Michael J. F. Blumer PhD, Department of Anatomy, Histology and Embryology, Division of Clinical and Functional Anatomy, Innsbruck Medical University, Müllerstrasse 59, A-6010 Innsbruck, Austria. T: +43 12 900371120; F: +43 512 900373112; E: email@example.com
Endochondral bone formation, the process by which most parts of our skeleton evolve, leads to the establishment of the diaphyseal primary (POC) and epiphyseal secondary ossification centre (SOC) in long bones. An essential event for the development of the SOC is the early generation of vascularized cartilage canals that requires the proteolytic cleavage of the cartilaginous matrix. This in turn will allow the canals to grow into the epiphysis. In the present study we therefore initially investigated which enzymes and types of cells are involved in this process. We have chosen the mouse as an animal model and focused our studies on the distal part of the femur during early stages after birth. The formation of the cartilage canals was promoted by tartrate-resistant acid phosphatase (TRAP) and membrane type-1 matrix metalloproteinases (MT1-MMP). In addition, macrophages and cells containing numerous lysosomes contributed to the establishment of the canals and enabled their further advancement into the epiphysis. As development continued, the SOC was formed, and in mice aged 10 days a distinct layer of type I collagen (= osteoid) was laid down onto the cartilage scaffold. The events leading to the establishment of the SOC were compared with those of the POC. Basically these processes were quite similar, and in both ossification centers, TRAP-positive chondroclasts resorbed the cartilage matrix. However, occasionally co-expression of TRAP and MT1-MMP was noted in a small subpopulation of this cell type. Furthermore, numerous osteoblasts expressed MT1-MMP from the start of endochondral ossification, whereas others did not. In osteocytogenesis, MT1-MMP has been shown to be critical for the establishment of the cytoplasmic processes mediating the communication between osteocytes and bone-lining cells. Considering the well-known fact that not all osteoblasts transform into osteocytes, and in accordance with the present data, we suggest that MT1-MMP is needed at the very beginning of osteocytogenesis and may additionally determine whether an osteoblast further differentiates into an osteocyte.
Most parts of our skeleton are formed by a process commonly referred to as endochondral ossification. In long bones the primary ossification centre (POC) develops first within the diaphysis, followed by the establishment of the secondary ossification centre (SOC) within the epiphysis. Basically, the formation of both ossification centres is quite similar, involving cartilage mineralization and resorption coupled with apoptosis of resident cells, angiogenesis, and finally de novo synthesis of the bone matrix. However, the development of the SOC differs in some fundamental aspects from that of the POC in that the vascularization occurs prior to chondrocyte hypertrophy and cartilaginous matrix mineralization (Roach, 2000; Álvarez et al. 2005a; Holmbeck & Szabova, 2006).
In murine species, the cartilage canals further erode the epiphyseal cartilage with subsequent development and give rise to the formation of the marrow cavity and several small ossification nuclei that finally join into a large SOC (Morini et al. 2004; Blumer et al. 2007). These processes are likewise governed by MMPs and additionally by aggrecanase, vascular endothelial growth factor (VEGF), and TRAP-positive chondroclasts as well as osteoclasts (Lee et al. 2001; Álvarez et al. 2005a; Blumer et al. 2008 for review). Chondroclasts are indistinguishable from osteoclasts as both cell types are multinucleated. The former resorb the calcified cartilage, whereas the latter resorb the mineralized bone matrix. Chondro-/osteoclasts derive from the macrophage-monocyte lineage, and their multinucleated phenotype is reached by fusion of the mononucleated cells (Roach, 2000; Takahara et al. 2004). In a previous in vivo study (Irie et al. 2001) it has been demonstrated that osteoclasts are capable of expressing MT1-MMP; however, it is still uncertain whether this is also valid for chondroclasts.
