To determine the role of the proteinase ADAMTS-1 in normal and accelerated catabolism of aggrecan in articular and growth plate cartilage of mice.
To determine the role of the proteinase ADAMTS-1 in normal and accelerated catabolism of aggrecan in articular and growth plate cartilage of mice.
Expression of ADAMTS-1 was determined using reverse transcriptase–polymerase chain reaction (RT-PCR) analysis of RNA isolated from microdissected chondrocytes from different zones of mouse growth plate and articular cartilage. Real-time RT-PCR for ADAMTS-4, ADAMTS-5, and ADAMTS-9 was performed on femoral head cartilage of wild-type (WT) and ADAMTS-1–knockout (KO) mice. Histologic and immunohistologic evaluation of growth plate and articular cartilage was performed in WT and KO mice from birth to 12 weeks of age. The effect of ADAMTS-1 ablation on cartilage proteoglycan loss was studied in antigen-induced arthritis (AIA). Aggrecan catabolism in WT and KO mice was studied in an in vitro model of cartilage degradation, by quantitation of glycosaminoglycan loss and histologic, immunohistologic, and Western immunoblot analyses.
ADAMTS-1 messenger RNA (mRNA) was expressed in normal mouse articular and growth plate cartilage and was up-regulated in terminal hypertrophic differentiation of growth plate chondrocytes. There was no difference in mRNA levels in the cartilage of WT compared with KO mice for the other potential aggrecanases ADAMTS-4, ADAMTS-5, or ADAMTS-9. ADAMTS-1–KO mice were significantly smaller than their WT littermates; however, no morphologic differences between the genotypes were evident in growth plate or articular cartilage from birth to skeletal maturity (12–16 weeks). Similarly, no difference in cartilage aggrecan content or presence of aggrecan degradation products was detected between WT and KO mice. There was no difference between WT and KO mice in the degree of synovial inflammation or depletion of cartilage aggrecan in AIA. There was no difference between WT and KO cartilage in either basal or stimulated aggrecan loss in vitro; however, subtle changes in the aggrecanase-generated aggrecan catabolites were observed in interleukin-1–treated cartilage.
Although ADAMTS-1 is expressed in articular and growth plate cartilage and is able to cleave aggrecan at physiologically relevant sites, our results indicate that it does not play a significant nonredundant role in normal cartilage and bone development and growth. Similarly, ablation of ADAMTS-1 offered no protection from accelerated aggrecanolysis in an inflammatory model of arthritis or in an in vitro model of early cartilage degradation. ADAMTS-1 does not appear to be a viable target for treatment of cartilage destruction in arthritis.
Accelerated proteolysis of aggrecan and the consequent loss of the glycosaminoglycan (GAG)–bearing region of the molecule from articular cartilage during arthritis is an early and central event in disease pathogenesis. Depletion of aggrecan renders the cartilage less resistant to mechanical compression and may be a prerequisite for subsequent matrix metalloproteinase (MMP)–driven destruction of the collagen network (1). Ultimately, erosion of articular cartilage results in unrecoverable joint dysfunction that may necessitate joint replacement surgery. Two predominant enzyme activities have been implicated in aggrecan proteolysis in arthritis, namely, MMPs and aggrecanases. However, based on the detection of specific aggrecan cleavage products, it seems highly likely that the proteolysis of aggrecan in human joint disease is due to aggrecanase (for review, see refs. 2 and3). Turnover of aggrecan is also implicated in normal physiologic processes, such as development of the primary and secondary ossification centers, and in the growth plates of long bones. As in pathologic articular cartilage, aggrecan fragments resulting from both MMP and aggrecanase activity (G1-PEN and G1-EGE, respectively) have been detected in these growth cartilages (4–6).
A number of aggrecanases, defined by their ability to cleave aggrecan at specific Glu–Xaa peptide bonds, have now been identified as members of the ADAMTS family of proteinases, which includes ADAMTS-1, ADAMTS-4, ADAMTS-5, ADAMTS-8, ADAMTS-9, and ADAMTS-15 (7–13). Although cleavage of aggrecan by aggrecanases has been clearly identified in the growth plate and in normal and arthritic articular cartilage, the identity of the responsible enzymes in normal versus pathologic situations and in the different tissues has yet to be resolved. ADAMTS-1, ADAMTS-4, ADAMTS-5, ADAMTS-9, and ADAMTS-15 are expressed in normal human articular chondrocytes (11, 14–16). There are conflicting data on expression change of these potential aggrecanases in osteoarthritic (OA) compared with normal human cartilage, however, with both increased (15–17) and decreased (11) levels being reported. ADAMTS-1, ADAMTS-4, and ADAMTS-5 proteins have been detected in normal and OA human articular cartilage, although from the few studies available, analysis of change in content with disease is not possible (16–18). There have been no studies of ADAMTS expression or protein in human growth cartilage. Increased expression of ADAMTS-5, but not ADAMTS-4, has been reported in rabbit hypertrophic growth plate chondrocytes and in response to thyroid hormone (19). Furin-activated ADAMTS-4 protein has been colocalized with aggrecanase-cleaved aggrecan in the growth plate adjacent to the secondary ossification center but not in the metaphysis (6). ADAMTS-1 is expressed in embryonic limb buds (20) and is present and up-regulated by parathyroid hormone in metaphyseal and diaphyseal bone (21).
