Specific antibody protection of the extracellular cartilage matrix against collagen antibody–induced damage




The type II collagen (CII)–specific monoclonal antibodies (mAb) M2139 and CIIC1 induce arthritis in vivo and degrade bovine cartilage explants in vitro, whereas mAb CIIF4 is nonarthritogenic and prevents arthritis development when given in combination with M2139 and CIIC1. To determine the nature of the protective capacity of CIIF4 antibody, we examined the effects of adding CIIF4 to cartilage explants cultured in vitro with M2139 and CIIC1.


Bovine cartilage explants were cultured in the presence of M2139 and CIIC1, with or without CIIF4. Histologic changes were examined, and chemical changes to collagens and proteoglycans were assessed by Fourier transform infrared microspectroscopy (FTIRM). Fresh cartilage and cartilage that had been freeze-thawed to kill chondrocytes cultured with or without the addition of GM6001, a broad-spectrum inhibitor of matrix metalloproteinases (MMPs), were compared using FTIRM analysis.


M2139 and CIIC1 caused progressive degradation of the cartilage surface and loss of CII, even in the absence of viable chondrocytes. CIIF4 did not cause cartilage damage, and when given with the arthritogenic mAb, it prevented their damage and permitted matrix regeneration, a process that required viable chondrocytes. Inhibition of MMP activity reduced cartilage damage but did not mimic the effects of CIIF4.


CII-reactive antibodies can cause cartilage damage or can be protective in vivo and in vitro, depending on their epitope specificity. Since CII antibodies of similar specificity also occur in rheumatoid arthritis in humans, more detailed studies should unravel the regulatory mechanisms operating at the effector level of arthritis pathogenesis.

Rheumatoid arthritis (RA) is a chronic autoimmune inflammatory disease that affects multiple joints, leading to articular damage and bone destruction (1). There is increasing evidence that B cells and secreted antibodies may play an important role in RA, and depleting B cells by use of monoclonal antibodies (mAb) to CD20 (rituximab) ameliorates RA. Since autoantibody levels are not completely depleted, even in RA patients undergoing B cell depletion therapy with rituximab (2), understanding the regulatory mechanisms at the effector level becomes all the more important in designing better treatments for RA patients.

An autoimmune response to native type II collagen (CII) develops in some patients, with IgG antibodies to CII demonstrable in the blood, cartilage, and synovium (3). In collagen-induced arthritis (CIA), which is induced by immunization of animals with CII (4, 5), there is destruction of the articular cartilage matrix as occurs in RA in humans, with accompanying T cell and B cell immune responses to CII that are seen as requisite for disease initiation and development (6). Certain anti-CII mAb produced from mice susceptible to CIA can, on passive transfer to naive mice, cause an acute arthritis known as collagen antibody–induced arthritis (CAIA) (7–10). Since CAIA occurs independently of the direct activity of B cells and T cells, it allows for the study of effector processes without consideration of events in the inductive phase (8).

The articular inflammation that characterizes the effector stage is customarily attributed to the formation and deposition of immune complexes and the activation of complement and Fc receptors (FcR) (11). Interestingly, cleavage of arthritogenic mAb at the hinge region in vivo (12) or selective removal of carbohydrate moieties in the CH2 domain in vitro (13) were shown to completely abrogate the development of antibody-mediated arthritis. Not all CII-specific antibodies are equally arthritogenic, however. CIIF4 mAb is inert after passive transfer and, indeed, is even protective against articular damage in vivo when transferred together with a combination of the normally potently arthritogenic mAb M2139 and CIIC1 (14), but the mechanism of this protection has not yet been identified.

Using in vitro systems based on cartilage explant cultures, we have previously shown that mAb M2139 and CIIC1, which are arthritogenic in vivo, adversely affect cartilage matrix integrity, whereas the nonarthritogenic mAb CIIF4 has no effect (15). In the present study, we extended these observations to show that in vitro, CIIF4 is counterdestructive/protective against the degradative effects of these arthritogenic mAb when used in combination. Notably, in the culture system we used, cartilage damage by antibody occurs independently of immune cells or their small-molecule mediators. Our enquiries were directed toward the mechanism(s) whereby the physical interaction of antibody with a collagen epitope is transduced to matrix degradation, and how mAb CIIF4 could interfere with such processes. We addressed these questions by assessing whether there was a requirement for living chondrocytes in the cultures for these effects of the CIIF4 mAb and whether attachment of CIIF4 to its epitope site might sterically interfere with the binding of matrix metalloproteinase 3 (MMP-3; stromelysin 1) at its catalytic site on CII.


