To study the relationship between the boundary-lubricating ability of synovial fluid (SF) and articular cartilage damage in a rabbit knee injury model, to correlate collagen markers of such damage with SF boundary-lubricating ability and elastase activity, and to examine the lubricating ability of SF, together with collagen markers of articular cartilage damage, under the inflammatory conditions of knee joint synovitis (KJS) and rheumatoid arthritis (RA).
SF was aspirated weekly from the affected knee joints of 10 adult rabbits following transection of the anterior and posterior cruciate ligaments. The boundary-lubricating ability of SF was determined in vitro using a previously described friction apparatus. Lubricin concentrations and type II collagen (CII) peptides were quantified by sandwich enzyme-linked immunosorbent assays (ELISAs). Levels of the C-terminal neoepitope 9A4 (derived from collagenase degradation of CI, CII, and CIII) and of epitope 5-D-4 of keratan sulfate (a marker of proteoglycan depletion) were quantified by inhibition ELISAs. Elastase activity was measured spectrophotometrically. The sensitivity of purified human lubricin to digestion by neutrophil elastase (NE) was examined by Western blotting.
The lubricating ability of SF from injured rabbit knees was significantly decreased at weeks 2 and 3 compared with week 1 after injury. Lubricin concentrations were significantly higher at week 1 than at weeks 2 and 3. CII peptide concentrations increased significantly at weeks 2 and 3 compared with week 1, while 9A4 neoepitope concentrations increased significantly at week 3 compared with weeks 1 and 2. There were no significant differences in epitope 5-D-4 concentrations among the 3 weeks. Elastase activity in SF increased significantly at weeks 2 and 3 compared with week 1. Elastase activity correlated significantly with diminishing lubrication at weeks 1, 2, and 3. SF from patients with KJS or RA exhibited deficient lubrication and elevated levels of CII peptides compared with SF from normal controls. NE was shown to completely degrade purified human lubricin in vitro.
Loss of boundary-lubricating ability of SF after injury is associated with damage to the articular cartilage matrix. This can be attributed to inflammatory processes resulting from the injury, particularly in the early phases. This association also exists in patients with acute knee injuries or progressive chronic inflammatory arthritis.
Boundary lubrication by synovial fluid (SF) refers to the ability of SF to reduce friction between apposed and pressurized cartilaginous surfaces independently of viscosity. Lubricin (∼227.5 kd), encoded by the gene PRG4, is a heavily glycosylated (50% weight/weight) mucinous glycoprotein secreted by synovial fibroblasts (1). It is considered the factor responsible for boundary lubrication of diarthrodial joints (2–4). Superficial-zone protein (SZP; ∼345 kd), another PRG4 gene product (1, 5), is expressed by chondrocytes located in the superficial zone of articular cartilage (6, 7). SZP is a component of the lamina splendens. The carbohydrate moieties of lubricin, and by extrapolation those of SZP, are considered to play an important role in lubrication, since glycosidase digestion of lubricin causes loss of boundary lubrication (3, 8).
Individuals with a history of knee joint injury are more likely to develop osteoarthritis (OA) than are those without a history of knee injury (9, 10). We recently reported that early signs of articular cartilage degradation, as indicated by type II collagen (CII) peptide release, appear in the SF from patients following knee injury (11). Presumably, the extent of articular damage depends, among other factors, on the friction and wear between the cartilaginous surfaces in the joint. Compromising boundary lubrication will cause increased friction, load amplification, and ultimately, greater cartilage damage.
We used the rabbit knee injury model, as described by Hulth et al (12), to determine the relationship between the boundary-lubricating ability of SF and articular cartilage damage. In vitro boundary-lubricating ability of SF, lubricin concentration, markers of cartilage degradation, and elastase activity in SF were measured, and correlations among these factors were determined. In particular, the in vitro lubricating ability was compared with the lubricin concentration as determined by a sandwich enzyme-linked immunosorbent assay (ELISA). The chosen epitopes reside in the functional elements of a boundary lubricant. The boundary-lubricating ability of SF was further examined in 2 populations of patients with inflammatory knee joint conditions: acute knee joint synovitis (KJS) and chronic inflammatory rheumatoid arthritis (RA). The boundary lubrication provided by SF under these 2 conditions was compared, as were collagen markers of articular cartilage degradation. The sensitivity of lubricin to neutrophil elastase (NE) was determined using Western blotting to analyze the time-dependent digestion of lubricin.