It is evident that the contribution of the different aggrecanases in normal and pathologic aggrecanolysis in articular and growth plate cartilage remains unclear. One way to address this issue is to study mice with a null mutation in one or more of the aggrecanases. Targeted disruption of the ADAMTS-4 gene in mice was recently described. The animals showed no musculoskeletal growth abnormalities, no abrogation of aggrecan catabolism in vitro, and the development of arthritis was not impaired in vivo (22). These results suggest that ADAMTS-4 is not the principal aggrecanase in murine cartilage. ADAMTS-1–knockout (KO) mice have also been described, and the phenotype includes developmental abnormalities in renal, adrenal, and adipose tissue and subfertility in females (23–25). These phenotypic abnormalities are consistent with the expression of ADAMTS-1 in the affected tissues (20, 26–28).
In light of the previous identification of ADAMTS-1 expression and/or protein in normal and pathologic articular cartilage (11, 14–16), limb buds (20), and metaphyseal bone (21), we undertook a more in-depth comparison of musculoskeletal development and normal and arthritic cartilage aggrecanolysis in ADAMTS-1–KO mice. In the present study, we examined the expression of ADAMTS-1 in growth plate chondrocyte maturation in wild-type (WT) mice. Bone and joint morphology and development were studied histologically and immunohistologically in WT versus ADAMTS-1–KO mice. To assess the potential contribution of ADAMTS-1 to cartilage destruction in joint disease, an inflammatory model of arthritis was evaluated. Finally, proteolysis of aggrecan was evaluated in WT and ADAMTS-1–KO cartilage using a novel in vitro model of early cartilage degradation.
Generation of the ADAMTS-1–KO mice using an exon 2 deletion construct and homologous recombination (Monash Institute of Reproduction and Development, in accordance with the institutional animal ethics regulations) has been previously described (25). This deletion brings a stop codon into frame, and there is almost complete ablation of the gene transcript, most likely due to nonsense-mediated decay of the messenger RNA (mRNA) (29). No ADAMTS-1 protein is secreted in this mouse (25), which shows a very similar phenotype to another independently generated ADAMTS-1–KO mouse (23), with a perinatal renal defect and subfertility in females associated with ovulation failure. In the present study, age-matched WT and KO mice were obtained from heterozygote breeding pairs and used for collection of tissues and arthritis studies. Comparisons of growth plate, bone, and joint morphology in WT and ADAMTS-1–KO animals were performed in sex-matched newborn, 3-week-old, and 12-week-old mice. Arthritis was induced in 10–12-week-old mice. In vitro aggrecan catabolism was studied in femoral head cartilage isolated from 23–26-day-old WT and ADAMTS-1–KO mice.
Femurs from 2-week-old WT mice were harvested immediately after death and dissected on ice to remove the majority of the cortical bone, leaving the metaphyseal surface of the growth plates intact. Dissected knees were snap frozen in OCT compound in liquid nitrogen–cooled isopentane and stored at −80°C. Serial 6-μm sagittal frozen sections were collected on RNase-free glass slides and air dried for 1 hour. Using an inverted microscope and an ophthalmic scalpel, cells from the proliferative (PR), prehypertrophic and early hypertrophic (PH), and late hypertrophic (LH) zones of the femoral growth plate and the articular cartilage were harvested separately into sterile Eppendorf tubes. Total RNA was isolated using the PicoPure RNA isolation kit (Arcturus, Mountain View, CA). Contaminating genomic DNA was removed by on-column DNase digestion (RNase-free DNase set; Qiagen, Chatsworth, CA).
