Acute joint injury leads to increased risk for osteoarthritis (OA). Although the mechanisms underlying this progression are unclear, early structural, metabolic, and compositional indicators of OA have been reproduced using in vitro models of cartilage injury. This study was undertaken to determine whether glycosaminoglycan (GAG) loss following in vitro cartilage injury is mediated by cellular biosynthesis, activation of enzymatic activity, or mechanical disruption of the cartilage extracellular matrix.
Immature bovine cartilage was cultured for up to 10 days. After 3 days, groups of samples were subjected to injurious mechanical compression (single uniaxial unconfined compression to 50% thickness, strain rate 100% per second). GAG release to the medium was measured, and levels were compared with those in location-matched, uninjured controls. The effects of medium supplementation with inhibitors of biosynthesis (cycloheximide), of matrix metalloproteinase (MMP) activity (CGS 27023A or GM 6001), and of aggrecanase activity (SB 703704) on GAG release after injury were assessed.
GAG release from injured cartilage was highest during the first 4 hours after injury, but remained higher than that in controls during the first 24 hours postinjury, and was not affected by inhibitors of biosynthesis or degradative enzymes. GAG release during the period 24–72 hours postinjury was similar to that in uninjured controls, but the MMP inhibitor CGS 27023A reduced cumulative GAG loss from injured samples between 1 day and 7 days postinjury. Other inhibitors of enzymatic degradation or biosynthesis had no significant effect on GAG release.
Injurious compression of articular cartilage induces an initially high rate of GAG release from the tissue, which could not be inhibited, consistent with mechanical damage. However, the finding that MMP inhibition reduced GAG loss in the days following injury suggests a potential therapeutic intervention.
Degenerative joint diseases such as osteoarthritis (OA) result in degradation of articular cartilage, characterized by release of macromolecular constituents from the tissue and subsequent loss of tissue integrity and mechanical properties (1). However, the precise etiology of the disease remains unknown. Consistent with observations that the in vivo mechanical environment of cartilage influences its cellular biosynthesis (2, 3), abnormalities in joint loading due to obesity, joint laxity (as in ligament transection ), or altered joint geometries (e.g., acetabular dysplasia ) have been recognized as risk factors for OA (6).
Additionally, traumatic joint injury leads to increased risk of OA (6, 7), though the precise mechanism by which joint injury leads to disease is poorly understood. In studies of synovial fluid from joints after trauma, it has been found that in the hours following injury, the level of the zymogen form of stromelysin (matrix metalloproteinase 3 [MMP-3]) in the synovial fluid was increased up to 40-fold over normal levels, and elevated levels of MMP-3 persisted for up to 20 years following injury (8–11). Further analysis of synovial fluid revealed evidence of MMP and aggrecanase activities in mediating proteoglycan cleavage and subsequent cartilage destruction after injury and in OA (12–14).
Joint injury is a complicated phenomenon, involving high amplitudes and complex modes of loading, and potentially, multiple joint tissues including cartilage, subchondral bone, ligament, tendon, and vascularized tissues such as synovium. To study the basic mechanisms involved in biologic responses to joint injury, in vitro models of acute cartilage injury have been developed by several groups, allowing more precise control of the tissue geometry, loading patterns, and incubation conditions. Because of the variation in protocols for “injurious compression,” it can often be difficult to compare experimental results directly. However, high levels of cartilage compression in vitro can result in cell death (necrosis, apoptosis, or both) (15–22), release of cartilage proteoglycans (15, 17, 19, 20), increased tissue water content and swelling (19, 22, 23), decreasing mechanical functionality (16, 19–21), and impaired response to further mechanical stimulation (16). Furthermore, cells remaining viable after injury exhibited increased levels of messenger RNA for matrix-degrading enzymes (24), and pulse-chase radiolabeling experiments have indicated that the matrix surrounding these remaining viable cells may undergo accelerated degradation (20).
Many of these characteristics are reminiscent of early stages of degenerative joint disease (1). Although OA progression in vivo occurs over a period of years, these trauma-induced changes appear on a relatively short time scale (hours to days) following in vitro injury. It is possible that the short-term effects of applied loading may involve direct damage to the cartilage extracellular matrix, initiation of cell-mediated tissue destruction, and accelerated transport of degraded matrix from the tissue, any or all of which may be important in long-term tissue changes. The objective of this study was to determine whether acute release of proteoglycan components of the extracellular matrix during the first hours and days following in vitro cartilage injury is mediated by alteration of the balance between anabolic and catabolic processes, changes in proteolytic enzyme activity, mechanical disruption of the cartilage extracellular matrix, or a combination of these processes.
