SEARCH

SEARCH BY CITATION

Keywords:

  • skeletal development;
  • MMPs;
  • MMP-13;
  • MMP activation;
  • anti-neoepitope antibodies;
  • type II collagen;
  • collagen fragments;
  • COL2;
  • CTX-II;
  • collagen biomarkers;
  • angiogenesis and endothelium

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSION
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

In long bone development, the evolution of the cartilaginous anlagen into a secondary ossification center is initiated by the formation of canals. The excavation to create the canals is achieved through lysis of the two major cartilage components, aggrecan, and the type II collagen (COL2) fibril. The present study examines the lysis of the fibril. Because it is known that matrix metalloproteinases (MMPs) cleave COL2 in vitro at the Gly775-Leu776 bond, it has been reasoned that, if such cleavage is detected in relation to the canals, it can be concluded that a collagenase is involved. Furthermore, because MMPs undergo change in domain structure with activation resulting in propeptide domain loss then, if such a loss is revealed in relation to the cleavage of COL2, this MMP is likely involved. The collective findings reveal that COL2 is attacked at the afore-described susceptible peptide bond at the surface of cartilage canals and, that MMP-13 cleaves it. Developmental Dynamics 238:1547–1563, 2009. © 2009 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSION
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

The skeleton evolves by means of two well-known mechanisms, described under the headings, “intramembranous” or “endochondral” bone formation (Jiang et al.,2002; Ballock and O'Keefe,2003). However, other less well understood mechanisms are also known to contribute. For example, some cartilages described as transient “ghost cartilages,” disappear before cartilage replacement in situ by intramembranous bone, tendons, or ligaments (Holmbeck et al.,2003; Holmbeck and Szabova,2006). Furthermore, at specified times in development, the cartilage of epiphyses of long bones disappear albeit selectively as described below. Surprisingly, the latter disappearance also requires the demise of the cartilage skeletal component. However, in contrast to “ghost cartilages,” only the central core of the cartilaginous epiphysis can be described as transient.

The disappearance of the epiphyseal, central core (which remains uncalcified throughout this evolution), takes place in the first 10 days in a young rat's life (Lee et al.,2001,2006). The removal (which necessarily requires the demise of type II collagen [COL2] fibrils) is achieved in the proximal end of the tibiae (that is, the current test model), through a series of attacks that involve cartilage canals which are shown to form after birth at 5–6 days of age, grow, and then join together to form a marrow space (days 7–8). Canal formations allow for the entry of blood vessels and osteoprogenitor cells. However, bone formation does not begin until the excavations are completed at approximately 9–10 days of age. Therefore, the current study of canal formation offers a unique opportunity to examine a model of anlagen remodeling that incorporates blood vessels and their growth (angiogenesis) but excludes bona fide bone formation (for review, see Roach et al.,1998; Lee et al.,2001,2006).

It is recognized that the degradation of COL2, the major collagen of cartilage, is crucial to canal formation, but the involved enzyme(s) and their specific role(s) are incompletely understood. It is also known that COL2 has the capacity to self-associate into complex fibrils. The monomers of COL2 consists of three identical α chains (α1(II)) (Eyre et al.,2002). These chains associate to form a central helical domain flanked by two nonhelical extensions, termed amino or carboxy domain telopeptides. After the onset of fibril formation, the telopeptides are involved in intermolecular cross-links that are catalyzed by lysyl oxidase(s) (Eyre et al.,2008). Stability is imparted to the fibril, and then the tissue, as a consequence of the cross-linking and additionally, by the covalent linkages of COL2 to minor collagens (that are also components of the fibril).

Among the 23 members of the MMP subfamily, five members share the ability in vitro and at neutral pH to cleave the triple helical domain of isolated COL2 molecules (Page-McCaw et al.,2007; Krane and Inada,2008; Murphy and Nagase,2008). The cleavage, at a specific Gly775-Leu776 bond, results in characteristic one-quarter and three-quarter fragments (Miller et al.,1976). Included in the five are three collagenases: collagenase-1 (MMP-1), collagenase-2 (MMP-8), and collagenase-3 (MMP-13; Mitchell et al.,1996; Billinghurst et al.,1997); one gelatinase: gelatinase A (MMP-2; Patterson et al.,2001); and one membrane type MMP: MT1-MMP (MMP-14; Ohuchi et al.,1997). All of these enzymes also attack a broad range of other substrates in vitro, making it a challenging task to definitively assign their roles in a tissue (Nagase et al.,2006).

MT1-MMP is the only MMP enzyme, based upon loss of function mouse models, to have been implicated in endochondral bone formation, intramembranous bone formation, “ghost cartilage” replacement, and cartilage canal excavations (for review, see Krane and Inada,2008). (Putative roles for MT1-MMP will be described below.) While loss of MT1-MMP function has been shown to modestly effect the development of the embryo, profound effects are observed after birth on both the skeleton and the soft tissues. Indeed, many MT1-MMP–deficient animals die prematurely (Holmbeck et al.,1999; Zhou et al.,2000). It is relevant that postnatal MT1-MMP–deficient animals show major defects in long bone epiphysis development and cartilage canal excavations (Holmbeck et al.,1999; Zhou et al.,2000). The responsible mechanisms remain to be explained.

Tissue alterations as a consequence of MT1-MMP deficiency have been attributed to the loss of collagenolytic activity for which the animals cannot compensate (Holmbeck et al.,2004; Itoh,2006). Accordingly, MT1-MMP has acquired the name “tethered collagenase” (Holmbeck et al.,2004). By comparing wild-type and MT1-MMP–deficient animals (either as isolated cells or intact tissue), it has been concluded that MT1-MMP may function as a tethered collagenase in fibroblasts (Holmbeck et al.,2004; Madsen et al.,2007), in living and apoptotic chondrocytes (Holmbeck et al.,2004; Wagenaar-Miller et al.,2007), in long bone osteocytes (for maintenance of osteocyte canalicular networks; Holmbeck et al.,2005), or in endothelial cells (during angiogenesis; Zhou et al.,2000; Chun et al.,2004). It is presently unknown which MT1-MMP–bearing cell(s) is/are involved in the excavation of cartilage canals.

While MT1-MMP can function as a collagenase, it has been shown in cell lines and in isolation to also exhibit the ability to amplify extracellular matrix proteolysis (Knauper et al.,1996b). MT1-MMP molecules form a complex requiring at least two MT1-MMP molecules (Itoh and Seiki,2006; Ra and Parks,2007). Once MT1-MMP complexes with a TIMP-2 molecule, it in turn, binds to a latent MMP-2 proenzyme molecule. The resulting molecular complex resides on the cell surface. A second MT1-MMP then activates the bound MMP-2. This activation results in the loss of the MMP-2 propeptide domain and a new amino terminus, thus defining a “neoepitope” that distinguishes the functional enzyme (Fig. 1; Murphy et al.,1999). However, MT1-MMP can also activate MMP-13 (Knauper et al.,1996b; Cowell et al.,1998). Although the activation process is less well known, it has been shown that the hemopexin domain of MMP-13 is essential for activation by MT1-MMP and that TIMP-2 is not involved (Knauper et al.,1996b,2002).

thumbnail image

Figure 1. Model postulated for proteolytic based matrix metalloproteinase (MMP) activation cascade. MT1-MMP activated intracellularly, is shown bound to the cell plasma membrane as active enzyme. MT1-MMP plays a central role in the activation of the soluble enzymes MMP-2, MMP-13, and MMP-9. MT1-MMP cleaves MMPs allowing them to generate a cascade of autolytic or heterolytic cleavages. In each case, peptide bond cleavage within the proenzyme leads to propeptide domain loss and a functional, mature enzyme. The unique amino-terminal neoepitope is identified for each functional enzyme. Finally, the functional collagenase enzymes known to cleave type II collagen (COL2) at the specific Gly775- Leu776 bond (three of the four enzymes shown), are represented as white coloured boxes. Diagram adapted from Murphy et al. (1999).

