In situ hybridization studies allowed for the localization of three MMPs and the angiogenic factor VEGF during secondary ossification. MMPs were widely expressed during ossification of the secondary center, whereas expression of VEGF was restricted to later stages.
Introduction: The spatiotemporal expression patterns of the matrix metalloproteinases gelatinase-B (MMP-9), collagenase-3 (MMP-13), and membrane-type 1 metalloproteinase (MMP-14) and the angiogenic peptide vascular endothelial growth factor (VEGF) were studied during development of the proximal epiphysis of the rat tibia.
Materials and Methods: Cell expression was analyzed by in situ hybridization. Studies on osteoclastic activity, matrix mineralization, cell proliferation, and vascular progression were also performed.
Results: MMP-9, MMP-13, and MMP-14 were expressed in discrete perichondrial cells that gave way to sites of intrachondral canal formation. High expression levels for the three MMPs were found at the blind ends of advancing intrachondral canals and at the expanding borders of the marrow space. Signals for MMP-9 and MMP-13 were in close proximity but did not overlap, whereas MMP-14 was expressed in both MMP-9+ and MMP-13+ cells. VEGF was not expressed during formation of intrachondral vascular canals but was observed in hypertrophic chondrocytes during formation of the bone marrow cavity.
Conclusions: Expression of MMPs and VEGF are constant events during development of the secondary ossification center. We propose that MMPs are involved in targeting proteolytic activity during epiphyseal development. VEGF is not expressed during early formation of vascular canals, but it may have a role in the formation of the bone marrow cavity.
Development of bones through a cartilage model involves formation of primary and secondary centers of ossification. Primary centers are formed at the middle of cartilage templates, whereas secondary centers develop later at the template ends.(1) The overall cellular strategies in developing primary and secondary ossification centers are quite similar, because both processes involve cartilage resorption, angiogenesis, and bone matrix deposition. However, there are some substantial differences between them. Ossification of the primary center begins with the formation of a bony collar around the midshaft of the cartilage template, and this is followed by hypertrophy of cartilage, calcification of the extracellular matrix, and invasion by capillaries and osteogenic cells.(2) Ossification proceeds in a single direction, from the center of the diaphysis to the metaphysis, with a fairly uniform speed during a relatively long period. In contrast, ossification of the secondary center is not preceded by the formation of bony collar or hypertrophy and mineralization of cartilage.(3) A first event is the generation of canals that run through the uncalcified hyaline cartilage of the epiphysis and constitute a path for invading vessels and osteogenic precursor cells.(4,5) Once cartilage canals have been developed, ossification begins in the middle of the epiphysis and progresses according to a centrifugal-radial pattern. Development of primary and secondary centers of ossification results in the establishment of the growth plate, which is the structure responsible for longitudinal bone growth.
Most of the studies on bone development have focused on the primary center of ossification, and knowledge of the cellular processes involved in ossification of the secondary center is low. The mechanisms that trigger the formation of vascular canals in specific regions of the perichondrium are not known. Likewise, the steps directing cartilage resorption to very specific areas of the epiphysis are not understood.(6,7) Furthermore, understanding the mechanism by which ossification is subsequently inhibited in the articular and growth plate cartilages is limited. It has been reported that three matrix metalloproteinases (MMPs) are involved in resorption processes during bone development: gelatinase-B (MMP-9), collagenase-3 (MMP-13), and membrane-type 1 metalloproteinase (MMP-14).(8–10) It has been proposed that MMP-14 has a major role during the development of secondary ossification centers because the process is disrupted in MMP-14-deficient mice.(11) Likewise, the angiogenic factor vascular endothelial growth factor (VEGF) has been reported to couple remodeling of hypertrophic cartilage, ossification, and angiogenesis during development of the primary ossification center.(12) However, its role during ossification of the secondary center is not clear. On this basis and to gain a better understanding of the cellular events during the formation of the secondary ossification center, we have studied the spatiotemporal expression patterns of MMP-9, MMP-13, MMP-14, and VEGF. Analysis of markers of chondrocytic/osteoblastic differentiation (collagens I, II, X, and osteocalcin) and studies of osteoclastic activity, calcification, and cell proliferation were also performed.