Using the RiboAmp kit (Ambion, Austin, TX), linear amplification was performed on RNA in 2 rounds to yield 30–50 μg of antisense RNA, following the manufacturer's guidelines. Five nanograms of antisense RNA was reverse transcribed in a reaction volume of 20 μl using SuperScript III reverse transcriptase (Invitrogen, San Diego, CA) at 50°C for 1 hour. PCR amplifications were carried out in a total volume of 25 μl on a GeneAmp PCR system 2400 (Applied Biosystems, Foster City, CA) using 1 μl of RT reaction, 10× reaction buffer, 1 mM MgCl2, 100 mM of each dNTP, 2 μM of each primer, and 1 unit of AmpliTaq gold DNA polymerase (Applied Biosystems). Primers were designed using the 3′-untranslated complementary DNA sequences for ADAMTS-1, aggrecan, type X collagen, alkaline phosphatase, and GAPDH (Table 1). Following an initial thermal activation for 7 minutes at 95°C, amplification consisted of 25 cycles of denaturation for 30 seconds at 94°C, annealing for 30 seconds at 55°C, and extension for 30 seconds at 72°C, followed by a final extension step of 5 minutes at 72°C. Ten-microliter aliquots of the PCR products were electrophoresed in 1.25% agarose gels stained with ethidium bromide and visualized with an Eagle Eye II still video system (Stratagene, La Jolla, CA) under ultraviolet light.
|Target (GenBank accession no.)||Primer|
|(BC050834)||(bp 4545–4564)||(bp 4815–4835)|
|(L07049)||(bp 6452–6472)||(bp 6604–6623)|
|Type X collagen||TGTGTGCCTTTCAATCGAGTG||TCCGGGCTTTAATAAGTGAGG|
|(X67348)||(bp 7910–7930)||(bp 8039–8059)|
|(X13409)||(bp 2175–2192)||(bp 2344–2364)|
|(M32599)||(bp 989–1009)||(bp 1143–1163)|
In order to determine whether there was a compensatory increase in other aggrecanases, TaqMan real-time PCR (Applied Biosystems) was used to detect the expression of mRNA for ADAMTS-4, ADAMTS-5, and ADAMTS-9 in the femoral head cartilage of age-matched (10-day-old) WT and ADAMTS-1–KO mice. Primer and probe sets were designed and synthesized using the Primer-by-Design service (Applied Biosystems). The sequences of the forward and reverse primers and the TaqMan probe, respectively, were as follows: for ADAMTS-4, 5′-CAAGCAGTCGGGCTCCTT-3′, 5′-GATCGTGACCACATCGCTGTA-3′, 5′-TCCATACCTGAATTTTTTG-3′; for ADAMTS-5, 5′-CCAGTTGTACAAAGATTATCGGAACCT-3′, 5′-GTTGCTCCTTCAGGGATCCT-3′, 5′-TCAGTATAACCCTTGCTTTTTT-3′; and for ADAMTS-9, 5′-AGCACACTGCGGTCATCAG-3′, 5′-ACAGACTGCAGTGGTTCAATGAAATA-3′, and 5′-AACGTGCCCATCATTC-3′. The primer/probe set for amplification of 18S ribosomal RNA (rRNA) was purchased from Applied Biosystems. Multiplex PCR was performed in an ABI Prism 7700 sequence detection system (Applied Biosystems) with primer/probe sets to the target (ADAMTS) and reference (18S rRNA) in the same reaction. The relative amount of mRNA expression between WT and KO mice was determined by the Δ threshold cycle (Ct) method (30), using the formula 2, where ΔCt is calculated by subtracting the Ct value of reference RNA from the Ct value of the target RNA.
Knee joints (mid-tibia to mid-femur) were harvested from newborn, 3-week-old, and 12-week-old WT and ADAMTS-1–KO mice, and the majority of the surrounding muscle and soft tissue was removed. Joints were fixed overnight at 4°C in 10% neutral buffered formalin, then decalcified for 24 hours at 4°C on a rocking platform with 10% formic acid containing 5% formalin. Decalcified limbs were washed in tap water and then equilibrated for several hours in phosphate buffered saline (PBS) containing 0.01M EDTA prior to embedding in paraffin. Sagittal sections (6μ) were adhered to Superfrost Plus slides (Menzel Glaser, Braunschweig, Germany) at 60°C overnight, dewaxed, and stained with hematoxylin and eosin (H&E) or toluidine blue with a fast green counterstain, as previously described (31). The distribution of type X collagen in growth plates of 3-week-old mice was visualized by immunohistochemistry using a polyclonal rabbit antiserum against recombinant type X collagen (a generous gift from Dr. Richard Wilson, Murdoch Children's Research Institute, Parkville, Victoria, Australia). Briefly, 6μ sections were dewaxed, rehydrated in graded ethanol, and endogenous peroxidase activity was blocked by incubation in 3% H2O2 in water for 5 minutes. Sections were treated with 0.1 units/ml protease-free chondroitinase ABC (ICN Biochemicals, Irvine, CA) in 0.1M Tris acetate, pH 7.0, containing 0.01M EDTA for 2 hours at 37°C. After blocking with normal goat serum, sections were incubated with primary antibody overnight at 4°C. Localization was performed using the Vectastain Elite ABC kit (Vector, Burlingame, CA) and diaminobenzidine (3 minutes) according to the manufacturer's instructions. Slides were then counterstained with Mayer's hematoxylin. The type X collagen antiserum was used at a 1:2,500 dilution; preimmune rabbit sera at the same dilution were used as a negative control.