Download figure to PowerPoint

In summary, whereas the identity of the cellular source(s) for a putative collagenase(s) remain(s) speculative in canal excavations, two models can be envisioned for MT1-MMP involvement in cartilage lysis at the canal surface. The first predicts a direct role (MT1-MMP cleaves COL2). The second model in contrast, presupposes an indirect involvement which means, therefore, that another collagenase (that is soluble) is involved in the putative attacks on fibrillar COL2 (Fig. 1). The latter interpretation, by necessity predicts, furthermore, that either MMP-2 or MMP-13 could fulfill this role. Thus, either one or both of these MMPs enzymes could be involved. The purpose of the current work is to identify and to characterize the collagenase that putatively functions during canal excavations.

The present work reports the generation of anti-neoepitope antibodies prepared against specified MMPs (indicative of their activation) or against anticipated COL2 fragments (that are indicative of MMP enzymatic attack). After confirming their specificity, these antibodies have been applied to the surface of cartilage canals. The combined anti-neoepitope approaches (directed either against different collagenase enzymes or various sites for COL2 cleavage on the fibril) have previously not been done in parallel in a tissue. The results (some acquired with the aid of electron microscopy), show that MMP-13 is the only collagenase enzyme capable of producing the COL2 fibril cleavage that has been uncovered at the susceptible Gly775-Leu776 peptide bond at the surface of cartilage canals. Therefore, the collective findings lend support to an indirect model for collagenolysis. Furthermore, the cellular source(s) is/are revealed thereby clarifying how canal excavations are regulated. Finally, the attacks on the fibril have been revealed with new precision in situ, and one outcome has been extended to search for, and quantify, one COL2 fragment in the urine. The collective results reveal that the sought-after fragment is detected at both the canal surface and in the urine, thereby suggesting that the global break-up of epiphyses during postnatal development may contribute to elevated COL2 fragments reported here in the urine of young animals of corresponding age.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSION
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

The approach has been to first identify the tissue sites where COL2 may be subjected to proteolytic attack in situ. This has been achieved in two steps. The first has been to prepare antibodies to identify MMP-generated COL2 breakdown fragments. The second has been to apply these antibodies to the tissue and to resolve the resulting reactions with the aid of light and electron microscopic immunohistochemical techniques.

Anti–COL2 Fragment Neoepitope Antibodies: Characterization and Tissue Immunostaining

Sodium dodecyl sulfate/polyacrylamide gel electrophoresis and western blotting.

As reported by Billinghurst et al. (1997), the enzymatic cleavage of helical COL2 by the recombinant, activated form of the MMP-13 enzyme takes place in two stages. The first cleavage, at the Gly775-Leu776 bond, yields one-quarter long fragments with a new amino terminus (LAGQR). These fragments are then subjected to a second cleavage that removes a tripeptide and results in a slightly shorter one-quarter form that presents a new amino terminus (QRGIV), hence the term “one-quarter short fragments.” We prepared antibodies against peptides that were identical to the sequences of one-quarter fragment amino termini. In addition, antibodies were prepared to the sequence EKGPDP, which detects a carboxy neoepitope present on fragments indicative of C-telopeptide domain cleavage (CTX-II; Eyre et al.,1996; Atley et al.,1998; Christgau et al.,2001). The latter is known to be a direct product of MMP cleavage in general and the MMP-13 enzyme, in particular. It is found in urine as a stable end-product of degradation (O'Kane et al.,2006; Charni-Ben Tabassi et al.,2008). Collectively, the anti–COL2 neoepitope antibodies we have prepared immunostained TCB COL2 fragments but did not immunostain intact α1(II) chains confirming in each case, that a COL2 neoepitope was required for staining of the immunoblots (Figs. 2, 3).

thumbnail image

Figure 2. Characterization of anti–type II collagen (COL2) anti-neoepitope antibodies using time course cleavage of human triple helical COL2 by rHuMMP-13 to generate long and short COL2 one-quarter fragments. A: Coomassie Blue stained samples separated under reducing conditions with sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS/PAGE) reveal undigested COL2 (lane 1) and digested COL2 for 0.5 (lane 2), 1 (lane 3), or 3 hr (lane 4). Right margins indicate the positions of the α1 (II) chains and the single forms of TCA and TCB chain fragments produced by MMP-13 cleavage of intact COL2. B,C: SDS/PAGE immunoblotting reveals immunostaining in TCB chain fragments.

Download figure to PowerPoint

thumbnail image

Figure 3. A,B: Matrix metalloproteinase (MMP) -13 digestion of COL2 characterized by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS/PAGE) stained by Coomassie Blue (A) and Western blot (B). The anti-EKGPDP (CTX-II) antibodies immunostain the carboxy terminal neoepitope in TCB fragments.

Download figure to PowerPoint

Histology of the canal blind end.

In the hindlimb proximal tibiae under study, the cartilage canals (usually two to four in number) emerge in the epiphysis in postnatal animals at 5–8 days of age. These structures begin as small depressions at the surface and thereafter develop into canals that grow in a radial direction through excavations at their tip or blind end (for review, see Lee et al.,2001). The canal base (or growing end of the canal) presents a modified cartilage layer described as the “preresorptive cartilage layer.” In proximal tibial epiphyses (the epiphyses under study), this layer measures approximately 8.3 μm ± 2.0 (range, 3.3–14.3). It exhibits a uniform loss of metachromasia indicative of aggrecan loss. It also contains terminal chondrocytes and collagen fibrils observed at various stages of degeneration (Lee et al.,2001,2006; Davoli et al.,2001). The preresorptive cartilage layer is defined by two borders, a distal border located adjacent to normal cartilage and a proximal border related to the cells in the lumen of the canal blind end (Figs. 4A, 5A). Along the distal border, the fibrillar collagen appears typical, whereas the fibrils observed at the proximal border appear swollen and fragmented before replacement by matrix debris and then complete resorption. (The latter takes place at the blind end of the cartilage canal.) Because epiphyses normally stain darkly with the dye Toluidine blue, the abrupt loss of staining at the distal border dramatically emphasizes the beginning of the preresorptive cartilage layer. Typically, the epiphysis contains scattered epiphyseal chondrocytes that exhibit no arrangement into cell columns (Fig. 4A, arrowhead). Moreover, the epiphyseal cells poised next to the preresorptive layer (soon to become terminal within the layer) are relatively small, distinctive chondrocytes that are rich in glycogen, deficient in type X collagen protein but do exhibit pale stained nuclei that immunostain with anti-aquaporin 5 antibodies (Fig. 6A–D). (Aquaporins are integral membrane proteins that have a predicted role in osmotically driven fluid transport. The presence of an aquaporin within cell nuclei is the only feature of the epiphyseal chondrocyte, so far shared between it and the hypertrophic chondrocyte of the primary growth plate.) Furthermore, whereas epiphyseal chondrocytes present as terminal cells within the preresorptive layer, the chondrocytes of the primary growth plate become terminal along the metaphyseal border of the primary growth plate (Fig. 6E–H). Our collective observations draw attention to the epiphyseal chondrocytes in the path of the advancing canals where their distinctiveness is confirmed based on cell structure and staining properties (Roach and Clarke, 1999,2000; Lee et al.,2001). A notable new finding, however, is the absence of immunostaining with anti–type X antibodies, which reveals that the specified epiphyseal cells appear to become terminal without expressing type X collagen protein.

thumbnail image

Figure 4. A–N: The canal blind end stained with the dye Toluidine blue (A) or with various immunoglobulins to reveal COL2 cleavage fragments (C–E) or matrix metalloproteinase (MMP) proteinases (G–N) or nonimmune IgGs to serve as controls (B,F). All of the panels represent a series of related sections prepared from the same tissue of a 5-day-old tibia after exposure to various antibodies (as identified above), or to nonimmune IgG (B,F). The latter (controls) are unstained confirming the specificity of the reactions. The yellow line in A and G denotes the distal border of the preresorptive cartilage layer (PRL). In contrast (also in the same panels), the cellular processes of canal cells (arrows) define a proximal border of the preresorptive cartilage layer. TC, denotes terminal chondrocyte; NC, normal cartilage; types α, β, and γ cells are identified in the canal lumen (A); arrowhead observed at right lower corner points to a typical, epiphyseal chondrocyte. In the remaining panels, a brown reaction product is indicative of antibody binding (C–E,G–N). In J, the heading MMP-2 P/A denotes Pro/Active enzyme recognition by the applied antibodies. MT1-MMP (G–I) reactivity is indicated in luminal cell processes ([UPWARDS ARROW], compare ascending arrows in A, to those shown in G, H, and I). TC, denotes terminal chondrocytes (A, B, F, and G). In B and F, CL denotes canal lumen. Scale bars = 10 μm in A (applies to F,G), 10 μm in B (applies to C–E,H–N).

