- Email: [email protected]
Plasminogen activators are involved in the degradation of bone by osteoclasts
Bone 43 (2008) 915–920
Contents lists available at ScienceDirect
Bone j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / b o n e
Plasminogen activators are involved in the degradation of bone by osteoclasts Vincent Everts a,⁎, Evis Daci b, Wikky Tigchelaar-Gutter c, Kees A. Hoeben c, Sophie Torrekens b, Geert Carmeliet b, Wouter Beertsen d a Department of Oral Cell Biology, Academic Centre for Dentistry Amsterdam (ACTA), Universiteit van Amsterdam and Vrije Universiteit, Research Institute MOVE, Van der Boechorststraat 7, 1081 BT Amsterdam, The Netherlands b Laboratory of Experimental Medicine and Endocrinology, K.U. Leuven, Leuven, Belgium c Department of Cell Biology and Histology, Academic Medical Centre (AMC), Universiteit van Amsterdam, Meibergdreef 15, 1105 AZ Amsterdam, The Netherlands d Department of Periodontology, Academic Centre for Dentistry Amsterdam (ACTA), Universiteit van Amsterdam and Vrije Universiteit, Louwesweg 1, 1066 EA Amsterdam, The Netherlands
a r t i c l e
i n f o
Article history: Received 23 August 2007 Revised 24 June 2008 Accepted 1 July 2008 Available online 21 July 2008 Edited by: J. Aubin Keywords: Osteoclast Bone degradation Plasminogen activators Proteolytic enzymes Integrins
a b s t r a c t Osteoclastic bone degradation depends on the activity of several proteolytic enzymes, in particular to those belonging to the classes of cysteine proteinases and matrix metalloproteinases (MMPs). Yet, several findings suggest that the two types of plasminogen activators (PA), the tissue- and urokinase-type PA (tPA and uPA, respectively) are also involved in this process. To investigate the involvement of these enzymes in osteoclastmediated bone matrix digestion, we analyzed bone explants of mice that were deficient for both tPA and uPA and compared them to wild type mice. The number of osteoclasts as well as their ultrastructural appearance was similar for both genotypes. Next, calvarial and metatarsal bone explants were cultured for 6 or 24 h in the presence of selective inhibitors of cysteine proteinases or MMPs and the effect on osteoclast-mediated bone matrix degradation was assessed. Inhibition of the activity of cysteine proteinases in explants of control mice resulted in massive areas of non-digested demineralized bone matrix adjacent to the ruffled border of osteoclasts, an effect already maximal after 6 h. However, at that time point these demineralized areas were not observed in bone explants from uPA/tPA deficient mice. After prolonged culturing (24 h), a comparable amount of demineralized bone matrix adjacent to actively resorbing osteoclasts was observed in the two genotypes, suggesting that degradation was delayed in uPA/tPA deficient bones. The activity of cysteine proteinases as assessed in bone extracts, proved to be higher in extracts from uPA/tPA−/− bones. Immunolocalization of the integrin αvβ3 of in vitro generated osteoclasts demonstrated a more diffuse labeling of osteoclasts derived from uPA/tPA−/− mice. Taken together, our data indicate that the PAs play a hitherto unrecognized role in osteoclast-mediated bone digestion. The present findings suggest that the PAs are involved in the initial steps of bone degradation, probably by a proper integrin-dependent attachment to bone. © 2008 Elsevier Inc. All rights reserved.
Introduction Bone degradation by osteoclasts involves two major steps: (i) dissolution of mineral, and (ii) digestion of the matrix. Dissolution of mineral is accomplished by acidification of the extracellular space at sites where the osteoclast is firmly attached to the mineralized surface. This resorption area, the ruffled border area, is unique in that it is secluded by the osteoclast from the rest of the tissue that surrounds the cell. Because of the presence of several membrane-associated ion pumps, the osteoclast is able to lower the pH in the extracellular space. This acidic pH results in the dissolution of the mineral crystallites exposing the matrix constituents of the bone. The second major step is accomplished by a number of proteolytic enzymes secreted by the osteoclasts. Several findings suggest that this enzymatic digestion depends on consecutive activities of different enzymes. First, cysteine ⁎ Corresponding author. Fax: +31 20 444 8683. E-mail address: [email protected] (V. Everts). 8756-3282/$ – see front matter © 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.bone.2008.07.004
proteinases functioning optimally at an acid pH are active, followed by matrix metalloproteinases which are active at a more neutral pH [1]. The main cysteine proteinase active at the site of resorption is cathepsin K [2–7]. Recently, however, we have shown that in addition to cathepsin K also other, yet unknown, cysteine proteinases participate in bone matrix digestion [8]. Moreover the specific contribution of these different enzymes is depending on the bone site undergoing resorption: osteoclasts present in long bones employ mainly cysteine proteinases, whereas those active in the skull use besides these enzymes also matrix metalloproteinases [9]. Although it is generally accepted that the enzyme classes mentioned are the most important ones used by the osteoclast to resorb mineralized tissues, several findings strongly suggest that also other enzymes may participate. In this respect the serine proteinases plasminogen activators (PAs), urokinase-type and tissue-type PA, have been suggested as plausible candidates since Grills and coworkers showed the presence of both forms of plasminogen activator in the cytoplasm of osteoclasts [10]. Yet, it remains unclear whether these serine proteinases are actually involved in bone resorp-
916
V. Everts et al. / Bone 43 (2008) 915–920
tion. Some findings suggest that tissue-type PA (tPA) may stimulate resorption by activating the formation of osteoclasts [11]. Others, however, were not able to find any clue as to a possible function of the PAs. Daci et al. [12] and Tumber et al. [13] demonstrated a role of plasminogen activators in the osteoblast-mediated digestion of collagen and noncollagenous proteins. They proposed that one or both PAs play a role in cleaning the bone surface by osteoblasts prior to osteoclastic resorption. Plasmin generated from plasminogen by the action of the PAs has been shown to have a series of functions. This serine proteinase degrades non-collagenous proteins such as fibronectin, osteocalcin and laminin, but also the core protein of proteoglycans [14]. The enzyme is also considered to play a role in the release and activation of growth factors, e.g. TGFβ, during digestion of the bone matrix [15]. Finally, plasmin has been shown to activate enzymes among which several MMPs [16]. Yet, several analyses, using either selective inhibitors of serine proteinases or cells obtained from mice deficient for the PAs, seem to indicate that PAs do not participate in the sequence of events leading to bone matrix resorption by osteoclasts. Since these findings provide no explanation for the high levels of PAs in osteoclasts, we investigated their possible role in more detail using an ultrastructural approach we introduced some time ago [17]. In these studies we showed that interfering with the activity of cysteine proteinases and/or MMPs resulted in the occurrence of non-digested bone matrix adjacent to the osteoclasts. Under these conditions the osteoclasts