MATERIALS AND METHODS
Medium supplement ITS-A was from Invitrogen (Carlsbad, CA). Cycloheximide (CHX) and APMA were from Sigma (St. Louis, MO). Recombinant human interleukin-1α (IL-1α) was from R&D Systems (Minneapolis, MN). The broad-spectrum hydroxamate MMP inhibitor GM 6001 (GM; also called Galardin or Ilomastat ) was from Chemicon (Temecula, CA). MMP inhibitor CGS 27023A (CGS) (26) and aggrecanase inhibitor SB 703704 (AGG) (27) were obtained from GlaxoSmithKline (King of Prussia, PA). The calcein AM–ethidium homodimer LIVE/DEAD viability/cytotoxicity kit was from Molecular Probes (Eugene, OR). Other materials were obtained as previously described (19).
Tissue harvest and initial culture.
A time line of the experiments is shown in Figure 1. Samples of immature (1–3-week-old) bovine articular cartilage were prepared from the patellofemoral grooves of 11 bovine calf knee joints on the day of slaughter, as previously described (19). Briefly, osteochondral samples were collected with a 9-mm–diameter stainless steel coring bit, then placed into a sledge microtome, where the superficial 100–300 μm of cartilage was removed and discarded. Two subsequent 1-mm cartilage slices, from the middle zone, were created parallel to the articular surface, and from these, 3-mm–diameter disks (4 disks per slice) were made with a stainless steel dermal punch. This technique allowed creation of 4 experimental groups per harvest, containing location- and depth-matched samples. To allow cartilage metabolism to stabilize following harvest, all samples were cultured for 3 days prior to mechanical loading in medium (low-glucose Dulbecco's modified Eagle's medium with 20 μg/ml ascorbate, 0.1 mM nonessential amino acids, 0.4 mM additional L-proline, 10 mM HEPES, 100 μg/ml streptomycin, 100 units/ml penicillin, and 0.2 μg/ml amphotericin B) supplemented with 10% (volume/volume) fetal bovine serum. All cultures were performed in a 37°C, 5% CO2 atmosphere.
In vitro injury.
On day 3 following harvest, serum-supplemented medium was replaced with serum-free medium (99% medium, 1% ITS-A [v/v]), and this was used for the duration of the experiment. For inhibitor experiments, culture medium was also supplemented with one of the following: CHX (100 μg/ml) to inhibit protein translation, 1 of 2 broad-spectrum hydroxamate MMP inhibitors (GM [10 μM] or CGS [10 μM]), or AGG (20 μM). Samples were allowed to equilibrate in medium with or without inhibitors for 6 hours before being either subjected to in vitro injury or left uninjured in a free-swelling state. Mechanical injury consisted of 1 uniaxial radially unconfined compression to 50% thickness, at 1 mm per second (strain rate 100% per second), followed by immediate release of load (Figure 1B), using an incubator-housed loading apparatus described previously (28). After injury, each sample was assigned a deformation score based on the shape of the cartilage specimen under visual inspection (0 = unchanged [cylindrical]; 1 = slightly noncircular or flattened; 2 = grossly ellipsoidal). Tissue was then returned to culture in fresh serum-free medium with or without inhibitors.
Postinjury tissue culture and sample analysis.
Typically, medium was collected and changed at 24 hours and 72 hours postinjury (Figure 1A), and collected medium was stored at −20°C prior to biochemical analysis. In one experiment, medium was also collected at 4 hours postinjury, and in 2 others, culture was continued for 7 days postinjury. In order to determine protein and glycosaminoglycan (GAG) biosynthesis, the medium of some samples was supplemented with 10 μCi/ml L-5-3H–proline and 5 μCi/ml Na235SO4 for either 24 hours (hours 0–24 postinjury) or 48 hours (hours 24–72 postinjury).