MATERIALS AND METHODS
Male Sprague-Dawley rats were obtained from the animal facility building of the University of Oviedo. Rats were killed on days 3, 5, 8, 11, 14, 19, 24, and 35 after birth under institutionally approved animal protocols. We used four animals per age group. All animals were injected intraperitoneally with 5-bromo-2′deoxy-uridine (BrdU; 100 mg/kg body weight; Sigma) 1 h before death. Tibias were isolated and cut through the sagittal plane of the epiphysis into two equal sized parts, obtaining four tibial halves from each animal. Two tibial halves were embedded in paraffin to obtain sections that were used in both histochemical and in situ hybridization studies. One tibial half was processed for mineralization studies, and the last tibial half was processed to obtain semithin sections.
Tissues for histochemical and in situ hybridization studies were fixed by immersion in 4% paraformaldehyde at 4°C for 12 h, rinsed in PBS, decalcified in 10% EDTA (pH 7.0) for 48 h at 4°C, dehydrated through a graded ethanol series, cleared in xylene, and embedded in paraffin. Sections were cut at a thickness of 5 μm and mounted on Superfrost Plus slides (Menzel-Glaser). Serial sections were used for (1) trichrome staining, (2) detection of TRACP activity, (3) localization of proliferating cells, and (4) in situ hybridization studies. Weigert's hematoxylin/Alcian blue/picrofuchsin was used to distinguish cartilage matrix (blue) from bone matrix (red). TRACP activity, a marker of the osteoclast lineage,(13) was determined by incubation with 50 mM sodium acetate (pH 5.2) containing 0.15% Napthol-AS-TR-phosphate, 50 mM sodium tartrate, and 0.1% Fast Red T.R. (Sigma). Proliferating cells were identified immunocytochemically. Sections were placed in 2N HCl for 30 minutes at 37°C for DNA denaturation, washed, and incubated with an anti-BrdU monoclonal antibody (1:20; Sigma). Immunostaining was performed using a mouse ExtrAvidin peroxidase staining kit (Sigma).
Angiogenic activity was quantified by the estimation of the labeling index (percentage of BrdU-labeled cells) in endothelial cells. Two different vascular regions were considered: vessels located between 0 and 100 μm from the cartilage border (vascular front) and vessels located between 100 and 200 μm from the cartilage border (vascular bed). These two vascular regions were studied in vascular canals on the proximal side of bone marrow cavity and on the distal side of the bone marrow cavity. For this purpose, 100 nucleated profiles of endothelial cells were randomly chosen per zone, and the number of BrdU-labeled cells was counted to calculate a percentage. A mean value for each zone was computed on a per-animal basis in the group, and these values were used to calculate a mean and SD per region in each age group. Data were compared between different zones and/or age groups by use of one-way ANOVA, and when differences were significant (p < 0.05), differences between specific means were tested by the Newman-Keuls multiple test.