Download figure to PowerPoint

thumbnail image

Figure 5. A–E: Electron micrographs revealing the normal structure of the canal blind end (A,B) or electron microscopic immunostaining of COL2 cleavage fragments (E) or matrix metalloproteinase (MMP) -13 neoepitopes (D) along the proximal border of the preresorptive cartilage layer or immunostaining with nonimmune IgG, as a control (C). In A, preresorptive cartilage layer (PRL) denotes the preresorptive cartilage layer whose range is defined by the single bracket ending above as proximal border and below, as distal border. Cellular features are emphasized, the degenerating terminal chondrocyte (TC) with asymmetrical cell membrane (APM) is identified (the APM is approximately twice normal thickness and unlike most of the remaining plasma membrane, it retains direct contact with the extracellular matrix) and, canal cells, specifically the cell processes (CP) of the β cell (A, boxed in area) and, those of α cells (B, arrows); both invade the proximal border of the preresorptive cartilage layer. Panels D and E (arrows), respectively, reveal immunostaining of the cellular processes (CP) with antibodies identified against relevant targets. Nonimmune IgG control (C) shows no immunostaining. The antibodies applied in D and E produce electron dense staining on cell processes (CP), on fibrillar collagen (emphasized by parallel arrows, D) but not on matrix debris (compare with D, E, and C). In A, the asterisks identifies the prominent nucleolus of a β cell and WBC, denotes, a white blood cell. Scale bars = 2 μm in A, 1 μm in B, 1 μm in C (applies to D,E).

Download figure to PowerPoint

thumbnail image

Figure 6. A–H: Five-day-old proximal tibia comparing epiphyseal chondrocytes (A–D) to primary growth plate chondrocytes (E–H) after routine staining with dyes (A,B,E,F) or immunostaining with antibodies (C,D,G,H). CL denotes canal lumen in A, C, and D. The box in A identifies the location of the chondrocytes featured in panel B. The upward pointing arrows in C, D, and H identify various cell nuclei. HC, panel E, denotes hypertrophic chondrocyte. Scale bars = 50 μm in A, 20 μm in B (applies to C,D), 20 μm in E (applies to F–H).

Download figure to PowerPoint

Three types of mononucleated cells have been identified within the canal lumen at the canal base in confirmation of an earlier report published from this laboratory (Lee et al.,2006). These are listed in order of frequency: (i) cells named “type α” that are the regular endothelial cells lining canal blood vessels (Figs. 4A, 5A,B), and are joined as expected by tight junctions; (ii) cells named “type β” cells (Figs. 4A, 5A) that are attached by junctional complexes to type α cells next to which they are often located; (iii) cells named “type γ” cells (Fig. 4A) that bear some resemblance to the cathepsin-B–rich cells resorbing the primary growth plate and were named “septoclasts” (Lee et al., 1995). Finally, osteoclasts were only occasionally encountered in the canal lumen and, therefore, are unlikely to play a major role in local resorption events as was previously reported by Cole and Wezeman (1987).

COL2 breakdown products at the canal blind end.

Frozen tissue sections have been prepared from animals of proximal tibiae of rats aged between 5 and 8 days. The sections have been immunostained with anti–COL2 neoepitope antibodies and examined first by light and then by electron microscopy techniques using horseradish peroxidise–diaminobenzidine (HRP-DAB) techniques to reveal the tissue sites for antibody binding (Lee et al.,1998). The results respectively show that the anti–COL2 one-quarter long, the anti–COL2 one-quarter short, and the anti–CTX-II antibodies all immunostain the canal blind end (Fig. 4C–E). The absence of reactivity with nonimmune IgG confirms the specificity of these reactions (Fig. 4B). In all cases, the immunostaining corresponds to the modified cartilage tissue described under the name “preresorptive layer” (Lee et al.,2001).

Electron microscopic examination of the same immunoreactions has been achieved by briefly exposing HRP-DAB–treated tissue sections to osmium tetroxide (1% v/v in 0.1 M sodium cacodylate; Lee et al.,1998) before dehydration and embedding in Epoxy resin. The results reveal that although nonimmune IgG produces no staining (Fig. 7A), the anti–COL2 fragment antibodies stain fibrillar collagen in the preresorptive layer. The approach has been to examine the features of the immunoreactions at the cartilage layer borders (Fig. 5A). Along the distal border, small dark, dot-like reactions are revealed by both anti–COL2 fragment antibodies. (A representative example of the COL2 one-quarter long reactions is shown in Fig. 7B.) The dots, indicative of intrahelical COL2 domain cleavage, exhibit a periodicity of 56 nm ± 5.6 (mean ± SD; n = 100 fibrils). Of interest, the observed staining pattern differs in specific ways for the anti–CTX-II antibodies (compare arrows in Fig. 7B,C). Whereas anti–CTX-II antibodies also immunostain fibrils, these antibodies stain material revealed between the fibrils. The collective data acquired along the distal border demonstrate that (i) a collagenase attacks the COL2 fibrils, (ii) the enzyme is soluble, (iii) the attacks are periodic, (iv) the one-quarter fragment neoepitopes are confined to the fibril, and finally, (v) the CTX-II neoepitopes are distinctively revealed on the fibril and at interfibrillar sites where they may represent stable COL2 tissue products.

thumbnail image

Figure 7. Electron microscopic immunostaining of type II collagen (COL2) cleavage fragments and various forms of matrix metalloproteinase (MMP) -13 along the distal border of the preresorptive layer. A: Nonimmune IgG controls show no reaction. B–E: The antibodies applied elicit an orderly deposition of electron dense dots along collagen fibrils, as emphasized by parallel arrows. Scale bar = 0.5 μm.

Download figure to PowerPoint

Proximal limit of the preresorptive cartilage layer.

Whereas the immunostaining with the anti–CTX-II antibodies remains intense along both borders of the layer (Fig. 4E), the discrete, dot-like, periodic fibrillar immunostaining observed previously (with the anti–COL2 one-quarter long antibodies) is replaced by a coating now highlighted by dark HRP-DAB reactions observed along the entire fibril length (compare Fig. 7B to Fig. 5E). In addition, canal cell processes (CP) are immunostained with the anti–COL1 one-quarter long fragment antibodies (Fig. 5E; arrows). Because the same processes are unstained with nonimmune IgG, the specificity of these reactions is confirmed (compare arrows in Fig. 5E and 5C). Hence, some collagenolysis is associated with the processes of canal cells. However, given the bulk of the COL2 intrahelical cleavage uncovered here between the two borders of the preresorptive layer, only a soluble collagenase could achieve these cuts.

MT1-MMP enzyme.