In endochondral bone formation the non-resorbed cartilage struts provide a scaffold onto which osteoblasts start to deposit osteoid (= non-mineralized type I collagen). Osteoblasts further differentiate into osteocytes as soon as they are completely embedded into the mineralized bone matrix but this transformation process is still not fully understood (Franz-Odendaal et al. 2006 for review). However, it has been shown that in rats osteocytogenesis is among others regulated by MT1-MMP (Filanti et al. 2000), and in the cortical bone of mice the investigations of Holmbeck et al. (2005) provide evidence that this proteolytic enzyme cleaves type I collagen. This cleavage is needed for a normal osteocyte phenotype, characterized by its cytoplasmic processes through which the communication between osteocytes and bone lining cells becomes possible. We speculate that MT1-MMP is also expressed at the very onset of endochondral bone formation.
In this study we investigated the mouse femur during early stages of endochondral ossification. First, we clarified at which moment TRAP-reactive cells are noted in cartilage canals. Secondly, we verified whether macrophages can be detected in the canals, using F4/80 as a marker for mouse macrophages. Third, we investigated the expression pattern of MT1-MMP and TRAP within the SOC. We examined whether MT1-MMP is expressed in osteoblasts at the very beginning of endochondral bone formation and compared the results of the SOC with those obtained from the POC. In addition, we tried to clarify whether the chondroclasts express MT1-MMP.
Materials and methods
Mice were obtained from the central laboratory animal facilities of the Innsbruck Medical University. Developmental stages day (D) 5, 8, 10 after birth were used, and two animals per age group were investigated in this study. Mice were anesthetized with CO2, and killed by cervical displacement. Subsequently, the legs were amputated and the femur was isolated from the tibia. The distal part of the femur was examined.
Light and transmission electron microscopy
The bones of the mice were fixed in 2.5% glutaraldehyde, 2% paraformaldehyde buffered in sodium cacodylate (0.1 m), pH 7.4 for 4 h at room temperature and rinsed in the same buffer. They were postfixed in 0.5% osmium tetroxide, 1% potassium hexacyanoferrate III in distilled water overnight at 4 °C, rinsed in distilled water and decalcified in 3% ascorbic acid in sodium chloride (0.15 m) for 12 h at 4 °C. This was followed by dehydration in graded ethanol series and embedding in Spurr's epoxy resin. Ribbons of consecutive semithin sections (2 µm) were cut on a Reichert Ultracut S microtome (Leica Microsystem, Wetzlar, Germany) with a histo-jumbo-diamond knife (Diatome, Biel, Switzerland) (Blumer et al. 2002) and stained with toluidine blue for 20 s at 60 °C. Complete series of ultrathin sections (80 nm) were cut on a Reichert Ultracut S microtome with an ultra-diamond knife, mounted on dioxane-formvar coated copper slot-grids and stained with an aqueous solution of uranyl acetate (1%) for 30 min at 20 °C followed by lead citrate for 3 min at 20 °C. The ultrathin sections were examined with an electron microscope 10A (Zeiss, Oberkochen, Germany).
Tissue preparation for enzyme histochemistry and immunohistochemistry
Bones were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS, 0.1 m) for 4 h, rinsed in the same buffer and decalcified as described before. Subsequently, the bones were dehydrated in graded isopropanol and xylene series and embedded in paraffin. Serial sections (6–7 µm) were made on an HM 355S microtome (Microm, Walldorf, Germany) and collected on SuperFrost®Plus slides.
Enzyme histochemistry (HC)
Tartrate-resistant acid phosphatase (TRAP) is an established marker for the chondro-/osteoclast lineage. To show TRAP activity, sections were deparaffinized, rinsed in PBS and incubated with a solution containing 50 nm sodium acetate (pH 5.2), 0.15% Naphthol-AS-TR-phosphate, 50 nm sodium tartrate, and 0.1% Fast Red T.R. (Sigma Aldrich Chemie Gmbh, Taufkirchen, Germany) for 30–40 min at room temperature. Subsequently, sections were rinsed in distilled water.
For type I collagen, a rabbit anti-human polyclonal antibody (1 : 100 in antibody diluents) (LF-67 from Prof. L. Fisher, National Institutes of Health, Bethesda, MD, USA) was used. Sections were deparaffinized, rinsed in PBS, and endogenous peroxidase activity blocked with 0.5% H2O2 in 30% methanol for 15 min in the dark. The subsequent protocol comprised primary antibody incubation overnight at 4 °C and saturation of the unspecific sites with 10% normal goat serum (NGS). This was followed by the application of a secondary antibody (goat anti-rabbit IgG/HRP-conjugated 1 : 500 in PBS) (catalogue no. P0448, DakoCytomation, Glostrup, Denmark) for 3 h at room temperature, washing in PBS and chromogenic detection of the antigen–antibody complex with 0.05% diaminobenzidine (DAB), 0.01% H2O2 in distilled water in the dark for approximately 10 min.