In situ hybridization was performed as previously reported.(14) Sections were hybridized to [35S]labeled antisense riboprobes and subsequently exposed to photographic emulsion at 4°C for several days, developed, fixed, cleared, and counterstained with 0.02% toluidine blue. Sections hybridized with a labeled-sense riboprobe were used as negative controls. Either sense or antisense [35S]uridine triphosphate-labeled RNA probes were synthesized from the corresponding linearized DNA using the appropriate RNA polymerases. Because resolution with digoxigenin-labeled probes is better than that with radiolabeled probes, additional in situ hybridization studies using digoxigenin-labeled probes were performed to better characterize the positive cell types. Probes were prepared with the DIG RNA labeling mix (Boehringer Mannheim), and the hybridized probes were detected with the alkaline phosphatase-coupled anti-DIG antibody (Boehringer Mannheim). Probes for in situ hybridization were as follows: probe for mouse α1 type I collagen gene was a 240-bp fragment amplified from the 3′ flanking region of the gene with the oligonucleotides 5′-GTCTGCTTCGTGTAAA-CTCCC and 5′-GGTAAGGTTGAATGCACTTTT and subcloned into pBluescript vector (GenBank accession no. X15896). The type II collagen probe was a 550-bp PstI fragment from the amino-terminal portion of rat proα1(II) chain cloned in PGEM 3Zf-vector(15) (GenBank accession no. K02804). Probe for murine osteocalcin consisted in a 212-bp fragment amplified from embryo RNA with the oligonucleotides 5′-TCTCTCTGCTCACTCTGCTGG and 5′-AGCAGGGTTAAGCTCACACT and subcloned into pBluescript vector (GenBank accession no. S67456). Type X collagen probe was a 650-bp HindIII fragment containing 360 bp of noncollagenous (NC1) domain and 290 bp of 3′-untranslated sequence of the mouse type X collagen gene, subcloned into the HindIII site of pBluescript(16) (GenBank accession no. NM_009925). Rat collagenase-3 probe was a 314-bp fragment corresponding to nucleotides 350-653 of the rat collagenase-3 gene(17) (GenBank accession no. M60616). The gelatinase B probe was generated by subcloning into pBluescript a 1353 BamHI fragment obtained by RT-PCR from embryo RNA with the following oligonucleotides: 5′-TGGCACCATCATAACATCACCT and 5′-AGAAGAAAATCTTCTTGGGCTG (GenBank accession no. NM_031055). The MMP-14 probe was a 640-bp PCR fragment from positions 1060-1700 in the human cDNA sequence (sequence data available from GenBank under accession no. D26512). The VEGF probe was a 450-bp PCR fragment from positions 1035-1485 in the human cDNA sequence subcloned into pBluescript (GenBank accession no. NM_003376).
For studies of matrix mineralization, tissues were fixed in 4% paraformaldehyde in PBS for 5 h at 4°C, dehydrated in acetone, embedded in Durkupan-ACM (Sigma), and sectioned at 2 μm on a Reicher Ultracut E ultramicrotome. The von Kossa staining was used to detect mineralization by setting sections in 1% AgNO3 for 60 minutes at room temperature and fixed with 5% sodium hyposulfite.
Semithin sections were obtained from tissues fixed in 2% glutaraldehyde and 0.7% ruthenium hexamine trichloride (RHT; Strem Chemicals) in 0.05 M cacodylate buffer, pH 7.4, for 3 h at 4°C. They were postfixed in 1% osmium tetroxide and 0.7% RHT in cacodylate buffer for 2 h. After washing, they were dehydrated with acetone, embedded in Durkupan-ACM (Sigma), sectioned at 0.5 μm on a Reicher Ultracut E ultramicrotome, and stained with toluidine blue.
MMPs are first expressed at discrete perichondrial cells that give way to sites of intrachondral canal formation
The proximal epiphysis of the tibia was completely cartilaginous at the third postnatal day (Fig. 1A). Expression of MMP-9, MMP-13, and MMP-14 was detected in a low number of perichondrial cells located at the superior intercondylar surface (Figs. 1B-1D). In contrast, no VEGF+ cells were detected in the epiphysis at this stage. Analysis of serial sections hybridized with the different probes showed that the three MMPs were expressed in close proximity but because of the small size of perichondrial cells, it was not possible to ascertain whether the three MMPs were co-expressed or expressed by different cells located close to each other.