Monoclonal antibodies.

The mAb we used were derived from hybridomas developed from CII-immunized mice, as described previously (10). All of the mAb were affinity purified from culture supernatants using γ-bind plus columns. M2139 and CIIC1 are arthritogenic mAb that bind to separate well-defined conformational epitopes of native CII: the J1 epitope (MPGERGAAGIAGPK; amino acids [aa] 551–564) and the C1 epitope (ARGLTGRPGDA; aa 359–369). CIIF4 is a nonarthritogenic mAb that binds to the conformational F4 epitope (ERGLKGHRGFT; aa 926–936) (16).

Analysis of the in vitro effects of monoclonal antibodies with the use of bovine cartilage explants.

Articular cartilage samples extracted from adult bovine metacarpophalangeal joints and cartilage shavings (5 × 5 × 1 mm) were cultured for up to 14 days in Dulbecco's modified Eagle's medium plus 20% (v/v) fetal calf serum and 25 μg/ml ascorbic acid, with the appropriate concentrations (25 μg/ml, 50 μg/ml, or 100 μg/ml) of mAb or in medium alone (15). Medium was changed every 2 days, and fresh ascorbic acid and mAb were added at each change. Cartilage samples were tested in duplicate, and all experiments were performed at least twice.

On selected days, cartilage explants were harvested, fixed in 4% paraformaldehyde, and embedded in paraffin (oriented to allow sectioning across the full depth of the cartilage) for histologic analysis and Fourier transform infrared microspectroscopy (FTIRM) or were processed for indirect immunofluorescence to measure antibody penetration (15, 17). For histologic examination, 5-μm sections of tissue were stained with hematoxylin and eosin and with toluidine blue to examine proteoglycan loss.

Analysis of changes in the chemical composition of the cartilage by FTIRM.

Sections of paraffin-embedded tissue (5 μm) were placed onto MirrIR low-e microscope slides (Kevley Technologies), and adjacent sections were stained with toluidine blue. FTIR images were recorded with a Stingray Digilab FTS 7000 series spectrometer coupled to a UMA 600 microscope equipped with a 64 × 64 focal plane array detector or were recorded using the infrared beamline at the Australian Synchrotron. For each spectrum, 16 scans were co-added at a resolution of 6 cm–1. The spectra were analyzed using CytoSpec imaging software. A quality test was performed to remove spectra with poor signal-to-noise ratios and spectra containing obvious artifacts. The spectra were further analyzed using 2 different methods.

Initially, spectra were analyzed as described in our previous publications (15, 17), in which spectra from the antibody-exposed surface of the cartilage were compared with those from the interior of the cartilage. Raw chemical maps were generated from the integrated intensities of specific functional groups identified in the spectra, and 10 spectra from the surface of the explant and 10 from the interior were extracted from the raw chemical maps. The mean spectra for “surface” and “interior” were calculated to assess the effects of antibody penetration on the peaks characteristic of collagen and of proteoglycans. An FTIRM spectrum from tissue containing collagen characteristically contains a triplet of peaks at 1203 cm–1, 1234 cm–1, and 1280 cm–1. However, since this collagen triplet coincides with a peak at 1240–1245 cm–1 derived from the sulfated glycosaminoglycan side chains of proteoglycans (18, 19), analysis was directed particularly to the location of the amide 1 peak (1640–1670 cm–1), which represents total protein, which for cartilage, is primarily collagen. The amide 1 peak for the native triple-helical collagen is unusual, being >1660 cm–1, as compared with most other proteins, for which the peak is ∼1650 cm–1, and there is a characteristic shift to a lower wave number on denaturation (15, 20–25). For proteoglycans, analysis was based on the height of the peak at 1076 cm–1, within the region of 1175–960 cm–1 derived from carbohydrate moieties (15). Since according to the Beer-Lambert law, the concentration of a particular chemical is proportional to the absorbance of the particular band, this characteristic peak provides a direct measure of the proteoglycan content.