Because sheddases' can release membrane anchored MT1-MMP, the possibility has been considered that MT1-MMP may be the collagenase present as a cleaved, soluble form (Itoh and Seiki,2006). Therefore, MT1-MMP has been localized using the same HRP techniques described beforehand. However, the approach has only revealed definitive immunostaining in relation to canal cell processes (compare arrows in Fig. 4A and 4G). Hence, no MT1-MMP is present where most of the fibrils were shown to be cleaved.

Anti-MMP Neoepitope Antibodies: Characterization and Tissue Staining

Having established that COL2 fibrillar collagen is subjected to intrahelical domain cleavage at the Gly775-Leu776 bond between the two borders of the preresorptive cartilage layer, the next step has been to search for the collagenase that makes the cut. Hence, antienzyme neoepitope antibodies were prepared and, after characterization, applied to the surface of the canal blind end. As before, antibody binding has been visualized with the aid of light and electron immunohistochemical techniques.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis and western blotting.

Two putative soluble collagenases, MMP-2 and MMP-13, were investigated (Fig. 1). As reported previously, MMP-2 and MMP-13 activations are accompanied by the loss of amino acids at the amino terminus as a consequence of proteolytic processing (Stetler-Stevenson et al.,1989; Okada et al.,1990; Knauper et al.,1996a). Activation can be achieved by the organomercurial agent APMA (4-aminophenylmercuric acetate), which results in autolytic cleavage. It can also be achieved by a specific cell surface event as summarized in Figure 1 (Murphy et al.,1999). Amino-terminal sequencing of the proenzyme before and after activation revealed the new amino terminus for MMP-2 (89YNFFP…) and, for MMP-13 (86YNVFP…) (Stetler-Stevenson et al.,1989; Okada et al.,1990; Knauper et al.,1996a). Comparisons made between relevant species (human, rat, and mouse) showed complete conservation (Lee et al.,2006, Supp. Table S1, which is available online). Antipeptide antibodies have been prepared to detect the new amino termini of MMP-2 or MMP-13. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS/PAGE) and Western blotting techniques demonstrate that the resulting antibodies immunostain the neoepitope of the active MMP-13 form but do not immunostain the larger, latent proenzyme form. Finally, the anti–MMP-13 antibodies do not cross react with the N-terminal sequences that are known to characterize MMP-2 neoepitope forms (Figs. 1, 8A,B), thereby confirming their specificity.

thumbnail image

Figure 8. Characterization of anti–matrix metalloproteinase (MMP) neoepitope antibodies using partial, autoactivated forms of rHu MMP-13 (A) or rHu MMP-2 (B) enzymes. A: Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS/PAGE) and Western blot of the rHuMMP-13 enzyme (lane 1) immunostained by antibodies described previously (Davoli et al.,2001) reveals the proenzyme (60 kDa), the processed active form (48 kDa), and the intermediate form (55 kDa). Left margin indicates low molecular weight standards. Anti–MMP-13 enzyme antibodies prepared to the peptide YNVFPggc, reveal staining of the neoepitope in the active enzyme band (lanes 4–8). Anti–MMP-13 proenzyme antibodies prepared against the peptide HPVTLAGILKKSTVgc stain both the rHu proenzyme band (lane 2) and the rat proenzyme (lane 3). The rat proenzyme, obtained from a guanidinium hydrochloride extract prepared from the rat growth plate (Lee et al.,2001), presents two bands indicative of glycosylated or unglycosylated proenzyme. B: The active form of the rHu MMP-2 is identified as a 62-kDa band (lane B1) (Hipps et al.,1991), and its neoepitope is recognized by the anti–MMP-2 enzyme antibodies prepared to the peptide YNFFPggc (lanes B5–9). The absence of any reactivity in lanes B2–4 confirms that the anti–MMP-13 antibodies do not immunostain the active forms of MMP-2.

Download figure to PowerPoint

MMP-2 enzyme.

The application of sheep anti–MMP-2 antibodies (shown to recognize pro and activated forms of MMP-2 by immunoblotting, Fig. 8B, lane 1) uncovers MMP-2 protein in the preresorptive cartilage layer (Fig. 4J). Surprisingly, when the anti–MMP-2 neoepitope antibodies are alternatively applied, these antibodies do not immunostain the canal blind end (Fig. 4K). In the hopes of reconciling these two observations, anti–MMP-2 proenzyme antibodies have been applied. These antibodies show the MMP-2 proenzyme in chondrocytes and canal cells, revealing these cells are involved in MMP-2 synthesis. However, no proenzyme was detected in the preresorptive cartilage layer (Fig. 4L). Accordingly, the collective data demonstrate that (i) the MMP-2 proenzyme is synthesized at this cartilage remodeling site and, (ii) no MMP-2 neoepitope is shown present but rather that, extracellular MMP-2 is present either as the incompletely processed proenzyme form or it has undergone additional autocatalytic cleavages, which result in neoepitope loss.

MMP-13 enzyme.

To continue the search for the collagenase shown to be active in the tissue, anti–MMP-13 anti-neoepitope antibodies have been prepared and applied to the canal surface. These antibodies result in the appearance of a dark, brown, narrow band at the blind end of canals (Figs. 4M, 9E–H) and the wall, enclosing the forming marrow space which is thinly but sharply stained (Fig. 9I,J). No immunostaining is produced with nonimmune IgG thereby confirming the specificity of the reactions (Fig. 9A–D). Furthermore, direct comparisons made between the activated MMP-13 enzyme in Figure 4M and the COL2 one-quarter fragments or CTX-II fragments (respectively, shown in Fig. 4C and 4E), makes it clear that the MMP-13 neoepitope and the COL2 fragments are localized in precisely the same sites.

thumbnail image

Figure 9. Epiphyses immunostained with anti–matrix metalloproteinase (MMP) neoepitope antibodies or nonimmune IgG. A–H: The tissue shown in panels A–D, E–H, E′–H″, or E″–H″ has been prepared sequentially as sections from the epiphyses of the same 5-day-old tibia. I–J: An older 7-day-old tibia. A–D: Upper panels confirm the absence of color in nonimmune controls. Upward pointing arrowheads (▴) identify cartilage canals and specifically the canal blind end, which is the dedicated site for cartilage remodeling. Descending arrows, in contrast (E,I,E″) identify the metaphyseal border of the primary growth plate, which is immunostained. The epiphyses are stained with the anti–MMP-13 neoepitope antibodies at 5 or 7 days of age. For comparison, the serial sections (E′,F′,G′,H′) have been treated with anti–MMP-2 neoepitope antibodies. The tissue is unstained. In contrast, the anti–MMP-9 neoepitope antibodies described in previous work (Lee et al.,2006) stained the canal blind end (E″,F″,G″,H″). Scale bar = 400 μm in A,E,I,E′E″, 100 μm in B,C,F,G,J,F′,G′F″,G″; 50 μm in D,H,H′,H″).

Download figure to PowerPoint

With the aid of the electron microscope, it has been possible to identify with increased precision the tissue target. Antibody immunostaining along the distal border reveals the MMP-13 enzyme along the COL2 fibril (Fig. 7D) where the reactions appear discrete at regular intervals (58 nm ± 11; mean ± SD; n = 100 fibrils). In contrast, at the proximal border, the activated enzyme is revealed by HRP-DAB reactions that appear as dark, large globular clumps. Their periodicity (observed on swollen fibrils) is partially retained (compare the arrows in Figs. 7D to 5D).

The described results demonstrate that (i) the MMP-13 enzyme bearing the neoepitope binds to the fibril, (ii) the binding is periodic, (iii) this binding is enhanced before complete resorption, and lastly, (iv) the binding on the fibril, irrespective of border, appears to correspond to the sites where intrahelical COL2 is cleaved.

MMP-9 enzyme.