On termination of culture, radiolabeled tissue samples were washed 4 times during a period of 2 hours with phosphate buffered saline plus 1 mM proline and 1 mM Na2SO4 to remove unincorporated radiolabel, then solubilized with 500 μg/ml proteinase K. A portion of this digest was assayed for radioactivity by liquid scintillation counting. Selected tissue samples that were not digested for biosynthesis analysis were inspected for qualitative distribution of viable cells, using LIVE/DEAD dyes and epifluorescence microscopy (Diaphot TMD; Nikon, Melville, NY). With this method, the calcein AM component is processed into a fluorescent product by live cells, while the ethidium homodimer component fluoresces only when bound to DNA and is excluded from viable cells by the intact cell membrane.
Medium collected from postinjury cultures was analyzed for sulfated GAG content using the dimethylmethylene blue assay, with shark chondroitin sulfate used as the standard (29). In addition, medium fractions from some radiolabeled samples were pooled and analyzed for macromolecular 35S-sulfate content using PD-10 (Sephadex G-25) column chromatography, in order to determine whether GAGs were synthesized and released during the pulse-labeling period (postinjury) or were present prior to injury. Based on the assumption that GAG disaccharides in immature bovine cartilage are, on average, monosulfated (30), the known specific activity of the radiolabel was used to convert moles of incorporated 35S-sulfate to moles of GAG disaccharide. The moles of GAG disaccharide were converted to mass using the molecular weight of a chondroitin sulfate disaccharide (458 gm/mole).
In studies of injury, effects of injury and inhibitors were evaluated by two-way analysis of variance (ANOVA) followed by comparisons of selected hypotheses, with Dunn-Sidak correction for multiple comparisons (31). In studies of inhibitor efficacy and biosynthesis, data were analyzed by ANOVA with animal source included as a random effect. To investigate possible relationships between peak stress and GAG release, linear regression analysis was performed. A nonparametric test of variable association (Spearman's rank correlation) was performed using the variables peak stress, deformation score, and slice depth to determine if these quantities were related. Statistical computations were performed using SYSTAT 9.0 (SPSS, Chicago, IL). Data are expressed as the mean ± SEM.
Determination of inhibitor efficacies in cartilage explant culture system.
The efficacy of MMP inhibitors was verified by incubating free-swelling, location-matched cartilage samples in culture medium with or without MMP inhibitors (10 μM) with 1 mM APMA to activate endogenous MMPs (32). In the absence of inhibitors, 24-hour treatment with 1 mM APMA induced an 18–20-fold increase in GAG released from free-swelling cartilage samples (n = 5) (P < 0.001), as shown in Figure 2. Inclusion of either inhibitor (n = 5 per group) eliminated >99% of this increase, returning GAG release to near control levels (12–17% above control; P < 0.05). The presence of either MMP inhibitor without APMA (n = 5) had a mild effect on 35S-sulfate or 3H-proline incorporation (<15% reduction), but this was not statistically significant (P > 0.2).
Aggrecanase inhibition was assessed by incubating free-swelling samples in culture medium with or without AGG (20 μM) with 2 ng/ml IL-1α to induce aggrecanase-mediated tissue degradation (33). After 3-day treatment with IL-1α, GAG release was 94% greater than in control cultures (n = 4 per group) (P < 0.001) (Figure 2). When AGG was present during culture, the increase in GAG release was only 28% above that in IL-1α–stimulated controls (n = 4) (P < 0.005). Thus, it appears that AGG, while not completely abolishing the effect of IL-1α on GAG release, had an inhibitory effect on aggrecanase-mediated matrix degradation. The inclusion of AGG without IL-1α reduced protein and GAG biosynthesis by uninjured cartilage (n = 5) by 25% and 29%, respectively (P < 0.05).
Effects of injurious compression on cartilage cell viability and biosynthesis.
Samples of immature bovine articular cartilage were subjected to compression to 50% of their original thickness over 0.5 seconds, a strain rate of 100% per second. Compression at this strain rate induced peak stresses of 22 ± 0.5 MPa within the tissue (as measured directly by the load cell of the compression instrument ), and 6-hour pretreatment with inhibitors of biosynthesis or catabolic activity (n = 54 total) had no significant effect on the magnitude of the peak stress compared with that of untreated samples (P = 0.6). Macroscopic inspection of the cartilage samples after compression revealed that although no samples exhibited gross fissuring due to loading, ∼25% were deformed into ellipsoidal geometry by injury (deformation score of 1 or 2), and these generally did not regain their initial cylindrical shape during the subsequent 3 days of postinjury culture. This deformation score was not significantly correlated with peak stress (r = 0.28, P = 0.05) or with slice depth (r = 0.22, P = 0.12).