Immunolocalization of membrane-type 1 metalloproteinase (MT1-MMP) was performed with a rabbit anti-human polyclonal antibody (1 : 75 in antibody diluents) (catalogue no. sc-30074, Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA). The sections were handled as described before and after blocking of the endogenous peroxidase activity, heat-induced epitope/antigen retrieval with citric buffer (1 : 10, pH = 6.0) was performed for 7 min in a microwave oven (400 W). After a cooling interval of 20 min, the sections were washed in PBS, incubated with the MT1-MMP antibody overnight at 4 °C and rinsed. This was followed by saturation of the unspecific sites NGS, and the visualization of the antigen-antibody complex was done as mentioned before.
Macrophages were localized with the rat anti-mouse F4/80 monoclonal antibody (1 : 75, 1.5% NGS in PBS) (Santa Cruz Biotechnology, catalogue no. sc-52664). The sections were deparaffinized, endogenous peroxidise was quenched and subsequently unspecific sites were blocked with 1.5% NGS. They were incubated with the F4/80 antibody overnight at 4 °C, rinsed in PBS and detection of the antigen-antibody complex performed following the protocol of the rat ABC staining system (Santa Cruz Biotechnology, catalogue no. sc-2019), using a goat anti-rat biotinylated secondary antibody (4 h), an AB enzyme reagent/HRP conjugated (1 h) and DAB (10 min).
Double labelling (immunohistochemistry and histochemistry)
To confirm the presence of F4/80-positive macrophages and TRAP cells, IHC was done first, and TRAP detection performed thereafter.
To confirm the presence of MT1-MMP- and TRAP-positive cells, TRAP detection was carried out first because heat-induced epitope/ antigen retrieval involved exposure of the sections to a high temperature (about 100 °C) that evidently destroyed the enzyme. After staining for TRAP was noted, the sections were rinsed in PBS, and IHC was performed as described before. Similarly, to identify type I collagen and TRAP-positive cells, HC was done first and IHC thereafter. This showed a nicer labelling pattern than if IHC was done first.
All sections were counterstained with Gils’ haematoxylin. Negative controls were obtained by substituting the primary antibody with antibody diluents. These sections yielded no labelling.
Light microscopy (LM) and transmission electron microscopy (TEM)
Structure of cartilage canals and the secondary ossification centre (SOC)
In the distal femoral epiphysis of mice, cartilage canals were seen for the first time 5 days after birth, starting off as invaginations of the perichondrium. They were not branched, and the electron translucent lumen of the canals contained vessels and mesenchymal cells. In addition, cells with numerous lysosomes that contained fibrils, some of them cross-banded, were uniformly distributed in the canal lumen (Fig. 1). In 8-day-old mice, additional canals were encountered; they remained non-branched and until this moment the epiphysis comprised only resting, non-calcified cartilage. At day (D) 10, the histology of the epiphysis had altered. The canals and their vessels were now intensively ramified within the calcified hypertrophic zone, whereas the segments of the canals within the resting zone remained non-branched and contained the same cell types as described before (Fig. 2A). Within the branched segments, mesenchymal and lysosomal cells were present likewise. In addition, multinucleated chondroclasts with ruffled borders and osteoblasts with a well-developed rough endoplasmic reticulum covered the cartilage scaffold, which was surrounded by an electron dense layer (Fig. 2B,C). This layer was equivalent to the newly formed type I collagen, which had also been demonstrated by Blumer et al. (2007; see Figs 4D and 5C). The osteoblasts extended short cytoplasmic processes towards the collagen seam (Fig. 2C). Thus, in mice aged 10 days the first evidence of endochondral bone formation was noted in the ramified segments of the canals, leading to the establishment of several small ossification nuclei that joined into a large SOC during preceding development. We have previously generated three-dimensional models of the mouse distal femur to elucidate these events (Blumer et al. 2007; Fig. 1E,F). However, to simplify matters we will designate the ossification nuclei as the secondary ossification centre (SOC) in the present study.