It is known from in vitro experimentation with purified proteins that gelatinases not only degraded collagen that had been cleaved by collagenases but also worked synergistically to facilitate the overall rate of collagen degradation (Harris and Krane, 1972; Murphy et al.,1982). In the present study, we have confirmed the presence of active forms of MMP-9 within the preresorptive layer (Davoli et al.,2001; Lee et al.,2006). Here for the first time, furthermore, we have applied the anti-neoepitope approach to the study of the MMP-2 enzyme and shown the absence of any counterpart active, MMP-2 forms within the preresorptive cartilage layer. Hence, two soluble MMP proteinases (MMP-9 and MMP-13) were identified here and shown to exhibit domain structure change indicative of activation. Consistent with the results shown earlier, however, only one of the activated proteinases shown present, that is, MMP-13 qualifies as a collagenase (Fig. 1). The collective data, therefore, suggest MMP-13 is the collagenase enzyme that cleaves the fibrils.

MMP-13 Regulation by Local Proenzyme Synthesis and Activation

MMP-13 neoepitope and cells.

To begin to determine the role of cells in regulating collagenolysis of COL2, the approach has been to first identify the MMP-13 neoepitope in relation to nearby cells (Figs. 5, 10, 11). As summarized in these various figures, the MMP-13 neoepitope is shown present on the plasma membrane of canal cell processes (Fig. 5D; specifically endothelial and β cell processes), along the unusual asymmetrical plasma membrane of the terminal chondrocytes described earlier (Fig. 11B–D; where the content of close-by endoplasmic reticulum may also be stained, Fig. 11D), or in the cytoplasm of some β cells (Fig. 10H). The diverse distribution of the MMP-13 cellular neoepitope suggests that at least three kinds of cells are involved in some aspect of MMP-13 processing.

thumbnail image

Figure 10. A–H: Canal blind end immunostained with anti–matrix metalloproteinase (MMP) -13 neoepitope antibodies (B,D,H), anti-MMP-13 proenzyme antibodies (C,E,G), or nonimmune IgG (A,F) to compare the respective cellular staining observed in close by sections of the canal base at 5 days of age. The preresorptive cartilage layer is prominently immunostained with both the anti-neoepitope and the anti-proenzyme antibodies (compare B and D to C and E), although the latter staining is confined mainly to the distal half of the cartilage layer (E, blue asterisks). The black arrows in C identify some prominently immunostained canal cells referred to as “β cells.” The single, white arrow below identifies an immunostained epiphyseal chondrocyte (C). In E, observed at higher magnification, the immunostaining of “α cells” is revealed. In G and H of electron micrographs, the symbol β (placed on the cell nuclei) denotes β cell, α denotes the endothelium, and γ identifies a rare septoclast or osteoclast (G). The circle in G circumscribes an electron dense reaction product that is indicative of MMP-13 proenzyme in the β cell Golgi apparatus. The immunostaining of the secretory organelles of “α cells” is also confirmed by arrows. RBC, denotes red blood cells (F). CP, denotes immunostained cell processes (E). Arrowhead (D) identifies the immunostained asymmetrical plasma membrane of a terminal chondrocyte. Finally, in the electron micrograph of H (revealing the nature of the β cell immunostaining observed in D), dark reaction products indicative of the MMP-13 neoepitope are identified by arrows in small vesicles and within the cell endosome (E). Scale bar = 50 μm in A–C, 10 μm in D–G, 1 μm in H.

Download figure to PowerPoint

thumbnail image

Figure 11. A–D: The matrix metalloproteinase (MMP) -13 neoepitope (B,C,D) or nonimmune IgG (A) staining of terminal chondrocytes (TC) within the in the preresorptive cartilage layer (PRL). (In the examples shown, whereas the cartilage layer is well defined by electron microscopy, the canal lumen appears empty because of tissue damage. Although canal cells have been torn, their processes are retained in the layer after fracture as identified by the asterisks in A–C.) A: Nonimmune IgG control shows no reactivity. B,D: The black arrows identify the intensely immunostained asymmetrical plasma membrane of the terminal chondrocyte (B,D), whereas the white arrows in B identify an asymmetrical membrane which has formed but it is not immunostained. D: Immunostaining within the r-ER cisternae of the terminal chondrocyte is also indicated. CL, canal lumen; PRL, preresorptive layer; TC, terminal chondrocyte; ER, endoplasmic reticulum. Scale bar = 1 μm in A,B, 0.5 μm in C,D.

Download figure to PowerPoint

MMP-13 proenzyme and cells.

To identify the cellular source(s) for the extracellular MMP-13 enzyme, antibodies have been applied with defined specificity for the MMP-13 propeptide domain (Fig. 8A, lanes 2 and 3). Three cell types have been shown to be immunostained and thus involved in MMP-13 synthesis; the cells are listed according to their frequency: (i) the cells named “type α,” which are the regular endothelial cells lining canal blood vessels (Fig. 10E,G), and are joined as expected by tight junctions; (ii) the cells named “type β” cells (Fig. 10C,E,G) that are attached by junctional complexes to type α cells; and (iii) the occasional chondrocyte (one example of an immunostained epiphyseal cell is demonstrated by the white arrow shown in Fig. 10C). Surprisingly, in addition, we have also observed intense, polarized immunostaining indicative of MMP-13 proenzyme associated either with canal cell processes or, the asymmetrical plasma membrane of the terminal chondrocyte (data not shown). Finally, the MMP-13 proenzyme is observed extracellularly in the preresorptive cartilage layer (Fig. 10E, blue asterisks), where it exhibits fibrillar periodicity (Fig. 7E). The collective data indicate that (i) different cells are involved in MMP-13 synthesis (principally within the canal lumen) and that (ii) some MMP-13 proenzyme finds its way to extracellular COL2 fibrils where it binds.

CTX-II Concentrations in Urine Vary Inversely With Animal Age and Growth Status

Since our earlier immunolocalization results have identified CTX-II as a product of COL2 degradation, we searched for available quantitative data on amounts of CTX-II known to be present in the urine of young, postnatal animals (Hoegh-Andersen et al.,2004; Ishikawa et al.,2004). Because this information was not available, we acquired voluntary, spot urine samples from the offspring of 5 C57BL6 litters that were part of an active, mouse breeding colony. Ten or more urine samples have been acquired at specified ages (5, 9, 26, and 61 days of age, Fig. 12A). The respective mean values (expressed as CTX-II μg/mmol creatinine ± SEM) were 574.3 ± 47.26 (n = 10), 467.8 ± 24.45 (n = 12), 45.99 ± 4.27 (n = 12), and 8.54 ± 0.39 (n = 10). One-way analysis of variance (Kruskal-Wallis test) revealed significance (P< 0.0001). Dunn's multiple comparison tests determined no significant difference at 5 and 9 days of age, but significance was acquired with increasing age as the urine CTX-II concentration decreased. We conclude that CTX-II levels are elevated in the urine of postnatal animals.

thumbnail image

Figure 12. A,B: Correlation of urine CTX-II levels (A) or tail length (B) with age. Spot urine samples from normal mice were collected. CTX-II levels were measured by enzyme-linked immunosorbent assay and normalized to creatinine. A: Circles denote mean of individual normalized measurements (determined in duplicate). Color dots reflect individual donors represented according to sex (red, female; blue, male). B: Tail length is expressed as mean ± SEM.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSION
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Two parameters have been examined to determine the particulars of fibrillar COL2 cleavage in vivo, the presence of (i) COL2 fragment breakdown neoepitopes and (ii) neoepitopes that are acquired by specified collagenases upon activation. Both approaches rely on anti-neoepitope antibody immunostaining techniques. Whereas the former approach is well known, the latter is new to the study of collagenases in tissues (Mort and Buttle,1999; Lee,2006; Lee et al.,2006). This study represents the first to uncover a direct change of domain structure of one MMP collagenase in situ and to link the molecular architecture of the processed enzyme to COL2 breakdown fragments in space and time.