Arthritis was induced in knee joints of 10–12-week-old WT (n = 8) and KO mice (n = 8) by intraarticular injection of methylated bovine serum albumin (mBSA; Sigma, St. Louis, MO) in mice preimmunized with mBSA (32). All procedures were approved and performed in accordance with the Royal Children's Hospital animal care and ethics committee guidelines. Briefly, animals received an intradermal injection of 100 μg mBSA in 100 μl Freund's complete adjuvant split between 2 sites at the base of the tail. After 10 days, animals were administered intraarticular mBSA (10 μl of 20 mg/ml mBSA in 0.9% sterile saline or vehicle alone) into the left and right knee joints. Animals were killed 7 days after intraarticular injection, and knee joints were harvested, processed, and stained with H&E and toluidine blue/fast green, as described above. Sagittal 6-μm sections through the central weight-bearing region of the medial femorotibial joint were scored by 4 observers who were blinded to the treatment and genotype. Sections were scored for synovitis (0–3 scale), synovial exudate (0–2 scale), pannus (0–1 scale), cartilage proteoglycan loss (0–3 scale), cartilage erosion (0–3 scale), and bone erosion (0–2 scale). An average of the 4 observers' scores for each section was generated, and the mean scores (individual parameters and total) for all WT and KO joints were compared using the Mann-Whitney U test for unpaired nonparametric data.
Femoral head cartilage was isolated from freshly killed 23–26-day-old WT and KO mice under sterile conditions, using forceps to induce a capital femoral physeal fracture. The femoral heads from WT and KO mice were pooled separately, cultured (37°C, humidified 5% CO2/95% air) for 3 days in HEPES buffered Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal calf serum, and then washed 3 times for 5 minutes in serum-free DMEM. Individual femoral heads were then cultured for 3 or 7 days in 0.4 ml serum-free DMEM with or without 10−5M retinoic acid (RA; Sigma) or in 10 ng/ml recombinant human interleukin-1α (IL-1α; Sigma) with no changes in medium.
At the termination of culture, femoral heads were blotted dry, weighed, and either extracted in 4M guanidine hydrochloride (GuHCl) containing proteinase inhibitors (0.01M EDTA, 0.1M 6-aminohexanoic acid, 0.005M benzamidine, 0.5 mM phenylmethylsulfonyl fluoride, 0.01M N-ethylmaleimide final concentration) for 48 hours at 4°C (33) or snap frozen in OCT compound, as described above. Medium was made to 0.01M EDTA to inhibit metalloproteinases and frozen at −20°C until further analysis. Frozen sections (6μ) of femoral heads cultured with or without RA or IL-1 from WT and KO mice were mounted on Superfrost Plus slides, air dried, and fixed for 15 minutes with 10% neutral buffered formalin. After washing in PBS, sections were either stained with toluidine blue/fast green or immunolocalized with polyclonal anti-PEN (34) at 0.6 μg/ml IgG final concentration or with a polyclonal anti-EGE (34) at 1.1 μg/ml IgG. In addition to preimmune sera, the specificity of the neoepitope antibody staining was verified by preabsorption with the immunizing peptide compared with an irrelevant peptide.
The 4M GuHCl extracts of cultured femoral heads were dialyzed with ultrapure water for 16 hours at 4°C, and the residue remaining after extraction was digested with papain (33). GAG content in the dialyzed extracts, papain digests, and conditioned medium was measured by the 1,9-dimethylmethylene blue (DMMB) assay, as previously described, using chondroitin sulfate C from shark cartilage (Sigma) as a standard (33). Differences in the release of sulfated GAG into the medium (expressed as the percentage of total GAG in the medium plus extract plus papain digest) associated with culture treatment and genotype were measured using a 2-factor analysis of variance and Bonferroni Dunn post hoc analysis.