demineralize bone but fail to digest the matrix. This approach was shown to be very sensitive in revealing the role of different enzymes in the osteoclastmediated digestion of bone matrix [1,9,17]. To elucidate the role of PAs in osteoclast-mediated bone matrix resorption we analyzed bone samples of mice lacking both tPA and uPA as described by Daci et al. [12]. Materials and methods Materials The general cysteine proteinase inhibitor E-64 was obtained from Sigma Chemical (St. Louis, MO). Medium 199 and fetal calf serum were from Gibco (Gibco Lab, Grand Island, NY). The MMP-inhibitor CT1166 was a kind gift from Dr. A.J. Docherty (R&D Celltech, Slough, UK) and has been described previously [18,19]. The cathepsin K substrate ZLeu-Arg-4-methoxy-β-naphtylamide was from Bachem AG (Bubendorf, Switzerland). Mice The uPA/tPA double knockout and WT mice (appropriate littermates were used) were bred in the animal housing facilities (Proefdierencentrum Leuven, Belgium) under conventional conditions and described in detail by Daci et al. [12]. Genotyping was performed by PCR of genomic DNA extracted from tail biopsies as described [12]. The experiments were approved by the ethical committee of the Katholieke Universiteit Leuven. Tissue culture Calvariae (frontal and parietal bones) and metatarsals of 5-day-old mice were aseptically removed and were cultured in 250 μl medium 199, containing 2.5% fetal calf serum. The bone explants were cultured in the presence of the broad-spectrum cysteine proteinase inhibitor E-64 (40 μM), and/or the matrix metalloproteinase inhibitor CT1166 (10 μM; [9]. The explants were cultured for 6 or 24 h and then processed for microscopic examination. Microscopy Calvarial and long bone explants were fixed in 1% glutaraldehyde and 4% formaldehyde in 0.1 M sodium cacodylate buffer (pH 7.4) and
without decalcification further processed for light and electron microscopic examination as described previously [17]. Semithin transverse sections (1–2 μm) were cut and stained with methylene blue. Ultrathin sections were cut with a diamond knife, contrasted with uranyl and lead and examined in a Zeiss EM-10 electron microscope. Morphometric analyses Non-cultured and cultured bone samples were morphometrically analyzed. These analyses included (i) the number of osteoclasts adjacent to the bone surface, and (ii) the surface area of the resorption pits as described previously [17]. In various studies we demonstrated that osteoclastic bone resorption is strongly inhibited by selective inhibitors of proteolytic enzymes [1,9,17,20–22]. In the presence of inhibitors of cysteine proteinases or matrix metalloproteinases osteoclasts are able to demineralize the bone but fail to digest the resulting demineralized bone matrix. Data were expressed as mean μm2 demineralized bone matrix (DBM) per osteoclast ± SD or SEM of five to seven explants. Cysteine proteinase activity Long bones (humeri) and calvariae were collected from wild type and the uPA/tPA deficient mice. The bones were cleaned of periosteal tissue, chopped into small pieces, sonicated (3 × 5 s) on ice, and overnight extracted in sodium acetate buffer (0.1 M, pH 5.3; 0.2% Triton X-100), at 4 °C, and again sonicated and centrifuged at 15,000 rpm (15 min). The supernatant was collected and frozen at −20 °C. Extracts were subjected to a fluorometric assay to determine cathepsin K activity. 10 μl aliquots of extracts were incubated in 50 μl phosphate buffer (100 mM, pH 6.0) containing 1 mg/ml of enzyme substrate (Z-Leu-Arg4-methoxy-β-naphtylamide), 1.3 mM EDTA, 1 mM dithiothreitol, and 2.67 mM L-cysteine. Incubation was performed at 37 °C, and readings were every 10 min. Substrate hydrolysis was determined using a multilabel counter (λex = 355 nm; λem = 430 nm). The extracts were analyzed for their protein content by using a BCA-kit and the analyses of enzyme activity were performed with samples equalized according to their protein level. In vitro osteoclast assays Bone marrow hematopoietic cells (BMC) were isolated from femora and tibiae of 4 to 6-week-old WT and uPA/tPA−/− mice and subjected to Ficoll-Paque (TM) plus (Stemcell Technologies, Grenoble, France) gradient purification. Cells at the gradient interface were collected and cultured overnight in α-MEM containing 10% heat-inactivated FCS, and supplemented with 10 ng/ml recombinant M-CSF (R&D Systems, Europe LTD, Abingdon, UK). After 24 h non-adherent cells were harvested and plated at 1.25 × 105 cells/cm2 in α-MEM containing 20 ng/ml M-CSF and 100 ng/ml RANKL (Peprotech EC, London, UK) (day 1). Medium was changed every two days and on day 6, the cells were stained for tartrate resistant acid phosphatase activity (TRACP). In order to assess resorption of bone tissue, osteoclasts were generated on dentin slices starting from BMC as described above. At days 8–10, cells were removed from the dentin slices with 0.25 M ammonium hydroxide and mechanical agitation. To visualize the resorption pits, dentin slices were stained with Mayer's hematoxylin. Immunostaining Osteoclasts, cultured on glass or on calcium-phosphate-coated discs (osteologic disc, BD Biosciences) for 6 days, were fixed with 4% formaldehyde/PBS and permeabilized with 0.1% Triton X-100 in PBS for 5 min at ambient temperature. After washing, cells were blocked with 5% BSA in PBS supplemented with 0.1% Tween and incubated
V. Everts et al. / Bone 43 (2008) 915–920
917
with anti-integrin β3 rabbit polyclonal antibody (ab47584; Abcam, Cambridge, UK) followed by incubation with goat anti-rabbit-488 antibody (Invitrogen, Carlsbad, CA). After washing with PBS, samples were mounted with fluorescent mounting medium containing TOPRO (Invitrogen). Statistical analysis Data were statistically analyzed using the Kruskal–Wallis nonparametric ANOVA test followed by Tukey–Kramer's multiple comparison tests. Effects were considered statistically significant when p b 0.05 (two-tailed). Results Morphological aspects Light and electron microscopic evaluation of bone samples from mice lacking uPA/tPA revealed, apart from a higher amount of mineralized matrix surrounding the osteocytes as described previously [23], no differences in the morphology as compared to bones from wild type mice. Also the number of osteoclasts per surface area as well as the morphology of these cells proved to be similar for both phenotypes. Effect of proteinase inhibitors In order to assess whether bone matrix degradation by osteoclasts was affected in the absence of uPA and tPA, calvarial bone explants were cultured in the presence of E-64, a cysteine proteinase inhibitor or CT1161, an inhibitor of matrix metalloproteinases. In previous studies we have shown that this procedure resulted in the occurrence
Fig. 2. Calvarial bone explants of wild type mice (WT,+/+) and uPA/tPA deficient mice (−/−) were cultured for 6 or 24 h in the presence of the cysteine proteinase inhibitor E64 (40 μM). (A–D) Electron micrographs of areas of demineralized bone matrix (DBM) adjacent to osteoclasts (oc) of explants cultured for 6 (A, B) or 24 h (C, D) in the presence of E-64. The double-sided arrows indicate the width of the demineralized areas. (A, C) explants of wild type mice; (B, D) explants of uPA/tPA−/− mice. Bo: bone. Bar: 5 μm. (E) Graph showing the volume density of demineralized bone matrix (DBM) next to the osteoclasts. The values are expressed as mean μm2 DBM per osteoclast (±SEM, n = 6). Note that already after 6 h the amount of DBM is maximal in the control explants.