Analysis was also performed using second derivatives, which measures changes in the slope of the spectra, where minima peaks in the second derivative spectra correlated with the maxima peaks in the nonderivatized spectra (26). This provided more information about the changes in cartilage components by allowing the resolution of bands that contribute to inflection points or shoulders in the original spectra, which represent the total of the spectra of all of the chemical components scanned within each pixel. It also eliminated problems with sloping baselines arising from loss of cartilage components, but at the expense of loss of information about relative amounts of components. To analyze changes across the cartilage, unsupervised hierarchical cluster analysis, a technique that identifies areas with similar spectral properties (27), was performed on the chemical maps derived using the second derivative spectra.

Requirement for living chondrocytes in damage or protection of cartilage.

To determine whether the changes observed in the presence of the mAb to CII required degradative enzymes or other factors produced by chondrocytes, experiments were conducted on explants in which chondrocytes were killed by freeze-thawing the cartilage explants 3 times in liquid nitrogen (28). Experiments were conducted on explants cultured with the combination of the mAb M2139 and CIIC1, with or without CIIF4, and on control explants cultured without mAb, which were then cultured as normal for 7 or 14 days (15). The results obtained were compared with those obtained in the same experiments in which cartilage contained living chondrocytes.

Contribution of matrix metalloproteinases to cartilage destruction in vitro.

The changes observed in cartilage cultured with arthritogenic mAb (see Results) resembled those induced by chondrocyte-derived degradative enzymes, particularly the MMPs, including collagenases (MMPs 1, 8, and 13), gelatinases (MMPs 2 and 9), and stromelysins and aggrecanases (MMPs 3, 7, 10, 11, 12, and 18) (29). Since the epitope for mAb CIIF4 in the C-terminal region of CII in the aggregated collagen fibrils is close to the cleavage site of MMP-3 in the N-terminal telopeptide of CII (30), the protective effect of CIIF4 could be attributed to steric inhibition of MMP-3 activity. Hence, to examine the effects of MMPs on the cultures, experiments were performed using the commercially available broad-spectrum MMP inhibitor GM6001 (Millipore). GM6001 was added to cultures in varying concentrations (15 μM, 25 μM, and 35 μM) together with 25 μg/ml each of mAb M2139 and CIIC1. Controls without mAb were included, and the effects were assessed histologically and by FTIRM.

Statistical analysis.

Statistical analyses were performed using Statistica for Windows, version 4.5 (StatSoft). Analysis of variance (ANOVA) or the nonparametric Kruskal-Wallis ANOVA by ranks was performed to determine whether there were significant differences between all groups in the experiment. Student's t-test or the Mann-Whitney U test was used to analyze differences between specified groups. P values less than 0.05 were considered significant.


Effect of anti-CII mAb on cartilage morphology in explant cultures.

Cartilage explants cultured in medium alone or with CIIF4 concentrations of up to 100 μg/ml appeared normal throughout the culture period, and the matrix stained strongly with toluidine blue (Figure 1A). The mAb bound to the cartilage, and on day 7, CIIF4 had penetrated 32 ± 12 μm (mean ± SD), M2139 plus CIIC1 had penetrated 45 ± 16.5 μm, and M2139 plus CIIC1 plus CIIF4 had penetrated 29 ± 4 μm. The use of mixtures of M2139 and CIIC1 resulted in profound changes in the explant structure, including marked loss of toluidine blue staining, gradual loss of the matrix integrity at the cartilage surface, with increasing mAb penetration, and chondrons containing several cells, suggestive of hyperplasia as a compensatory response to mAb-mediated damage (Figure 1B), and more so with increasing concentrations of mAb or with increasing times in culture.

Figure 1.

Toluidine blue–stained cartilage samples cultured with various monoclonal antibodies (mAb). A, Cartilage sample cultured for 14 days with CIIF4, showing even staining and regular distribution of chondrocytes. B, Cartilage sample cultured for 14 days with a combination of arthritogenic mAb M2139 and CIIC1, showing uneven loss of proteoglycan staining and clumping of chondrocytes. C, Appearance of fibroblast-like cells on the base of the culture plate of cartilage explants cultured with M2139 and CIIC1. D, Cartilage sample cultured for 14 days with the combination of M2139, CIIC1, and CIIF4 mAb. (Original magnification × 200.)