Immunohistochemistry (IHC) and enzyme histochemistry (HC)
Staining pattern of cartilage canals
Serial sections through a cartilage canal 5 days after birth demonstrated that mononucleated TRAP cells were present at its most apical tip as well as its lateral wall. Furthermore, several macrophages were distributed within the lumen of the canal but the examination of numerous canals provided evidence that these cells were never encountered at the tip of the canals (Fig. 3A–C). In mice aged 8 days the staining pattern for TRAP and F4/80 was similar when compared with D 5 (data not shown). At D 10, both cell types were noted in the non-branched segments and in those areas where the ramification of the canals started off. However, compared with D 5, the macrophages now stained more strongly and were more numerous in contrast to the TRAP-reactive cells (Fig. 3D,E). It should be emphasized that F4/80 labelling never overlapped with TRAP staining. MT1-MMP is essential for epiphyseal bone development (Holmbeck et al. 1999) and in accordance with a recent study on the rat tibia (Álvarez et al. 2005a) several canal cells in mice also stained with antibodies for this proteolytic enzyme. These cells were found in all developmental stages; they were regularly distributed within the cavity of the canals and also seen at the canal tip (Fig. 3F). Co-expression of TRAP and MT1-MMP could not be detected.
Staining pattern of the SOC
Within the SOC, intense staining for TRAP was associated with large multinucleated chondroclasts that accumulated on the mixed spicules composed of calcified cartilage matrix and type I collagen (= osteoid) (Fig. 4A,C). In addition, MT1-MMP-positive cells that resemble chondroclasts were encountered in the same area (Fig. 4D, inset). Moreover, the later enzyme was associated with small, mononucleated cells that lined the osteoid layer like a string of pearls (Fig. 4B,D). Based on our TEM observations these cells were regarded as osteoblasts (compare with Fig. 2C). However, not all osteoblasts labelled for MT1-MMP. Macrophages were detected throughout the marrow cavity but they were never seen in close contact with the cartilage matrix. Double labelling (F4/80, TRAP) clearly revealed those macrophages and TRAP cells as two distinct cell types (Fig. 4C). Furthermore, double labelling for TRAP and MT1-MMP provided no clear evidence that co-localization of both enzymes occurred within the SOC (Fig. 4E).
Staining pattern of the primary ossification centre (POC)
The POC was formed prior to the establishment of the SOC. Our results demonstrated that in mice aged 5, 8 and 10 days the POC was well developed and a distinct seam of type I collagen surrounded the cartilage scaffold (Fig. 5A, inset). Intensive TRAP staining was noted on the scaffold surface and was also associated with the terminal row of the hypertrophic chondrocytes. The TRAP-positive cells were clearly multinucleated chondroclasts (Fig. 5A,C). Similar to the staining pattern within the SOC, MT1-MMP was noted in several osteoblasts but was also observed in chondroclasts that were located in the same position as the TRAP cells (Fig. 5B,D). However, we want to emphasize that the number of the MT1-MMP-positive multinucleated cells was decisively lower when compared with that of the TRAP cells. Double labelling (MT1-MMP, TRAP) showed that the majority of the chondroclasts expressed only TRAP; however, a few chondroclasts expressed both enzymes (Fig. 5E). Chondroclasts that labelled exclusively for MT1-MMP were not encountered when double labelling was carried out. Macrophages and TRAP-positive mononucleated cells were noted in the marrow cavity. In accordance with our observations, in the cartilage canals, as well as in the SOC, F4/80 and TRAP staining did not overlap (Fig. 5A,C inset). The staining pattern for TRAP, F4/80 and MT1-MMP was similar in all developmental stages.
Within the last few years, convincing studies have emerged implicating proteolytic activity as pivotal for a normal development of the skeleton and maintenance of bone (Holmbeck et al. 1999; 2005; Álvarez et al. 2005a; Holmbeck & Szabova, 2006; Malemud, 2006; Blumer et al. 2008 for reviews). Our present findings on the mouse femur are in line with these observations and demonstrate that enzymes such as TRAP and MT1-MMP are involved in the development of the SOC and POC, respectively. Furthermore, F4/80-positive macrophages appear to play a role in these processes.