Fig. 1. Electron micrographs of osteoclasts obtained from calvarial bone samples of wild type (A) and uPA/tPA deficient (B) mice. The explants were cultured for 24 h in the presence of the cysteine proteinase inhibitor E-64 (40 μM). Note the area of demineralized bone matrix (DBM) adjacent to the osteoclast (oc). Bo: bone. Bar: 5 μm.
of areas of demineralized non-digested bone matrix adjacent to the osteoclasts. The advantage of this approach is that quantification of the demineralized areas was proven to give a reliable measure of the bone resorbing activity of the osteoclast. In the presence of either proteinase inhibitor, WT osteoclasts were hampered in their ability to digest the bone matrix. Also in the absence of functional PAs, blockage of the activity of cysteine proteinases (or MMPs) resulted in areas of non-digested bone matrix adjacent to actively resorbing osteoclasts (Fig. 1). This effect in the PA deficient bones was, however, delayed compared to WT bones and became only apparent after prolonged culturing. Quantitative analysis
918
V. Everts et al. / Bone 43 (2008) 915–920
Fig. 3. Calvarial explants of uPA/tPA deficient mice and wild type mice were cultured for 6 h in the absence or presence of the cysteine proteinase inhibitor E-64 (40 μM; CPi) or the matrix metalloproteinase inhibitor CT1166 (10 μM; MMPi). The volume density of demineralized bone matrix (DBM) adjacent to the osteoclasts was analyzed and is expressed as mean μm2 DBM per osteoclast (OC) (±SEM, n = 6).
of the amount of non-digested bone matrix following a 6 h culture period in the presence of the cysteine proteinase inhibitor revealed a significant lower amount of demineralized non-digested bone matrix in uPA and tPA deficient (uPA/tPA−/−) bone explants (Fig. 2). However, this effect was no longer observed after 24 h of culturing (Fig. 2). Subsequently we incubated calvarial explants for 6 h with either the cysteine proteinase inhibitor or the MMP-inhibitor CT1166. In line with previous findings [1], the matrix metalloproteinase inhibitor did not result in an increased level of non-digested bone matrix adjacent to the osteoclast and inactivation of uPA/tPA did not change this effect (Fig. 3). Yet, consistent with the findings shown above, in the presence of the cysteine proteinase inhibitor, the uPA/tPA deficiency resulted in a seven-fold decreased amount of non-digested bone matrix compared to WT calvariae cultured with the cysteine proteinase inhibitor. In several studies we already showed that calvarial osteoclasts differed from long bone osteoclasts with respect to the enzymes used for the degradation of bone matrix [1,8,9,17]. Therefore we also analyzed long bone (metacarpal) explants. Nevertheless, comparable data were obtained using long bones, showing a lower amount of demineralized non-digested bone matrix adjacent to uPA/tPA−/− osteoclasts at the 6 h time point in the presence of the cysteine proteinase inhibitor compared to WT osteoclasts in the same setting (Fig. 4). Next we analyzed the amount of demineralized bone matrix devoid of osteoclasts. Previously we showed that osteoclasts move from site to site leaving behind areas of demineralized non-digested
Fig. 4. Metacarpal bone explants of uPA/tPA deficient mice and wild type mice were cultured for 6 h with the cysteine proteinase inhibitor E-64 (40 μM). The volume density of demineralized bone matrix (DBM) adjacent to the osteoclasts was analyzed and is expressed as mean μm2 DBM per osteoclast (OC) (±SEM, n = 6).
Fig. 5. The amount of demineralized bone matrix (DBM) not occupied by osteoclasts was analyzed. At the 6 h time point a significantly lower amount of DBM was found in the bone explants of the uPA/tPA deficient mice. This effect was not longer significant at the 24 h time point. Data are expressed as mean μm2 DBM without osteoclast (OC) per OC (±SEM, n = 6).
bone matrix [22]. These areas are subsequently cleaned by bone lining cells [22]. In the presence of the cysteine proteinase inhibitor, the uPA/ tPA−/− calvarial bone explants showed at the 6 h time point a significant lower amount of demineralized bone matrix not occupied by osteoclasts (Fig. 5). This effect was no longer apparent after a culture period of 24 h. Finally, we assessed the number of active osteoclasts characterized by a ruffled border in contact with the bone surface by counting the number of osteoclasts flanking demineralized versus mineralized bone matrix in explants cultured with the cysteine proteinase inhibitor. At the 6 h time point, most of the osteoclasts in bones from WT mice were associated with a demineralized area and only few osteoclasts lined the mineralized bone surface (Fig. 6). The reverse proved to be true for the explants obtained from uPA/tPA−/− mice where a significantly higher number of cells were associated with a normal, mineralized, bone surface. Again, the difference seen at the 6 h time point was no longer apparent after a 24 h culture period.
Fig. 6. Number of osteoclasts adjacent to a demineralized area of bone matrix (DBM: yes) or in contact with mineralized bone surface (DBM: no). Calvarial bone explants of uPA/tPA deficient mice (−/−) and wild type mice (+/+) were cultured for 6 or 24 h in the presence of the cysteine proteinase inhibitor E-64 (40 μM) and the number of osteoclasts was assessed in bone sections of similar length. Note that at the 6 h time point most wild type osteoclasts border a DBM whereas most osteoclasts of the deficient bones border mineralized bone. At the 24 h time point the numbers are similar for both phenotypes. Data are expressed as mean number of osteoclast per section (±SEM, n = 6).
