells.
For in vitro analysis, there are numerous ways to quantitate apoptosis. However, because of the differences in the various methods used, it is possible that this contributes to the variation in detection of apoptosis among the disparate studies (Table 2). For example, the trypan blue exclusion assay is a cell viability assay based on the ability of live cells to exclude the dye; because cells in the early stages of apoptosis retain their membrane integrity for several hours, it might underestimate the extent of apoptotic cell death. The MTT assay is a colorimetric assay that is based on the ability of live cells to utilize a tetrazolium salt and measures the activities of various dehydrogenase enzymes indispensable for cell survival. Both the trypan blue exclusion assay and the MTT assay lack the specificity to discriminate between apoptosis and necrosis.
Table 2. Methods for detection of apoptosis in cultured cells*
IS = in situ histologic study; ELISA = enzyme-linked immunosorbent assay.
Trypan blue exclusion; Hoechst 33342; propidium iodide
Exclusion of vital dye
May underestimate the extent of apoptotic cell death due to maintenance of membrane integrity after cell death; not specific for apoptosis
MTT assay; alamar blue assay
Metabolic activity of live cell
Rapid and easy; not specific for apoptosis
Cytospin preparation of cells; electron microscopy exam (IS)
Morphologic assessment of cell death
The most specific measure of apoptosis; limited field of examination; subject to interpretation error
Relatively specific for apoptosis; some cell types may not exhibit DNA fragmentation during cell death, resulting in underestimation of cell death
Annexin V staining
Change in cell surface molecule
Specific for apoptosis
Although the annexin V–propidium iodide (PI) fluorescence-activated cell sorter (FACS) analysis is considered a specific and objective method for quantitation of apoptosis, reported inconsistencies might stem from errors in collecting a small amount of light, dead cells, which are hard to be spun down by ordinary 1,500–revolutions per minute, 5-minute centrifugation. The DNA fragmentation enzyme-linked immunosorbent assay (ELISA) is based on a quantitative sandwich enzyme immunoassay using antibodies against DNA and histones, and determines mono- and oligonucleosomes in the cytoplasmic fraction of cell lysates. It is reported to be specific for detection of apoptosis.
De facto, all of these methods have their own drawbacks. Detection of apoptotic bodies is too restrictive, and although electron microscopy appears to be the gold standard, it is complicated and does not allow the investigation of large numbers of samples or cells. The ISEL or TUNEL technologies, which are based on the detection of DNA strand breaks within tissue sections, are prone to technical errors and highly dependent on various technical parameters such as proteolytic digestion and TdT-enzyme concentration (18–21). Therefore, careful evaluation of the technique and correlation with the morphologic appearance of apoptotic bodies is necessary, at least in some of the positive cells. If the method is not extremely carefully applied, the results appear to be nonspecific for apoptosis (18, 19, 21). We and others have recently shown that this is particularly true in fetal and adult cartilage (22, 23). Frozen and thawed chondrocytes can show up to 92% TUNEL positivity (23). Moreover, frozen sections of human articular cartilage have often revealed far too high rates of apoptotic cell death (6, 24). On the basis of these observations taken together, one has to be very careful in order to avoid confusion about the cellular phenomena that actually take place. The optimal strategy would be to combine a panel of the different technical approaches (25).
An excellent in vivo control system for detection of apoptotic cells in cartilage is the fetal growth plate cartilage (for review, see ref. 26). If not properly controlled, most hypertrophic chondrocytes and even some of the cells in the proliferative zone appear to be positive in the TUNEL reaction (refs. 27–29 and Aigner T: unpublished observations). Under the appropriate conditions, however, strong labeling for apoptotic cells is observed selectively in the lowest hypertrophic zone (22, 30, 31), which is consistent with ultrastructural data (32, 33).