MATERIALS AND METHODS
Rabbit injury model.
The rabbit injury model was an adaptation (13) of that described by Hulth et al (12). A total of 10 NZW rabbits were used for this study. Institutional Animal Care and Use Committee approvals were obtained prior to performing the surgery. The right hind knee of each rabbit was subjected to anterior and posterior cruciate ligament transection. The left hind knee was subjected to a sham operation. Each rabbit underwent operative knee joint aspirations, under aseptic conditions, at 1-week intervals for 3 weeks. The presence of posttraumatic effusions obviated the need for lavage with saline. Approximately 600 μl of SF was collected from each rabbit at weekly intervals. The SF was immediately stored at –20°C. Lavage fluid was also aspirated from the sham-operated knee at weekly intervals for 3 weeks. To do this, a total of 1.0 ml of sterile saline was injected into the knee joint. The joint was mobilized through 10 full-arc flexion/extension ranges of motion, and the lavage fluid was aspirated and stored at –20°C. Normal rabbit SF (30 μl) was obtained postmortem by aspirating the knees of 8 NZW rabbits.
Boundary-lubricating ability of SF.
The boundary-lubricating ability of SF was measured using a friction apparatus, as reported by Davis et al (14). The apparatus is shown in Figure 1. Lubricant (200 μl) was applied between a bearing of latex and a ring of polished glass with a contact area of 1.59 cm2. Under a pressure of 0.35 × 106 N/m2, the latex was oscillated against the polished glass with an entraining velocity of 0.37 mm/second. The bearing system was axially loaded within a gimbal system free to rotate around 2 perpendicular horizontal axes. The friction apparatus recorded displacements of the gimbal system around the vertical loading axis through a linear voltage displacement transducer, in which the output was directly proportional to the frictional torque (F). A bearing load of 70N was related to the coefficient of friction (μ) via Amonton's law, F = μN.
The μ value of SF was recorded at room temperature and was preceded by a baseline measurement of the μ value of normal saline. Lubrication was manifested by a reduction in the μ value of SF relative to the μ value of normal saline. Negative Δμ values indicate lubrication, while positive Δμ values indicate friction. Addition of 200 μl of lubricant was followed by bringing the bearing surfaces close enough so that the solution wet both surfaces. After 5 minutes for equilibration, the latex-coated bearing was brought to rest on the glass as it was oscillating. Voltage measurements were recorded at 1, 3, and 5 minutes. At this point, the surfaces were separated for 2 minutes and then brought back together for 3 additional 5-minute sessions. The 3- and 5-minute μ values of the last 2 5-minute sessions were recorded and averaged, and the data were later combined with those from another duplicate experiment, providing 8 distinct measurements of the μ value.
SF from patients with KJS and RA.
SF was obtained from 67 emergency department patients who were diagnosed as having KJS (15). Inclusion criteria for patients with KJS were monarticular knee joint effusions judged to be clinically significant and requiring arthrocentesis in the absence of radiographic evidence of fracture. These patients, who were treated by emergency physicians, had reported joint swelling (either following mild trauma or for reasons they could not recall). Exclusion criteria included a diagnosis of occult fracture, history of sickle cell anemia or hemophilia, patellar bursitis, gout, pseudogout, and samples contaminated with blood. Samples with any evidence of infectious synovitis following culture for 48 hours were also excluded. SF from 45 patients diagnosed as having RA was obtained during routine rheumatologic evaluation. Samples of normal human SF were obtained from human donors undergoing allograft cartilage transplant surgery and from subjects postmortem. These subjects had no evidence of gross degenerative joint disease, and their samples were provided as a gift from Dr. Martin Lotz (Scripps Research Institute, La Jolla, CA). Human subject research approvals were obtained from the Institutional Review Boards at Rhode Island Hospital, Providence, RI, and Scripps Research Institute.
Sandwich ELISA for lubricin.