V. Everts et al. / Bone 43 (2008) 915–920
919
Cysteine proteinase activity in bone extracts Extracts of bones were analyzed for their cysteine proteinase activity using a fluorescent substrate selective for this class of enzymes. Bone extracts obtained from uPA/tPA−/− mice proved to contain a significant higher level of activity (Fig. 7). Osteoclasts generated in vitro In order to analyze whether the above described delay in resorption was due to the osteoclast itself or to other cells present in the bone microenvironment in the vicinity of the osteoclast (e.g. bone lining cells), osteoclasts were generated in vitro. In addition to the resorbing activity of these cells we analyzed the localization of the integrin αvβ3. Osteoclasts were readily generated by using bone marrow cells of both genotypes. After 6 days of culturing in the presence of M-CSF and RANKL numerous TRACP positive dentin-resorbing osteoclasts were found. By comparing the two populations of osteoclasts a difference in size was noted. Osteoclasts generated from uPA/tPA−/− marrow were almost 5-times larger than those generated from wild type cells (WT: 1347 ± 1203 μm2; uPA/tPA−/− 6210 ± 6440 μm2). Such a difference in size, however, was not seen between osteoclasts present in the bone explants in which the cells were formed in vivo. Analysis of resorption by osteoclasts of the two genotypes generated on dentin slices and cultured either with or without the cysteine proteinase inhibitor E64, did not reveal any difference. Immunolocalization of β3 integrin Since several data indicate that the receptor for uPA, uPAR, may bind to the vitronectin receptor (αvβ3 integrin, 27,28) we analyzed the presence and localization of the β3 integrin by in vitro generated osteoclasts. This integrin subunit was highly expressed by osteoclasts generated from both genotypes (Fig. 8). However, the distribution of β3 integrin was markedly different between WT osteoclasts and those generated from uPA/tPA−/− mice. WT osteoclasts showed a belt-like β3 integrin staining, whereas cells of the deficient mice were characterized by a more diffuse labeling (Fig. 8). Assessment of the percentage of cells with this difference in localization revealed that 69% of the wild type cells had a clear belt-like localization, whereas this was 42% for the uPA/tPA−/− cells. Discussion The data presented in this study are the first to demonstrate a role for plasminogen activators in the process of osteoclastic bone resorp-
Fig. 7. Cysteine proteinase activity in extracts of calvarial bone explants obtained from uPA/tPA deficient mice and wild type mice. Activity was assessed in the extracts using a fluorescent substrate. Data are expressed as mean fluorescence (arbitrary units) ± SD, n = 5.
Fig. 8. Immunolocalization of β3-integrin subunit in osteoclasts generated in vitro. Bone marrow cells of both genotypes were cultured in the presence of M-CSF and RANKL and after 6 days the presence and localization of β3-integrin subunit was visualized with an anti-β3 antibody. Confocal scanning microscopy showed differences in the localization of the integrin subunit between wild type (A, C) and uPA/tPA−/− (B, D) mice: in the cells of the latter genotype the integrin is more diffusely localized. Cells were counterstained with DAPI to stain the nuclei. Note the difference in size of osteoclasts derived from wild type precursors (A) and those from deficient mice (B). Bar: A,B 60 μm; C,D 15 μm.
tion. Our findings clearly reveal that bone resorption by osteoclasts is delayed in the absence of these serine proteinases. This effect was especially detected when the activity of cysteine proteinases was blocked. Culturing wild type (control) bone explants in the presence of a cysteine proteinase inhibitor for 6 h resulted in a high level of nondigested bone matrix adjacent to the osteoclasts. This effect was not observed in the explants obtained from the uPA/tPA deficient mice indicating that due to the lack of plasminogen activators the dissolution of mineralized bone is delayed. How exactly serine proteinases are involved in this process is not known yet. Data presented in the literature indicate a role for these serine proteinases in at least four processes that are related to the bone resorption sequence: (i) migration of osteoclasts and/or their precursors to the resorption site [11,13,24], (ii) resorption of non-mineralized collagen protruding from the bone surface prior to osteoclast resorption [12,13,24], (iii) activation of enzymes like several MMPs [25], and (iv) degradation of noncollagenous proteins [12,26]. Our data do not seem to support the first possibility. If migration of mature osteoclasts was affected by the absence of the PAs, one would expect to observe an unhindered resorption by the osteoclasts attached to the mineralized bone surface. Yet, at the 6 h time point the demineralized area per osteoclast was significantly lower in the uPA/tPA deficient bones (see Fig. 2). Since migration is limited during this relatively short incubation period [22], the findings rather indicate that resorption itself is affected. Moreover, if migration was indeed disturbed the difference in bone resorption would also have been apparent at the 24 h time point. This proved not to be the case; the number of resorption pits was similar for both genotypes, thus indicating that migration along the bone surface does not depend on the activity of PAs. Are our findings explained by the observed role of the PAs in collagen degradation by osteoblasts/bone lining cells? This seems very unlikely since, as outlined in the previous paragraph, the effect we noted in the absence of the PAs was seen adjacent to osteoclasts already
920
V. Everts et al. / Bone 43 (2008) 915–920
present at that site and not depending anymore on the activity of osteoblasts/bone lining cells. Another possible explanation implies that the PAs are, by an unknown mechanism, involved in activating cysteine proteinases, comparable to their role in MMP activation. However, our data on the activity of cysteine proteinases in the bone extracts do not support this notion. On the contrary, a higher activity was found in the PA deficient bone extracts, suggesting rather a compensatory mechanism when the PAs are deficient. Finally, the PAs might be involved in the initial steps of the osteoclast-mediated bone resorption process. This possibility is supported by our findings showing that the increase in non-mineralized bone areas induced by the cysteine proteinase inhibitor was absent at an early time point in PA deficient bone explants and became only apparent at later time points. A similar delay was seen in a previous study in which we interfered with MMP activity [1]. These observations were explained by assuming a sequence of proteolytic activities: first digestion by cysteine proteinases at a low pH, followed by a slight increase in pH sufficient for MMPs to exert their activity. The exact function of the PAs during the initial step of osteoclastic bone degradation remains still to be elucidated, but the findings strongly suggest a PA-mediated effect at the onset of this process likely occurring prior to the acidification of the resorption site. Since the latter process of acidification only happens when the osteoclast is firmly attached to the bone surface, we propose that the PAs are involved in this attachment/sealing process. It remains to be elucidated whether for this process the enzymatic activity is needed or whether it depends on the binding of the PAs to their receptor. Since it has been shown that the receptor for uPA, uPAR, may interact with the vitronectin receptor (αvβ3 integrin [27,28], an integrin essential for osteoclast adhesion to the bone surface [29], the sealing process may be hampered in the absence of this PA. Our present findings appear to support this assumption since the osteoclasts of the deficient mice showed a more diffuse labeling pattern of the β3 integrin subunit. Collectively, our data suggest a role for the plasminogen activators in integrin-mediated attachment to the bone surface; we propose that in the absence of the PAs osteoclasts are hampered in their attachment resulting in a somewhat delayed resorption. Acknowledgments G. Carmeliet was funded by grant G.0229.04 of FWO (Fund for Scientific Research Flanders), Vlaanderen, Belgium. References [1] Everts V, Delaissé JM, Korper W, Beertsen W. Cysteine proteinases and matrix metalloproteinases play distinct roles in the subosteoclastic resorption zone. J Bone Miner Res 1998;13:1420–30. [2] Bromme D, Okamoto K, Wang BB, Biroc S. Human cathepsin O2, a matrix proteindegrading cysteine protease expressed in osteoclasts. Functional expression of human cathepsin O2 in Spodoptera frugiperda and characterization of the enzyme. J Biol Chem 1996;271:2126–32. [3] Delaissé JM, Andersen TL, Engsig MT, Henriksen K, Troen T, Blavier L. Matrix metalloproteinases (MMP) and cathepsin K contribute differently to osteoclastic activities. Microsc Res Tech 2003;61:504–13. [4] Furuyama N, Fujisawa Y. Distinct roles of cathepsin K and cathepsin L in osteoclastic bone resorption. Endocr Res 2000;26:189–204. [5] Godat E, Lecaille F, Desmazes C, Duchene S, Weidauer E, Saftig P, et al. Cathepsin K: a cysteine protease with unique kinin-degrading properties. Biochem J 2004; 383: 501–6.