Portions of femoral head extracts and conditioned medium from an equivalent wet weight of cartilage were adjusted to 0.1M Tris acetate, pH 6.5, and digested for 2 hours at 37°C with 0.01 units chondroitinase ABC (Seikagaku Kogyo, Tokyo, Japan) per 10 μg GAG. Samples were then dialyzed with ultrapure water for 16 hours at 4°C, freeze dried, dissolved in sodium dodecyl sulfate (SDS)–polyacrylamide gel electrophoresis sample buffer containing 2 mM dithiothreitol, and boiled for 5 minutes. Proteins were separated on 7.5% or 10% Laemmli gels and transferred to polyvinylidene difluoride membrane. Membranes were probed with various antibodies, including polyclonal anti-G1 domain (provided by Professor Tim Hardingham, University of Manchester, Manchester, UK), polyclonal anti-EGE and anti-PEN to the aggrecanase- and MMP-generated neoepitopes in aggrecan, respectively (34), monoclonal antibodies 2B6 and 3B3 to chondroitinase-generated chondroitin 4-sulfate and chondroitin 6-sulfate, respectively (35), and monoclonal antibody 8A4 to link protein (36) (provided by Professor Bruce Caterson, Cardiff University, Cardiff, Wales), monoclonal antibody 9A4 to the collagenase-generated neoepitope in collagen (37) (provided by Dr. Peter Mitchell, Pfizer, New York, NY), and polyclonal anti–MMP-3 and anti–MMP-13 (38) (provided by Professor Gillian Murphy, Cambridge University, Cambridge, UK).
Proteins were visualized using horseradish peroxidase–conjugated secondary antibodies (Dako, Carpinteria, CA) and an enhanced chemiluminescence kit (Amersham, Arlington Heights, IL). To validate the G1, G1-PEN, and G1-EGE Western blot analyses, an extract of cartilage from WT mice that contained 10 μg of sulfated GAG was digested overnight at 37°C with 484 μg/ml MMP-3 (a gift from Professor Gillian Murphy, Cambridge University) in buffer that contained 50 mM Tris HCl, pH 7.5, 100 mM NaCl, and 10 mM CaCl2. The digest was boiled, and doubling dilutions, beginning with 15-μl aliquots, were loaded onto SDS gels for Western blotting. The blots were analyzed by scanning densitometry and Quantity One software (Bio-Rad, Richmond, CA).
All data were analyzed using the StatView software package for Macintosh (SAS, Cary, NC). Results are expressed as the mean ± SEM. P values less than 0.05 were considered significant.
Using RT-PCR, ADAMTS-1 mRNA was detected in chondrocytes isolated from the mouse growth plate and articular cartilage (Figure 1). The expression of ADAMTS-1 in the growth plate increased as the cells underwent maturation and hypertrophy, as compared with the unchanged expression of the housekeeping gene GAPDH. The microdissection techniques that had separated PR, PH, and LH zone chondrocytes were validated by analyzing genes known either to remain constant between maturational zones of the growth plate (aggrecan) or to increase with hypertrophy (type X collagen and alkaline phosphatase) (Figure 1). To determine whether ablation of ADAMTS-1 caused a compensatory increase in the expression of other potential aggrecanases, RT-PCR was used to measure the levels of mRNA for ADAMTS-4, ADAMTS-5, and ADAMTS-9 extracted from femoral head cartilage. There was no significant increase in mRNA expression of ADAMTS-4, ADAMTS-5, or ADAMTS-9 in the cartilage of ADAMTS-1–KO mice compared with WT littermate controls (Table 2).
|Wild-type ΔCt||Knockout ΔCt|
|ADAMTS-4||17.2 ± 0.2||18.9 ± 0.4|
|ADAMTS-5||16.6 ± 0.4||17.9 ± 0.5|
|ADAMTS-9||16.0 ± 0.5||17.0 ± 0.3|
ADAMTS-1–KO mice were significantly smaller than their WT littermates at 3 weeks of age, with a mean ± SEM body weight of 10.3 ± 0.5 gm compared with 12.8 ± 0.4 gm, respectively (P = 0.0005). Although the KO animals were proportionally smaller than the WT mice, ADAMTS-1 was expressed in growth plates and articular cartilage, and we therefore examined these tissues in WT and ADAMTS-1–KO animals of different ages (Figure 2). No morphologic differences in proximal tibial or distal femoral growth plates were observed between WT and KO mice (Figure 2A). Similarly, no difference in proteoglycan (toluidine blue staining) in the growth plate or longitudinal calcified cartilage trabeculae in the metaphysis was observed (Figure 2B). No difference in cartilage total GAG content between WT versus KO mice (26.7–27.2 μg GAG/mg wet weight; n = 6) was confirmed using the DMMB assay in papain digests of isolated femoral heads. Histologically, the vascular invasion and development of secondary centers of ossification and articular cartilage were unaffected by ADAMTS-1 ablation (results not shown). Chondrocyte maturation and hypertrophy were unaffected in the growth plates of KO mice, as determined morphologically and by detection of normal type X collagen content and distribution (Figure 2C).