After 3 days of culture following injury, the distribution of viable cells was qualitatively assessed using a cell viability kit and epifluorescence microscopy. In uninjured samples, a homogeneous population of live cells was observed across the tissue, with scattered cells staining as nonviable. In the injured tissue, however, large areas of devitalized cells were observed, particularly at the periphery of the samples and near the few blood vessels within the immature tissue. These populations of stained cells were not quantified in this study; however, detailed analyses of the effect of this injury protocol on cell apoptosis and necrosis in calf articular cartilage have recently been performed in another study (34).
Biosynthesis in injured and uninjured cartilage was assessed by 48-hour culture in the presence of radiolabeled precursors for GAG (35S-sulfate) and protein (3H-proline) biosynthesis during the period 24–72 hours postinjury. Both 35S-sulfate incorporation and 3H-proline incorporation into injured tissue were ∼50–60% below that observed in uninjured samples, consistent with previously reported values with this compression protocol (16).
GAG release occurs within 24 hours of injury and is not due to increased GAG biosynthesis.
During the first 24 hours following injury, the amount of GAG detected in the culture medium of injured cartilage samples was nearly double that observed in location- and depth-matched samples cultured under free-swelling conditions (10.4 ± 0.5 μg per sample versus 5.3 ± 0.1 μg per sample; P < 0.001) (Figure 3A). In a separate group of experiments, it was found that 33 ± 4% of the GAG released during the first 24 hours appeared in the medium during the first 4 hours after injury (Figure 3B), suggesting an initial burst of GAG release shortly following mechanical insult.
To determine whether GAG released from the tissue during this time was due to an immediate anabolic tissue response followed by release of newly synthesized proteoglycan or to degradation of existing matrix, some samples were cultured in the presence of 35S-sulfate immediately after loading, so that newly synthesized macromolecules in the medium could be assayed. Of the total GAG released to the culture medium during the 24 hours following injurious compression, the proportion that was determined to be newly synthesized (i.e., containing 35S-sulfate) was <1%, indicating that the majority of the released GAG was synthesized prior to injury and not due to increased GAG biosynthesis in response to injury. This result was further studied by culturing postinjury cartilage in medium containing CHX, a potent inhibitor of protein translation. Although CHX treatment reduced 35S-sulfate incorporation in free-swelling and injured cartilage by >99% (data not shown), there was no reduction in the amount of GAG released from injured tissue.
Another possibility was that the GAG released following injury was generated by serum-stimulated biosynthesis of proteoglycans during the initial 3 days of culture between harvest and injury, and the subsequent loss of this population from the tissue. To test this, some samples were incubated with medium with 10% FBS and CHX during the preinjury period, as well as during postinjury culture. Again, CHX treatment did not reduce the GAG loss at either the 1-day (P = 0.68) or the 3-day (P = 0.81) postinjury time point (n = 6 per group).
To examine whether GAG loss during this acute period after injury could be related to increases in the activities of degradative enzymes within the tissue, MMP and aggrecanase activities were inhibited by addition of chemical compounds to the culture medium. Although these compounds were proven to be effective in reducing GAG loss from chemically stimulated cartilage samples in this system, none were capable of significantly reducing injury-induced GAG loss in the first 24 hours following compression (Figure 3C).
GAG release at 1–3 days postinjury occurs more slowly and can be attenuated by an MMP inhibitor, but not by an aggrecanase inhibitor.
During the period 24–72 hours after compression (experiment days 4–6; Figure 1A), the amount of GAG released from injured samples (12.0 ± 0.3 μg per sample; n = 70–71) was similar to that released from free-swelling controls (11.4 ± 0.3 μg per sample) (P = 0.15), as seen in Figure 4A. In the injured samples (n = 62), the relationship between peak stress during injury and GAG release during the first 3 days postinjury was not significant (P = 0.38) (Figure 5).