It was not surprising to us that the acquisition of a new amino-terminus would be revealed for MMP-13 in situ (Mort and Buttle,1999; Lee,2006; Lee et al.,2006). After all, we had validated previously the loss of the MMP-9 prodomain with the acquisition of MMP-9 activity as proof of concept in the same tissue (Lee et al.,2006). However, given the similarity of the amino acid residues between MMP-2 and MMP-13, it was surprising to us that the anti–MMP-13 neoepitope antibodies described here, allowed us to immunostain the MMP-13 neoepitope (YNVF) but not the MMP-2 neoepitope (YNFF). We attribute the discrimination to the presence of two, large repeating phenylalanine residues present in the MMP-2 sequence. The second phenylalanine residue may preclude the MMP-2 enzyme from fitting into the anti–MMP-13 antibody binding pocket. As a consequence, the specificity of the MMP-13 neoepitope antibodies has afforded us a unique opportunity to evaluate the role of MMP-13 in anlagen remodeling.

Until this study, most investigators, including us, had presumed that the COL2 of the cartilage was subjected to proteolysis. The possibility that the MT1-MMP enzyme was cutting COL2 had been proposed (Holmbeck,2004; refer also to Holmbeck et al.,1999; Zhou et al.,2000). However, neither this, nor any other information, was decisive enough to link any collagenase to the dramatic changes exhibited by collagen fibrils, such as their extensive degradation in the 3- to 14-μm-thick preresorptive layer described by Davoli et al. (2001).

Previous knowledge had been obtained from knockout mice. However, the elimination of a functional MMP-13 enzyme in this manner had little or no apparent effect on secondary ossification center formation (Inada et al.,2004; Stickens et al.,2004). Only when the loss of the MMP-13 gene was combined with the loss of the MMP-9 gene in double knockout mice, was the development of the epiphysis affected.

The current study allowed us to examine canal excavation without interfering with normal gene function. The collective results revealed involvement of MMP-13. The approaches we have applied have shown where the MMP-13 enzyme was produced, underwent activation and exerted effects (Fig. 13). First, the anti–MMP-13-proenzyme antibodies revealed the proenzyme in four sites (Fig. 13D). One was the cytoplasm of the “α cell.” The other was the cytoplasm of the “β cell.” Another was the asymmetrical membrane (APM) of the terminal chondrocyte (as identified in Fig. 13A). The first two were taken to be MMP-13 producer cells. (It remains unclear whether or not the terminal cell is the cellular source for the MMP-13 proenzyme uncovered on its own asymmetrical plasma membrane.) The other proenzyme site was the preresorptive layer, where electron microscopic examination pinpointed it to collagen fibrils (Fig. 7E). The interpretation was that, some MMP-13 escaped cell-associated activation and migrated into the preresorptive layer where it bound to fibrillar collagen. That the anti–MMP-13 proenzyme antibodies produced reactions within the preresorptive layer was surprising to us. This stemmed from the perception in the literature that collagenase activation was thought to precede collagenase binding to substrate (as reviewed by Hembry et al.,1986). Herein, however, immunoreactions indicative of the proenzyme appeared along fibrils that approached a periodicity of 67 nm. It was concluded that MMP-13 proenzyme was binding to specific sites on the fibril. However, this binding was found to be confined to the distal half of the preresorptive cartilage layer that is, where the fibrils exhibited no morphological evidence indicative of fibril demise (Fig. 13D). These combined observations suggested that the binding of the MMP-13 proenzyme was an early event in the break-up of the fibril.

thumbnail image

Figure 13. A–D: A summary model of the immunostaining reactions observed with various antibodies (B–D) or nonimmune IgG (A). The reactions have been placed into the context of the canal base by showing the features of the fibrils in the cartilage that is normal (below) and by depicting the gross changes that have been observed in the fibrils within the preresorptive cartilage layer (above). The latter are based on observed changes determined using routine electron microscopy techniques. Also integrated into the model, are previous results which have revealed the dramatic removal of aggrecan core protein by matrix metalloproteinase (MMP) cleavage at the distal border of the preresorptive layer (Lee et al.,2001). Furthermore, the relevant canal cells are shown at the top of the figure. The new results are depicted in B, C, and D where the immunostaining reactions indicative of type II collagen (COL2) one-quarter long, MMP-13 neoepitope, or MMP-13 proenzyme are shown with the help of red shading. The latter is applied to the cells or the matrix to appropriately indicate the distribution of the epitope target named at the top of each panel.

Download figure to PowerPoint

Most striking was the localization of the anti–MMP-13 neoepitope antibodies on collagen fibrils in the same location, indicating the presence of activated enzyme on the fibrils (Fig. 13C). Here, the spacing between immunoreactions also approached 67 nm, suggesting that the two enzyme forms may bind at the same places on the fibril. The binding was interpreted as follows. The active enzyme, like the proenzyme, relies upon interaction between the hemopexin-like domain and the collagen α chain (Murphy and Knauper,1997). Moreover, upon activation, the enzyme's cleft is exposed, making it possible for the substrate to be suitably positioned for hydrolysis (Overall,2002; Chung et al.,2004). Hence, although the latter interaction offered more stability than the former one in vitro, both enzyme forms possessed the capacity to bind to the collagen in vivo (Murphy and Knauper,1997). Because both enzyme forms appeared to be clustered in the same locale, this suggested that both were positioned where the susceptible Gly775-Leu766 bonds might be exposed on the outside surface of the fibril (Fig. 13B; Welgus et al.,1980).

In our search for evidence to support the cleavage of the fibrils' COL2, it was known that MMP-13 gave rise to characteristic COL2 one-quarter fragments (Billinghurst et al.,1997) and CTX-II fragments (Eyre et al.,1996; Atley et al.,1998; Christgau et al.,2001). Anti-neoepitope antibodies against these COL2 degradation products generated reactions in the preresorptive cartilage layer at the same location as the MMP-13 enzyme. Also in the electron microscope, the fragments were distributed along collagen fibrils, exactly like the activated enzyme. Because the fragments were localized in exactly the same sites as the MMP-13 enzyme, we propose they must have been produced by the enzyme. The collective data suggest that MMP-13 was responsible for the extensive damage to the fibrils that had been previously described in the layer by Davoli et al. (2001).

The point was raised previously (Krane and Inada,2008), that fibrillar collagen may be subjected to surface attacks that do not necessarily result in the demise of the fibril. The afore-described results we have obtained along the distal border of the preresorptive cartilage layer would tend to support this view, because here, at least from a morphological perspective, the fibrils appear normal. However, we have also observed that before the fibrils lose their form and are replaced by matrix debris (Figs. 5C–E, 13A), they exhibit swelling and fragmentation (as depicted along the proximal border in Fig. 13A). We interpret these findings to mean that the changes observed in fibril structure are the affects of MMP-13 attacks initiated at the distal border but brought about by the continuing exposure of the fibril to the MMP-13 enzyme over time (estimated to last for approximately 45 min), thereby resulting in the eventual fibril demise described at the proximal border (Fig. 13B,C).