Lubricin was detected using a sensitive rabbit polyclonal antibody (pAb) J108 generated against a polypeptide epitope (FESFERGRECDAQCKKYDK). This epitope, encoded by PRG4 exon 3, is in the amino terminus of lubricin, and it is present in all lubricin isoforms (1). Microtiter plates (Corning, Corning, NY) were coated overnight at 4°C with pAb J108 at a 1:10,000 dilution in 10 mM sodium carbonate buffer, pH 9.0. Each well received 100 μl. The plate was subsequently washed twice with phosphate buffered saline (PBS)–0.05% Tween 20, 250 μl/well. The plate was then incubated with 3% bovine serum albumin (BSA; Sigma-Aldrich, St. Louis, MO) in carbonate buffer, pH 9.0 (200 μl/well) for 2 hours at room temperature.
Human lubricin, purified by a method described previously (1) from SF obtained from patients undergoing total knee replacement, was used to construct the standard curve. Serial dilutions of human lubricin (5–0.025 μg/ml), normal rabbit SF, and SF from injured rabbit knees (100 μl/well) were incubated at room temperature for 1 hour. The plate was subsequently washed twice with PBS–0.05% Tween 20, 250 μl/well. The plate was then incubated with peanut agglutinin (PNA; from Arachis hypogaea) conjugated to peroxidase (Sigma-Aldrich) at a concentration of 0.25 μg/ml in PBS/1% BSA, 100 μl/well, for 1 hour at room temperature. The plate was washed twice with PBS–0.05% Tween 20, 250 μl, and twice with PBS, 100 μl/well. Fluorogenic substrate developing solution (100 μl) (Quanta Blu; Pierce, Rockford, IL) was added for 60 minutes, and the fluorescence was measured at an excitation wavelength of 330 nm and an emission wavelength of 460 nm using a Packard fluorocounter (Packard, Meriden, CT).
Western blot analysis of lubricin content in SF aspirated from injured rabbit knees.
Aspirated SF from injured rabbit knees was electrophoresed in equal volumes (15 μl) on precast sodium dodecyl sulfate 4–15% polyacrylamide gels (Bio-Rad, Hercules, CA) under reducing conditions. Normal rabbit SF (5 μl) as well as high molecular weight standards (Gibco BRL, Gaithersburg, MD) were electrophoresed simultaneously. Electrophoresis was performed at 150V for 90 minutes until the wave front exited from the bottom of the gel. Western transfer of gel components to nitrocellulose was carried out under semidry conditions at 20V for 40 minutes (Bio-Rad). The blot was blocked overnight at 4°C with 2% BSA in PBS, washed with PBS–2% Tween 20, and probed with PNA–peroxidase at a concentration of 5 μg/ml in PBS–2% Tween 20 for 60 minutes at room temperature. Following extensive washing with PBS–2% Tween 20 and PBS, chemiluminescent substrate (Pierce) was added, and immunopositive bands on BioMax film (Eastman Kodak, Rochester, NY) were detected in a darkroom.
Sandwich ELISA for CII peptides.
Two monoclonal antibodies against CII, 18:6:D6 and 14:7:D8, were used in this assay (16). Antibody 18:6:D6 was generated against CII CNBr cleavage peptide 9.7 coupled to ovalbumin. Antibody 14:7:D8 was generated using a 15–amino acid synthetic peptide (GPQGPRGDKGEAGEP) coupled to keyhole limpet hemocyanin and was conjugated to horseradish peroxidase using the EZ-Link Maleimide Activated Horseradish Peroxidase kit (Pierce). Antibody 14:7:D8 reacts with an amino acid sequence in CII as well as with types I, III, and V collagen (16). The assay was developed and performed as previously described (11).
Inhibition ELISA for 9A4 epitope.
Antibody 9A4 was kindly provided by Dr. Ivan Otterness, Pfizer Central Research, Groton, CT. Antibody 9A4 detects a C-terminal neoepitope derived from collagenase degradation of fibrillar types I, II, and III collagen (17). A proline-rich 9–amino acid peptide (AEGPPGPQG) to which antibody 9A4 has shown a high binding affinity (17) was synthesized using 9-fluorenylmethoxycarbonyl chemistry and used at a final concentration of 500 ng/ml in 50 mM sodium carbonate buffer, pH 9.5, to coat microtiter plates (overnight at 4°C). The assay was developed and performed as previously described (11).
