Stromelysin-1 (MMP-3) and Stromelysin-2 (MMP-10) Expression in Developing Human Bone: Potential Roles in Skeletal Development

Bone Vol. 23, No. 1 July 1998:7–12

Stromelysin-1 (MMP-3) and Stromelysin-2 (MMP-10) Expression in Developing Human Bone: Potential Roles in Skeletal Development S. BORD,1,3 A. HORNER,1 R. M. HEMBRY,2 and J. E. COMPSTON1 1

Department of Medicine University of Cambridge School of Clinical Medicine, Addenbrooke’s Hospital, Cambridge, UK Strangeways Research Laboratory, Cambridge, UK 3 Anglia Polytechnic University, Cambridge, UK 2

Introduction

Stromelysin, a member of the matrix metalloproteinase family, demonstrates wide substrate specificity with the ability to degrade proteoglycan, fibronectin, laminin, casein, and the nonhelical region of collagen. The two forms of stromelysin (SL), types 1 (MMP-3) and 2 (MMP-10), share 82% sequence homology, but exhibit differences in cellular synthesis and inducibility by cytokines and growth factors in vitro. However, the distribution of the two isoforms in bone has not been reported. We investigated the presence of SL-1 and SL-2 in human osteophytic and neonatal rib bone using immunohistochemistry and, combined with a new method of in situ zymography, determined the activity of the immunolocalized stromelysins. Latent SL-1 was strongly expressed in the extracellular matrix in fibrous tissue surrounding areas of endochondral ossification in osteophytes, and adjacent to the periosteum of fetal rib bone. Active SL-1 expression was detected in osteocytes and the matrix surrounding osteocytic lacunae. SL-2 showed intense cell-associated staining at sites of resorption in areas of endochondral ossification and in resorptive cells at the chondro-osseous junction, which correlated with enzyme activity detected by zymography. Within the rib, active SL-2 expression was localized in chondrocytes of the growth plate, whereas only occasional SL-1 signal was evident. Vascular areas showed strong SL-2 staining with some proteolytic activity. SL-2, but not SL-1, was strongly expressed in osteoclasts and most mononuclear cells within the marrow. At sites of bone formation both isoforms were expressed by osteoblasts with SL-1 also present in osteoid. These results demonstrate, for the first time, the differential expression of SL-1 and SL-2 in developing human bone, indicating specific roles for the two isoforms. In situ zymography demonstrates that SL-2 is produced in an active form with associated degradation, whereas SL-1, in a matrixbound proenzyme form, may act as a reservoir for later activation. (Bone 23:7–12; 1998) © 1998 by Elsevier Science Inc. All rights reserved.

Matrix metalloproteinases (MMPs) are a family of endopeptidases that degrade extracellular matrix and basement membrane components. They are synthesized and secreted by multiple cell types in a proform requiring extracellular activation, and their activity is tightly regulated by specific inhibitors (TIMPs). There is much evidence to suggest that these enzymes and their inhibitors play important roles both in normal developmental processes and pathological degradation.18 MMPs fall into three main groups according to their substrate specificity: the collagenases (MMP-1, -8, and -13) cleave native collagens types I, II, and III at a single locus; the gelatinases (MMP-2 and -9) cleave denatured collagens and type IV; and the stromelysins (MMP-3 and -10) have a broad substrate specificity including proteoglycans, types III, IV, and V collagens, casein, and fibronectin.2 Stromelysin-1 (SL-1, MMP-3) and stromelysin-2 (SL-2, MMP-10) are closely related, the two isoforms sharing 82% sequence homology23; however, their expression appears to be differently regulated. SL-1 was initially isolated from rabbit synovial fibroblasts9 and SL-2 identified in a human adenocarcinoma cDNA library.17 SL-1 production is stimulated in many cell types by growth factors, cytokines, tumor promoters, and oncogenes, whereas SL-2 appears to be largely unaffected by these agents. In vitro studies have shown that SL-1 is produced by stimulated human16 and mouse osteoblasts,24 two osteosarcoma cell lines (MG63 and U2OS),21 and isolated rabbit osteoclasts.14 We have reported the presence, in situ, of SL-1 in human osteoblasts, but not osteoclasts.4 Immunohistochemistry, although demonstrating the presence of MMP protein, is restricted by currently available antibodies that do not differentiate between latent and proteolytically active forms of the enzyme nor between the free and inhibitor-complexed enzyme. We have used in situ substrate zymography to demonstrate latent and active SL-1 and SL-2 in human osteophytic and neonatal rib bone.