Because changes in fibril structure can be correlated with the presence of the activated MMP-13 enzyme binding to the fibril, it is considered important to review the studies that contribute to our understanding of MMP-13 activation. Because “α and β cells” (tentatively described collectively as “vascular cells” in Fig. 13A) express the MMP-13 proenzyme and appear to also express MT1-MMP (Alvarez et al.,2005), the interpretation was that the MMP-13 proenzyme produced by the vascular cells was bound to the cell processes where it was subjected to activation (compare canal cell reactions in Fig. 13D to 13C). Because MT1-MMP appears to be present on these cell processes as was shown in Figure 4F, then we conclude that the necessary components (MT1-MMP and MMP-13) are likely in place to lend support to an indirect role for MT1-MMP in COL2 lysis. That the activation of MMP-13 is a step in canal excavations may be surprising to some. Indeed, in vitro experimentation using isolated endothelial cells cultured on type I collagen gels has emphasized that endothelial cells require MT1-MMP but not MMP-13 (or other MMPs), to migrate or to sprout under the conditions of study (Chun et al.,2004). However, the current work alternatively suggests that, as far as the vascular invasion of the canal blind end is concerned, MT1-MMP and MMP-13 are both involved. Furthermore, strong evidence is presented here that MMP-13 rather than MT1-MMP actually attacks the bulk of COL2 fibrils. Because we have observed the MMP-13 proenzyme in “α” (endothelial) cells, we conclude that the endothelium contributes at least some of the MMP-13 shown to be involved. This conclusion then raises the question as to the role of the “β cell.” It should be mentioned that, in previous work reported from this laboratory, we attempted to assign identity to “β cells” using antibodies to detect classic markers of endothelial cells, pericytes, or markers indicative of the macro-osteoclast lineage (Lee et al.,2006). However, the “β cells” remained unstained. Second, because “β cells” are intimately associated with the endothelium and stain intensely with the anti–MMP-13 proenzyme antibodies, we can conclude that these so far unidentified cells appear to function as “helper” cells in canal formation. Because of the close physical contact between “α” and the cells called “β,” we have not discounted the possibility that “β cells” may be activated endothelial cells. Therefore, to advance our understanding of basic mechanisms of long bone development, further testing awaits to determine whether or not the MMP-13 of the endothelium directly participates in the COL2 fibril lysis that is required for vascular invasion of cartilage.

We have also reported here that anti–MMP-13 neoepitope antibodies bind to the asymmetrical plasma membrane of the terminal chondrocyte (Fig. 13C). We have observed the immunostaining of the same tissue site with anti–MMP-13 proenzyme or anti–COL2 one-quarter fragment antibodies (see terminal cells in Fig. 13D and 13B, respectively). Indeed these reactions drew our attention to the cellular condensations that resembled at least in part, accounts previously reported by Farnum and Wilsman (1987) in the primary growth plate. Although the focus of the latter study was placed upon hypertrophic chondrocytes, some of the features seemed to be shared with the terminal epiphyseal chondrocyte. The current work reveals that a similar polarized contact occurs as a step during the terminal differentiation of the epiphyseal chondrocyte. Because the MMP-13 neoepitope is revealed in these cells on their asymmetrical membranes and sometimes in closely associated rER cisternae, the significance of the observations is worthy of consideration. On one hand, it is known that, when MMP-13 activation takes place in the rough endoplasmic reticulum (rER), this leads to autodegradation of the enzyme (Kennedy et al.,2005). Hence, the MMP-13 neoepitope revealed in the ER of the terminal cell may reflect the active enzyme that will degrade as this cell prepares to die (Fig. 11D). However, on the outside of the same terminal cell, we have detected the MMP-13 proenzyme. We have also determined that COL2 one-quarter fragments are generated at the surface of the same asymmetrical membrane (Fig. 13B), just as they were observed in association with canal cell processes (Fig. 5E). We interpret these combined observations to mean that the MMP-13 enzyme undergoes activation on the asymmetrical membrane and that COL2, and possibly other substrates, are cleaved as a consequence of this activation. The collective data suggest that the occurrences we have described on the APM somehow affect the fate of the terminal chondrocyte.

In conclusion, whereas the COL2 one-quarter long fragments were shown to be critical in establishing that a collagenase was involved in fibril lysis, it was surprising to us to find that the anti–CTX-II antibodies intensely immunostained the same preresorptive cartilage layer, indicating that CTX-II fragments were also prominent. CTX-II fragments found in the urine of adults have been targeted for a biomarker assay for cartilage collagen breakdown. The tissue sources for CTX-II in urine are unknown but have been linked to degradation in various anatomical sites including articular cartilage, regions of the skeleton where calcified cartilage is subjected to resorptions or to other tissues enriched in type II collagen such as the hyaline cartilage of the ribs or the intervertebral discs (Matyas et al.,2004; O'Kane et al., 2005). The precise tissue sources will eventually be identified. Given the elevated levels of CTX-II observed in animal urine after birth, we propose that the remodeling of the uncalcified cartilaginous anlagen should be examined in the future, as a potential tissue source for these fragments. While the MMP-13 enzyme generates CTX-II fragments in vitro, MMP-9 can also produce it. Hence, as shown in Figure 14, the CTX-II we report here in the tissue could reflect the activity of either enzyme because we have shown both enzymes are present at the canal base as the mature enzyme forms.

thumbnail image

Figure 14. Model proposed to summarize the features and outcomes of type II collagen (COL2) cleavage at the canal blind end. Pink dots (middle panel) signify the susceptible Gly775_Leu776 bond.

Download figure to PowerPoint

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSION
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Study of the development of long bones has confirmed the disappearance of the epiphyseal central core in conjunction with blood vessels and their growth, the demise of collagen fibrils and the terminal differentiation of central core chondrocytes. MMP-13 is shown here to play a key role in each of these processes. Study of the MMP-13 enzyme has made it possible to identify the cells producing this enzyme (the “α” [endothelial] and “β” cells [the latter of which is physically related to the endothelium] and possibly, some epiphyseal chondrocytes) and, to detect the MMP-13 main tissue target (the preresorptive layer). In addition, the cartilage resorption associated with cartilage canals has been shown to proceed in a series of discrete areas referred to as remodeling sites. Theses sites include aggrecan- and collagen-cleaving proteases. Evidence presented here shows that MMP-13 is responsible for the collagen lysis taking place thereby imparting clarity to the role of MMP enzymes in general, and MMP-13 in particular, in this tissue in development. Moreover, as a consequence of the current work the life history of the MMP-13 enzyme in this tissue has been uncovered and this knowledge provides an understanding as to how this enzyme exerts affects in situ and at a series of discrete sites to mediate COL2 lyses (Fig. 14). Various COL2 fragments have been uncovered making it possible to document COL2 fibril lyses, and to use this information to assist in the identification of the involved collagenase and to raise awareness to the fact that one fragment (CTX-II) is generated consistently at cartilage remodeling sites (it may also contribute to levels shown present in urine). Finally, the association has been established between MMP-13 and the terminal chondrocyte, suggesting that MMP-13 may also have a role in mediating some aspect of terminal cell function (as localized to the asymmetrical plasma membrane of only these specified epiphyseal chondrocytes). In conclusion, key roles for MMP-13 are suggested by the various approaches we have applied here.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSION
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Preparation of Rabbit Antipeptide Antibodies

New Zealand White rabbits were immunized as a service provided by the Diagnostic and Research Support Services Laboratory (Animal Resources Centre, McGill University; Lee et al.,2001). Immunizing peptides were synthesized using Fmoc chemistry, purified by reverse chromatography and coupled to ovalbumin using a bifunctional reagent. Antisera were affinity purified using Sulfolink coupling gel (Pierce, Rockford, IL; Lee et al.,2001,2006). Peptides were as follows (numbering shown begins after the signal peptide; g, denotes glycine spacer; c, the cysteine used for coupling): MMP-2 proenzyme (31CPKESCNLFVLK- DTLKgc; accession no. P08253), MMP-2 (81YNFFPggc), MMP-13 proenzyme (30HPVTLAGILKKSTVTSgc; accession no. P23097), MMP-13 (86YNVFPggc), COL2 one-quarter long (776LAGQRggc; accession no. P02458), COL2 one-quarter short (779QRGIVggc), COL2 C-telopeptide (CTX-II) (cggEKGPDP1035), and type X collagen (YNRQQHYDPRSGIFTCKIPGIYYFSYggc).