A clear relationship between the location of MMP+ perichondrial zones at the third postnatal day and the sites of intrachondral canal development by the age of 5 days was observed (Figs. 1E-1H). Once formed, canals progressed through the unmineralized resting cartilage toward the middle of the epiphysis (Fig. 1I), where cartilage showed signs of incipient hypertrophy, and the first foci of calcified matrix were detected by von Kossa staining (Fig. 1J). Canals contained buds of mesenchymal-type cells and blood vessels (Fig. 1K). Numerous BrdU-labeled cells were located inside the canals, although labeling was also observed in the perichondrium and in chondrocytes surrounding cartilage canals (Fig. 1L). BrdU labeling was especially high in endothelial cells of blood vessels located close to the cartilage interface (Table 1). Chondrocytes surrounding the canals showed expression of type II collagen (Fig. 1M) but did not express type X collagen (Fig. 1N), consistent with their resting phenotype. Cells inside canals expressed low levels of type I collagen (Fig. 1O), whereas osteocalcin was not detected (Fig. 1P).
Table Table 1.. Labeling Index (%) of Endothelial Cells in Vessels Located Between 0 and 100 μm From the Cartilage Border and Vessels Located Between 100 and 200 μm From the Cartilage Border in Three Different Epiphyseal Areas: Intrachondral Canals (IC), the Proximal Side of Bone Marrow Cavity (PM), and Distal Side of the Bone Marrow Cavity (DM).
Analysis of TRACP activity showed the existence of numerous positive cells located at the inner border of the canals, close to cartilage matrix (Fig. 1Q). At this stage, TRACP+ cells were mononucleated and primarily located at the advancing tips. MMP-9, MMP-13, and MMP-14 were concomitantly expressed at similar locations during canal formation, but analysis of serial sections showed that the three MMPs did not have the same distribution patterns. Expression of MMP-9 was localized to a few perivascular cells located at the margins of the canals (Fig. 1R). Expression of MMP-13 was mainly observed in perivascular cells clustered in small foci at the interface with resting cartilage near the blind end of canals (Fig. 1S). Expression of MMP-14 was also found at the blind ends of canals, but it presented a more extensive pattern compared with that of either MMP-13 or MMP-9 (Fig. 1T). Positive cells were not restricted to the margin but extended at a distance from the interface region. Comparative analysis of serial sections showed that cells expressing MMP-13 and cells expressing MMP-9 had a close spatial relationship but there was no overlap. On the other hand, cells positive for MMP-13 were also positive for MMP-14. Likewise, most of cells expressing MMP-9 co-expressed MMP-14. Finally, there was a distinct population of cells located at a distance from the interface region that only expressed MMP-14. On the basis of the expression profile, three populations of MMP-expressing cells were identified in cartilage canals: cells co-expressing MMP-13 and MMP-14, cells co-expressing MMP-9 and MMP-14, and cells expressing only MMP-14. Likewise, MMP-9 was expressed in a subset of TRACP+ cells. Finally, expression of the three MMPs studied was not detected in chondrocytes or endothelial cells at this stage.
Formation of the marrow cavity involves increased MMP expression and the onset of VEGF expression
From day 5 onward, invading canals gradually extended centripetally into the cartilage. By 8 days of age, the canals reached the middle of the epiphysis, began to spread transversely, and fused with each other to form a central cavity that enlarged centrifugally (Fig. 2A). Formation of the epiphyseal cavity was associated with extended mineralization of the cartilaginous matrix detected by von Kossa staining (Fig. 2B) and increased cell proliferation visualized by BrdU labeling (Fig. 2C). Multinucleated cells showing osteoclast-like structure appeared at the cartilage-osseous interface (Fig. 2D). In addition, small mononucleated cells forming aggregations of 5-10 cells were found next to terminal hypertrophic chondrocytes (Fig. 2D). These cells had a cytoplasmic process that projected from the cell apex toward the septa of the last row of hypertrophic chondrocytes and displayed a visible association with small resorption tips (Fig. 2D).
Formation of the marrow cavity was associated with a marked increase of the number of MMP-expressing cells. MMP-9 (Figs. 2E-2F), MMP-13 (Fig. 2G), and MMP-14 (Fig. 2H) were expressed at similar locations within the developing marrow space. Positive cells were mainly located at the margin between the marrow cavity and the surrounding cartilage. Positive cells also appeared scattered throughout the central cavity and in the remaining cartilage canals located at the periphery (Figs. 2E-2H). The surrounding chondrocytes were hypertrophic and expressed type X collagen (data not shown). Cells inside the central cavity expressed type I collagen but no signs of osteocalcin expression were detected at this stage (data not shown).