Although injurious compression did not affect the magnitude of GAG release during this period, it was possible that anabolic and catabolic processes within the tissue were differently affected, so that the mechanisms by which GAG was released could differ between controls and injured samples. Inhibition of biosynthesis and catabolic enzyme activities during the period 24–72 hours postinjury resulted in a different pattern of GAG release than inhibition during the first 24 hours (Figure 4B). Similar to findings in the acute period, inclusion of CHX, the aggrecanase inhibitor AGG, or the MMP inhibitor GM had no effect on GAG release from injured tissues (P = 0.67, P = 0.16, and P = 0.75, respectively). However, with addition of the MMP inhibitor CGS (n = 19), GAG release from injured tissue was reduced to a level 20% below that in untreated, injured tissue (P < 0.005). Addition of CGS had no effect on the cumulative GAG release from free-swelling tissue samples through 7 days of medium supplementation, but significantly reduced cumulative GAG release from injured samples over the same time period (Figure 6).
The results presented here provide insight into the mechanisms of extracellular matrix degradation in an in vitro articular cartilage injury model. The release of GAGs from cartilage that had been subjected to injurious compression was observed to occur at an initially high rate (during the first 24 hours postinjury), which decreased to a rate approximating that from uninjured samples during the subsequent several days. Chemical inhibition of protein biosynthesis, as well as broad inhibition of MMPs and aggrecanases, did not affect the earliest GAG release, but MMP inhibition significantly reduced GAG release during the subsequent 1–7-day period following injury.
These observations strongly suggest that the initially high rate of GAG release during the acute period within 24 hours of injury is related to mechanical disruption of the matrix, rather than to cell-mediated processes involving biosynthesis. However, although none of the inhibitors used in this study reduced GAG release during this early period, it is possible that other enzymes not affected by these inhibitors could have been activated by the applied compression. Analysis of the conditioned culture media revealed that the majority (>95%) of the sulfated GAG released from the tissue during the first several days after injury was not due to an anabolic response of the tissue to injurious compression, but rather represented the degradation of a preexisting population of matrix molecules. Although no macroscopic fissures were noted after injury, the shape change of some samples from a cylindrical to an ellipsoidal geometry, and the persistence of this shape change throughout the duration of the culture, could provide further evidence of damage at the molecular level, particularly to the collagen network. Indeed, swelling of cartilage samples, usually taken as an indication of collagen network damage, has been reported previously from studies using this injury protocol (16). Injury-induced damage to the collagen network could lead to diffusive release of proteoglycans without fragmentation of the core protein, or perhaps, the release of proteoglycan aggregates (35).
In the application of protocols for injurious compression, the mechanical parameters are final strain, strain (or stress) rate, and peak stress, any two of which are independent. In the current study, we defined the final strain (50%) and the strain rate (100% per second), and measured the peak stress resulting from this compression. Interestingly, our study revealed no significant relationship between peak stress during injury and the amount of GAG released over the first 3 days postinjury (Figure 5), implying that the relationship between biochemical markers of degradation and mechanical injury parameters is complex. Although the concept of a “threshold stress” in cartilage injury has been proposed (22), it seems clear that other parameters must also be considered. For example, it is not clear whether the cellular effects of mechanical injury are due predominantly to tissue stress (which, in this model of applied strain, is a function mainly of the rate of load application and tissue permeability, producing hydrostatic pressure within the tissue) or tissue strain (affecting cellular morphology and possibly cell–matrix interactions).
In addition, the inhibition of cellular biosynthesis using CHX had no effect on GAG release in the first 3 days postinjury, consistent with an initial report (36) and further supporting the notion that released GAG was not newly synthesized. Being a broad inhibitor of protein translation, CHX should also inhibit synthesis of degradative enzymes (MMPs, aggrecanases, and cathepsins) and their inhibitors (tissue inhibitors of metalloproteinases). Taken together with the observation that treatment with an MMP inhibitor reduced GAG release at later time points (1–7 days postinjury), these findings indicate that GAG release may be mediated by a mechanism involving the activity of a preexisting pool of MMPs. However, it is not clear from the studies performed here whether these MMPs act directly to degrade matrix constituents or are members of a more extensive cascade of tissue destruction involving activation of other proteases (37).