These studies suggested that ADAMTS-1 did not appear to have a critical function in aggrecan turnover in the process of bone and joint development and growth. To evaluate whether ADAMTS-1 might play a role in aggrecanolysis in joint disease, we studied cartilage breakdown in AIA, where aggrecanase cleavage is responsible for the early loss of aggrecan from cartilage (39). Following mBSA immunization, intraarticular saline failed to induce an inflammatory reaction or proteoglycan loss from cartilage in either WT or KO mice (Figure 3A, a and b). In contrast, intraarticular injection of mBSA induced marked synovitis, with exudation and neutrophil influx into the joint cavity accompanied by moderate (Figure 3A, c and d) to severe (Figure 3A, e and f) loss of proteoglycan (toluidine blue staining) from the uncalcified tibial and femoral articular cartilage. There was no significant difference in total arthritis scores (data not shown) or in individual scores for synovial exudate, synovitis, or cartilage proteoglycan loss (Figure 3B) between WT and ADAMTS-1–KO mice. No significant pannus or bone erosion was histologically evident in arthritic joints of either genotype (results not shown).
Although no difference in aggrecan loss was evident in AIA, it was possible that the marked inflammation associated with this model may have masked any subtle differences in aggrecan turnover between WT and ADAMTS-1–KO mice. We therefore examined aggrecan degradation in isolated femoral head cartilage in vitro under controlled and catabolically stimulated conditions. There was no difference in basal GAG release from femoral head cartilage from WT versus ADAMTS-1–KO mice on day 3 (Figure 4A) or day 7 (data not shown). GAG release increased significantly (P < 0.0001) in response to RA (56–58% of total) and IL-1 (57–64% of total), with no significant difference between genotypes on either day 3 (Figure 4A) or day 7 (data not shown). There was a reduction in toluidine blue staining in both the articular and uncalcified growth plate cartilage in RA- and IL-1–treated femoral heads compared with controls (Figure 4B). This GAG loss coincided with increased abundance of G1-EGE, but not G1-PEN, in the articular cartilage but not the growth plate of both RA- and IL-1–stimulated femoral heads (Figure 4B). In contrast, G1-PEN was primarily localized in the growth plate cartilage and appeared more abundant in cultures stimulated with RA (Figure 4B). No difference was detected in the GAG loss or the staining intensity or distribution of aggrecan neoepitopes in WT versus KO mice.
To further evaluate aggrecan proteolysis in this explant system, the metabolites released into the medium and present in the cartilage extracts were analyzed by Western blotting (Figure 5). The polyclonal anti-G1 domain antibody detected large molecular mass G1-bearing species (>250 kd) representing “intact” aggrecan, and smaller products proteolytically processed from the C-terminus (Figure 5A). Stimulation with RA and IL-1 decreased the content of intact aggrecan and, particularly in the case of IL-1, increased the intensity of a band at ∼75 kd in both WT and ADAMTS-1–KO cartilage. This 75-kd G1 metabolite colocalized with the aggrecanase-generated G1-EGE fragment, which was slightly increased in RA-treated and markedly increased in IL-1–treated cartilage (Figure 5A). RA treatment induced an increase in the G1-PEN fragment, which migrated at ∼60 kd. Aggrecan released into the culture medium consisted of the same intact and small molecular mass G1-bearing fragments as seen in the extracts (Figure 5B) along with chondroitin sulfate–bearing metabolites that lacked G1 (ranging in size from ∼90 kd to 250 kd) (Figure 5C). Western blots from separate experiments suggested that there was less G1-EGE in both extracts (Figure 5A) and medium (Figure 5B) of IL-1–treated ADAMTS-1–KO compared with WT cartilage cultures (KO/WT G1-EGE, as determined by scanning densitometry was 17–65% and 10–30% in extracts and medium, respectively).
The intensity of the 60- and 70-kd bands observed with anti-G1, particularly in RA-treated cultures, did not always match that detected by anti-PEN and anti-EGE antibodies, respectively. The relationship between mass of antigen and signal derived from ECL Western blots was compared for anti-G1 and anti-PEN antibodies. WT cartilage extracts were digested with an MMP at a high concentration to generate G1-PEN, and doubling dilutions of the digest were analyzed by Western blotting and densitometric scanning (Figure 5D). The results show that the line slope for the anti-G1 antibody is relatively flat, and only modest increases in the ECL signal are produced for each doubling in the mass of antigen.
In contrast, the line slope for the anti-PEN antibody was significantly steeper over the unsaturated range, and marked increases in the ECL signal were produced for doublings of antigen. This result may partly explain why G1 epitope in the RA-treated samples (Figures 5A and B, lanes 2 and 5) does not reflect the marked increase in PEN epitope (Figures 5A and B, lanes 14 and 17) detected with the anti-PEN antibody. Similar differences in the line slope of the anti-G1 antibody compared with the anti-EGE antibody account for differences in EGE band intensities detected with the 2 antibodies (Figures 5A and B). However, we cannot rule out that primary cleavage by other enzymes, such as m-calpain or cathepsin B (40, 41), or subsequent peptidase cleavage of G1-PEN and G1-EGE could generate similarly sized G1 metabolites that are not recognized by the neoepitope antibodies.