Technical pitfalls in the detection of apoptosis in OA adult articular cartilage
Despite the appearance of many empty lacunae in OA cartilage at the histologic level, careful investigation has shown only a minor, albeit significant, increase in empty lacunae in OA lesions (22). Thus, cell disintegration is certainly a phenomenon in the destruction of OA cartilage, but further studies are necessary to define the cell content of articular cartilage in relation to disease stage and age. At present, many contradictory findings exist (34–36), which might be partly due to the presence of swelling and collapse phenomena commonly observed in aged and diseased cartilage.
Empty lacunae presumably arise in human OA cartilage largely because of technical artifacts (4, 22). Theoretically, it cannot be excluded that cell death might have occurred after cell division, resulting in cartilage lacunae containing cell debris, together with one viable cell. However, this would be of little pathogenetic relevance, because it merely indicates cell turnover in single lacunae without any overall cell loss. Intriguing, however, would be a mechanism in which, as described by Roach and colleagues for chick hypertrophic chondrocytes (37), cell proliferation might occur with subsequent apoptotic cell death of one daughter cell, while the other showed a stably transformed phenotype. Although, to date, there is no experimental evidence for similar events in human articular (OA) cartilage, this might implicate a point of no return for the disease process, leading to a fixed “OA” cellular phenotype.
Contradictory reports exist on the fraction of “apoptotic” TUNEL-positive cells in normal and OA cartilage (6–8, 22, 25, 38), ranging from clearly <1% (22) to up to ∼20% (24). In most cases, no correlation between cell morphology and TUNEL staining could be established (25). Measurement of apoptosis by TUNEL staining in degenerated articular cartilage (or perhaps any degenerated tissue) is clearly prone to overestimation if not carefully performed. For example, the percentage of apoptotic cardiomyocytes in end-stage heart failure was initially reported to be 5–35% (39). This high rate was later reduced to 0.23% after careful morphologic evaluation for nuclear condensation of TUNEL-positive cardiomyocytes by confocal microscopy (40). In a recent study, counting of apoptotic cells that were characterized morphologically by a condensed nucleus with shrunken cytoplasm or fragmented nucleus was performed, rather than relying primarily on the TUNEL reaction (38). Interestingly, although the pattern of TUNEL staining in these experiments correlated with that of apoptotic cell distribution, the percentage of TUNEL-positive cells was roughly 10 times higher than that of apoptotic cells.
Since apoptosis, in most cases, is completed quickly once the degradation phase that is heralded by DNA fragmentation begins, a high rate of apoptosis reflected by the TUNEL staining result (which, in degenerating cartilage, was as high as 12% in previous studies) would theoretically result in cartilage degradation within a very short period of time, which is not consistent with the prolonged disease course of OA. In addition, ultrastructural examination must be carried out with strictly defined criteria. Many features described in the literature as “compatible” with apoptosis are neither specific for, nor fully indicative of, this process. True apoptotic bodies are nuclear (41), and not cytoplasmic exclusion bodies (6) or cells with condensed chromatin or cytoplasm (25).
Altogether, the experimental evidence clearly suggests that apoptosis occurs in OA cartilage, but at a very low rate. The relative contribution of apoptotic cell death in the pathogenesis of a degenerative disease is difficult to assess because of the chronic nature of the disease process, and at any one time, only a limited number of apoptotic cells can be detected. This is well illustrated by neuronal apoptosis in Alzheimer's disease. Only 0.02–0.04% of brain neurons in patients with Alzheimer's disease display apoptotic morphology and cytoplasmic immunoreactivity for activated caspase 3 (42), and this low rate is considered compatible with the progression of neuronal degeneration in this chronic disease. It would certainly be an oversimplification to state that apoptotic chondrocyte death is the main pathogenetic mechanism in OA. The implication of chondrocyte death, i.e., cellular disintegration, should, rather, be interpreted in the context of its relationship with matrix degradation. For example, a previous report shows that matrix vesicles released from apoptotic nitric oxide (NO)–treated chondrocytes produce pyrophosphate and can precipitate calcium, suggesting that the functional capacity of such vesicles may contribute to the pathologic calcification of cartilage (43).