Inhibition ELISA for epitope 5-D-4 of keratan sulfate.
Microtiter plates were coated overnight at 4°C with chondroitinase ABC–treated (18) bovine aggrecan monomer A1D1 to a final concentration of 1 μg/ml in 50 mM sodium carbonate buffer, pH 9.5. The assay was developed and performed as previously described (11).
Elastase activity in SF.
Hydrolysis of a low molecular weight synthetic substrate, N-succinyl trialanine p-nitroanilide (SANA; Sigma-Aldrich), which releases p-nitroaniline, was measured spectrophotometrically at 410 nm (19) and used as a measure of SF elastase activity. Pancreatic porcine elastase type I (6.8 units/mg; Sigma-Aldrich) was used to construct a standard curve. One unit of the enzyme is defined as 1.0 μmole of SANA hydrolyzed per minute at 25°C. Elastase activity of SF samples was expressed as microunits per milliliter. To inhibit metal-dependent elastase activity in the SF, samples were preincubated with 10 mM EDTA solution for 3 hours at 37°C, and elastase activity was measured as previously described (11).
Total protein levels.
Total protein concentrations in SF samples were determined colorimetrically using the Micro BCA protein assay reagent kit (Pierce) as previously described (11).
Lubricin digestion by NE.
Purified human lubricin at a final concentration of 50 μg/ml in 100 mM Tris HCl, 100 mM CaCl2, pH 8.8, was incubated at 37°C with human NE (Sigma-Aldrich) at a final concentration of 0.5 units/ml. One unit is defined as the release of 1 nmole of p-nitrophenol per second at 37°C. The reaction mixture was subsequently sampled at 30 minutes and at 1, 2, 4, 6, 12, and 24 hours, and the reaction was stopped by adding phenylmethylsulfonyl fluoride (Sigma-Aldrich) at a final concentration of 1 mM. Lubricin incubated in the absence of elastase served as a negative control. Electrophoresis and Western transfer were performed as described above. Probing with pAb J108 was conducted at a 1:5,000 dilution in PBS–2% Tween 20 for 60 minutes at room temperature. Following washing with PBS–2% Tween 20, peroxidase-linked goat anti-rabbit IgG was added at a dilution of 1:10,000 for 60 minutes. Following extensive washing with PBS–2% Tween 20 and PBS, chemiluminescent substrate was added. Immunopositive bands were detected as described above. Probing with PNA–peroxidase was performed in a manner similar to that described above.
The SF μ values, the concentrations of lubricin, CII peptides, 5-D-4 epitope, and 9A4 epitope, and the levels of elastase activity were represented by scatterplot graphs. Student's unpaired 2-sample t-test was used to test for differences in boundary-lubricating ability, lubricin concentrations, and markers of cartilage damage in aspirated SF following injury. Boundary-lubricating abilities of SF from patients with different inflammatory conditions were represented by box plots. The horizontal line within each box represents the median, the box represents the interquartile range, the error bars represent the 10th and 90th percentiles, and the individual solid circles represent outlying values. Student's t-test was used to detect significant differences. Correlations between elastase activity, boundary lubrication, and SF levels of lubricin were performed using Spearman's correlation.
Rabbit injury model.
Boundary-lubricating ability of SF in vitro.
The boundary-lubricating ability of SF from injured rabbit knees, as indicated by the SF μ values, Δμ values, and lubricin concentrations, is shown in Table 1 and Figure 2. Mean μ values shown in Table 1 were calculated from 2 separate experiments, each with 4 distinct measurements of μ. The recorded μ values of normal saline (used as controls in the lubrication experiments) exhibited less variation compared with recorded μ values of SF. The variation in Δμ values was due to the variation in SF μ values and not normal saline μ values.