From day 9 to day 10, cartilage degradation was associated with a migration of cells from the surface of the cartilage, and fibroblast-like cells began to appear on the base of the plate (Figure 1C). Such cells, which were taken to represent chondrocytes undergoing dedifferentiation as a result of matrix loss from the cut surface of the tissue, also appeared in control cultures by day 14, but their numbers were quite sparse. The addition of 50 μg/ml of CIIF4 to cultures containing 50 μg/ml of the arthritogenic mAb reversed the degenerative changes, as judged by a reduced loss of toluidine blue staining of proteoglycans at the cartilage surface, normal architecture of the tissue, albeit with a slight increase in multicellular chondrons, and even increased proteoglycan synthesis, particularly at the cartilage surface, after 14 days in culture (Figure 1D). Also, the appearance of fibroblast-like cells on the base of the plates was reduced to that seen in normal cultures.

Role of chondrocytes in antibody-mediated cartilage damage in vitro.

By light microscopy, freeze-thawed cartilage cultured for up to 14 days without mAb or with CIIF4 appeared normal, with rounded chondrocytes distributed within a matrix that stained evenly with toluidine blue. Histologically, freeze-thawed cartilage cultured with the combination of mAb M2139 and mAb CIIC1 showed changes similar to or greater than those observed in the cartilage containing living chondrocytes (Figure 1B), although the chondrocyte pairing and clumping seen in the living cartilage was not observed in the freeze-thawed tissue. The addition of CIIF4 to cultures of freeze-thawed cartilage did not protect from the damaging effects of the arthritogenic mAb.

Chemical changes in cartilage matrix as measured by FTIRM.

Histologic differences were confirmed by FTIRM, a technique that provides information on the nature of the chemical changes that have occurred. Spectra obtained from 5 samples of cartilage cultured for 14 days with up to 100 μg/ml of CIIF4 resembled spectra from control cartilage cultured without mAb. In each case, the amide 1 peak at the cartilage surface after incubation with CIIF4 was located above wave number 1660 cm–1 (range 1662–1670), which was comparable with that in the controls (1662–1666 cm–1), and the areas under the curve for the amide 1 and proteoglycan peaks were similar to those in the controls. Spectra derived from the cartilage surface and the cartilage interior were similar (Table 1 and Figure 2A).

Table 1. Location of the amide 1 peak and the height of the proteoglycan peak at the surface and in the interior of cartilage explants cultured in the presence and absence of arthritogenic mAb*
Treatment, vital status of cells, and locationDay 7Day 14
Amide 1 peak, median (range) cm–1Proteoglycan peak, maximum ± SD absorbanceAmide 1 peak, median (range) cm–1Proteoglycan peak, maximum ± SD absorbance
  • *

    Fourier transform infrared microspectroscopy was used to determine the location of the amide 1 peak and the height of the proteoglycan peak at 1,076 cm–1 at the cartilage surface, where the monoclonal antibody (mAb) penetrates, and in the interior of cartilage explants cultured for 7 or 14 days under the indicated conditions (see Materials and Methods for details). Analyses were performed in cartilage containing living chondrocytes or in cartilage in which the chondrocytes had been killed by repeated freezing and thawing.

  • Cartilage damage extends to the interior.