Analysis of serial sections hybridized with different [35S]labeled probes showed that the three MMPs did not have the same distribution pattern. MMP-9 was mainly expressed in scattered cells located at the inner border of the marrow cavity, close to the surrounding hypertrophic cartilage (Fig. 2I). MMP-13+ cells showed a tight spatial relationship with those positive for MMP-9 at the border of the marrow cavity, but no overlap between the two probes was observed (Fig. 2J). Cells positive for MMP-14 were also found at the margin of the marrow cavity, but they extended at a distance from the chondro-osseous margin (Fig. 2K). Analysis of TRACP activity showed positive cells scattered along the expanding border of the marrow cavity (Fig. 2L). Hybridization with digoxigenin-labeled probes resulted in a better resolution that allowed different MMP-expressing cells to be more readily distinguished (Figs. 2M-2O). Thus, transcripts for MMP-9 were seen in two cell types: small-mononucleated cells and large-multinucleated cells having cytoplasmic processes (Fig. 2M). These positive cells were scattered along the chondro-osseous margin of the marrow cavity, similar to those observed in the previous experiment using [35S]labeled probe (Fig. 2I). MMP-13+ cells were small in size, mononucleated, and formed small aggregations of 5-10 cells located close to terminal hypertrophic chondrocytes. The size, shape, and position of these MMP+ cells, as well as their mode of aggregation, were comparable with those mononucleated cells having cytoplasmic processes extending toward septa of hypertrophic cartilage discernible in semithin sections (Fig. 2P). Finally, there was a large number of cells expressing MMP-14 at the chondro-osseous margin, most of them being small in size and mononucleated, although some large multinucleated cells were also found (Fig. 2O). Analysis of serial sections hybridized with the different probes showed that the expression patterns of MMPs in the marrow cavity were basically the same as in cartilage canals: MMP-9+ and MMP-13+ cells were in close vicinity, but they did not overlap. Likewise, transcripts for MMP-14 were present in both MMP-13+ and MMP-9+ cells, whereas there was a population of cells that only expressed MMP-14. TRACP+ cells in adjacent sections corresponded to the location of MMP-9 expression, although there was a population of cells that expressed MMP-9 and were TRACP−.
Expression of VEGF was first detected in the epiphysis during the early stages of the formation of the central marrow cavity. At this stage, VEGF+ cells formed a continuous rim around the central cavity that was restricted to hypertrophic cartilage (Figs. 2Q-2S). Positive cells were never found inside the marrow cavity, and no difference was observed in VEGF expression between hypertrophic chondrocytes of the epiphysis and those of the metaphyseal growth plate (Figs. 2Q-2S). The onset of VEGF expression was associated with an increase in angiogenic activity, quantified as the percentage of BrdU-labeled endothelial cells (Table 1). BrdU labeling was markedly increased in the region adjacent to the interface with VEGF-expressing hypertrophic chondrocytes (Fig. 2T; Table 1), whereas it significantly decreased in the region located further from the chondro-osseous interface.
Development of epiphyseal polarity is associated with changes in the expression patterns of both MMPs and VEGF
Invasion of hypertrophic cartilage by the marrow space proceeded uniformly from the center to the periphery during early stages of marrow cavity formation, and this resulted in circumferential enlargement. However, differences in the progression of the ossification process between the proximal chondro-osseous interface (that facing the articular surface) and the distal chondro-osseous interface (that facing the metaphysis) appeared by the age of 11 days. Ossification was unchanged at the proximal side, but it became modified at the distal side. Modifications at the distal side included attenuation of the hypertrophic cartilage, changes in mineral deposition, and modifications in microvascular organization. The amount of hypertrophic cartilage progressively decreased at the distal side and, by day 14, resting cartilage outlined most of the distal chondro-osseous junction, and the shape of the marrow cavity changed from spherical to hemispherical (Fig. 3A). Extensive mineralization occurred on the distal side, and this resulted in the formation of an incipient epiphyseal bone plate (Fig. 3B). BrdU labeling was markedly decreased in the region adjacent to the distal chondro-osseous interface, whereas it remained unchanged at proximal and lateral regions (Fig. 3C). Capillaries at the distal cartilage-bone interface appeared in close proximity to resting cartilage, but they did not form a proper vascular front because they presented relatively wide lumina, variable contours (Fig. 3D), and very low BrdU labeling (Table 1). No significant differences in the percentage of BrdU-labeled endothelial cells between vascular vessels located close to cartilage and those located further from the chondro-osseous interface were found in the distal side at this age (Table 1).