Table 1. Coefficients of friction (μ values) of synovial fluid (SF) aspirated from injured rabbit knees 1 week after injury and recorded μ values of normal saline (NS) used as controls*
There was a significant increase in the μ value of SF at week 2 (t = 3.39, P = 0.003 versus week 1) and week 3 (t = 4.54, P < 0.001 versus week 1), with no significant difference between weeks 2 and 3 (t = 0.28, P = 0.786) (Figure 2A). The majority of the SF aspirates from the injured knee retained their ability to lubricate, as indicated by a negative Δμ value. A total of 10% of the SF samples were nonlubricating at week 1. At weeks 2 and 3, 70% and 80% of samples, respectively, demonstrated declining lubricating ability compared with week 1. The Δμ value of SF from the injured rabbit knees significantly increased at week 2 (t = 2.89, P = 0.01 versus week 1) and week 3 (t = 4.61, P < 0.001 versus week 1), with no significant difference in SF lubrication between weeks 2 and 3 (t = 0.56, P = 0.561) (Figure 2B). The detection range of the lubricin sandwich ELISA was 5–0.05 μg/ml. Lubricin concentrations were significantly higher at week 1 compared with weeks 2 (t = 3.22, P = 0.006) and 3 (t = 3.57, P = 0.004), with no significant difference between levels at weeks 2 and 3 (t = 0.11, P = 0.920) (Figure 2C). Analysis of SF aspirated from the excised joints of normal rabbits revealed a mean ± SD lubricin concentration of 280 ± 25 μg/ml. A significant correlation existed between μ values and lubricin concentrations in SF from injured rabbit knees at weeks 1, 2, and 3 after injury (R2 = 0.524, P = 0.003) (analysis not shown). The boundary-lubricating ability of saline lavage fluid from the sham-operated knee was not assayed due to dilution of lubricin, which could have falsely elevated the μ values.
Western blot analysis of rabbit SF lubricin content was qualitatively performed using PNA–peroxidase. Under conditions of sodium dodecyl sulfate–polyacrylamide gel electrophoresis and Western transfer, SF aspirated from rabbits 1, 3, 5, 6, and 7 showed immunopositive bands with PNA, while SF aspirated from rabbits 2, 4, 8, 9, and 10 had no reactivity. A representative decrease in detected lubricin at weeks 2 and 3 compared with week 1 is illustrated in Figure 2D. Levels of lubricin in some animals were too low to be detectable by PNA Western blotting, or changes over time could not be determined using PNA as a primary probe. A polyclonal antibody was not useful in these Western blots because the anti-rabbit IgG secondary antibody caused high background by detecting endogenous IgG molecules.
Markers of articular cartilage degradation.
CII peptide detection by sandwich ELISA. The detection range of the CII sandwich ELISA was 2.5–0.005 μg/ml. At week 1 after injury, a total of 70% of the SF from the injured rabbit knees had detectable quantities of CII peptides by sandwich ELISA compared with 90% at week 2 and 100% at week 3 (Figure 3A). There were no detectable quantities of CII peptides in the lavage fluid aspirated from the sham-operated knees or in the SF aspirated from the knee joints of normal rabbits. The CII peptide levels showed significant elevations at week 2 (t = 2.07, P = 0.035 versus week 1) and week 3 (t = 3.79, P < 0.001 versus week 1). There was no significant difference in the levels of CII peptides between weeks 2 and 3 (t = 1.79, P = 0.061). When CII peptide concentrations were normalized to total protein concentrations (data not shown), the significant elevations of CII peptide concentrations at week 2 (t = 2.25, P = 0.027 versus week 1) and week 3 (t = 3.88, P = 0.001 versus week 1) were still apparent.
Epitope 9A4. The detection range of the 9A4 inhibition ELISA was 2–0.02 μg/ml. At week 1 after injury, a total of 90% of the SF from the injured rabbit knees had detectable quantities of 9A4 epitope compared with 100% at both week 2 and week 3 (Figure 3B). There were no detectable quantities of 9A4 epitope in the lavage fluid aspirated from the sham-operated knees. The 9A4 epitope levels were significantly elevated at week 2 (t = 2.834, P = 0.007 versus week 1). The 9A4 epitope levels at week 3 were significantly higher than those at week 1 (t = 3.94, P = 0.001) and week 2 (t = 3.36, P = 0.003).