No mAb    
  Surface1666 (1639–1670)0.37 ± 0.061662 (1662–1662)0.39 ± 0.06
  Interior1666 (1639–1670)0.37 ± 0.041666 (1635–1670)0.41 ± 0.05
CIIF4 100 μg/ml    
  Surface1666 (1639–1670)0.32 ± 0.081662 (1654–1670)0.38 ± 0.03
  Interior1666 (1639–1670)0.40 ± 0.051662 (1650–1670)0.30 ± 0.06
M2139 25 μg/ml + CIIC1 25 μg/ml    
  Surface1658 (1643–1658)0.28 ± 0.051643 (1639–1650)0.15 ± 0.02
  Interior1666 (1662–1670)0.32 ± 0.081651 (1639–1666)0.11 ± 0.02
M2139 25 μg/ml + CIIC1 25 μg/ml + CIIF4 50 μg/ml    
  Surface1666 (1654–1670)0.30 ± 0.091658 (1639–1670)0.55 ± 0.06
  Interior1662 (1647–1662)0.25 ± 0.051666 (1646–1677)0.37 ± 0.04
M2139 25 μg/ml + CIIC1 25 μg/ml    
  Surface1643 (1631–1658)0.16 ± 0.031635 (1631–1643)0.15 ± 0.04
  Interior1666 (1658–1666)0.39 ± 0.031654 (1646–1658)0.22 ± 0.03
M2139 25 μg/ml + CIIC1 25 μg/ml + CIIF4 50 μg/ml    
  Surface1647 (1643–1650)0.07 ± 0.021639 (1631–1639)0.05 ± 0.18
  Interior1666 (1666–1670)0.29 ± 0.011650 (1643–1654)0.39 ± 0.05
Figure 2.

Fourier transform infrared microspectroscopy spectra of cartilage sections cultured with various antibodies. The results shown are the mean of 10 measurements taken from the central areas (thick line) and near the surface of the tissue (thin line). Shown in each panel is the location of the amide 1 peak, which represents the total protein content of the tissue, in the region 1600–1700 cm–1, and the height of the peak at 1076 cm−1, representing proteoglycans. A, Spectra of cartilage cultured for 14 days with 100 μg/ml of monoclonal antibody (mAb) CIIF4. B, Spectra of cartilage cultured for 10 days with 25 μg/ml each of mAb M2139 and CIIC1. C, Spectra of cartilage cultured for 10 days with 50 μg/ml each of mAb M2139 and CIIC1. D, Spectra of cartilage cultured for 10 days with 25 μg/ml each of mAb M2139 and CIIC1 with the addition of 50 μg/ml of mAb CIIF4.

In contrast to the lack of effect of CIIF4 on the FTIRM spectra of cartilage, there were striking differences in spectra obtained from cartilage cultured in the presence of CIIC1 and M2139 (Table 1 and Figure 2B). Beyond the region of penetration by mAb, the spectra were generally similar to those of controls cultured without mAb, but spectra from the surface of the cartilage showed substantial changes that increased with time in culture or with higher concentrations of mAb (Figure 2C). There was a progressive shift in the location of the amide 1 peak from >1660 cm–1 to as low as 1643 cm–1 or 1639 cm–1, which is suggestive of progressive denaturation of the collagen, and these changes gradually became evident deeper within the cartilage (Table 1).

Initially, there was loss of the proteoglycan peak, but no loss of protein, as judged by the area under the curve for the amide 1 peak, but with increasing time in culture, there was a substantial decrease in the amide 1 peak and, indeed, across the whole spectrum, indicating a total loss of matrix. The data indicated that the arthritogenic mAb caused initial denaturation of collagen and loss of proteoglycans, with ongoing progressive denaturation and loss of collagen over the period of culture. These changes were antibody mediated and did not require viable chondrocytes or new synthesis of degradative enzymes because similar or even greater effects were seen in cartilage that had been freeze-thawed (Table 1).

The addition of CIIF4 to cultures with combinations of the arthritogenic mAb counteracted these changes (Table 1 and Figure 2D), although the protective effect was somewhat delayed. On day 7, there was often some degradation, and the changes observed in explants cultured with CIIF4 in combination with any of the arthritogenic mAb were not significantly different from those seen in explants cultured with the arthritogenic mAb alone. However, by days 10–14 in cultures that contained CIIF4, the position of the amide 1 peak at the cartilage surface had normalized to that for the control cartilage (above wave number 1660 cm–1), and there was an equivalent normalization, or even an increase, in the proteoglycan content at the cartilage surface, as measured by the height of the proteoglycan peak at 1076 cm–1. In contrast to the degradative effects of the combination of arthritogenic mAb, which did not require living chondrocytes, the protective effect of CIIF4 was observed only when the cartilage contained viable cells (Table 1).

FTIRM analysis using second derivatives.