[6] Gowen M, Lazner F, Dodds R, Kapadia R, Feild J, Tavaria M, et al. Cathepsin K knockout mice develop osteopetrosis due to a deficit in matrix degradation but not demineralization. J Bone Miner Res 1999;14:1654–63. [7] Saftig P, Hunziker E, Wehmeyer O, Jones S, Boyde A, Rommerskirch W, et al. Impaired osteoclastic bone resorption leads to osteopetrosis in cathepsin-Kdeficient mice. Proc Natl Acad Sci U S A 1998;95:13453–8. [8] Everts V, Korper W, Hoeben KA, Jansen ID, Bromme D, Cleutjens KB, et al. Osteoclastic bone degradation and the role of different cysteine proteinases and matrix metalloproteinases: differences between calvaria and long bone. J Bone Miner Res 2006;21:1399–408. [9] Everts V, Korper W, Jansen DC, Steinfort J, Lammerse I, Heera S, et al. Functional heterogeneity of osteoclasts: matrix metalloproteinases participate in osteoclastic resorption of calvarial bone but not in resorption of long bone. FASEB J 1999; 13:1219–30. [10] Grills BL, Gallagher JA, Allan EH, Yumita S, Martin TJ. Identification of plasminogen activator in osteoclasts. J Bone Miner Res 1990;5:499–505. [11] Hoekman K, Lowik CW, van de RM, Bijvoet OL, Verheijen JH, Papapoulos SE. The effect of tissue type plasminogen activator (tPA) on osteoclastic resorption in embryonic mouse long bone explants: a possible role for the growth factor domain of tPA. Bone Miner 1992;17:1–13. [12] Daci E, Udagawa N, Martin TJ, Bouillon R, Carmeliet G. The role of the plasminogen system in bone resorption in vitro. J Bone Miner Res 1999;14:946–52. [13] Tumber A, Papaioannou S, Breckon J, Meikle MC, Reynolds JJ, Hill PA. The effects of serine proteinase inhibitors on bone resorption in vitro. J Endocrinol 2003;178:437–47. [14] Webber MM, Waghray A. Urokinase-mediated extracellular matrix degradation by human prostatic carcinoma cells and its inhibition by retinoic acid. Clin Cancer Res 1995;1:755–61. [15] Jennings JC, Mohan S. Heterogeneity of latent transforming growth factor-beta isolated from bone matrix proteins. Endocrinology 1990;126:1014–21. [16] Murphy G, Ward R, Gavrilovic J, Atkinson S. Physiological mechanisms for metalloproteinase activation. Matrix Suppl 1992;1:224–30. [17] Everts V, Delaissé JM, Korper W, Niehof A, Vaes G, Beertsen W. Degradation of collagen in the bone-resorbing compartment underlying the osteoclast involves both cysteine-proteinases and matrix metalloproteinases. J Cell Physiol 1992;150:221–31. [18] Kerkvliet EH, Jansen ID, Schoenmaker TA, Docherty AJ, Beertsen W, Everts V. Low molecular weight inhibitors of matrix metalloproteinases can enhance the expression of matrix metalloproteinase-2 (gelatinase A) without inhibiting its activation. Cancer 2003;97:1582–8. [19] Creemers LB, Jansen ID, Docherty AJ, Reynolds JJ, Beertsen W, Everts V. Gelatinase A (MMP-2) and cysteine proteinases are essential for the degradation of collagen in soft connective tissue. Matrix Biol 1998;17:35–46. [20] Everts V, Beertsen W, Schroder R. Effects of the proteinase inhibitors leupeptin and E-64 on osteoclastic bone resorption. Calcif Tissue Int 1988;43:172–8. [21] Everts V, Korper W, Docherty AJ, Beertsen W. Matrix metalloproteinase inhibitors block osteoclastic resorption of calvarial bone but not the resorption of long bone. Ann N Y Acad Sci 1999;878:603–6. [22] Everts V, Delaissé JM, Korper W, Jansen DC, Tigchelaar-Gutter W, Saftig P, et al. The bone lining cell: its role in cleaning Howship's lacunae and initiating bone formation. J Bone Miner Res 2002;17:77–90. [23] Daci E, Everts V, Torrekens S, Van Herck E, Tigchelaar-Gutterr W, Bouillon R, et al. Increased bone formation in mice lacking plasminogen activators. J Bone Miner Res 2003;18:1167–76. [24] Leloup G, Lemoine P, Carmeliet P, Vaes G. Bone resorption and response to calcium-regulating hormones in the absence of tissue or urokinase plasminogen activator or of their type 1 inhibitor. J Bone Miner Res 1996;11: 1146–57. [25] Murphy G, Stanton H, Cowell S, Butler G, Knauper V, Atkinson S, et al. Mechanisms for pro matrix metalloproteinase activation. APMIS Acta Pahol Microbiol Immunol Scand 1999;107:38–44. [26] Dallas SL, Rosser JL, Mundy GR, Bonewald LF. Proteolysis of latent transforming growth factor-beta (TGF-beta)-binding protein-1 by osteoclasts. A cellular mechanism for release of TGF-beta from bone matrix. J Biol Chem 2002;277: 21352–60. [27] Wei Y, Waltz DA, Rao N, Drummond RJ, Rosenberg S, Chapman HA. Identification of the urokinase receptor as an adhesion receptor for vitronectin. J Biol Chem 1994;269:32380–8. [28] Nip J, Rabbani SA, Shibata HR, Brodt P. Coordinated expression of the vitronectin receptor and the urokinase-type plasminogen activator receptor in metastatic melanoma cells. J Clin Invest 1995;95:2096–103. [29] Helfrich MH, Nesbitt SA, Horton MA. Integrins on rat osteoclasts: characterization of two monoclonal antibodies (F4 and F11) to rat beta 3. J Bone Miner Res 1992;7:345–51.