Key Words: Metalloproteinases; Stromelysin; Bone; Osteoblasts; Osteoclasts; Chondrocytes.

Materials and Methods Tissue The study was conducted on neonatal ribs and osteophytes. Neonatal ribs were removed at postmortem (28 –37 wk gestation) and osteophytes were obtained following routine shoulder and hip joint replacement surgery. Informed written consent was

Address for correspondence and reprints: Dr. Sharyn Bord, Department of Medicine, Level 5, Addenbrooke’s Hospital, Cambridge CB2 2QQ, UK. E-mail: [email protected] © 1998 by Elsevier Science Inc. All rights reserved.

7

8756-3282/98/$19.00 PII S8756-3282(98)00064-7

8

S. Bord et al. Stromelysin in bone development

given and approval obtained from the local ethics committee. Following removal the tissue was immediately placed on ice and then immersed in 5% polyvinylalcohol (PVA) for 2 min, snap frozen in liquid nitrogen, and stored at 270°C. Frozen sections were obtained using a Bright cryostat (Huntingdon, UK) with a cabinet temperature of 230°C, equipped with a slow-drive hightorque motor with automatic speed control and a highly polished tungsten carbide knife. Histology Undecalcified, unfixed 9-mm-thick frozen sections were stained with Diff-Quik (Baxter Dade AG) or 1% toluidine blue for general morphology, and von Kossa with a van Geisen counterstain to assess matrix mineralization. Alkaline phosphatase staining to identify osteoblasts was detected using a coupled reaction with a-naphthyl acid phosphate and fast red TR.6 Tartrateresistant acid phosphatase (TRAP) staining in osteoclasts and preosteoclasts was demonstrated by reactivity with AS-BI phosphate and sodium tartrate, postcoupled with fast garnet.15 Antibodies Well-characterized specific polyclonal antibodies to human SL-11 and human SL-213 were raised in sheep. Because the antiserum to SL-1 partially cross reacts with SL-21 additional precautions were taken to remove cross-reacting epitopes. The peptide to which the SL-2 antibody was raised was combined with the antibody to either SL-1, SL-2, or NSS (overnight at 4°C) in a concentration range of 0 –100 mg/mL and then used to stain sections as described below. The immunoglobulin (IgG) concentrations of the antisera were maintained at 10 mg/mL in all preparations. Immunoglobulins for the antisera were prepared by triple ammonium sulfate precipitation. Pooled normal sheep serum IgG (NSS) was used as a control. A biotinylated rabbit antisheep IgG (Vector Laboratories) was used as secondary antibody. Immunohistochemistry Sets of serial sections 9 mm thick were picked off onto glass slides, coated with 2% 3-aminopropyltriethoxysilane (APES), air dried for 10 min, fixed in 4% paraformaldehyde for 22 min at room temperature, and washed in phosphate-buffered saline (PBS). Immunolocalization was carried out using an indirect immunoperoxidase system. Endogenous peroxidase and nonspecific binding were blocked by incubation with ImmunoPure peroxidase suppressor (Pierce, 22 min) and extensive washing followed by 20% normal rabbit serum (12 min) prior to the primary antibody incubation (10 mg/mL, overnight at 4°C in a humid chamber). Following further washing the second antibody was applied (5 mg/mL, 40 min at room temperature). The sections were washed and the signal amplified using the avidin-biotin complex (ABC Vector Laboratories) for 30 min at room temperature. Sites of antigen binding were visualized using 3,39-diaminobenzidine (DAB) as a chromogen. Some sections were lightly counterstained with hematoxylin or methylgreen to detect nuclei. Sections were mounted in aqueous mount (90% glycerol in PBS) and observed under bright-field microscopy on an Olympus BH-2 microscope. Photographs were taken using Ektachrome 64 film. Substrate Gels Substrate gels were prepared using an adaptation of the method of Galis et al.12 FITC-labeled casein (Sigma, 1 mg/mL in distilled water) was mixed 1:1 with a solution of 1% agarose melted in 50