Epitope 5-D-4 of keratan sulfate. The detection range of the 5-D-4 inhibition ELISA was 2–0.1 μg/ml. At week 1 after injury, a total of 90% of the SF from the injured rabbit knees had detectable levels of 5-D-4 epitope compared with 100% at both week 2 and week 3 (Figure 3C). There were no detectable levels of 5-D-4 epitope in the lavage fluid aspirated from the sham-operated knees. There were no significant differences in the levels of 5-D-4 epitope among the 3 weeks after injury.
The detection range of elastase activity was 1–0.05 μunits/ml. At week 1 after injury, a total of 70% of the SF from the injured rabbit knees had detectable levels of elastase activity compared with 80% at both week 2 and week 3 (Figure 4). There was a significant increase in elastase activity at week 2 (t = 4.24, P = 0.001 versus week 1) and week 3 (t = 3.29, P = 0.002 versus week 1). However, there was no significant difference in elastase activity between weeks 2 and 3 (t = 1.67, P = 0.061). There was no detectable elastase activity in the lavage fluid aspirated from the sham-operated knees.
To further investigate the source of elastase activity in the SF samples, EDTA was added to abolish metal-dependent elastase activity (Figure 4). The mean inhibition of elastase activity by EDTA was calculated to be 76% at week 1, 45% at week 2, and 24% at week 3.
Correlations of elastase activity with boundary-lubricating ability and markers of collagen release in the SF from the injured rabbit knees were calculated to assess whether boundary lubrication and articular cartilage damage might be related. There was a significant correlation between elastase activity and diminishing boundary lubrication at weeks 1, 2, and 3 after injury (R2 = 0.469, P = 0.023) (analysis not shown). Furthermore, there was a significant correlation between elastase activity and 9A4 epitope levels at weeks 1, 2, and 3 after injury (R2 = 0.593, P = 0.005) (analysis not shown).
SF from human knee joints with acute or chronic inflammation.
Boundary-lubricating ability of human SF in vitro.
The boundary-lubricating ability of SF from KJS and RA patients and cartilage transplant donors (normal controls) is presented in Figure 5. The RA SF samples failed to lubricate, as shown by a positive Δμ value across all patients. The SF from patients with KJS exhibited a large variation in lubricating ability compared with SF from normal controls. Overall, the Δμ value of KJS SF was significantly higher than that of SF from normal controls (t = 4.23, P < 0.001). The Δμ value of RA SF was significantly higher than the Δμ values of SF from patients with KJS (t = 4.05, P < 0.001) or from normal controls (t = 6.06, P < 0.001).
CII peptide detection by sandwich ELISA. A total of 77% of KJS SF samples had detectable quantities of CII peptides, compared with 95% of RA SF samples (data not shown). The CII peptides in SF from patients with KJS showed a wide range of concentrations (0.07–3.37 μg/ml) and a mean concentration of 0.94 μg/ml (95% confidence interval [95% CI] 0.86–1.20), while RA SF aspirates had a mean CII peptide concentration of 0.17 μg/ml (95% CI 0.14–0.19). SF obtained postmortem from normal subjects had significantly lower CII peptide concentrations (<50 ng/ml) (11) compared with SF aspirates from patients with KJS or RA. The CII peptide concentration showed a significant elevation in KJS SF compared with RA SF (t = 4.95, P < 0.001) (data not shown). When CII peptide concentrations were normalized to total protein levels, the significant elevation in KJS CII peptide concentrations compared with RA CII peptide concentrations (P < 0.01) was still apparent.
Lubricin degradation by human NE.
Lubricin was degraded in a time-dependent manner by NE, as revealed by diminishing intensity of an ∼227-kd band compared with control (Figure 6). Lubricin was stable at 37°C, as evidenced by a strong immunopositive band after incubation without NE for 24 hours. Western blotting and probing with pAb J108 and PNA–peroxidase produced similar results; however, loss of the J108 epitope appeared to occur before complete loss of the protein's central mucin-like domain. This is consistent with NE cleaving lubricin in the protein's terminal globular domains, which are outside the central, highly O-linked glycosylated, mucin-like domain.