Formation of the cartilage canals
Early generation of cartilage canals is a key feature during normal epiphyseal development and permits the formation of the SOC. In doing so, a machinery of several metalloproteinases (MMP-9 = gelatinase B, MMP-13 = collagenase-3, and MT1-MMP = MMP 14) cleave the components of the non-calcified cartilage matrix, clearing a path for vessels and bone-forming cells to invade the epiphysis (Vu et al. 1998; Holmbeck et al. 1999; Zhou et al. 2000; Lee et al. 2001; Davoli et al. 2001; Álvarez et al. 2005a; Melton et al. 2006; Blumer et al. 2007, 2008 for review). Surprisingly, MMP-9 as well as MMP-13 has been shown to affect canal formation only minimally, and the corresponding knock-out mice ultimately develop a skeleton of normal appearance (Vu et al. 1998; Stickens et al. 2004). However, mice lacking both enzymes have a dramatically delayed vascular recruitment and formation of the SOC, and finally exhibit shortened bones, indicating a synergy between MMP-9 and MMP-13 during cartilage matrix remodelling (Stickens et al. 2004). Notably, MT1-MMP deficiency has even stronger effects on the epiphyseal bone formation as it completely disrupts the formation of cartilage canals as well as the development of the SOC (Holmbeck et al. 1999, Zhou et al. 2000). Consequently, the MT1-MMP knock-out mice exhibit an abnormally structured growth plate, probably leading to its impaired function, reflected by a reduced longitudinal growth of the long bones (Holmbeck et al. 1999). Taken together, these findings provide clear evidence for an absolute requirement of MMPs during particular stages of epiphyseal development. Consistent with previous observations on the rat tibia (Álvarez et al. 2005a) our findings show that MT1-MMP is expressed in mice during early stages of canal formation (D 5, 8 and 10). In addition, mononucleated TRAP-positive cells are found at the same moment, and they are regarded as the precursors of the multinucleated chondroclasts (Roach, 2000). The exact role of the TRAP enzyme during the generation of the cartilage canals is still unclear and, at the moment, we cannot determine which components of the cartilage matrix the TRAP enzyme actually attacks, but it is remarkable that TRAP-deficient mice also exhibit abnormalities in the growth plate architecture and finally have a similar phenotypic appearance as MT1-MMP null mice and MMP-9/MMP-13 double knock-out mice (Hayman et al. 1996, 2000; Suter et al. 2001; Hollberg et al. 2002; Hayman & Cox, 2003; Roberts et al. 2007). Thus, it appears very probable that the formation of the cartilage canals and the development of epiphyseal bone are somehow altered in the TRAP knock-out mouse, and it is therefore the aim of our future studies to ascertain how the deletion of the TRAP enzyme affects these processes. However, here we draw the conclusion that TRAP is essential for normal epiphyseal development and longitudinal growth of the bones. Our immunohistochemical findings also demonstrate F4/80-positive macrophages within the cavity of the canals but noticeably, these cells are never encountered at the very apex of the canals. Macrophages do not label for TRAP and therefore they appear to represent a cell type of their own, in agreement with a recent study (Li et al. 2007). MT1-MMP-labelled cells are evenly scattered within the lumen of the canals and are also present at the tip. MT1-MMP staining does not overlap with TRAP labelling and we assume that these cells represent an additional cell type. Furthermore, our TEM observations demonstrate cells with numerous lysosomes within the lumen of the canals. These lysosomes contain fragments of fibrils occasionally cross-banded and thus resembling type I collagen. In our recent investigation (Blumer et al. 2007) we could show that the canal mesenchymal cells belong to the bone lineage and start to express type I collagen, the major bone-relevant protein 5 days after birth. In accordance with the latter study, the expression of this protein clearly increases during proceeding development. Collectively, the present data suggest that MT1-MMP and TRAP are instrumental for the lysis of the non-calcified components of the cartilage matrix, whereas the lysosomal cells resorb the decomposed remnants and presumably also portions of the newly synthesized type I collagen. In doing so, a suitable environment is created for further advancement of cartilage canals into the epiphysis, a requirement for a normal bone development. However, at the moment we cannot determine whether the lysosomal cells belong to the TRAP-, MT1-MMP- or F4/80-positive cells.