There is currently no answer as to what extent apoptosis should be observed in degenerative cartilage in order for it to play an important pathogenetic role. Nevertheless, chondrocyte apoptosis plays a role in the pathogenesis of OA not because it is widespread in OA cartilage, but because chondrocyte apoptosis exists in cartilage with advanced OA, and the frequency of apoptosis is increased compared with that in normal cartilage.
In contrast to the other cartilage zones, the calcified cartilage layer shows a significant number of empty lacunae. In fact, the calcified cartilage layer represents a major portion of the remaining cartilage in late-stage OA. This might, in part, explain the increasing numbers of empty lacunae reported within high-grade lesions (2, 4, 5). Despite the extensive presence of empty lacunae (1), TUNEL-positive cells were rare in calcified cartilage (22, 25), suggesting that the detection of actual apoptotic/necrotic cells is not a prerequisite for the presence of apoptosis/necrosis within articular cartilage. The rather high numbers of apoptotic cells (25) and empty lacunae (22) support the concept that the ongoing calcification of articular cartilage in OA reflects the restart of mechanisms during the disease process that are physiologically seen in the fetal growth plate, such as onset of chondrocyte hypertrophy with the expression of type X collagen (44, 45), matrix calcification, and, finally, apoptotic cell death of the terminally differentiated chondrocytes (32, 33, 46).
Mechanism of chondrocyte apoptosis
Chondrocyte apoptosis is a complex process, and it will be of much importance to investigate the factors that induce it. Fetal growth plate cartilage seems to provide a very useful material for such an investigation, because apoptotic events are constitutively present and the geography of events is clear. Mansfield and colleagues and Rajpurohit and colleagues showed that inorganic phosphate ions increase chondrocyte apoptosis in a dose-dependent manner (47, 48). Increases in phosphate and calcium ions could well be one mechanism responsible for chondrocyte apoptosis in the fetal growth plate, as well as in the lower zones of OA cartilage, because both areas show progressive matrix calcification.
Another very successful way to investigate apoptotic cell death is through the use of in vitro culture systems. Thus, cultured chondrocytes undergo apoptosis in response to various stimuli, including serum deprivation (49), or in response to treatment with Fas ligand or anti-Fas/CD95 antibodies (43, 50, 51), NO-donor sodium nitroprusside (50), staurosporine/dihydrocytochalasin B (52), ceramide, and retinoic acid (53). However, as always, one needs to be careful about which cells to culture and under which conditions to obtain physiologically relevant results. Thus, fetal chondrocytes behave differently from adult chondrocytes, species differences have to be taken into consideration, and differences in culture conditions, such as monolayer or alginate beads, might strongly modulate cell behavior.
The role of NO synthase (NOS) has been one focus of interest in the pathogenesis of OA, even before its role in induction of chondrocyte apoptosis was reported (50). Because OA cartilage is thought to produce a large amount of NO, it could serve as a powerful initiator of chondrocyte apoptosis in OA cartilage. NO has been implicated in the inhibition of neosynthesis cartilage matrix macromolecules such as aggrecan, in enhancing matrix metalloproteinase (MMP) activity, and in reduction of interleukin-1 (IL-1) receptor antagonist synthesis. Therefore, induction of chondrocyte apoptosis could also serve as an additional indicator of the importance of NO for mediating cartilage degradation (54). Selective inhibition of inducible NOS (iNOS) was suggested to have positive effects on the progression of lesions in an experimental canine OA model (55). Nevertheless, with the plethora of detrimental effects NO exerts on articular cartilage, the benefit from inhibiting chondrocyte apoptosis by iNOS inhibitor is difficult to define.