The in vitro culture system was further characterized by analyzing the release of other matrix proteins and proteinases (Figure 6). There was an increased release of link protein from both WT and ADAMTS-1–KO cartilage in response to RA and IL-1 stimulation (Figure 6A). Release of collagen from femoral heads on either day 3 or day 7 was minimal, as determined by hydroxyproline analysis (data not shown). However, collagenase-cleaved collagen was detected with antibody 9A4, and there was increased release of these fragments into the culture medium of cartilage stimulated with RA from both WT and ADAMTS-1–KO mice (Figure 6B). Increased levels of MMP-3 (Figure 6C) and MMP-13 (data not shown) were detected in IL-1– but not RA-stimulated WT and ADAMTS-1–KO cartilage cultures.
Proteolysis at both the N341–F and the E373–A peptide bonds in the interglobular domain of aggrecan has been reported in normal and arthritic articular cartilage (for review, see ref. 2). Cleavage at the latter site through the action of the aggrecanases appears to be the principal event in early pathologic aggrecan loss. In contrast, while both cleavage events have been detected in growth cartilage, the relative importance of these events in bone and cartilage growth and development remains unclear (4–6). Furthermore, there are conflicting data on the expression and localization of aggrecanases both in growth cartilage and normal and diseased articular cartilage. ADAMTS-5, but not ADAMTS-4, is expressed in hypertrophic growth plate chondrocytes in vitro (19), but G1-EGE colocalizes with furin-activated ADAMTS-4 and not ADAMTS-5 protein in growth cartilage in vivo (6). Similarly, both increased and decreased ADAMTS-1, ADAMTS-4, and ADAMTS-5 expression has been reported in human OA cartilage (15–17). Defining the proteinases responsible for aggrecan proteolysis in normal versus pathologic aggrecan turnover has important implications for the treatment of arthritic disease.
This is the first study to demonstrate that ADAMTS-1 is expressed in the chondrocytes of the growth plate and, furthermore, that it is up-regulated with chondrocyte hypertrophy. Attempts to demonstrate ADAMTS-1 protein in these cartilages using immunohistochemistry or Western blotting were unsuccessful (data not shown), reflecting the insensitivity of the methods used and the relatively low abundance of ADAMTS enzymes in cartilage (22). Nevertheless, the increase in ADAMTS-1 mRNA we observed parallels that shown for ADAMTS-5 mRNA in rabbit growth plate chondrocytes undergoing hypertrophic differentiation and maturation in vitro and contrasts with the unchanged and very low expression of ADAMTS-4 in these cells (19).
Increased expression of ADAMTS-1 has been reported in bone and osteoblasts in response to parathyroid hormone (PTH), PTH-related protein (PTHrP), and vitamin D3 (21). The increase in ADAMTS-1 in the hypertrophic chondrocytes may therefore be related to the increased expression of the PTH/PTHrP receptor that occurs in prehypertrophic and early hypertrophic chondrocytes (19, 42), although maximal ADAMTS-1 expression was not evident until late-stage hypertrophy. In addition to its proteolytic activity (7, 43, 44), ADAMTS-1 also has anti-angiogenic properties possibly associated with its ability to sequester vascular endothelial growth factor (26, 45, 46). ADAMTS-1 in the lower growth plate could therefore function to inhibit or delay vascular invasion or promote turnover of aggrecan or other matrix components, both of which could be significant for normal growth plate maturation. However, no apparent effects on chondrocyte maturation or growth plate development, as detected by morphology and type X collagen synthesis, were observed in ADAMTS-1–KO mice. This is similar to the lack of growth plate phenotype in ADAMTS-4–KO mice (22) and could be due to redundancy and compensation by other enzymes, such as ADAMTS-5, which are similarly up-regulated by thyroid hormones in the growth plate (19).
However, it is noteworthy that we did not demonstrate aggrecanase-generated G1-EGE in WT or ADAMTS-1–KO growth plates, even after catabolic stimulation that was able to increase this fragment in articular cartilage (Figure 4B). This may suggest that while ADAMTS enzymes are expressed in growth plate chondrocytes, their proteolytic activity is tightly regulated, and minimal aggrecanase cleavage of aggrecan occurs. Glasson et al (22) did demonstrate G1-EGE in growth plates of WT mice but not ADAMTS-4–KO mice; however, this was predominantly cell-associated rather than matrix staining and the mice were mature (14–18 weeks). Cleavage of aggrecan by aggrecanases in growth plates of young mice during their growth phase has not been demonstrated. Similarly, G1-EGE was localized to the secondary center of ossification but not the metaphyseal border of the primary growth plate in rats (5, 6). The proteolytic target of ADAMTS in the growth plate may be cartilage oligomeric matrix protein or fibromodulin (47, 48) rather than aggrecan. Alternatively, secondary proteolysis by MMPs or cathepsin B in the growth plate (but not articular cartilage) could convert the G1-EGE into G1-PEN.
