Cartilage oligomeric matrix protein specific antibodies are pathogenic

Animals

COMP-deficient 129/Sv mice [20] were backcrossed for 10 generations to B10.Q mice to introduce the A^q allele in the MHC locus. B10.Q mice with a mutated Ncf1 (Ncf1 ^m1J denoted as Ncf1*) have been described previously [21]. To obtain mice with COMP-deficiency and Ncf1 mutation (COMP-/-Ncf1^*/*), COMP-deficient B10.Q mice were intercrossed with Ncf1 mutant mice to get COMP-deficient and Ncf1 mutated heterozygous mice. These heterozygous mice were intercrossed and the offspring were investigated for COMP deficiency and Ncf1 mutation by PCR and DNA sequence respectively [21, 22]. For passive transfer experiments (BALB/c × B10.Q) F1 male mice (known as QB) were used; these showed higher sensitivity to antibody-induced arthritis [7]. All the mice were kept and bred in a climate-controlled environment with a 12-h light/dark cycle. Mice were housed in polystyrene cages containing wood shavings and fed standard rodent chow and water ad libitum in the animal house of Medical Inflammation Research, Sweden. All the experiments were performed using age-matched mice between 8 and 10 weeks and blinded to the investigator. Malmö/Lund and Stockholm ethical committees approved all the experiments.

Production of recombinant COMP proteins

The corrected cDNA for mouse COMP [22] served as the template for PCR amplification of the following recombinant COMP fragments (Figure 1). In addition to full-length his-tagged mouse COMP (pCOMP, nucleotide 82-2289), the fragments produced were; (1) mCOMP, monomeric COMP, that is, only lacking the coiled-coil repeats, nucleotide 238-2289; (2) EGF1-TIII8, EGF-like repeats and TSP3 repeats, nucleotide 238-1569; (3) TIII1-CG, TSP3 repeats and C-terminal globular domain, nucleotide 820-2289; (4) EGF1-4, EGF-like repeats, nucleotide 238-819; (5) TIII1-8, TSP3 repeats, nucleotide 820-1569, and (6) CG, C-terminal globular domain, nucleotide 1570-2289 [NM_016685.2, GenBank]. The mouse COMP recombinant fragment cDNAs were amplified using 5'-phosphorylated upstream primers in combination with downstream primers adding a stop codon followed by an Nhe I site. The PCR products were digested with Nhe I and inserted into the Pvu II and Nhe I sites of the expression vector pCEP4-TAGZyme. This plasmid was constructed by inserting the sequence 5'-GGTACCGCCTGCCGCCTGCCT GCCTGCCACTGAGGGTTCCCAGCACCATGAGGGCCTGGATCTTCTTTCTCCTTTGCCTGGCCGGGAGGGCTCTGGCAGCCCGACATCACCATCACCATCACCAGCTGAAGCTT-3' between the Kpn I and HinDIII sites of plasmid pCEP4, yielding an expression vector with a BM40 signal peptide followed by a hexahistidine tag compatible with the TAGZyme removal system (Qiagen, Copenhagen, Denmark). All the PCR-generated fragments and their cloning sites were confirmed by nucleotide sequencing.

The resulting recombinant COMP plasmids were transfected into 293-c18 cells (ATCC CRL-10852) using the FuGene6 reagent (Roche, Basel, Switzerland ) and transfected cells were selected with hygromycin and grown to confluence in Dulbecco's modified Eagle's medium with 10% fetal calf serum (Gibco, Life Technologies Inc, Camarillo, CA, USA). Secretion of recombinant proteins into the medium was confirmed by SDS-PAGE of samples from transfected and non-transfected cells. Serum-free culture medium was harvested, and his-tagged recombinant mouse COMP proteins were purified using Ni²⁺-metal chelating and MonoQ ion exchange chromatography. Purity and size of all the proteins were assessed by SDS-PAGE and colloidal Coomassie Brilliant Blue G250 staining with or without prior reduction of disulfide bonds.