The mechanism of NO-mediated chondrocyte apoptosis has recently been partially elucidated (56). Chondrocyte apoptosis induced with 1 mM of sodium nitroprusside is completely blocked by caspase 3 inhibitor and caspase 9 inhibitor. However, a higher dose of sodium nitroprusside at 2 mM causes caspase-independent cell death, implying occurrence of necrosis at this concentration. Upstream, sodium nitroprusside–generated NO induces expression of cyclooxygenase 2, possibly through the extracellular signal–regulated kinase 1/2 and p38 kinase pathways, resulting in an increase in release of prostaglandin E2 (PGE2) in human OA chondrocytes (57). The role of PGE2, however, in induction of chondrocyte apoptosis is debated. In one study, normal bovine articular chondrocytes exhibited DNA fragmentation by PGE2 (56), while a separate study using human OA chondrocytes reported that PGE2 sensitizes chondrocytes to NO-induced apoptosis, rather than inducing apoptosis directly (57). Whether susceptibility to PGE2-mediated apoptosis is species-dependent or varies between normal and OA chondrocytes is a subject of further research.
NO-induced chondrocyte apoptosis via PGE2 is also important from the clinical standpoint, considering the widespread prescription of nonsteroidal antiinflammatory drugs to the OA patient population, and their effect on the maintenance of OA cartilage. Meanwhile, many of these results rely on in vitro systems, and a recent report raises the possibility that traditional NO-donor compounds might mediate chondrocyte death via toxic products other than NO. Therefore, these results may have to be taken cautiously (58). In addition, the generation and effect of NO might depend considerably on the manner of NO production, delivery, and achieved levels of concentration, which may be difficult to reproduce in an experimental condition.
Among the mediators of cartilage degradation, IL-1 and tumor necrosis factor α (TNFα) seem to be very potent players. In a variety of cell types, signaling pathways by catabolic cytokines such as TNFα and IL-1 involve activation of sphingomyelinase, which degrades the membrane phospholipid sphingomyelin into phosphocholine and the intracellular messenger ceramide (59). Ceramide stimulates messenger RNA (mRNA) expression of various MMPs in human fibroblasts and keratinocytes, and the signaling cascade is reported to involve mitogen-activated protein kinases and activator protein 1 (60, 61). Recently, Sabatini et al reported ceramide-stimulated proteoglycan degradation in rabbit cartilage explant cultures in a dose-dependent manner (62). This effect was inhibited by the MMP inhibitor, batimastat, and mRNA expression of MMPs 1, 3, and 13 was stimulated. Intriguingly, ceramide also caused chondrocyte apoptosis at the same dose range that stimulated MMPs. An inhibitor of ceramide generation, fumonisin B1, reportedly does not affect sodium nitroprusside–induced chondrocyte death, indicating that ceramide generation is independent of NO-induced chondrocyte death (57).
Whether stimulation of MMP expression by ceramide and induction of apoptosis by ceramide are concomitant events in the same cell or are alternative responses that depend on the differentiation and activation state of the chondrocyte is a subject of future research. Of note, chondrocytes are resistant to induction of apoptosis by IL-1 treatment alone (50). Induction of chondrocyte apoptosis by TNFα is debatable, because 2 recent reports revealed different results. Aizawa et al (63) reported that TNFα caused a dose-dependent 1.5–2-fold increase in cell death in both hypertrophic and nonhypertrophic chondrocytes from chick embryo sterna, whereas another report did not reveal DNA fragmentation at a rather high concentration (100 ng/ml) of TNFα in human OA chondrocytes (51). Much of the controversy may derive from the differences in apoptosis susceptibility between species, between young and old, and between normal and diseased chondrocytes.
TNF receptor–mediated signaling leads to nuclear factor κB (NF-κB) activation, and this pathway is responsible for protection of various cell types from the cytotoxic effects of TNFα (64). TNF-receptor signaling simultaneously stimulates pathways for apoptosis induction via receptor-associated proteins that activate caspases and pathways for suppression of cell death via NF-κB. Thus, the balance of these 2 signaling pathways determines the fate of cells after stimulation with TNFα. Investigators in our laboratory have observed that human OA chondrocytes undergo apoptosis when treated with TNFα plus the proteasome inhibitor, MG132, or the transcription inhibitor, actinomycin D, but not when treated with TNFα alone (65). Considering the effects of proteasome inhibitors on a cell system, it is not certain whether its effect on NF-κB activation was directly related to its effect on chondrocyte apoptosis. Although we observed inhibition of TNF-induced NF-κB activation by MG132 in chondrocytes, this is not direct proof that NF-κB inhibition is responsible for MG132/TNF-mediated apoptosis (Aigner T, Kim HA: unpublished data).