Contents lists available at ScienceDirect
Bone j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / b o n e
Plasminogen activators are involved in the degradation of bone by osteoclasts Vincent Everts a,⁎, Evis Daci b, Wikky Tigchelaar-Gutter c, Kees A. Hoeben c, Sophie Torrekens b, Geert Carmeliet b, Wouter Beertsen d a Department of Oral Cell Biology, Academic Centre for Dentistry Amsterdam (ACTA), Universiteit van Amsterdam and Vrije Universiteit, Research Institute MOVE, Van der Boechorststraat 7, 1081 BT Amsterdam, The Netherlands b Laboratory of Experimental Medicine and Endocrinology, K.U. Leuven, Leuven, Belgium c Department of Cell Biology and Histology, Academic Medical Centre (AMC), Universiteit van Amsterdam, Meibergdreef 15, 1105 AZ Amsterdam, The Netherlands d Department of Periodontology, Academic Centre for Dentistry Amsterdam (ACTA), Universiteit van Amsterdam and Vrije Universiteit, Louwesweg 1, 1066 EA Amsterdam, The Netherlands
a r t i c l e
i n f o
Article history: Received 23 August 2007 Revised 24 June 2008 Accepted 1 July 2008 Available online 21 July 2008 Edited by: J. Aubin Keywords: Osteoclast Bone degradation Plasminogen activators Proteolytic enzymes Integrins
a b s t r a c t Osteoclastic bone degradation depends on the activity of several proteolytic enzymes, in particular to those belonging to the classes of cysteine proteinases and matrix metalloproteinases (MMPs). Yet, several findings suggest that the two types of plasminogen activators (PA), the tissue- and urokinase-type PA (tPA and uPA, respectively) are also involved in this process. To investigate the involvement of these enzymes in osteoclastmediated bone matrix digestion, we analyzed bone explants of mice that were deficient for both tPA and uPA and compared them to wild type mice. The number of osteoclasts as well as their ultrastructural appearance was similar for both genotypes. Next, calvarial and metatarsal bone explants were cultured for 6 or 24 h in the presence of selective inhibitors of cysteine proteinases or MMPs and the effect on osteoclast-mediated bone matrix degradation was assessed. Inhibition of the activity of cysteine proteinases in explants of control mice resulted in massive areas of non-digested demineralized bone matrix adjacent to the ruffled border of osteoclasts, an effect already maximal after 6 h. However, at that time point these demineralized areas were not observed in bone explants from uPA/tPA deficient mice. After prolonged culturing (24 h), a comparable amount of demineralized bone matrix adjacent to actively resorbing osteoclasts was observed in the two genotypes, suggesting that degradation was delayed in uPA/tPA deficient bones. The activity of cysteine proteinases as assessed in bone extracts, proved to be higher in extracts from uPA/tPA−/− bones. Immunolocalization of the integrin αvβ3 of in vitro generated osteoclasts demonstrated a more diffuse labeling of osteoclasts derived from uPA/tPA−/− mice. Taken together, our data indicate that the PAs play a hitherto unrecognized role in osteoclast-mediated bone digestion. The present findings suggest that the PAs are involved in the initial steps of bone degradation, probably by a proper integrin-dependent attachment to bone. © 2008 Elsevier Inc. All rights reserved.
Introduction Bone degradation by osteoclasts involves two major steps: (i) dissolution of mineral, and (ii) digestion of the matrix. Dissolution of mineral is accomplished by acidification of the extracellular space at sites where the osteoclast is firmly attached to the mineralized surface. This resorption area, the ruffled border area, is unique in that it is secluded by the osteoclast from the rest of the tissue that surrounds the cell. Because of the presence of several membrane-associated ion pumps, the osteoclast is able to lower the pH in the extracellular space. This acidic pH results in the dissolution of the mineral crystallites exposing the matrix constituents of the bone. The second major step is accomplished by a number of proteolytic enzymes secreted by the osteoclasts. Several findings suggest that this enzymatic digestion depends on consecutive activities of different enzymes. First, cysteine ⁎ Corresponding author. Fax: +31 20 444 8683. E-mail address: [email protected] (V. Everts). 8756-3282/$ – see front matter © 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.bone.2008.07.004
proteinases functioning optimally at an acid pH are active, followed by matrix metalloproteinases which are active at a more neutral pH [1]. The main cysteine proteinase active at the site of resorption is cathepsin K [2–7]. Recently, however, we have shown that in addition to cathepsin K also other, yet unknown, cysteine proteinases participate in bone matrix digestion [8]. Moreover the specific contribution of these different enzymes is depending on the bone site undergoing resorption: osteoclasts present in long bones employ mainly cysteine proteinases, whereas those active in the skull use besides these enzymes also matrix metalloproteinases [9]. Although it is generally accepted that the enzyme classes mentioned are the most important ones used by the osteoclast to resorb mineralized tissues, several findings strongly suggest that also other enzymes may participate. In this respect the serine proteinases plasminogen activators (PAs), urokinase-type and tissue-type PA, have been suggested as plausible candidates since Grills and coworkers showed the presence of both forms of plasminogen activator in the cytoplasm of osteoclasts [10]. Yet, it remains unclear whether these serine proteinases are actually involved in bone resorp-
916
V. Everts et al. / Bone 43 (2008) 915–920
tion. Some findings suggest that tissue-type PA (tPA) may stimulate resorption by activating the formation of osteoclasts [11]. Others, however, were not able to find any clue as to a possible function of the PAs. Daci et al. [12] and Tumber et al. [13] demonstrated a role of plasminogen activators in the osteoblast-mediated digestion of collagen and noncollagenous proteins. They proposed that one or both PAs play a role in cleaning the bone surface by osteoblasts prior to osteoclastic resorption. Plasmin generated from plasminogen by the action of the PAs has been shown to have a series of functions. This serine proteinase degrades non-collagenous proteins such as fibronectin, osteocalcin and laminin, but also the core protein of proteoglycans [14]. The enzyme is also considered to play a role in the release and activation of growth factors, e.g. TGFβ, during digestion of the bone matrix [15]. Finally, plasmin has been shown to activate enzymes among which several MMPs [16]. Yet, several analyses, using either selective inhibitors of serine proteinases or cells obtained from mice deficient for the PAs, seem to indicate that PAs do not participate in the sequence of events leading to bone matrix resorption by osteoclasts. Since these findings provide no explanation for the high levels of PAs in osteoclasts, we investigated their possible role in more detail using an ultrastructural approach we introduced some time ago [17]. In these studies we showed that interfering with the activity of cysteine proteinases and/or MMPs resulted in the occurrence of non-digested bone matrix adjacent to the osteoclasts. Under these conditions the osteoclasts demineralize bone but fail to digest the matrix. This approach was