Immunoblotting

Purified proteins, collagen digests, and guanidinium extracts of rat growth plate were resolved by SDS/PAGE and stained with Coomassie blue or, to demonstrate epitope specificity, transferred to nitrocellulose membrane for immunodetection. After blocking with 1% BSA/PBST, membranes were incubated with primary antibodies overnight at 4°C. After incubation with alkaline phosphatase conjugated secondary antibodies, the reactive bands were detected with the substrate NBT/BCIP (Promega, Madison, WI). Specificity to the COL2 fragments was demonstrated by rHuMMP-13 digestion of COL2. Human COL2 and MMP-13 were generously provided by Dr. David Eyre (Department of Orthopaedics and Sports Medicine, University of Washington) and Dr. Gillian Murphy (University of Cambridge, England), respectively. The collagen was resuspended in digestion buffer (0.05 M Tris, 0.01M CaCl2, 0.15 M NaCl, pH 7.5) and incubated at 25°C in the presence or absence of MMP-13. The digestions were stopped at 0, 0.5, 1, and 3 hr with the addition of 20 mM ethylenediaminetetraacetic acid (EDTA). Antibodies to the one-quarter long and one-quarter short cleavage fragments were applied at 10 μg/ml and 4 μg/ml, respectively. The specificity of CTX-II antibodies was assessed by the MMP-13 digestion of bovine type II collagen (Southern Biotech, Birmingham, AL). The digestions were stopped at 0, 2, and 3 hr and anti–CTX-II antibodies were applied at 4 μg/ml. RHu MMP-2 and rHuMMP-13 and sheep anti-human pro/active MMP-2 antibodies were generously provided by Dr. Gillian Murphy (Knauper et al.,1996a,2002). Rat MMP-13 proenzyme was prepared from a guanidinium extract of 21-day-old primary growth plate (see Lee et al.,2001, for details) and dialyzed into 10 mM sodium acetate, pH 6.0. Rabbit anti–MMP-13 (Davoli et al.,2001) and sheep anti–MMP-2 (Hipps et al.,1991) were applied at concentrations of 2 μg/ml. Antibodies were also applied to rat growth plate extract or rHu proteins to test antibody specificity against the MMP-13 proenzyme (4 μg/ml), MMP-13 (2 μg/ml), or MMP-2 (10 μg/ml).

Animals and Tissue Sample Preparation

Sprague-Dawley rat pups, aged 5 to 8 days (Charles River, Saint-Constant, Quebec) or C57BL6 mice (generated as part of a breeding program) were housed and handled as recommended by the Canadian Council on Animal Care. Rats were anesthetized with sodium pentobarbital (Somnotol, MTC Pharmaceuticals, Cambridge, Ontario), exsanguinated under anesthesia and perfused through the heart with periodate-lysine-paraformaldehyde (PLP) fixative (McLean and Nakane,1974). Proximal tibiae were removed, fixed for an additional 24 hr in PLP, decalcified in 10% EDTA, and embedded in a mixture of 5% sucrose/OCT compound. Cryosections were cut at either 8 μm (light microscopy) or 30 μm (electron microscopy), mounted on gelatin-coated slides, and stored at −20°C until analysis (Lee et al.,2001). Mouse tail length was determined with calipers (measuring from tail base to tip) and urine spot samples (postnatal days 4 through 65) were taken at a fixed hour every morning. The urine was promptly aliquoted so that analyses could be conducted in duplicate and frozen at −20°C for future biochemical analysis.

Light and Electron Microscopic Immunohistochemistry

Primary antibodies, with the exception of the anti–MT1-MMP (catalog no. IM39L, Calbiochem, San Diego, CA), the sheep anti–pro/activated MMP-2 (Hipps et al.,1991), the anti-aquaporin 5 (Alpha Diagnostic International, San Antonio, TX), and the nonimmune rabbit IgG (Jackson Immuno Research, West Grove, PA) were prepared in our laboratory (see above for details). For light microscopy immunohistochemistry, primary antibody target and dilutions were as follows: MMP-13 proenzyme (1.25 μg/ml), MMP-13 (1.25 μg/ml), MMP-9 (0.5 μg/ml; Lee et al.,2006), MMP-2 proenzyme (1.25 μg/ml), MMP-2 (5 μg/ml), pro/activated MMP-2 (10 μg/ml; Hipps et al.,1991), COL2 one-quarter long (3 μg/ml), COL2 one-quarter short (3 μg/ml), CTX-II (1 μg/ml), MT1-MMP (55 μg/ml), aquaporin 5 (10 μg/ml), and type X collagen (6 μg/ml). Primary antibody targets and dilutions for electron microscopy immunohistochemistry were MMP-13 proenzyme and MMP-13 (1.25 μg/ml), anti–COL2 one-quarter long (3 μg/ml), and anti–CTX-II (1 μg/ml). Briefly, frozen sections were immersed in 4% formaldehyde to improve adhesion to the slide then pretreated with Chondroitinase ABC in the presence of proteinase inhibitors. Endogenous peroxidase activity was quenched with H2O2 before antigen localization using an indirect immunoperoxidase technique (Vectastain ABC kit, Vector Laboratories Inc., Burlingame, CA) and visualization with DAB substrate (Vector Laboratories). For light microscopic examination, 8-μm-thick sections were counterstained with undiluted Gill's hematoxylin and photographed on a Leica DM RB microscope using a Roper Scientific Coolsnap chilled CCD camera system. For electron microscopic immunohistochemistry, 30-μm-thick sections were osmicated, dehydrated in acetone, and embedded in epoxy resin after exposure to chromogen (DAB). Ultrathin sections were examined with a Philips 400 electron microscope at 80 kV without counterstaining (Lee et al.,2001).

Routine Microscopy

Tibiae were dissected from rats that had been perfused through the heart with a mixture containing 2.5% glutaraldehyde and 2% formaldehyde in 0.1 M cacodylate buffer (pH 7.3). After post-fixation in potassium-reduced osmium tetroxide, dehydration in acetone, and embedding in JEMbed epoxy resin, sections were prepared at a thickness of 1 μm and stained with 1% aqueous Toluidine blue (Fig. 6B,F) or with the periodic acid–Schiff technique to reveal glycogen (Fig. 6A,E; as described previously; Lee et al.,1982).

CTX-II Urinalysis

Urine CTX-II was measured using the Urine Pre-Clinical Cartilaps ELISA (IDS Nordic a/s, Herlev, Denmark) according to the manufacturer's instructions. The intra- and interassay coefficients of this assay are <4.6% and <8.1%, respectively. The CTX-II values were normalized to the concentration of creatinine. The latter was quantified with a commercial kit as described by the manufacturer (Creatinine Colorimetric Assay Kit, Oxford Biochemical Research, Oxford, MI).

Statistical Analysis

Data were expressed as mean values ± standard error of the mean (SEM). One-way analysis of variance was used for statistical comparisons among groups, and the Dunn test was used for all pairwise comparisons of the mean values of the different groups. The described calculations were made using GraphPad Prism version 5.01 for Windows (GraphPad Software, San Diego, CA, www.graphpad.com).

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSION
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

The authors thank Dr. Gillian Murphy (Department of Oncology, University of Cambridge, England) for generously providing rHu MMP-2 and MMP-13 enzymes and the antibodies prepared in sheep to human MMP-2. We also thank Dr. David Eyre (Department of Orthopaedics, University of Washington School of Medicine, Seattle, WA) for the generous gift of pepsin purified human, type II collagen. The peptides used in this study were prepared by Ms. Elisa de Miguel in the Core Biotechnology Facility of the Shriners Hospital for Children. The illustrations were prepared by Ms. Guylaine Bedard and Mr. Mark Lepik. E.R.L. was funded by the Shriners of North America.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSION
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. CONCLUSION
  7. EXPERIMENTAL PROCEDURES
  8. Acknowledgements
  9. REFERENCES
  10. Supporting Information

Additional Supporting Information may be found in the online version of this article.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.