Bone Vol. 23, No. 1 July 1998:7–12

mmol/L Tris (pH 7.2) containing 10 mmol/L CaCl and kept at 56°C in a dipping chamber. Prewarmed microscope slides were dipped into the casein solution, drained for a few seconds, and allowed to gel horizontally at room temperature. Prior to use, the slides were examined by fluorescence microscopy to insure there was an even coating of casein over the slide. Cryosections were picked off onto the slides, covered with 80 mL of Tris buffer at pH 7.2, 6.8, or 6.0, and incubated at 37°C for 16–20 h. Using fluorescence microscopy areas of lysis were identified as dark areas within the fluorescent coating and compared with SL-1 and SL-2 expression observed on serial sections by immunohistochemistry. To confirm that the lysis was due to MMP activity, some sections were treated with 1,10phenanthroline or 1 mmol/L EDTA, zinc chelators that inhibit MMP activity. Results Addition of the SL-2 peptide to the SL-2 antibody dose-dependently abolished staining, confirming its specificity. When the peptide was added to the SL-1 antibody a slight reduction in staining at the higher concentrations was observed, consistent with a weak interaction with SL-2.1 NSS with peptide was negative at all concentrations. Osteophytic Bone Histological features. The osteophytes exhibited histological features similar to those reported by ourselves4,5 and others.10,20 Vascular invasion was associated with the appearance of a narrow fibrocartilage outer collar, containing collagenous fibers with cartilage cells and scant cartilage matrix. This, together with an outer layer of fibrous tissue, surrounded the osteophyte (Figure 1a). A central area with bone formation and resorption was present, with mineralized bone and thick osteoid seams. Both endochondral and intramembranous bone formation were evident, as were woven and mature lamellar bone. Immunolocalization. Serial sections were assessed for SL-1 and SL-2 expression. Normal sheep serum, as detailed in Materials and Methods, was used as a negative control. In addition, omission of the primary or secondary antibodies was included to detect nonspecific binding of the second antibody or substrate. All these showed absence of staining (Figure 1b). Stromelysin-1. Within osteophytic bone stromelysin-1 was most evident as matrix-associated staining in the immature fibrous tissue (Figure 1c) and cartilage adjacent to areas of bone formation, with occasional cell-associated expression in hypertrophic chondrocytes. Sections stained with anti-SL-1 absorbed with SL-2 peptide also showed staining at these sites confirming the presence of SL-1. No expression was seen in vascular cells (Figure 1c). Within the bony areas positive staining was most frequently detected in osteoid at sites of bone formation. However, only occasional plump osteoblasts at these sites showed expression of SL-1. Many TRAP-positive osteoclasts were evident but all were negative for SL-1. Stromelysin-2. Intense SL-2 expression was seen in many cell types within osteophytic bone. Vascular cells in the fibrocartilage stained strongly (Figure 1d) as did chondrocytes in most areas of endochondral ossification. Osteoblasts that appeared to be actively involved with synthesis of new osteoid expressed SL-2, particularly those cells closest to the bone surface, whereas no staining was detected in the matrix. SL-2 signal was seen in mononuclear cells at sites of resorption and in some, but not all, osteoclasts. The osteoclasts were identified by morphology and TRAP positivity.