shown to be very sensitive in revealing the role of different enzymes in the osteoclastmediated digestion of bone matrix [1,9,17]. To elucidate the role of PAs in osteoclast-mediated bone matrix resorption we analyzed bone samples of mice lacking both tPA and uPA as described by Daci et al. [12]. Materials and methods Materials The general cysteine proteinase inhibitor E-64 was obtained from Sigma Chemical (St. Louis, MO). Medium 199 and fetal calf serum were from Gibco (Gibco Lab, Grand Island, NY). The MMP-inhibitor CT1166 was a kind gift from Dr. A.J. Docherty (R&D Celltech, Slough, UK) and has been described previously [18,19]. The cathepsin K substrate ZLeu-Arg-4-methoxy-β-naphtylamide was from Bachem AG (Bubendorf, Switzerland). Mice The uPA/tPA double knockout and WT mice (appropriate littermates were used) were bred in the animal housing facilities (Proefdierencentrum Leuven, Belgium) under conventional conditions and described in detail by Daci et al. [12]. Genotyping was performed by PCR of genomic DNA extracted from tail biopsies as described [12]. The experiments were approved by the ethical committee of the Katholieke Universiteit Leuven. Tissue culture Calvariae (frontal and parietal bones) and metatarsals of 5-day-old mice were aseptically removed and were cultured in 250 μl medium 199, containing 2.5% fetal calf serum. The bone explants were cultured in the presence of the broad-spectrum cysteine proteinase inhibitor E-64 (40 μM), and/or the matrix metalloproteinase inhibitor CT1166 (10 μM; [9]. The explants were cultured for 6 or 24 h and then processed for microscopic examination. Microscopy Calvarial and long bone explants were fixed in 1% glutaraldehyde and 4% formaldehyde in 0.1 M sodium cacodylate buffer (pH 7.4) and
without decalcification further processed for light and electron microscopic examination as described previously [17]. Semithin transverse sections (1–2 μm) were cut and stained with methylene blue. Ultrathin sections were cut with a diamond knife, contrasted with uranyl and lead and examined in a Zeiss EM-10 electron microscope. Morphometric analyses Non-cultured and cultured bone samples were morphometrically analyzed. These analyses included (i) the number of osteoclasts adjacent to the bone surface, and (ii) the surface area of the resorption pits as described previously [17]. In various studies we demonstrated that osteoclastic bone resorption is strongly inhibited by selective inhibitors of proteolytic enzymes [1,9,17,20–22]. In the presence of inhibitors of cysteine proteinases or matrix metalloproteinases osteoclasts are able to demineralize the bone but fail to digest the resulting demineralized bone matrix. Data were expressed as mean μm2 demineralized bone matrix (DBM) per osteoclast ± SD or SEM of five to seven explants. Cysteine proteinase activity Long bones (humeri) and calvariae were collected from wild type and the uPA/tPA deficient mice. The bones were cleaned of periosteal tissue, chopped into small pieces, sonicated (3 × 5 s) on ice, and overnight extracted in sodium acetate buffer (0.1 M, pH 5.3; 0.2% Triton X-100), at 4 °C, and again sonicated and centrifuged at 15,000 rpm (15 min). The supernatant was collected and frozen at −20 °C. Extracts were subjected to a fluorometric assay to determine cathepsin K activity. 10 μl aliquots of extracts were incubated in 50 μl phosphate buffer (100 mM, pH 6.0) containing 1 mg/ml of enzyme substrate (Z-Leu-Arg4-methoxy-β-naphtylamide), 1.3 mM EDTA, 1 mM dithiothreitol, and 2.67 mM L-cysteine. Incubation was performed at 37 °C, and readings were every 10 min. Substrate hydrolysis was determined using a multilabel counter (λex = 355 nm; λem = 430 nm). The extracts were analyzed for their protein content by using a BCA-kit and the analyses of enzyme activity were performed with samples equalized according to their protein level. In vitro osteoclast assays Bone marrow hematopoietic cells (BMC) were isolated from femora and tibiae of 4 to 6-week-old WT and uPA/tPA−/− mice and subjected to Ficoll-Paque (TM) plus (Stemcell Technologies, Grenoble, France) gradient purification. Cells at the gradient interface were collected and cultured overnight in α-MEM containing 10% heat-inactivated FCS, and supplemented with 10 ng/ml recombinant M-CSF (R&D Systems, Europe LTD, Abingdon, UK). After 24 h non-adherent cells were harvested and plated at 1.25 × 105 cells/cm2 in α-MEM containing 20 ng/ml M-CSF and 100 ng/ml RANKL (Peprotech EC, London, UK) (day 1). Medium was changed every two days and on day 6, the cells were stained for tartrate resistant acid phosphatase activity (TRACP). In order to assess resorption of bone tissue, osteoclasts were generated on dentin slices starting from BMC as described above. At days 8–10, cells were removed from the dentin slices with 0.25 M ammonium hydroxide and mechanical agitation. To visualize the resorption pits, dentin slices were stained with Mayer's hematoxylin. Immunostaining Osteoclasts, cultured on glass or on calcium-phosphate-coated discs (osteologic disc, BD Biosciences) for 6 days, were fixed with 4% formaldehyde/PBS and permeabilized with 0.1% Triton X-100 in PBS for 5 min at ambient temperature. After washing, cells were blocked with 5% BSA in PBS supplemented with 0.1% Tween and incubated
V. Everts et al. / Bone 43 (2008) 915–920
917
with anti-integrin β3 rabbit polyclonal antibody (ab47584; Abcam, Cambridge, UK) followed by incubation with goat anti-rabbit-488 antibody (Invitrogen, Carlsbad, CA). After washing with PBS, samples were mounted with fluorescent mounting medium containing TOPRO (Invitrogen). Statistical analysis Data were statistically analyzed using the Kruskal–Wallis nonparametric ANOVA test followed by Tukey–Kramer's multiple comparison tests. Effects were considered statistically significant when p b 0.05 (two-tailed). Results Morphological aspects Light and electron microscopic evaluation of bone samples from mice lacking uPA/tPA revealed, apart from a higher amount of mineralized matrix surrounding the osteocytes as described previously [23], no differences in the morphology as compared to bones from wild type mice. Also the number of osteoclasts per surface area as well as the morphology of these cells proved to be similar for both phenotypes. Effect of proteinase inhibitors In order to assess whether bone matrix degradation by osteoclasts was affected in the absence of uPA and tPA, calvarial bone explants were cultured in the presence of E-64, a cysteine proteinase inhibitor or CT1161, an inhibitor of matrix metalloproteinases. In previous studies we have shown that this procedure resulted in the occurrence
Fig. 2. Calvarial bone explants of wild type mice (WT,+/+) and uPA/tPA deficient mice (−/−) were cultured for 6 or 24 h in the presence of the cysteine proteinase inhibitor E64 (40 μM). (A–D) Electron micrographs of areas of demineralized bone matrix (DBM) adjacent to osteoclasts (oc) of explants cultured for 6 (A, B) or 24 h (C, D) in the presence of E-64. The double-sided arrows indicate the width of the demineralized areas. (A, C) explants of wild type mice; (B, D) explants of uPA/tPA−/− mice. Bo: bone. Bar: 5 μm. (E) Graph showing the volume density of demineralized bone matrix (DBM) next to the osteoclasts. The values are expressed as mean μm2 DBM per osteoclast (±SEM, n = 6). Note that already after 6 h the amount of DBM is maximal in the control explants.