Rat COMP purification

Purification of COMP from the Swarm rat chondrosarcoma was performed as previously described [17]. Briefly, 50 g rat chondrosarcoma was homogenized and centrifuged, the pellet was extracted in 150 mM NaCl, 50 mM Tris, 10 mM EDTA, pH 7.4, supernatants were harvested, and COMP was purified using DEAE-Sepharose fast-flow media column (GE Healthcare, Uppsala, Sweden), COMP-containing fractions were evaluated by SDS-PAGE. Pooled fractions were then exchanged to 150 mM NaCl, 10 mM Tris, and 2 mM EDTA, pH 7.4 buffer. Pooled fractions were concentrated and negatively purified using HiTrap heparin-Sepharose column, HiTrap Q Sepharose column, heparin column (GE Healthcare, Uppsala, Sweden) and Q column. The purity and size of COMP fractions were assessed by SDS-PAGE under reducing and non-reducing conditions.

Synthetic peptides

Sequence alignment identified 15 amino acid differences between the mature peptide sequences of rat and mouse COMP. The following peptides, corresponding to the rat COMP sequence and containing the amino acids differing from mouse COMP (underlined), were synthesized using methods previously described in detail [8]. P1: DVRELLRHRVKEITFLK (in coiled-coil repeats); P2: PGLSVRPVALCAPGSCF (in EGF1); P3: SCFPGVVCTETATGARC (in EGF1); P4: GVGLTFAKTNKQVCTDI (in EGF2); P5: QPGFVGDQRSGCQRRGQ (in EGF3); P6: PSPCHEKADCILERDGSRS (in EGF4); P7: GQEDVDRDRIGDACDPD (in TIII2); P8: NPDQRNSDKDKWGDAC D (in TIII3); P9: DACDNCRSQKNDDQKDT (in TIII4); P10: NDDQKDTDRDGQG DACDDDI (in TIII4); P11: DTIPEDYERHRLRRA (C-terminal peptide). Peptide identities were confirmed by mass spectrometry and fast protein liquid chromatography was used to maintain purity at > 98% (Hermann GbR Synthetische Biomoleküle, Freiburg, Germany).

Generation of COMP-specific B cell hybridomas

Hybridomas producing anti-COMP mAbs were generated as described earlier [23]. Briefly, COMP^-/-Ncf1^*/* and COMP^+/+Ncf1^*/* mice were immunized at the base of the tail with 100 μg of rat COMP emulsified in complete Freund's adjuvant (CFA; Difco, Detroit, MI, USA), followed by an i.p. booster with 50 μg of rat COMP in PBS. Three to five days after the booster dose, spleen or inguinal lymph cells were fused with SP2/0 or NSO-bcl2 myeloma cell lines. Two weeks after the fusion, all the primary plates were screened for antibody production by ELISA using rat COMP. Cells from positive wells were sub-cloned 5 times by limiting dilution and expanded for further use. Monoclonal antibodies were purified from the supernatants by affinity chromatography using protein G column (GammaBind plus Sepharose; GE Healthcare, Uppsala, Sweden) and bound IgG was eluted in 0.1 M glycine (pH 2.8) and neutralized with one-sixth volume of 1 M Tris-HCl (pH 8.5). The eluted fractions were dialyzed against PBS buffer extensively. Monoclonal antibodies were quantified by measuring absorbance at 280 nm and the purity of the antibodies was assessed by SDS-PAGE. The endotoxin content in all mAbs preparations was found to be below detection limit (less than 1 EU/mg) as analyzed with the Limulus amebocyte lysate (Pyrochrome) method (Cape Cod Inc., Falmouth, MA, USA).