The development of morphological differences between proximal and distal sides of the marrow cavity was associated with changes in the expression pattern of the MMPs. Expression levels of MMP-9 (Figs. 3E-3F), MMP-13 (Fig. 3G), and MMP-14(Fig. 3H) were markedly decreased in the region adjacent to the distal chondro-osseous interface, whereas they remained basically unchanged at proximal and lateral regions. Decreased expression of MMPs was especially evident in regions where resting cartilage directly outlined the marrow cavity (Figs. 2E-2H). Cells inside the central cavity expressed both type I collagen and osteocalcin, but positive cells were located at a distance from the resorption front (data not shown). In the same way, clear proximo-distal differences in VEGF expression were found from day 14 onward. VEGF was expressed at proximal and lateral regions of the chondro-osseous interface, whereas no expression was found at the distal interface (Figs. 3I-3K). VEGF signal was restricted to sites where hypertrophic chondrocytes outlined the marrow cavity, whereas there was no expression where bordering cartilage was at the resting stage (Fig. 3L).
Progression of the ossification process accentuated epiphyseal polarity by forming a transverse plate of bone at the distal side. This structure was clearly defined at the age of 35 days (Figs. 3M-3N). The formation of the transverse plate implied the end of cartilage resorption at the distal border, although resorption remained at the proximal side where hypertrophic cartilage was still present (Fig. 3M). Capillary sprouts formed a well-defined vascular front with considerable BrdU labeling (Fig. 3O; Table 1), and numerous resorptive cells could also be observed at the proximal side (Fig. 3P). The formation of the bone plate at the distal border was associated with a marked decrease in MMP expression, whereas expression was retained at the proximal side. In the same way, VEGF was only expressed at the proximal chondro-osseous interface, where some hypertrophic chondrocytes persisted.
It is well known that the proteolytic activity of MMPs is essential for normal bone development. Many studies on the expression patterns of MMPs during endochondral ossification have been reported.(10,18) Because most of these studies focused on primary ossification, our study was focused on the comparison of the expression of MMPs and VEGF during primary versus secondary ossification. Results obtained showed that expression of MMP-9, MMP-13, and MMP-14 began simultaneously at discrete locations in the perichondrium that gave way to sites of intrachondral canal formation and then continued at the blind end of advancing canals. Expression was also increased at the borders of the marrow cavity and decreased in nonexpanding edges of the marrow cavity. This spatiotemporal expression pattern strongly suggests that the three MMPs are required to generate the profound remodeling of the extracellular matrix that is essential for ossification of the epiphysis. Likewise, our results showed that VEGF was not expressed during formation of intrachondral vascular canals and thus is not responsible for the beginning of secondary ossification. Nevertheless, VEGF was expressed in hypertrophic chondrocytes during formation of the bone marrow cavity. Overall, the results show that the expression patterns of MMPs and VEGF during secondary ossification have substantial similarities with that reported during primary ossification,(19) but some significant differences also exist.