The results obtained using second derivative spectra confirmed those obtained by comparison of direct spectra, and second derivative minima were, as expected, located similarly to the peak maxima on the original spectra. In freeze-thawed cartilage, the use of second derivatives confirmed that the major change in chemical composition in the cartilage cultured with the arthritogenic mAb was in the location of the amide 1 peak, which was consistent with collagen denaturation, and there was little or no change in most components of the matrix, as judged by the lack of changes in other regions of the spectra (Figure 3). However, analysis of the second derivatives provided more information than was obtained using the primary spectra, for which the cartilage contained living chondrocytes. Whatever the treatment used, the spectra derived from living cartilage were more complex than those derived from freeze-thawed cartilage, particularly in the region above 1500 cm–1 (Figure 4). The most prominent and consistent feature was the appearance of a second derivative minimum at 1632–1635 cm–1, either as a separate peak or as a shoulder on another peak in the amide 1 region. This component could not be identified but was consistent with likely chondrocyte metabolic activation. It was seen particularly in spectra from all samples of CIIF4-treated living cartilage, as well as in untreated controls and in cartilage treated with arthritogenic mAb, although less prominently.

Figure 3.

Fourier transform infrared microspectroscopy (FTIRM) of freeze-thawed cartilage sections cultured for 7 days with monoclonal antibodies (mAb) M2139 and CIIC1, showing a comparison of the results obtained by direct analysis and by unsupervised hierarchical cluster analysis (UHCA) using second derivative spectra. Both techniques showed peaks located in similar positions, but the direct spectra showed loss of cartilage components. A, Light microscopy image of an unstained cartilage section on a Kevley reflective slide (original magnification × 200). B, Infrared image processed by UHCA using second derivative spectra with 4 clusters (4 different colors). Pixels with poor signal-to-noise ratios are shown as black. C, Mean FTIRM spectra obtained by direct analysis of 10 pixels from the interior and exterior of the cartilage. D, Second derivative spectra corresponding to the clusters shown in B. Pixels colored blue were judged to be artifacts and were removed during analysis.

Figure 4.

Fourier transform infrared microspectroscopy (FTIRM) of a section of living cartilage cultured for 7 days with monoclonal antibodies (mAb) M2139 and CIIC1 plus mAb CIIF4, showing a comparison of the results obtained by direct analysis and by unsupervised hierarchical cluster analysis (UHCA) using second derivative spectra. Both techniques showed the normal appearance of the cartilage spectra in the presence of CIIF4, but the use of second derivatives highlighted the appearance of a peak at ∼1632–1635 cm–1, in the region of living chondrocytes that did not appear in the freeze-thawed cartilage samples. A, Light microscopy image of unstained cartilage section on a Kevley reflective slide (original magnification × 200). B, Infrared image processed by UHCA using second derivative spectra with 4 clusters (4 different colors). Pixels with poor signal-to-noise ratios are shown as black. C, Mean FTIRM spectra obtained by direct analysis of 10 pixels from the interior and exterior of the cartilage. D, Second derivative spectra corresponding to the clusters shown in B. Pixels colored green were judged to be artifacts and were removed during analysis.

Assessment of the role of MMPs in the cartilage damage mediated by anti-CII mAb.

To assess whether MMPs affected the matrix destruction observed in the presence of arthritogenic mAb, and whether MMP-3 could determine the counterdestructive effect of CIIF4, we investigated the use of the broad-spectrum MMP inhibitor GM6001 (15, 25, and 35 μM) in combination with M2139 and CIIC1 (25 μg/ml each) in bovine cartilage explant cultures. At all concentrations, the addition of the MMP inhibitor GM6001 reduced the cartilage damage seen in response to the arthritogenic mAb, but the changes differed from the protective effects of CIIF4. Fragmentation at the surface of the antibody-treated cartilage was seen histologically, but it had a fibrillar appearance and followed the orientation of collagen fibrils within the tissue, being parallel with the joint surface, but perpendicular to the cut surface of the cartilage. The substantial loss of proteoglycans normally seen in cartilage treated with the arthritogenic mAb did not occur, and the intensity of the toluidine blue staining was maintained (Figure 5). By FTIRM, there was little evidence of collagen denaturation, as judged by the location of the amide 1 peak, which remained above 1660 cm–1, and the area under the proteoglycan peak was similar to that of the controls (data not shown). Similar effects were seen in freeze-thawed cartilage. Taken together, these results suggest that blocking MMP activity in the cartilage prevented degradation of the damaged collagen fibrils and loss of proteoglycan but did not prevent the initial damage mediated by antibodies.