Another well-known apoptosis-inducing pathway is CD95/CD95 ligand–mediated signal transduction. Because chondrocytes in human OA and cultured chondrocytes express Fas receptor, activation of this pathway was postulated to mediate chondrocyte apoptosis in human OA (38, 66). The observation that Fas ligand is detected in the synovial fluid of OA patients, where it derives from infiltrating lymphocytes (67), is further evidence for Fas/Fas ligand–mediated chondrocyte apoptosis in OA pathogenesis. However, direct induction of Fas-mediated apoptosis in cultured chondrocytes is not without controversy (43, 68, 69).
Investigators in our laboratory failed to detect significant death of chondrocytes treated with agonistic CD95 antibody alone, by either MTT assay or trypan blue exclusion assay (70). When one of us analyzed cell death using MTT, trypan blue, annexin V–PI FACS, and DNA-fragmentation ELISA simultaneously, however, it was found that the DNA-fragmentation ELISA detected a small, but consistent, increase in DNA fragments in anti-Fas–treated chondrocytes (Aigner T: unpublished observations). Although it is reported that the process of DNA fragmentation and cell death may be independent in some apoptotic processes (71), the significance of DNA fragmentation in the absence of metabolic disturbance (indicative of cell death) is uncertain. Another possibility, of course, is that the DNA-fragmentation ELISA is the most sensitive assay for detection of little amounts of chondrocyte death, such as that induced by anti-Fas treatment. Again, for the best result, it is appropriate to combine 2 or more assays and evaluate apoptosis both morphologically and biochemically.
The issue of Fas receptor–mediated chondrocyte apoptosis is important because of the recent interest in the therapeutic implication of apoptosis induction in synoviocytes with Fas-signaling molecules, e.g., in rheumatoid arthritis. There is concern about the effects of this type of treatment on adjacent joint tissues. Thus, the susceptibility of chondrocytes to Fas-mediated apoptosis is of concern regarding its impact upon articular cartilage integrity. Similar to our results with TNFα, coincubation of anti-Fas and proteasome inhibitor MG132 induces significant cell death, and the role of NF-κB activation in chondrocyte survival is postulated (51). Again, however, the direct role of NF-κB inhibition should be assessed with a specific method, such as transfection of the super-repressor form of inhibitor of NF-κB. With regard to Fas/Fas-ligand signaling, there is an inherent barrier to receptor engagement on chondrocytes in articular cartilage, because chondrocytes are not in close proximity to each other and the dense matrix hinders free movement of Fas ligand from the synovial space into the cartilage substance. In this regard, matrix depletion occurring in OA cartilage may have a role in induction of chondrocyte apoptosis.
Matrix, mechanical stress, and chondrocyte death (Figure 1)
It is generally accepted that except for circulating blood cells, most cells require attachment to one another or to the extracellular matrix for proper growth, function, and survival. Otherwise, they are committed to apoptosis, and this dependence of cell growth and survival on substrate attachment is known as anchorage dependence. Since loss of proteoglycan and perturbation of the collagen network are 2 of the main pathologic features of cartilage degradation, this provides an ample environment for the disturbance of chondrocyte anchorage to the extracellular matrix. This is particularly true because the pericellular cartilage matrix is severely altered in OA articular cartilage (72, 73).