It could be considered that MMP-9, MMP-13, and MMP-14 have a major role in eroding epiphyseal cartilage to allow accommodation of blood vessels and osteogenic cells.(20) In this regard, overlap between their expression, both temporally and spatially, suggests the existence of functional interaction among them. Accordingly, it has been reported that MMP-9 and MMP-13 have cooperative effects resulting in an increase of collagenolytic activity.(10,21) In the same way, it has been proposed that MMP-9 and MMP-13 cooperate during degradation of the unmineralized septa of hypertrophic chondrocytes.(22) Furthermore, studies on the double-null mice, MMP-9−/− and MMP-13−/−, show that these MMPs act synergistically in the initiation and development of the primary and secondary sites of ossification.(18) Nevertheless, recent studies have shown that MMP activity does not only serve to erode the extracellular matrix (ECM) but also plays a significant role in regulating more subtle functions such as vascular invasion, bioavailability of growth factors, chondrocyte apoptosis, and recruitment of osteoprogenitor cells.(18) Present data on the expression pattern of MMPs are consistent with the possibility that ECM remodeling by MMPs could be a basic process underlying the synchronized differentiation of chondrocytes, osteoblasts, and endothelial cells during development of the secondary ossification centers.
The in situ hybridization studies presented here showed the localization of three different populations of MMP-expressing cells at zones of active cartilage resorption: cells co-expressing MMP-13 and MMP-14, cells co-expressing MMP-9 and MMP-14, and cells expressing MMP-14. Cells co-expressing MMP-13 and MMP-14 presented surface protrusions extending toward cartilaginous septa where resorption tips were microscopically recognizable. Comparable surface protrusions in other cell types have been proposed to have specific membrane domains with active MMP-14 involved in the process of MMP activation.(23) The fact that cells co-expressing MMP-13 and MMP-14 were TRACP− implies that they belong to a cell lineage different from the osteoclast line. This is interesting in that most studies to date have reported MMP-14 expression in osteoclasts.(24,25) The existence of cells expressing MMP-14 belonging to a cell lineage different from the osteoclast line is consistent with previous studies in which mice lacking MMP-14(11,26) have more severe skeletal abnormalities than mice lacking both osteoclasts and bone marrow macrophages.(27,28) The results indicate that MMP-14 must be expressed in cells other than osteoclasts. Cells co-expressing MMP-9 and MMP-14 as well as some of the cells expressing MMP-14 alone were TRACP+. In this way, both MMP-9 and MMP-14 have been considered as osteoclast-associated genes.(29) Because it has been reported that different genes are expressed at distinct times during osteoclast differentiation, it could be considered that TRACP+ cells expressing MMP-14 alone and TRACP+ cells co-expressing MMP-9 and MMP-14 may represent different stages of osteoclastogenesis. MMP-9-expressing cells located at the chondro-osseous junction of the metaphyseal growth plate have been postulated to be chondroclasts, a hypothesized cell type of undefined origin whose function is to resorb cartilage.(30,31)
MMP-9 has been reported to be specifically involved in proteolysis of nonmineralized cartilage.(32) This fact, together with its expression at the advancing tips of cartilage canals, strongly suggests a role in the formation of cartilage canals. However, it has been reported that invasion at the secondary site of ossification was only minimally affected in mice lacking MMP-9,(31) and therefore, other enzymes may compensate for its function. In contrast, mice deficient in MMP-14 failed to develop intrachondral canals,(11,26) indicating that this enzyme plays a predominant role in this process.
Cells do not typically produce MMPs unless they are needed for remodeling, and therefore, regulation of MMP activity is primarily at the level of gene transcription.(33) However, MMP activity also depends on the activation of latent proenzymes (with the exception of MT-MMPs) by other proteolytic enzymes such as plasmin and other MMPs.(34) The activated enzyme can be inhibited by general inhibitors, such as α-2-macroglobulin, or locally by tissue inhibitors of MMPs (TIMPs) that bind MMPs with high affinity.(35) Results obtained in this study show that the level of MMP expression differs greatly in cells located next to each other, for example, in the blind ends and lateral walls of intrachondral canals or the expanding and nonexpanding borders of the marrow space. The occurrence of positive and negative cells at close proximity means that expression of MMPs could be turned off or on rapidly at specific areas. This fits with the assumption that regulation of MMPs during ossification of the secondary center takes place at the transcriptional level. Nevertheless, regulation at other levels cannot be excluded.