Bone Vol. 23, No. 1 July 1998:7–12

S. Bord et al. Stromelysin in bone development

9

Figure 1. Serial sections of osteophytic bone. (a) Low-power view of a section stained with Diff-Quik shows the vascular cartilage (V) and fibrous tissue (F) adjacent to the bone (B). (b– d) Sections stained by indirect immunoperoxidase with DAB as a chromogen using (b) NSS IgG, (c) anti-SL-1 IgG, and (d) anti-SL-2 IgG. (b) Low-power view of a control section immunolocalized with NSS IgG. No detectable staining is seen in any of the areas. (c) Higher power view of section stained with anti-SL-1 IgG showing boxed area in (b). Intense matrix-bound SL-1 expression is apparent in the fibrous tissue, but no staining is seen staining in the vascular area (arrows). (d) Higher power view of adjacent section stained with anti-SL-2 IgG shows positive expression in many vascular cells (arrows), but absence of staining in the fibrous tissue. Bars 5 50 mm (a– d). Figure 2. Serial sections of neonatal rib stained by indirect immunoperoxidase with (a) anti-SL-1 IgG and (b– d) anti-SL-2 and detected with DAB substrate. (a) A chondro-osseous junction with periosteum (P) and loosely woven bone (B) adjacent to chondrocytes (arrows). Matrix-associated SL-1 staining is seen in the periosteum, whereas the chondrocytes are negative. (b) Distinct cell-associated SL-2 signal is detected in the chondrocytes (arrows) and there is low expression in cells of the periosteal covering (arrowheads). (c) An osteoclast (arrow) close to the bone surface (B) in the primary spongiosa exhibits strong SL-2 expression. Methylgreen counterstain demonstrates the nuclei of the osteoclast and surrounding cells (arrowheads). (d) Osteoblasts at a bone forming surface are immunoreactive for SL-2. The cells closest to the bone (B) show the most intense signal (arrows). Bars 5 20 mm (a– d).

Substrate gels. Areas of lysis were evident on most sections. These areas were compared with the serial sections on which immunolocalization for SL-1 and SL-2 was performed. The correlation of SL-1 protein expression and lysis was low, but, in contrast, there was a strong correlation between SL-2 expression

and casein lysis. No lysis was detected in areas where intense matrix-bound SL-1 expression was evident; however, there were distinct bands of lysis around the SL-2-positive hypertrophic chondrocytes. Many osteoclasts were evident and those that were immunoreactive for SL-2 also showed proteolytic activity on the

10

S. Bord et al. Stromelysin in bone development

casein substrate. Vascular cells showed immunoreactive SL-2 signal with associated proteolysis in about 20% of these positive cells, indicating there may be selective activity at the same location. Neonatal Rib Histological features. Serial sections were cut both longitudinally and cross sectionally through the rib samples. Longitudinal sections showed the growth plate enveloped by a periosteal collar. Stromelysin-1. Stromelysin-1 expression in the rib was less pronounced than in osteophytic bone. Staining was predominately matrix-associated, particularly at the chondro-osseous junction where the loosely woven bone and connective tissue stained strongly. Most samples showed some immunoreactive SL-1 in the periosteum, particularly in the area close to the chondro-osseous junction (Figure 2a). Only occasional staining was observed in the chondrocytes. Some osteoblasts on trabecular surfaces showed positive signal, whereas all osteoclasts were negative. Many osteocytes stained positively, particularly those close to remodeling surfaces. Stromelysin-2. Intense SL-2 signal was evident in the hypertrophic region of the growth plate, but decreased through the proliferative zone, with only low levels in the resting chondrocytes. Toward the chondro-osseous junction that surrounded the cartilage, the hypertrophic chondrocytes at the peripheral edge showed intense signal (Figure 2b). Cell-associated expression was seen in the cells of the perichondrium and the periosteum (Figure 2b). Within the primary spongiosa cartilage remnants were remodeled by osteoclasts, some of which expressed SL-2 (Figure 2c). Alkaline phosphatase-positive osteoblasts associated with areas of new bone formation were immunoreactive for SL-2, with those cells closest to the osteoid bone surface staining most intensely (Figure 2d). Substrate gels. Sections of neonatal rib on the casein substrate gel gave very similar results to the osteophyte sections. Proliferating and hypertrophic chondrocytes with intense SL-2 expression (Figure 3a) also showed rings of lysis (Figure 3b). Sections treated with ethylene-diamine tetra-acetic acid (EDTA), an inhibitor of MMP activity, showed no degradation of the casein substrate (Figure 3c). Osteoblasts in areas of bone formation were also SL-2 positive with associated lysis. Marked lysis was also detected at sites of resorption both in the osteophyte and in the chondro-osseous junction in the rib, where sites of lysis colocalized with the cell-associated SL-2 signal. Osteoclasts close to the bone surface were negative for SL-1 (Figure 3d), but strongly positive for SL-2 (Figure 3e), and demonstrated areas of lysis that became more distinct when the sections were incubated in lower pH buffer, the greatest amount of lysis around the osteoclasts being observed at pH 6.0 (Figure 3f). The addition of 1,10-phenanthroline or EDTA blocked most of the lysis at all pH values (Figure 3g). No lysis was observed in association with SL-1 signal in the matrix. Lysis did occur in some osteocytes and correlated with SL-1 but not SL-2 expression (Figure 3h, i). Discussion These results demonstrate, for the first time, the differential expression of SL-1 and SL-2 in developing human bone. The use of in situ zymography enabled distinction of active and inactive stromelysin, and demonstrated that immunoreactive SL-2 was

