A number of model systems have provided some insights. For instance, IL-1-degraded cartilage tissues and cell suspensions are frequently employed as model systems to study cartilage metabolism under osteoarthritis-like conditions [1-4] and to assess the efficacy of various drug therapies [5-9]. IL-1 signals cells via cell surface receptors to upregulate proteases, matrix metalloproteinases and aggrecanase, while downregulating matrix metalloproteinase inhibition and proteoglycan synthesis [10,11]. The ultimate effect of these metabolic changes include damage to the ECM, loss of ECM components (especially proteoglycan), and alteration of the normal balance of catabolic and anabolic processes that regulate cartilage ECM composition.

With regard to the repair phase, following IL-1-induced cartilage damage, relatively little is known, but partial to complete recovery appears possible [1,3]. In studies of repair, investigators focused on glycosaminoglycan (GAG) – an important and abundant cartilage macromolecule - and used histologic, biochemical, and/or radiolabel incorporation methods to measure GAG concentration ([GAG]) and/or GAG synthesis rates at selected time points. Although these data suggest that repair might be possible, our understanding is limited because of the uncertainty in the temporal changes in matrix composition for a given sample. This uncertainty is a direct consequence of the inherently destructive nature of histologic and biochemical methods; these techniques therefore do not permit monitoring of the spatial and temporal evolution of matrix macromolecules within individual samples/animals. Having to infer temporal spatial information from a cross-section of samples confounds efforts to conduct a detailed analysis of responses in these model systems.

The delayed Gadolinium-Enhanced Magnetic Resonance Imaging of Cartilage (dGEMRIC) method overcomes these problems in that it is nondestructive and can be used to assess quantitatively the spatial distribution of cartilage [GAG] [12,13]. We and others previously reported on the use of dGEMRIC to monitor GAG regeneration in living cartilage tissue over time in trypsin-depleted bovine cartilage [14] and in cell-laden polymer matrices (tissue engineered cartilage) [15]. Those studies clearly demonstrated that dGEMRIC is feasible for long-term [GAG] monitoring, and that its sensitivity is sufficient to permit observation of the spatiotemporal evolution of [GAG]. Each of those studies began with a homogeneously proteoglycan-deficient matrix, in which GAG was replenished over a 5- to 8-week period to levels considered to be within the normal range.

With respect to the spatial heterogeneity in recovery rates, we are not aware of any histologic (or other) data describing the apparent dependence of the rate of [GAG] replenishment on the initial state of the ECM. We previously showed that spontaneous recovery from complete trypsin-induced GAG loss occurs uniformly throughout the tissue, showing no significant spatial heterogeneity in recovery rates and nearly complete recovery to the initial state in approximately 5 weeks [14]. Those data suggest that the capacity for cells to synthesize new matrix is uniform throughout the tissue. Thus, the differential response seen here is presumably due to the state of the ECM immediately after IL-1 exposure.

Here, we consider [GAG] as a surrogate for defining the ECM state immediately following IL-1 exposure. [GAG] itself is one direct measure of ECM state. In the context of the present study, [GAG] might also serve as a surrogate for the state of other ECM macromolecules, such as collagen. Given the broad spectrum of IL-1-induced enzymes, we consider that regions of the tissue where [GAG] is more severely affected by IL-1 may also be regions that experience greater damage to the collagen network, as compared with regions that are more resistant to the effects of IL-1. Our finding that the regions that experienced the least recovery were those most severely affected by IL-1 could be a consequence of these same regions sustaining more significant damage to the ECM network. This conclusion is consistent with findings regarding IL-1-induced degradation in rabbits, in which recovery rates decreased with apparent severity of degradation [1].

Interestingly, it is in the latter phase of recovery that a discrepancy in recovery rates becomes evident. During the early phases of recovery, the rate of [GAG] accumulation in the regions with the lowest [GAG] was comparable to the rates of recovery in regions with the highest [GAG]. The mechanism for this heterogeneity is unclear. It could be a manifestation of a corresponding distribution of IL-1-induced changes in chondrocyte metabolism or viability [23], or it could be a manifestation of a corresponding distribution of damage to the collagen scaffold, which in turn limits the ability to replenish [GAG]. Kruijsen et al. [24] showed that both the severity and chronicity of antigen-induced inflammation determined the degree of chondrocyte killing in their in vivo murine model of arthritis. Their studies showed that chondrocyte death was most highly correlated with the degree of joint inflammation present 14 days after arthritis induction. That finding suggests that sustained exposure to IL-1, a proinflammatory agent, may also cause chondrocyte death. However, the fact that the early recovery phase in the present study showed no heterogeneity makes cell death a less likely cause of limited [GAG] replenishment.