Formation of the SOC and POC
In both ossification centres the staining patterns for TRAP, MT1-MMP and F4/80 are quite similar. TRAP as well as MT1-MMP labelling is associated with multinucleated chondroclasts, but multinucleated TRAP cells are clearly more numerous. Double labelling (MT1-MMP, TRAP) occasionally results in an overlapping staining pattern that is, however, distinctly observable only within the POC. Regarding the function of both enzymes, it is well established that TRAP mediates the resorption of the calcified cartilage as well as the bone matrix (Roach, 2000; Hollberg et al. 2002; Hayman & Cox, 2003). MT1-MMP is required for cleavage of type I (found in bone) and type II collagen (found in hyaline cartilage), respectively (Holmbeck et al. 1999, 2005). Within the epiphysis and the diaphysis, newly formed type I collagen is laid down onto the calcified cartilage scaffold, and we assume that during the formation of the ossification centres both enzymes are needed for the disintegration of the existing cartilage matrix as well as for the modelling of the new bone matrix. Immunoreactivity for MT1-MMP has also been demonstrated in the osteoclasts of the mandible of a 3-week-old rat in vivo where it appears to be important for the migration of osteoclasts as well as the rinsing of the resorption lacunae (Irie et al. 2001). Unfortunately, the authors have not elucidated whether the expression of MT1-MMP applies for all of these cells. However, our findings based on double labelling (MT1-MMP, TRAP) suggest that two populations of chondroclasts exist during early endochondral ossification: one population that expresses only TRAP and an additional small subset of multinucleated cells that expresses both enzymes simultaneously.
In the processes of bone development osteoblasts mature to osteocytes that are finally buried within the mineralized bone matrix. However, not all osteoblasts transform into osteocytes and those osteoblasts that do not become embedded into bone may either switch to inactive, bone-lining cells or undergo cell death (Franz-Odendaal et al. 2006 for review). Osteoblasts express a number of bone relevant proteins, and Franz-Odendaal et al. (2006) have concluded that the decision as to which pathway is followed may depend on the genes which are up-regulated in osteoblasts. In the osteoblast-to-osteocyte ontogeny, the former cells start to generate cytoplasmic processes and their development and further maintenance in osteocytes depends on the proteolytic activity of MT1-MMP cleaving the pericellular type I collagen in the cortical bone. This leads to the establishment of a functional network of cell processes in which all bone cells remain connected. This network, however, does not evolve in MT1-MMP-deficient mice (Holmbeck et al. 2005). Consistent with the study of Filanti et al. (2000), our observations reveal that the enzyme is expressed in osteoblasts during endochondral bone formation. However, we could demonstrate that not all osteoblasts synthesize MT1-MMP and this expression pattern is noted at the very beginning (within the SOC) and during preceding endochondral ossification (within the POC). We therefore assume that MT1-MMP is required at the very early stages of osteocytogenesis to ensure the formation of the cytoplasmic processes and furthermore might ascertain the decision whether an osteoblast transforms into an osteocyte.
In summary, our results provide evidence that apart from MT1-MMP, TRAP-positive mononucleated cells and macrophages promote the formation of cartilage canals. Furthermore, the disintegration of the cartilaginous as well as bony matrix during endochondral bone formation appears to be triggered by TRAP-positive chondroclasts and likewise by a small subpopulation of this cell type expressing TRAP and MT1-MMP at the same moment. Our data further suggest that MT1-MMP is needed at the onset of osteocytogenesis and may be one of those critical genes that determine the fate of the osteoblasts.
The authors thank Prof. VMD. H. Dietrich for kindly providing the material for this study. We gratefully acknowledge the provision of anti type I collagen by Prof. PhD. L. Fisher (Bethesda, MD, USA). We specially thank E. Richter for assistance in the laboratory, Prof. PhD. K. Pfaller and T. Pérez for their valuable discussion and C. Siemon for carefully reading and correcting the manuscript.