Bone Vol. 23, No. 1 July 1998:7–12

largely active, whereas SL-1 was predominately latent and matrix bound. These results extend our earlier work that described the expression of MMPs and TIMP in human osteophytic bone; in that study, the presence of only the SL-1 isoform was investigated.5 Osteophytic bone provides a model of rapid bone turnover, in which endochondral ossification, vascular invasion, cellular proliferation, and new bone formation are seen, resembling the changes seen in the epiphyseal growth plate of long bones. However, it is pathological bone and so, in the present study, neonatal rib was included as an example of developing normal human bone. In both tissues, the patterns of immunolocalization of both SL-1 and SL-2 were very similar; however, generally, the intensity of staining was higher in the osteophytes than in the ribs, possibly reflecting the greater rate of activity and growth in osteophytic bone. Some interesting differences between SL-1 and SL-2 have been highlighted, which suggests differing functional roles for the two isoforms. Currently, the antibodies available for detection of MMPs do not distinguish between active, latent, or inhibitor-complexed forms, and therefore it is only possible to speculate on the function of these enzymes in bone. Electrophoretic techniques using gelatin and casein zymography are used to demonstrate active gelatinase and stromelysin. However, samples can only be derived from culture media, or cell or tissue homogenates, and thus do not provide in situ information. Similarly, fluorogenic peptides and fluorescent-labeled proteins can be used to identify active protein, but again, are restricted to prepared samples. Both SL-1 and SL-2 were present in osteoblasts on bone forming surfaces, although the number of SL-1-positive cells was lower, and the expression more variable than SL-2, with SL-2 staining most pronounced in the cells closest to the bone surface. Stromelysin has been shown to cleave types I and III procollagens,19 and osteoblastic SL-2 may play a role in degrading these procollagens and facilitating collagen fibril orientation prior to mineralization. Immunoreactive SL-1 was incorporated in the osteoid on bone surfaces. Stromelysin binds to collagen both in its latent and active form, but TIMP-1 binds only to the active enzyme.1 Our previous study showed that SL-1 expression in some areas was accompanied by low levels of TIMP-1, suggesting the matrixbound enzyme could be either active or in its latent proform. This is clarified by the current study which showed no lysis of the casein substrate at these sites, suggesting that the sequestered matrix-bound enzyme is latent. The latent form may play a role in bone resorption, during which it would be released and become available for activation. Stromelysin can autoactivate at low pH,19 and thus it is also possible that the acid environment of bone resorption could accomplish this activation and enable the enzyme to activate other local enzymes (e.g., collagenase) and factors stored in bone matrix (e.g., transforming growth factor-b [TGF-b]) TGF-b expression has been identified in similar locations, with the active form localized at sites of resorption (Horner, unpublished observations). As TGF-b is known to downregulate MMP synthesis and upregulate TIMP,27 this could act as a regulatory mechanism for MMP production and activation. This proposed function must be confined to SL-1 as there was no evidence of any matrix-bound SL-2 expression. Although the majority of immunoreactive SL-1 did not correlate with lysis of the substrate, it is of interest to note that SL-1 expression in some osteocytes was associated with proteolytic activity on the casein substrate. A difference in the expression patterns between the isoforms was seen in osteoclasts. Although there was no evidence of SL-1, intense staining of SL-2 was detected in some osteoclasts close to