A limitation (and obvious next step) to this study is that we do not have independent information about the integrity of the collagen matrix. Magnetic resonance imaging techniques are actively being developed to image the collagen component of tissue, which can be incorporated into future studies [25]. Human osteoarthritic tissue is, by histologic measures, spatially heterogeneous in both collagen damage and degree of proteoglycan depletion. The notion that the differential ability to replenish GAG fully is related to the differential state of the collagen matrix has been suggested by others, based in part on the finding that GAG depletion corresponds with regions that are positive for the col3/4 epitope, which is indicative of collagen damage [6,26]. Although we lack specific information on the state of the collagen matrix, the differences observed in the present study (and, indeed, the demonstrated ability to evaluate regional differences in response) suggest that dGEMRIC and model systems such as these may be useful for establishing a better understanding of the capacity of cartilage to repair osteoarthritis-like degradation.

The rate of GAG replenishment observed in this study compares well with published synthesis data. Using sulfate incorporation over periods of less than 24 hours, we and others have reported sulfate incorporation ratios between 0.06 and 0.13nmol/mg wet-weight/hour for young bovine cartilage [27-29]. These incorporation rates imply GAG synthesis rates of 6–13 mg/ml tissue water per week (assuming 1 sulfate per disaccharide, 502 g/mol disaccaride, and 0.8ml tissue water/g wet-weight). By comparison, we observed GAG accumulation rates of 4–14mg/ml per week, as inferred from the change in [GAG]. These values also compare well with the 2–7.3mg/ml per week rates of GAG replenishment seen in young bovine cartilage explants recovering from trypsin-induced GAG depletion [14]. Comparison of these GAG accumulation rates with the rate of GAG release into the culture medium clearly suggests that at least 75% of the newly synthesized GAG is retained by the tissue. (By contrast, in control tissue the amount of GAG synthesized is roughly equivalent to the amount released into the medium.)

We do not have the ability to determine the regional variations in synthesis and loss. Although we clearly observed regional variations in [GAG] accumulation, it is important to appreciate that these differences could arise by regional differences in synthesis or in loss, or both.

Looking more generally at attempts to evaluate [GAG] recovery in IL-1-degraded cartilage, our data are consistent with the temporal progression seen in other model systems in which recovery occurs and is measurable within the first few weeks after a [GAG]-depleting intervention. For example, Takegami et al. [9] reported [GAG] recovery in alginate cultures of human intervertebral disc cells pre-exposed to 0.5ng/ml IL-1 for 3 days. During the first 2 weeks of post-IL-1-exposure culture, those investigators observed [GAG] recovery rates of approximately 4%/day with very little change in [GAG] observed during the third week, when [GAG] levels reached about 85% of the control level. In an in vivo rabbit knee joint subjected to intra-articular injections of IL-1, Page Thomas et al. [3] used SO₄ uptake and toluidine blue staining to observe GAG losses of 25–60% in several cartilage sites within the knee, with gradual recovery over the subsequent 3–4 weeks. Arner [1] also examined in vivo GAG synthesis and accumulation in rabbits following intra-articular injections of IL-1. Using dimethylmethylene blue assay and sulfate incorporation, Arner found that both single and multiple injections of IL-1 led to an initial depression in GAG synthesis rate and a slight drop in tissue [GAG] for 4 days after IL-1 exposure. These changes were followed by enhanced synthesis (relative to controls), with a commensurate increase in tissue [GAG] over the subsequent 2 weeks as the content of GAG in the tissue approached 90% of control levels. In those in vivo systems it appears that GAG loss continues for 4–7 days after IL-1 exposure [1]. This is longer than the 1–2 days seen in the present study (in which GAG loss into the medium returned to control levels), and is probably a consequence of IL-1 clearing more quickly from the in vitro environment.

Thus, in widely different model systems – including that reported here – it appears that after an IL-1-induced GAG-depleting intervention tissue can reaccumulate GAG and does so most rapidly during the first few weeks. However, unlike the study described here, in which both spatial and temporal changes in [GAG] were monitored, in the studies described above it was not possible to derive more specific spatial information because the destructive nature of the [GAG] measurements required that time course information be inferred from averages of separate animals/samples harvested at different time points.