Antibody measurements

Enzyme immunoassays were performed as described earlier [22]. Briefly, 96-well flat-bottom ELISA plates (Nunc MaxiSorp; eBioscience, San Diego, CA, USA) were coated with 10 μg/ml of recombinant COMP proteins, or with 10 μg/ml synthetic COMP peptide in PBS at 4°C and left overnight. Plates were blocked with 1% bovine serum albumin in PBS containing Tween, incubated with the supernatants from hybridomas or sera, washed, and again incubated with goat anti-mouse IgG peroxidase-conjugated antibodies (Jackson Immuno-Research, West Grove, PA, USA) and ABTS tablets (Roche) as substrate. For isotype-specific assessment, biotinylated anti-IgM, -IgG1, -IgG2a, -IgG2b, or -IgG3 reagents (Jackson Immuno-Research, West Grove, PA, USA) were used for detection.

Passive transfer of antibodies and evaluation of arthritis

The anti-CII mAb (M2139) binding to the MPGERGAAGIAGPK epitope of the CB10 fragment (aa 551-564) of the CII molecule used in this study was produced as described earlier [6, 19]. The cocktail of M2139 and COMP mAbs and the combination of COMP mAbs were prepared by mixing equal concentrations of each of the sterile-filtered antibody solution to achieve a final amount of 9 mg. On day 5 (or day 7), all the mice received 25 μg/mouse lipopolysaccharide (LPS) from Escherichia coli serotype O26:B6 (Sigma-Aldrich, Saint Louis, MO, USA) intraperitoneally to enhance the severity of the arthritis developed. Arthritis development was monitored daily in a blinded fashion for 16 days using an extended scoring protocol [24]. Briefly, clinical arthritis is defined as swelling and redness in the joint and was scored as follows: 1 point was given for each swollen or red toe, 1 point for each swollen joint (metatarsal phalangeal joints, metacarpal phalangeal joints, proximal interphalangeal joints, and distal interphalangeal joints), and 5 points for a swollen ankle (maximum score per limb was 15 and maximum score per mouse was 60).

Histology

For histological assessments, knee joint and paws from adult mice were dissected, decalcified, dehydrated, and then paraffin-embedded, as described previously [23]. Sections (5 μm) were stained with hematoxylin/eosin or toluidine blue. For analysis of anti-COMP mAb reactivity with joint tissue in vivo, neonatal mice were injected with 100 μg biotinylated mAbs against COMP intraperitoneally. After 24 h, knee joints were snap frozen in isopentane on dry ice and stored at -70°C. Joint sections (7 μm) were fixed in 4% paraformaldehyde for 5 minutes, rinsed in PBS, incubated with Extravidin peroxidase (Sigma-Aldrich, Saint Louis, MO, USA) for 30 minutes and developed with diaminobenzidine (DAB Kit; Dako, Copenhagen, Denmark) for 8 to 9 minutes. To assess direct binding of anti-COMP mAbs to the tissue sections, limbs from naïve neonatal mice were harvested and snap frozen, cryo-sectioned, and sections of 7 μm were subjected to 5 μg/ml biotinylated mAbs against COMP for 40 minutes. Extravidin peroxidase and DAB were used for the detecting system.

Statistical analyses

Quantitative data are expressed as mean ± standard error of the mean (SEM). Arthritis incidence was analyzed using chi-square or Fisher's exact test (SAS Institute; Cary, NC, USA) and comparison of severity was performed using the non-parametric Mann-Whitney U-test. Analysis of variance (ANOVA) was used to analyze polyclonal Ab response statistics. Differences with P-values less than 0.05 were considered statistically significant.

Specificity of polyclonal antibodies to COMP

To perform epitope mapping of COMP, we expressed recombinant his-tagged full-length mouse COMP and a panel of overlapping COMP fragments in HEK 293-c18 cells (Figure 1A). All the recombinant proteins were purified using affinity and ion exchange chromatographic methods. For each recombinant protein, the expected apparent molecular weight and a purity of > 95% was verified by SDS-PAGE analysis (Figure 1B).