Recent studies have shed light on the role of the collagen framework in the maintenance of cartilage cell survival. In a mouse model lacking type II collagen, chondrocyte apoptosis increased markedly in articular cartilage during embryogenesis, providing direct evidence that type II collagen is crucial for supporting the survival of chondrocytes in articular cartilage (74). Cao et al reported that in a suspension-culture system, removal of collagen by collagenase resulted in chondrocyte apoptosis, whereas re-addition of collagen restored cell viability (75). Consistent with the findings of a previous report (76) on anti–β1 integrin antibody–induced chondrocyte apoptosis in chick sternal cartilage organ culture, Cao et al also showed that anti–β1 integrin antibody abolished collagen-mediated chondrocyte aggregation, which hampered the ability of collagen to restore chondrocyte viability.
Whether apoptosis or cell disintegration is primary or secondary to the destruction of cartilage matrix is difficult to answer. There is also the possibility that chondrocyte death and matrix loss form a vicious cycle, with the progression of one having detrimental effects on the other. Certainly, chondrocytes become somewhat fragile if the inter- or pericellular matrix is removed or deranged (75, 76), as occurs in OA cartilage (72, 73, 77). In fact, degradation products of pericellular matrix components, such as fibronectin, might directly stimulate chondrocytes, or alternatively, modulate their cellular phenotype or even initiate cellular death programs.
A mixture of all of these events is reflected, for example, in the high rates of cellular apoptosis of (OA) chondrocytes after isolation in culture (refs. 6 and7 and Aigner T: unpublished observations) and might well be an important factor in OA cartilage degeneration. This is also confirmed by in vitro experiments performed using collagenase digestion (78). Thus, 11% TUNEL-positive “apoptotic” cells of normal articular cartilage after enzymatic isolations (6) are certainly not representative of the in vivo situation, but clearly document that chondrocytes are fragile after isolation (matrix removal). Also, cell bubbling is a typical phenomenon immediately after cell isolation. However, chondrocytes have a high capacity to recover, and presumably not only in vitro, but also in vivo. Chondrocytes do withstand very adverse conditions in situ (e.g., anoxia) and appear to survive even if they show severe cellular derangement indicative of apoptotic or necrotic cell disintegration in other cells. Whatever is true, cellular disintegration in articular cartilage represents a point of no return, given the lack of cell supply within this tissue. Thus, major cell death inevitably leads to a failure of cartilage matrix turnover, because chondrocytes are the only source of matrix-component synthesis in articular cartilage.
Last, but not least, the pathogenesis of OA cannot be properly delineated without elucidation of biochemical signaling initiated by mechanical stress. The role of mechanical stress on induction of chondrocyte apoptosis has been reported recently by different investigators. Chondrocyte apoptosis is induced at compression stress as low as 4.5 MPa in a bovine cartilage explant, while other parameters of matrix degradation, such as disruption of the collagen fibril network, tissue swelling, release of glycosaminoglycan, and an increase in nitrite level, were apparent only at higher stress levels (79). Investigators in another study using bovine cartilage explants applied repetitive compressive loading and observed that above a threshold of 6 MPa, cell viability was inversely proportional to applied stress (80). Again, gross damage to cartilage occurred at much higher levels of stress than those that resulted in chondrocyte death. A recent study utilizing human and rabbit cartilage explants demonstrated that chondrocyte apoptosis induced by an impact load could be inhibited by caspase inhibitor, along with inhibition of glycosaminoglycan release (81). This finding introduced the possibility of modulating cartilage damage by a pharmacologic agent that inhibits chondrocyte cell death. Although these model systems of mechanical loading do not accurately reflect what happens in human articular cartilage in everyday life, they provide an explanation for the increased risk of OA posed by joint overloading.
Implications for the cell biology of OA cartilage and outlook
The cellular reaction pattern during the OA disease process is, at first sight, rather pleomorphic. However, it can be basically divided into 3 categories, which do not necessarily occur sequentially and which all might influence cartilage matrix maintenance by chondrocytes (Figure 2). First, chondrocytes activate or deactivate their synthetic-anabolic activity (for review, see refs. 82 and83). Second, chondrocytes undergo phenotypic modulation, implicating an overall severely altered gene-expression profile of the cells in the diseased tissue (for review, see refs. 82 and84). Third, the chondrocytes can undergo cell death, whether it is programmed or not, or they can proliferate in order to compensate for cell loss or in order to increase their synthetic activity, as cells do in many other tissues of the body. Thus, many studies (22, 34, 35, 85, 86) have clearly shown that there is (a very low) proliferative activity in OA chondrocytes, resulting in the chondrocyte clusters typical of OA cartilage. However, it is doubtful whether this is an efficient mode of tissue repair (86), since the cell clusters do not appear to add necessarily to matrix anabolism (87).