Bone Vol. 23, No. 1 July 1998:7–12

S. Bord et al. Stromelysin in bone development

11

Figure 3. Serial sections of neonatal rib showing: (a– c) chondrocytes in the growth plate; (d– g) an area of bone (B) with a highly resorbed edge, and osteoclasts (arrows) in close proximity; (h, i) osteocytes at a remodeling surface. (a) SL-2 expression detected using DAB chromogen is seen in most of the proliferating chondrocytes (arrowheads), with nuclei demonstrated by a methylgreen counterstain. (b) In situ gel zymography demonstrates the proteolytic activity of the chondrocytes. The cells are identified by their autofluorescence and dark areas of lysis in the fluorescent casein substrate can be seen in the territorial region of the chondrocytes (asterisk) and, to a lesser extent, in the interterritorial regions (arrowhead). (c) High-power view of a section treated with EDTA, an inhibitor of MMP activity. No lysis in the fluorescent gel coating is visible. The chondrocytes are identifiable by their slight autofluorescence (arrow). (d) A section immunolocalized for SL-1 shows no positive signal. (e) Immunoreactive SL-2 is clearly demonstrated in the osteoclasts (arrows) close to the bone surface. (f) Dark areas of lysis (asterisks) in the fluorescence-labeled casein substrate demonstrate proteolytic activity of the osteoclasts. The section was incubated in Tris buffer (pH 6.0). (g) A control section incubated in Tris buffer (pH 6.0) containing EDTA, an inhibitor of MMP activity, shows no lysis. (h) SL-1 expression immunolocalized in some osteocytes and surrounding lacunae (arrowheads) close to a remodeling surface. Osteocytes away from the surface show no staining. (i) Higher power view of boxed area in (h) demonstrates proteolytic lysis of the fluorescent substrate in the osteocytic lacunae (arrowheads). Bars 5 20 mm.

bone surfaces and associated with lysis of the casein substrate. Collagenase expression has been demonstrated in human and mouse osteoclasts3,4 and gelatinase B in osteoclasts from many species,4,14,25,26 and thus casein lysis by osteoclasts could possibly be due to any of these MMPs as well as to other proteolytic

enzymes. Incubation of the sections with EDTA, an inhibitor of MMP activity, prevented lysis, indicating that substrate degradation was due to MMP activity. Casein is a known substrate for stromelysin; because the most pronounced lysis occurred at pH 6.0 and gelatinase B and collagenase only act at neutral pH it

12

S. Bord et al. Stromelysin in bone development

seems likely the observed lysis was due to SL-2. However, other proteolytic enzymes may also be involved, possibly regulated by changing pH. Few human studies, particularly in bone, have investigated the presence of both SL-1 and SL-2. Saarialho-Kere et al.22 found that distinct populations of basal keratinocytes expressed stromelysin in chronic wounds from human patients. SL-1 was immunolocalized in basal keratinocytes adjacent to but distal from the wound edge, whereas SL-2 was seen only in the basal keratinocytes at the migrating front, suggesting that the two MMPs may have distinct roles in tissue repair. Several studies have documented the presence of SL-1 in animal growth plate chondrocytes,7,8 but the distribution of SL-2 has not been reported. Edwards et al.11 reported the expression of various MMPs, including SL-1, but not SL-2, in the subarticular zones of 7–14-week human fetal limbs and observed low levels, close to the limit of detection, of SL-1 in chondrocytes. Our finding of only occasional staining in chondrocytes may reflect the different source of tissue. In contrast, SL-2 was evident in most chondrocytes with associated substrate lysis. This lysis was prevented by EDTA, but this observation does not exclude the potential involvement of other MMPs, such as collagenase. As postulated for animal growth plates, stromelysin may modulate proteoglycan degradation prior to mineralization. SL-1 appears to play that role in the rabbit growth plate,8 but its apparent absence, and the presence of SL-2 in the human rib, may represent a species difference. The presence of differential staining patterns in the chondro-osseous junction in the rib indicates that SL-1 may be required for soft tissue degradation, allowing cell migration and making space for the expanding bone, whereas SL-2 may have a more specific role in resorption at precise cellular locations. This study demonstrates differential expression and possibly distinct functional roles for SL-1 and SL-2 in human bone development. Further work is now required to design specific substrates for other MMPs and thus provide information on their potential functional roles and interactions. Acknowledgments: J. E. C. is supported by the Wellcome Trust. R. M. H. is funded by the Medical Research Council. The authors are indebted to Chris Constant, Dr. Nick Coleman, and Dr. Wilf Kelsall for their invaluable supply of human bone.