In this study, we examined the temporal pattern of changes in boundary-lubricating ability of SF and in markers of cartilage degradation in rabbit knees following a surgically induced injury. The early biochemical changes that may precipitate OA secondary to acute knee injury are not entirely understood. We hypothesized that subsequent to injury, damage to the articular cartilage and loss of boundary-lubricating ability are early interrelated events. The loss of boundary lubrication by SF following injury would contribute to premature wear as a result of increased friction. Previous studies of either in vitro or in vivo cartilage injury models have established cartilage damage to be a consequence of excessive loading (20–24). Investigators of these models of injury did not examine changes to the boundary-lubricating ability of SF or its role in mitigating cartilage damage. Given the important chondroprotective properties that SF confers on articular cartilage, it is essential to understand the effects of acute injury on boundary lubrication.
The boundary-lubricating ability of SF was determined in an artificial bearing system under load, which isolated boundary lubricant conditions. This system has the advantage of reproducibility compared with cartilage-upon-cartilage bearing, and it mimics the results of this bearing, although μ values were higher overall (14). The lubricating ability of rabbit SF diminished after injury, as shown by the elevation in μ values at weeks 2 and 3 after injury. When measured by ELISA, lubricin concentrations in SF at weeks 2 and 3 after injury decreased compared with those at week 1 and compared with those in normal rabbit SF. Furthermore, μ values in SF correlated significantly with the lubricin concentration determined by sandwich ELISA.
Deterioration of the boundary-lubricating ability of SF could result from decreased lubricin synthesis or increased lubricin degradation. Support for the hypothesis of increased degradation of lubricin in the injured joint derives from measures of elastase and matrix metalloproteinase (MMP) activity in the SF. NE, the levels of which are elevated in inflammation, can degrade purified human lubricin, as demonstrated by the diminishing immunopositive bands of human lubricin following incubation with NE. This observation is supported by previous studies that established the sensitivity of SZP to degradation by NE (25).
Digestion of the N-terminal domain prior to the mucin-like domain, as evidenced in Figure 6, supports the utility of a sandwich ELISA. The assay of either epitope alone would be insensitive or could falsely show the complete digestion of lubricin. This may have occurred in our Western blot assays of SF aspirated from rabbits 2, 4, 8, 9, and 10, none of which showed reactivity, but all of which continued to lubricate at week 2. Similarly, SF from rabbit 1 continued to lubricate at week 2, but lubricin was undetectable by Western blot analysis using PNA–peroxidase (Figure 2D).
In addition to changes in boundary lubrication, we also found evidence of damage to the articular cartilage matrix following injury, as assessed by the release of CII peptides and the creation of neoepitope 9A4. The release of CII peptides in the SF indicates that CII degradation is an early event in damage of the matrix. Elevated 9A4 epitope levels point to active metalloproteinase activity following injury (26). Epitope 9A4 is produced when CI, CII, and CIII are degraded by the collagenase subfamily of metalloproteinases (MMPs) (26). Increased expression of collagenases has been reported following injury (27) and in traumatized synovial membranes (28). The SF from injured rabbit knees contained elevated levels of epitope 5-D-4 of keratan sulfate compared with lavage fluid from the sham-operated knees, providing corroborating evidence of the damage to articular cartilage, which may also be indicative of severity. Released CII peptides first increased and then decreased over weeks 1–3, while levels of proteoglycans (as detected via epitope 5-D-4 of keratan sulfate) did not change significantly, suggesting that the damage to articular cartilage in this model is perhaps superficial, but could evolve over time into arthritic lesions clinically consistent with OA of traumatic origin (10).
Elastase activity due to NE was compared with elastase activity due to metal-dependent metalloproteinases to further understand the biochemical factors that lead to deterioration of SF boundary-lubricating ability and matrix damage in the injury model. The significant and early contribution of metal-dependent elastase activity to the total elastase activity, represented by significant lowering of elastase activity with EDTA pretreatment (19), indicates a contribution by MMPs to the damage of the collagen matrix. This role was supported by the finding of elevated 9A4 epitope levels in the SF following injury. Purified NE has been shown to damage cartilage explants in vitro (29). NE, and not MMPs, can destroy the superficial layer of cartilage, which leads to MMPs having better access to CII molecules in less superficial layers of cartilage (30). We speculate that the convergent effects of NE and MMPs lead to the damage of chondroprotective mechanisms provided by lubricin, which then leads to damage of the superficial zone of cartilage and release of CII peptides into the SF following injury.