A key event during endochondral ossification is the control of angiogenesis. Blood vessels grow synchronously with cartilage resorption and bone deposition, implying the existence of local factors capable of coupling these processes. VEGF is a candidate for this function because receptors for VEGF have been reported in both endothelial cells and osteoclasts.(36) VEGF blockage caused suppression of blood vessel invasion, concomitant with impaired trabecular bone formation, at the metaphyseal growth plate cartilage.(12) Likewise, studies using mice expressing only the VEGF 120 isoform have provided evidence for multiple important roles of VEGF in both endochondral and intramembranous bone formation.(37) In this way, a recent study using an ex vivo system has shown that VEGF is involved in many processes during endochondral ossification but does not seem to participate to the same extent in intramembranous ossification.(38) In this study, VEGF was not detected in any cell type associated with formation of vascular canals, whereas hypertrophic chondrocytes bordering the early bone marrow cavity expressed VEGF. Furthermore, the onset of VEGF expression was temporally coincident with a significant change in vascular progression. Both VEGF expression and vascular invasion simultaneously ceased at the distal side of the epiphysis. These results differ somewhat from those recently reported by Morini et al.(39) for secondary ossification in the rat humerus. In that study, VEGF was detected by immunohistochemistry in some mesenchymal cells and resting chondrocytes as well as hypertrophic chondrocytes. Our results are in agreement with the hypothesis that VEGF is not involved in triggering vascularization of the epiphysis during cartilage canal formation but is actively responsible for neovascularization of hypertrophic cartilage during bone marrow expansion. These results lead to a model of two distinct angiogenic processes: angiogenesis during formation of cartilage canals, which is independent of VEGF, and angiogenesis during bone marrow expansion, which requires VEGF. These results are consistent with previous works indicating that vascularization of the chondroepiphysis comprises two different processes: quiescent angiogenesis and reactionary angiogenesis.(40) The first involves formation of discrete vascular networks that progress directionally without associated ossification, whereas the second is responsible for systematic vascularization throughout the central region of the epiphysis and is associated with ossification. Angiogenesis during these two processes shows marked differences that must be associated with distinct regulatory mechanisms. VEGF could be specifically involved in the control of reactionary angiogenesis, whereas the nature of the factor responsible for activation of quiescent angiogenesis remains to be studied. It is remarkable that many putative angiogenic factors have been identified in hypertrophic cartilage but not in resting cartilage, which on the contrary, is rich in anti-angiogenic activities.(41) Because the angiogenic response depends on a balance of activators and inhibitors, the possibility that angiogenesis during cartilage canal formation is caused in part by a local decrease of anti-angiogenic activity cannot be excluded. It has been proposed that one function of MMP-9 is to generate angiogenic activators or inactivate angiogenic inhibitors.(31) The results presented here showing that MMP-9 is expressed at discrete zones within the perichondrium that give way to sites of intrachondral canal formation are consistent with the hypothesis that MMP-9 itself or one of the substrates that could act as an angiogenic factor.
In conclusion, we show that the MMPs, MMP-9, MMP-13, and MMP-14, are all expressed during ossification of the secondary center with expression patterns that have a clear relationship to the sequence of events that take place during epiphyseal ossification. These results strongly suggest that the three MMPs functionally interact and are involved in targeting proteolytic activity during different stages of epiphyseal development. Likewise, VEGF is not involved in formation of vascular canals, but it is expressed during formation and vascularization of the bone marrow cavity.
This work was supported by funds from MCYT (Spain) MCT-00-BMC-0446. JA was supported by the Instituto Universitario de Oncologia del Principado de Asturias (IUOPA), and LC was supported by the Ministerio de Ciencia y Tecnologia (MCYT, FP2000-5486). We are grateful to Carlos López-Otin (University of Oviedo) for many helpful comments. Maria Quintana is acknowledged for assistance in the laboratory.