Figure 5.

A, Toluidine blue–stained cartilage sample cultured for 14 days with monoclonal antibodies (mAb) M2139 and CIIC1 in the presence of the broad-spectrum matrix metalloproteinase inhibitor GM6001. The intensity of the staining was maintained, but there was fragmentation at the joint surface (JS) and the cut surface (C) that approximately aligned with the expected direction of the collagen fibrils. B, Spectra obtained by Fourier transform infrared microspectroscopy. The proteoglycan peak at ∼1076 cm–1 was maintained, and the amide 1 peak remained above 1660 cm–1. Measurements were taken from the central areas (thick line) and near the surface of the tissue (thin line).


This study was designed to examine the effects on bovine cartilage explants of 3 mAb to native CII that bind to well-identified and structurally different conformational epitopes (16). Two of these mAb, M2139 and CIIC1, are arthritogenic upon passive transfer in vivo (9, 14), and whether used individually (15) or in combination, they caused progressive denaturation of CII and substantial loss from cartilage of both CII and proteoglycans in vitro. By contrast, the third mAb, CIIF4, which is nonarthritogenic and protective in vivo when given with other mAb (14, 17), had no adverse effect in vitro when used alone on cultured cartilage and countered the adverse effects of arthritogenic mAb, even promoting the regeneration of cartilage matrix components. Thus, this counterdestructive effect of CIIF4 in vitro parallels its antiarthritogenic effects in vivo. It is notable that whereas the destructive effects of the arthritogenic mAb were independent of the presence of viable chondrocytes, the protective effect of CIIF4 required viable chondrocytes, since it was abrogated by freeze-thawing of the cultured cartilage explant.

The effector role of autoantibodies in the development of arthritis has been extensively studied in the context of murine CAIA. The bound antibodies have been shown to trigger and enhance inflammation by activating the complement cascade and FcγR-bearing cells, with release of proinflammatory cytokines by mononuclear cells within the synovium, leading to recruitment of neutrophils and macrophages that amplify the response by the further release of cytokines and tissue-degrading enzymes (31). Moreover, preceding or in parallel with these nonspecific effects of immune complex cellular activation, there is increasing evidence that autoantibodies to CII can have specific destructive effects within the cartilage (15, 16, 17). In addition to the degradative effects on preformed cartilage seen in the present study, the arthritogenic mAb have been shown to impair the synthesis of new matrix and disrupt collagen fibril formation in chondrocyte cultures in vitro (17, 32–34), changes that would directly affect the processes of cartilage repair following damage to the matrix.

The mechanism by which CIIF4 exerts its effects is currently unknown. The nonarthritogenic mAb CIIF4 was derived in a manner similar to that of the arthritogenic mAb from a DBA/1 mouse immunized with CII, and it is an IgG2a mAb, as is the arthritogenic mAb CIIC1. It binds strongly to CII, as demonstrated by ELISA, as well as to cartilage, both in vitro and in vivo. Hence, CIIF4 should activate complement and bind to FcγR-bearing cells to trigger immune complex–mediated inflammation. Nonetheless, it is not arthritogenic, but is actually protective in vivo, reducing the arthritis in mice injected with otherwise arthritogenic mAb (14, 17). CIIF4 is reactive with a conformational epitope at the COOH-terminus of the CII triple helix (aa 932–936), and within the assembled collagen fibrils, it is close to one of the cleavage sites of MMP-3 (stromelysin 1) (14) that is located within the NH2 telopeptide region in the collagen fibril (30). Moreover, its effects differ from those of other mAb, in that they were only seen when viable chondrocytes remained in the cartilage.