To ascertain whether antibodies show reactivity towards individual full-length and truncated form of COMP proteins, microtiter plates were coated with either recombinant COMP molecules or COMP purified from rat chondrosarcoma, and used in enzyme immunoassays with sera obtained at 50 days after primary immunization. As shown in Figure 1C, reaction with recombinant COMP proteins was in the following order: pCOMP > mCOMP approximately = EGF1-TIII8 approximately = EGF1-4 > TIII1-CG > CG approximately = TIII1-8. Notably, of the single domain fragments, reactivity to EGF1-4 is considerably higher than to TIII1-8 or CG. However, in the native fold, the TSP3 repeats (TIII1-8) are wrapped around the C-terminal globule (CG) making direct comparison complicated. Nevertheless, the lower reactivity to the TIII1-CG fragment (the complete TSP3 and C-terminal globule fragment), compared to all the fragments containing the EGF repeats (pCOMP, mCOMP, EGF1-TIII8 and EGF1-4) underlines an important role of the EGF repeats in the immune response.

In COMP^+/+Ncf1^*/* mice, the incidence of arthritis was 80% with the day of disease onset at 36 to 38 days [17]. Since cross-reactions with autologous COMP appear to be of importance for the development of arthritis, we analyzed the antibody reactivity to the autologous mouse recombinant full-length COMP and COMP fragments. Immunized mice produced a rapid and strong IgG response as early as 14 days after immunization (Figure 1D). Antibody reactivity to both rat and mouse COMP showed a definite trend of increasing serum titres, from day 14 onwards with a peak response around day 50. Analysis of the kinetics of antibody response to each of the recombinant COMP fragments confirmed the important contribution of EGF repeats to the overall immune response. Antibody levels to the EGF1-4 fragment were higher than observed for TIII1-8, CG and TIII1-CG fragments at all time points tested in our experiments.

Generation and characterization of anti-COMP mAbs

For the detection of mAbs binding to different regions of COMP molecule, we performed ELISA with the recombinant COMP fragments. The binding properties of 4 representative mAbs to the truncated COMP fragments are shown in Figure 2. The 16B5 clone produced a specific signal only with pCOMP, which is a full length COMP molecule, but produced no signal with any of the other COMP fragments. Hence, we deduced that 16B5 binding site is either located at the N-terminal coiled-coil pentamerization domain or a conformational discontinuous epitope of pentameric COMP. The 15A11 clone represents mAb with a binding site located at the EGF-like repeat domain, which produced a specific signal with pCOMP, mCOMP, EGF1-TIII8 and EGF1-4 fragments but produced no signal with TIII1-8 and CG fragments. This antibody does, however, also show reactivity to the TIII1-CG fragment, although producing a lower signal. The 1B8 clone represents mAb having a binding site located at the TSP3 domain, which produced a specific signal with pCOMP, mCOMP, EGF1-TIII8, TIII1-CG and TIII1-8 fragments, but produced no signal with EGF1-4 and CG fragments. The 1D10 clone represents mAb binding to epitopes that are located at the C-terminal globular domain, which produced a strong specific signal with the CG fragment, bound the CG containing fragments pCOMP, mCOMP, TIII1-CG, but produced little or no signal with the EGF1-4, EGF1-TIII8 and TIII1-8 fragments. This mapping strategy using the recombinant COMP fragments was applied on each of the 18 mouse-specific mAbs. The mAbs binding sites were thereby narrowed down to one of the COMP domains. Among these 18 mAbs, 12 reacted with the EGF-like domain. Alignment of the rat and mouse COMP sequences revealed 15 amino acid differences between the mature peptide sequences. Considering that these differences could result in a strong immune response, we produced 11 synthetic peptides corresponding to the rat COMP sequence, each spanning one or two variant residues. To fine map the anti-COMP mAb epitopes, we performed ELISA on the 11 synthetic peptides with the hope that some of the mAbs would react to a linear peptide structure. As shown in Figure 3, the peptide PSPCHEKADCILERDGSRS, which locates at EGF-repeat 4, was found to bind the 15A11 mAb. None of the other mAbs tested (16B5, 12H8, 1B8, 2D3, 3F8, 6H4) showed binding to any of the peptides tested.