An interesting, but untested, hypothesis to explain the observations with MMP inhibitors is that injurious compression activates a preexisting, inactive population of matrix-degrading enzymes, without need for further cellular biosynthesis. Previous work (32), as well as the current study, indicate that APMA treatment of cartilage leads to a rapid and severe loss of GAG from cartilage explants, coincident with near-total inhibition of cellular biosynthesis, implying biosynthesis-independent, MMP-mediated GAG release. It remains to be determined whether injury can induce similar activation of latent MMPs. Interestingly, a different mode of cartilage injury (scalpel incision) has been reported to instigate an immediate matrix-based response involving mobilization of a preexisting population of basic fibroblast growth factor, allowing its diffusion to cells and initiating intracellular signaling events within minutes of incision (38). Thus, injury to the cartilage matrix itself may directly elicit a catabolic response. Another possibility is that acute removal of matrix components by mechanical injury results in increased susceptibility of the remaining matrix to enzyme-mediated degradation. Thus, constitutively low catabolic enzyme activity could result in greater tissue degradation, without increases in the amount or activity of the enzyme itself.
Of the 2 broad-spectrum MMP inhibitors used in this study, only 1 (CGS) appeared to have an influence on postinjury GAG release, while the other had no apparent effect. To our knowledge, this is the first report of the use of the MMP inhibitor GM 6001 (also known as Galardin or Ilomastat) in a cartilage explant system. In studies with isolated MMPs and substrates in solution, GM 6001 has increasing binding affinity for MMPs with decreasing pH (39), and in other tissue systems, may reduce production of inflammatory cytokines (including IL-1α and IL-1β, IL-6, and tumor necrosis factor α) (40). CGS 27023A, however, has previously been used in cartilage explant cultures, at concentrations similar to those used in the present study (41, 42). In those studies, CGS had little effect on IL-1α–stimulated GAG released to the culture medium, consistent with reports that IL-1α induces aggrecanase, not MMP, activity (33, 43). The cause of the differential effects of these inhibitors is unclear and may reflect differences in their activities in an organ explant culture environment; this has not been studied extensively.
In the current experiments, inhibition of aggrecanase activity using a soluble pharmacologic inhibitor had no effect on GAG release immediately following injury. However, this does not rule out a potential role of aggrecanase at other time points, particularly in cases of joint injury in vivo, where other tissues may participate in an inflammatory response involving catabolic cytokine release. Indeed, in vitro studies have previously demonstrated that mechanical injury and inflammatory cytokines function together in a synergistic manner (i.e., more than additively) to increase GAG loss from cartilage explants (36). Ongoing studies are investigating the effects of coincubation of injured cartilage with joint capsule and synovium, to elucidate interactions among the various tissues.
Studies of mechanical injury of cartilage have used various tissue sources, tissue ages, and applied loading protocols. While there are clear differences between the immature bovine tissue used in this study and adult bovine and human cartilage, calf tissue has been well characterized, and certain trends of mechanotransduction and injury responses have been replicated in adult cartilage (17, 21). However, the age and source of the tissue under study can affect the experimental results. For example, Kurz and coworkers showed that the extent and distribution of mechanically induced apoptosis in bovine cartilage varied significantly with age (44). In addition, it is well known that the mechanical properties (45) and cellular metabolism (46) can vary greatly with age, as well as with depth and tissue location. While injury studies have incorporated a range of applied strains, strain rates, and peak stresses, there appears to be no consensus on a single loading protocol that produces “injury.” Although experimental results may depend on specific mechanical parameters, certain trends (e.g., zonal cell death, decreased biosynthesis, release of matrix components, diminished mechanical properties) observed using our loading protocol (Figure 1B) are similar to results obtained by others using alternative injury protocols that involve high levels of stress and strain. Followup studies are aimed at determining how the mechanisms identified here are related to graded levels of injurious strain, strain rate, and peak stress.
The choice of sulfated GAG release as the indicator of tissue degradation was motivated by clinical observations in which decreased levels of tissue GAG (47), coupled with increases in proteoglycan fragments in synovial and other bodily fluids, can be an early indicator of OA. The activities of matrix-degrading enzymes in OA also produce fragments of other molecules, such as collagens (48, 49) and noncollagenous proteins (50). Study of these degradation products may provide complementary insight into the connection between joint trauma and OA development. Additionally, further study of the structure of the released GAGs, for example, investigation of proteoglycan fragment size or characteristic core protein cleavage neoepitopes (51), is ongoing, and may lead to improved understanding of the underlying mechanisms of the observed GAG release.