Our hypothesis was that CIIF4 exerts its protective effect by steric hindrance, blocking the cleavage of CII by MMP-3, which is produced by chondrocytes and by synovial fibroblasts, and is associated with cartilage degradation in osteoarthritis and in RA (35, 36). MMP-3 primarily targets proteoglycans and is not a conventional collagenase, but it acts as a telopeptidase by cleaving CII at a site inside its NH2-telopeptide crosslinking residue (30) and most likely also cleaving off the C-telopeptide. MMP-3 plays an essential role in the degradation of not only aggrecan, but also collagen fibrils in the cartilage (35), so that MMP-3–knockout mice are resistant to cartilage degradation in antigen-induced arthritis (37). The use of a broad-spectrum MMP inhibitor to block MMP-3 activity in the cultures, however, did not mimic the effect of the addition of the mAb CIIF4. Thus, the MMP inhibitor did not prevent cartilage damage as assessed histologically, and there were no apparent differences in the cultures that contained living or dead chondrocytes, indicating that the protective effect of CIIF4 was not merely a result of steric hindrance of MMP-3 activity, although steric hindrance of the binding of one or other of the arthritogenic mAb could not be discounted.

The use of the MMP inhibitor GM6001 in culture provided further information about the likely mechanism of damage by the arthritogenic mAb. By FTIRM, it appeared that the MMP inhibitor prevented both the denaturation of collagen and the loss of proteoglycans seen in cartilage cultured with the arthritogenic mAb, indicating that MMPs do play a role in the cartilage degradation induced by the antibodies. The fibrillar nature of the cartilage damage in the presence of the inhibitor suggests that the initial effect of the binding of the arthritogenic mAb on the surface of the collagen fibrils was to cause disaggregation of the collagen fibrils, but the lack of collagen denaturation or proteoglycan loss in the presence of the MMP inhibitor indicates that these enzymes play an important role in further cartilage degradation. The results suggest that the disruption of the collagen matrix causes the release of the MMPs known to be sequestered in the intact matrix (38, 39), and these MMPs are responsible for the ongoing loss of proteoglycans and collagen degradation, without any requirement for further MMP inhibitor synthesis.

A striking observation from these studies is the importance of living chondrocytes in the maintenance of cartilage integrity in the presence of arthritogenic mAb and in the protective effect of CIIF4. In each case, the arthritogenic mAb caused more damage to the freeze-thawed cartilage, as judged by the greater shift in the position of the amide 1 peak and the loss of proteoglycans on both day 7 and day 14. This augmented damage was also seen in the freeze-thawed cartilage even in the presence of CIIF4. The appearance of chondrocyte dimers, or clumps of cells in the antibody-treated cartilage, which were absent in the freeze-thawed cartilage, suggested that cell division was occurring in response to the treatment, and the appearance of an additional peak at 1632–1635 cm–1 in the living cartilage was consistent with cellular activation. Taken together, these observations suggest that there is a balance between damage and regeneration within the cartilage.

The use of FTIRM, a technique that allows the examination of localized chemical changes within tissue without any requirement of a priori knowledge of the likely mechanism has allowed definitive detection of the protective effect of CIIF4. Determination of the mechanism of the protective effect of CIIF4 will require the use of different techniques and is beyond the scope of the current study, but we are investigating the gene expression profiles in chondrocytes treated with CIIF4 and other antibodies.

The major question that arises from these studies is whether there is any connection with the pathogenesis of RA. RA is generally considered to be an immune complex–mediated disease, and its features are well accounted for by the formation and deposition in joint structures of immune complexes that are operative by complement activation and by Fc binding and Fcγ activation. In CIA in animals, the antigenic component is clearly CII, but in RA in humans, there is uncertainty (noting that an antibody response to ALL of the epitopes recognized by the mAb in this study also occur in RA). If antibodies to CII of the same specificity as the arthritogenic mAb also occur in human sera (14), the mechanisms of joint damage that occur in CAIA could also occur in RA. Interestingly, both proteoglycan and CII destruction and secretion into the synovium occur in RA (40–42). It is striking that antibodies to CII of the same specificity as CIIF4 also occur in human sera, but they are associated with osteoarthritis rather than rheumatoid arthritis (14), and it is tempting to speculate that such antibodies may also prevent the inflammatory damage seen in human RA.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Rowley had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Croxford, McNaughton, Holmdahl, Rowley.

Acquisition of data. Croxford, Crombie, Rowley.

Analysis and interpretation of data. Croxford, Crombie, McNaughton, Nandakumar, Rowley.


We thank Ian Mackay and Senga Whittingham for helpful discussions and editorial comments. Part of this research was undertaken using the infrared microspectroscopy beamline at the Australian Synchrotron (Clayton, Victoria, Australia).