References 1. Allan, J. A., Hembry, R. M., Angal, S., Reynolds, J. J., and Murphy, G. Binding of latent and high Mr active forms of stromelysin to collagen is mediated by the C-terminal domain. J Cell Sci 99:789 –795; 1991. 2. Birkedal-Hansen, H., Moore, W. G. I., Bodden, M. K., Windsor, L. J., Birkedal-Hansen, B., Decarlo, A., and Engler, J. A. Matrix metalloproteinases: A review. Crit Rev Oral Biol Med 4:197–220; (1993). 3. Blavier, L. and Delaisse´, J.-M. Matrix metalloproteinases are obligatory for the migration of preosteoclasts to the developing marrow cavity of primitive long bones. J Cell Sci 108:364 –365; 1995. 4. Bord, S., Horner, A., Hembry, R. M., Reynolds, J. J., and Compston, J. E. Production of collagenase by human osteoblasts and osteoclasts in vivo. Bone 19:35– 40; 1996. 5. Bord, S., Horner, A., Hembry, R. M., Reynolds, J. J., and Compston, J. E. Distribution of matrix metalloproteinases and their inhibitor, TIMP-1, in developing human osteophytic bone. J Anat 191:39 – 48; 1997. 6. Bradbeer, J., Lindsay, P. C., and Reeve, J. Fluctuation of mineral apposition rate at individual bone remodelling sites in human iliac cancellous bone: Independent correlations with osteoid width and osteoblastic alkaline phosphatase activity. J Bone Miner Res 9:1679 –1686; 1994. 7. Breckon, J. J. W., Hembry, R. M., Reynolds, J. J., and Meikle, M. C. Matrix metalloproteinases and TIMP-1 localization at sites of osteogenesis in the craniofacial region of the rabbit embryo. Anat Rec 242:177–187; 1995. 8. Brown, C. C., Hembry, R. M., and Reynolds, J. J. Immunolocalization of