Monoclonal antibodies specific for COMP bind cartilage

To test whether the mAbs bind with native COMP in cartilage, COMP mAbs were biotinylated and injected separately into neonatal COMP^+/+ mice and the presence in joints of biotinylated mAbs after 24 h was analyzed using cryo-sections of whole paws. The mAbs demonstrated specific binding to cartilage and synovial tissue, with different affinity to the joint cartilage tissue (Figure 4). The 1D10 and 15A11 mAbs gave detectable staining, showing accumulation on the lining of the normal articular cavity, most visibly along the cartilage surface and synovial tissue. In contrast, mAb 16B5 showed no appreciable staining. The specificity for COMP was verified by performing the same experiment in neonatal COMP^-/- mice, where no joint staining was observed (Figure 4).

Cartilage tissue is a dense ECM with a very high concentration of proteoglycans, suggesting that antibody penetration into the tissue would be limited. In agreement with this, the anti-collagen II (CII) mAb M2139 staining was restricted to the cartilage surface (Figure 4). The 1D10 and 15A11 mAbs stainings, although weaker than the CII staining, show the same pattern, with additional staining of the synovium and bone.

To verify the in vitro binding capacity of mAbs to cartilage, joint sections from untreated neonatal wild-type or COMP-/- mice were incubated with the individual biotinylated mAbs and binding of the antibody detected by indirect immunohistochemistry (Figure 5). As expected, mAb M2139 in this case gave uniform CII staining throughout the cartilage from wild-type mice. This was also the case for the three COMP mAbs that showed clear cartilage staining in wild-type mice (1D10, 15A11 and 12H8).

COMP-specific monoclonal antibodies induce and enhance severe acute arthritis in naïve mice

To investigate the pathogenic potential of COMP-specific mAbs, (BALB/c × B10.Q) F1 mice were first injected with single COMP-specific mAbs in combination with a sub-arthritogenic dose of M2139, a CII specific IgG2b mAb. This experimental design was chosen because most of the COMP-specific mAbs are of the poorly complement-fixing IgG1 isotype, and complement activation is one of the critical factors for development of arthritis [25, 26]. We have earlier reported that 4.5 mg of M2139 induced no visible arthritis [7]. This dose was therefore chosen to use in combination with different COMP-specific mAbs, to test whether these mAbs could contribute to tissue injury by enhancing the subclinical disease established by the CII-specific M2139 antibody alone. Indeed, combination of single COMP-specific mAbs with a sub-optimal dose of M2139 mAb did induce arthritis (Figure 6A). Among the different combinations tested, the 15A11 plus M2139, and the 3F8 plus M2139 showed the most efficient arthritogenic activity. In the 15A11 plus M2139 group, all six mice presented aggressive disease, which may indicate that they were most effective at complementing each other. In the 3F8 plus M2139 group, four out of six mice presented aggressive disease (Figure 6A). To confirm the clinical observations, paws from treated animals were subjected to histopathological examination after the termination of the experiments. In the tissue sections of arthritic mice, we could detect inflammatory infiltration in the synovium and articular cavity, synovial hyperplasia, pannus formation, bone and cartilage erosions, and furthermore proteoglycan loss was clearly evident (Figure 6B).

To further test whether a combination of antibodies to the most dominant epitopes of COMP was arthritogenic, anti-COMP mAbs 1B8, 1D10, 3F8, 15A11 and 16B5, which bind sites located at different domains of COMP, were mixed at equal concentrations and injected intravenously into 2-month-old mice. Clinical symptoms of arthritis developed in four out of six mice characterized by a rapid onset albeit with a low disease severity (Figure 6C). No signs of arthritis were observed in animals injected with control monoclonal antibodies.