An issue hardly addressed at all in any study so far is the relative importance of the discussed phenomena in relationship to the stage of the disease process. Thus, it is apparent that programmed cell death in aging or during early stages of the disease will be of greater relevance than in late stages, in which the joint surface is already terminally destructed. On the other hand, in normal cartilage and early stages, mechanisms potentially triggered by mediators such as IL-1 (50) might counteract cellular death programs, whereas such cellular mechanisms are no longer working in the advanced stages of disease.
Whichever type of cell death takes place in articular cartilage, it will be very important to prevent it, because it is detrimental to the maintenance of articular cartilage. Although the outcome of apoptosis, paraptosis, or oncosis/necrosis is the same and all lead to cell-deprived cartilage, obviously the knowledge of these different forms of cell death which involve different initiation and progression factors is essential in order to develop concepts of counteraction. Thus, it will be important for future research to characterize in more detail the events going on during cellular degeneration. One reason why we understand very little of the cellular reaction pattern in OA cartilage is that most of the involved genes have not yet been identified and characterized. Gene-expression profiling using complementary DNA (cDNA) arrays is an upcoming powerful technology that permits identification of differences in mRNA expression levels of large numbers of genes at the same time (88). Recently, a first, larger screen was performed on articular chondrocytes in our laboratory (89), although genes important for apoptotic cellular events were not directly assessed. However, the “inhibitor of apoptosis” was increased by IL-1 (Aigner T: unpublished data), potentially explaining why IL-1–stimulated chondrocytes are not prone to apoptotic cell death (90) despite a high production of NO, which is usually an initiator of apoptosis in chondrocytes (as discussed above). Also, IL-1 induced NF-κB, which is thought to have antiapoptotic effects on chondrocytes (90).
It will be very important to investigate the early stages of the disease process. Most studies, including ours, were done largely on peripheral areas of eroded tissue of late-stage OA material obtained at surgery. This always assumes that the low-to-moderate Mankin's grades (35) of late-stage OA in the joints are comparable with those for the central areas of cartilage in early or moderately advanced OA. This, however, is not proven and needs further investigation. In fact, in light of the findings by one of us in recent cDNA-array analyses (Aigner T: unpublished data), this is rather unlikely.
Thus, it will be important to establish animal models that give access to the early stages of the disease. This could involve in vitro model systems, such as localized laser injury which allows identification of the temporospatial distribution of cellular events (91). Physical loading might be another even more physiologically relevant stimulus (23), as well as the induction of NO synthesis (92). Less defined, but even closer to the natural course of the OA disease process, may be models such as the anterior cruciate ligament transection (ACLT) model, which shows significant levels of “apoptotic” cells, although the 29% of apoptotic cells described in this model (7) very likely reflects an overestimation of cell death resulting from increased fragility of OA chondrocytes after cell isolation. In a recent report, the results appear to show that treatment with hyaluronic acid in the ACLT model leads to significant reduction of cell apoptosis (93).
Animal models will allow more precise evaluation of the time course of cellular degeneration and its consequences, will enable identification of inducing factors, and most importantly, will reveal the best approach to interfere with ongoing cellular degeneration and thereby maintain cartilage-matrix integrity, the ultimate goal of any anti-OA therapy.
Overall, we have reached a reasonable level of understanding of the extent of cell death occurring in the disease process, but we are still at very early stages in our understanding of the mechanisms underlying this process and in developing means of intervening with this facet of cartilage destruction.