Bone Vol. 23, No. 1 July 1998:7–12

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

metalloproteinases and their inhibitor in the rabbit growth plate. J Bone J Surg 71-A:580 –593; 1989. Chin, J. R., Murphy, G., and Werb, Z. Stromelysin, a connective tissuedegrading metalloendopeptidase secreted by stimulated rabbit synovial fibroblasts in parallel with collagenase. J Biol Chem 260:12367–12376; 1985. Dodds, R. A. and Gowen, M. The growing osteophyte: A model system for the study of human bone development and remodelling in situ. J Histotechnol 17:37– 45; 1994. Edwards, J. C. W., Wilkinson, L. S., Soothill, P., Hembry, R. M., Murphy, G., and Reynolds, J. J. Matrix metalloproteinases in the formation of human synovial joint cavities. J Anat 188:355–360; 1996. Galis, Z. S., Sukhova, G. K., and Libby, P. Microscopic localization of active proteases by in situ zymography: Detection of matrix metalloproteinase activity in vascular tissue. FASEB J 9:974 –980; 1995. Hembry, R. M., Bagga, M. R., Reynolds, J. J., and Hamblen, D. L. Immunolocalisation studies of six matrix metalloproteinases and their inhibitors, TIMP-1 and TIMP-2, in synovia from patients with osteo- and rheumatoid arthritis. Ann Rheum Dis 54:25–32; 1995. Hill, P. A., Murphy, G. M., Docherty, A. J. P., Hembry, R. M., Millican, T. A., and Reynolds, J. J. The effects of selective inhibitors of matrix metalloproteinases (MMPs) on bone resorption and the identification of MMPs and TIMP-1 in isolated osteoclasts. J Cell Sci 107:3055–3064; 1994. Loveridge, N., Dean, V., Goltzman, D., and Hendy, G. N. Bioactivity of parathyroid hormone and parathroid hormone-like peptide: Agonist and antagonist activities of amino-terminal fragments as assessed by the cytochemical bioassay and in situ biochemistry. Endocrinology 128:1938 –1946; 1991. Meikle, M. C., Bord, S., Hembry, R. M., Compston, J. E., Croucher, P. I., and Reynolds, J. J. Human osteoblasts in culture synthesise collagenase and other matrix metalloproteinases in response to osteotrophic hormones and cytokines. J Cell Sci 103:1093–1099; 1992. Muller, D., Quantin, B., Gesnel, M. C., Millon-Collard, R., Abecassis, J., and Breathnach, R. The collagenase gene family in humans consists of at least four members. Biochem J 253:187–192; 1988. Murphy, G. and Reynolds, J. J. Extracellular matrix degradation. In: Royce, P. and Steinmann, B., Eds. Connective Tissue and Its Heritable Disorders. New York: Wiley-Liss; 1993; 287–316. Murphy, G., Cockett, M. I., Ward, R. V., and Docherty, A. J. P. Matrix metalloproteinase degradation of elastin, type IV collagen and proteoglycan. Biochem J 277:277–279; 1991. Resnick, D. and Niwayama, G. Degenerative disease of extraspinal locations. In: Resnick, D., Ed. Diagnosis of Bone and Joint Disorders (3rd ed.). Vol. 3. Philadelphia: Saunders; 1995; 1263–1372. Rifas, L., Fausto, A., Scott, M. J., Avioli, L. V., and Welgus, H. G. Expression of metalloproteinases and tissue inhibitors of metalloproteinase in human osteoblast-like cells: Differentiation is associated with repression of metalloproteinase biosynthesis. Endocrinology 134:213–221; 1994. Saarialho-Kere, U. K., Pentland, A. P., Birkedal-Hansen, H., Parks, W. C., and Welgus, H. G. Distinct populations of basal keratinocytes express stromelysin-1 and stromelysin-2 in chronic wounds. J Clin Invest 94:79 – 88; 1994. Sirum, K. L. and Brinckerhoff, C. E. Cloning of the genes for human stromelysin and stromelysin-2: Differential expression in rheumatoid synovial fibroblasts. Biochemistry 28:8691– 8698; 1989. Thomson, B. M., Atkinson, S. J., McGarrity, A. M., Hembry, R. M., Reynolds, J. J., and Meikle, M. C. Type I collagen degradation by mouse calvarial osteoblasts stimulated with 1,25-dihydroxyvitamin D3: Evidence for a plasminogen-plasmin-metalloproteinase activation cascade. Biochim Biophys Acta 1014:125–132; 1989. Witty, J. P., Foster, S. A., Stricklin, G. P., Matisian, L. M., and Stern, P. A. Parathyroid hormone-induced resorption in fetal rat limb bones is associated with production of the metalloproteinases collagenase and gelatinase-B. J Bone Miner Res 11:72–78; 1996. Wucherpfennig, A. L., Li, Y.-P., Stetler-Stevenson, W. G., Rosenburg, A. E., and Stashenko, P. Expression of 92 kD type IV collagenase/gelatinase B in human osteoclasts. J Bone Miner Res 9:549 –556; 1994. Zeng, G., McCue, H. M., Mastrangelo, L., and Millis, A. J. T. Endogenous TGFb is modified during cellular aging: Effects on metalloproteinase and TIMP-1 expression. Exp Cell Res 228:271–276; 1996.

Date Received: November 13, 1997 Date Revised: February 25, 1998 Date Accepted: February 25, 1998