ADAMTS-1 is expressed in normal and arthritic human articular chondrocytes (11, 14, 16). Although we demonstrated ADAMTS-1 expression in mouse articular chondrocytes, we did not observe any abrogation of aggrecan loss in AIA in ADAMTS-1–KO mice. This is consistent with evidence implicating ADAMTS-4 and ADAMTS-5 as the principal aggrecanases in pathologic conditions and with the finding that they are significantly more active than ADAMTS-1 against aggrecan in solution assays (7, 17, 49, 50). Furthermore, the in vitro release of aggrecan from the cartilage of ADAMTS-1–KO mice was identical to that of WT mice, suggesting that ADAMTS-1 has no significant role in this process. These findings with mouse cartilage are consistent with those of Tortorella et al (51), who showed that the soluble aggrecanase activity in conditioned medium from IL-1–stimulated bovine cartilage was not inhibited using an ADAMTS-1 antibody, but could be reduced by 75% and 15% using antibodies against ADAMTS-4 and ADAMTS-5, respectively. However, the lack of effect of ADAMTS-4 ablation on cartilage aggrecan metabolism or arthritis development (22) suggests that ADAMTS-5 may be the principal enzyme involved in these processes in mice.
In the present study, no compensatory increase in mRNA expression for other aggrecanases was found that would explain the lack of cartilage phenotype in the ADAMTS-1–KO mouse. Despite the lack of effect on GAG release, we did consistently observe less G1-EGE in IL-1–stimulated ADAMTS-1–KO cartilage cultures, suggesting that ADAMTS-1 may contribute to the generation of some of this neoepitope. This finding, in conjunction with a lack of change in GAG release, suggests alternative mechanisms for GAG loss in ADAMTS-1–KO cultures rather than compensation by other aggrecanases. There was no consistent difference in the release of intact G1-bearing aggrecan or the distribution of the C-terminal chondroitin sulfate–bearing catabolites between WT and ADAMTS-1–KO cartilage. Furthermore, the compensatory GAG release or potential loss of EGE epitope could not be explained by increased cleavage to generate PEN.
Culture of cartilage explants from humans and animals in which the mass of tissue is not limiting has been used to investigate proteolysis of aggrecan, collagen, and other matrix proteins. There is a need for a similar in vitro culture system using mouse cartilage to take advantage of the increasing availability of genetically modified mice to elucidate the mechanisms and pathways involved in cartilage catabolism. We have characterized a new and simple model in which cartilage and medium from a single femoral head (∼1 mg of tissue) from a 3–4-week-old mouse will provide sufficient aggrecan and aggrecan fragments for GAG analysis by the DMMB assay as well as 4–5 individual Western blots. Mouse cartilage responded in a similar manner to other species, not only with increased aggrecanase activity, but also with release of G1-bearing aggrecan and link protein in response to RA and IL-1, which is likely associated with increased hyaluronan release or breakdown (33, 52–55).
In contrast to other species, G1-PEN was increased in mouse cartilage, particularly following treatment with RA, due to the presence of growth plate as well as articular cartilage. Dose-response curves (not shown) demonstrated that GAG loss from mouse cartilage was maximal at 10 μM RA, whereas 1 μM elicits maximal GAG release from other species (33, 56–58). Collagen cleavage by collagenases was also increased in RA-treated cultures; however, this was not associated with increased MMP-13 even though RA has been shown to increase MMP-13 expression in rodent chondrocytes (59). This may be associated with insufficient sensitivity of the Western blot detection method in the present study. Alternatively, although MMP-13 appears to be the principal interstitial collagenase in rodents (60), MMP-14, which is also up-regulated by RA (59), may be responsible for the collagen cleavage observed in our explant cultures. Whether this collagenolytic activity occurs in the articular or growth plate cartilage was not determined.
In conclusion, although ADAMTS-1 is expressed in articular and growth cartilage and is able to cleave aggrecan at physiologically relevant sites, our results indicate that it does not play a significant nonredundant role in normal cartilage and bone development and growth. Furthermore, lack of protection against pathologic aggrecan loss in vivo or in vitro in the ADAMTS-1–KO mouse suggests that ADAMTS-1 is not a suitable target for modulating cartilage destruction in arthritis.