The significant correlation between elastase activity and diminishing boundary lubrication (increasing μ values) supports the notion of a causal relationship between inflammation and proteolytic degradation of lubricin, leading to damage to articular cartilage. In the later stages of injury, other factors, such as joint utilization and loss of interarticular surface congruence, contribute to the extent of damage, which may in turn complicate our ability to understand these cascading catabolic events.
The KJS patient population represents a clinical extension of the rabbit injury model. These patients have experienced a knee insult as a result of a blunt injury, a sprain, or a strain, or they may have had subclinical trauma. These patients had an elevated SF nucleated cell count indicative of active inflammation, which correlated with SF boundary-lubricating ability and was also associated with CII peptide release (15). At the opposite end of this spectrum is the population of patients with RA (representing progressive chronic inflammatory arthritis), in whom SF boundary-lubricating ability was completely lost and articular cartilage damage was evident. These findings indicate that the deterioration of SF boundary-lubricating ability (and its possible association with matrix damage) is not a unique feature of early stages of a knee injury, but is instead a common feature in inflammatory conditions, both acute and chronic. Furthermore, wear-induced damage, which adheres to a classic tribologic model, may also occur, but it probably does not adequately explain CII peptide release.
The results of this study delineate early changes to the boundary-lubricating ability of SF and to articular cartilage integrity following injury. The loss of boundary-lubricating ability of SF and the release of CII peptides into SF are both early and distinct events following injury. NE released from infiltrating nucleated cells and MMPs released from the articular matrix appear to be associated with these early changes in boundary lubrication by SF and release of CII peptides. In this model, we could not isolate the wear effects and pattern of cartilage degradation created by the loss of boundary lubrication. Injury models in general display contemporaneous inflammation, which can affect many extracellular matrix proteins digested by a myriad of enzymes. However, our efforts in this study were restricted to NE and MMPs.
A sandwich ELISA proved to be a sensitive and specific method for determining lubricin concentration. Rational epitope selection was directed toward a non–alternatively spliced N-terminal region and O-linked glycosylations. These moieties play roles in molecular adhesion to a surface and in repulsion force generation, respectively. Sandwich ELISA of these posited functional motifs in lubricin correlated highly with results of mechanical assays across SF samples with varying lubricin concentrations.
Another limitation of this study was the lack of normal rabbit SF from the contralateral control joints to establish normal lubricating ability. Scant SF aspirates from control joints also necessitated lavaging of the joint cavity with the resulting dilution of its components of interest, which could have rendered them undetectable. However, detection limits for the assays were established by internal controls, by our previous assays of normal human SF, and by retesting on normal rabbit SF from excised limbs. Given the high lubricin concentration by sandwich ELISA, normal rabbit SF should exhibit efficient lubricating ability, better than the SF from the injured rabbit knees.
We undertook the present study to clarify and develop our recent findings concerning traumatic synovial effusions observed in patients in an emergency department (15). The clinical population of patients in emergency departments has not previously been a focus of rheumatologic and orthopedic investigation, and these patients frequently do not receive subspecialty followup. Further study is needed to determine the long-term consequences of trauma resulting in monarticular knee joint effusions judged to be clinically significant and requiring arthrocentesis in the absence of radiographic evidence of fracture. Support for a classic lubrication and wear (tribologic) explanation of OA will necessarily await studies using animal models in which endogenous lubricin levels can be manipulated and/or in which purified lubricin can be therapeutically administered.
The authors thank Dr. Hans Barrach of Rhode Island Hospital for his assistance with the animal model and for providing RA SF samples. The authors thank Dr. Martin Lotz of Scripps Research Institute for providing normal human SF and Dr. Ivan Otterness of Pfizer Central Research for providing antibody 9A4.