Fig. 1. Electron micrographs of osteoclasts obtained from calvarial bone samples of wild type (A) and uPA/tPA deficient (B) mice. The explants were cultured for 24 h in the presence of the cysteine proteinase inhibitor E-64 (40 μM). Note the area of demineralized bone matrix (DBM) adjacent to the osteoclast (oc). Bo: bone. Bar: 5 μm.
of areas of demineralized non-digested bone matrix adjacent to the osteoclasts. The advantage of this approach is that quantification of the demineralized areas was proven to give a reliable measure of the bone resorbing activity of the osteoclast. In the presence of either proteinase inhibitor, WT osteoclasts were hampered in their ability to digest the bone matrix. Also in the absence of functional PAs, blockage of the activity of cysteine proteinases (or MMPs) resulted in areas of non-digested bone matrix adjacent to actively resorbing osteoclasts (Fig. 1). This effect in the PA deficient bones was, however, delayed compared to WT bones and became only apparent after prolonged culturing. Quantitative analysis
918
V. Everts et al. / Bone 43 (2008) 915–920
Fig. 3. Calvarial explants of uPA/tPA deficient mice and wild type mice were cultured for 6 h in the absence or presence of the cysteine proteinase inhibitor E-64 (40 μM; CPi) or the matrix metalloproteinase inhibitor CT1166 (10 μM; MMPi). The volume density of demineralized bone matrix (DBM) adjacent to the osteoclasts was analyzed and is expressed as mean μm2 DBM per osteoclast (OC) (±SEM, n = 6).
of the amount of non-digested bone matrix following a 6 h culture period in the presence of the cysteine proteinase inhibitor revealed a significant lower amount of demineralized non-digested bone matrix in uPA and tPA deficient (uPA/tPA−/−) bone explants (Fig. 2). However, this effect was no longer observed after 24 h of culturing (Fig. 2). Subsequently we incubated calvarial explants for 6 h with either the cysteine proteinase inhibitor or the MMP-inhibitor CT1166. In line with previous findings [1], the matrix metalloproteinase inhibitor did not result in an increased level of non-digested bone matrix adjacent to the osteoclast and inactivation of uPA/tPA did not change this effect (Fig. 3). Yet, consistent with the findings shown above, in the presence of the cysteine proteinase inhibitor, the uPA/tPA deficiency resulted in a seven-fold decreased amount of non-digested bone matrix compared to WT calvariae cultured with the cysteine proteinase inhibitor. In several studies we already showed that calvarial osteoclasts differed from long bone osteoclasts with respect to the enzymes used for the degradation of bone matrix [1,8,9,17]. Therefore we also analyzed long bone (metacarpal) explants. Nevertheless, comparable data were obtained using long bones, showing a lower amount of demineralized non-digested bone matrix adjacent to uPA/tPA−/− osteoclasts at the 6 h time point in the presence of the cysteine proteinase inhibitor compared to WT osteoclasts in the same setting (Fig. 4). Next we analyzed the amount of demineralized bone matrix devoid of osteoclasts. Previously we showed that osteoclasts move from site to site leaving behind areas of demineralized non-digested
Fig. 4. Metacarpal bone explants of uPA/tPA deficient mice and wild type mice were cultured for 6 h with the cysteine proteinase inhibitor E-64 (40 μM). The volume density of demineralized bone matrix (DBM) adjacent to the osteoclasts was analyzed and is expressed as mean μm2 DBM per osteoclast (OC) (±SEM, n = 6).
Fig. 5. The amount of demineralized bone matrix (DBM) not occupied by osteoclasts was analyzed. At the 6 h time point a significantly lower amount of DBM was found in the bone explants of the uPA/tPA deficient mice. This effect was not longer significant at the 24 h time point. Data are expressed as mean μm2 DBM without osteoclast (OC) per OC (±SEM, n = 6).
bone matrix [22]. These areas are subsequently cleaned by bone lining cells [22]. In the presence of the cysteine proteinase inhibitor, the uPA/ tPA−/− calvarial bone explants showed at the 6 h time point a significant lower amount of demineralized bone matrix not occupied by osteoclasts (Fig. 5). This effect was no longer apparent after a culture period of 24 h. Finally, we assessed the number of active osteoclasts characterized by a ruffled border in contact with the bone surface by counting the number of osteoclasts flanking demineralized versus mineralized bone matrix in explants cultured with the cysteine proteinase inhibitor. At the 6 h time point, most of the osteoclasts in bones from WT mice were associated with a demineralized area and only few osteoclasts lined the mineralized bone surface (Fig. 6). The reverse proved to be true for the explants obtained from uPA/tPA−/− mice where a significantly higher number of cells were associated with a normal, mineralized, bone surface. Again, the difference seen at the 6 h time point was no longer apparent after a 24 h culture period.
Fig. 6. Number of osteoclasts adjacent to a demineralized area of bone matrix (DBM: yes) or in contact with mineralized bone surface (DBM: no). Calvarial bone explants of uPA/tPA deficient mice (−/−) and wild type mice (+/+) were cultured for 6 or 24 h in the presence of the cysteine proteinase inhibitor E-64 (40 μM) and the number of osteoclasts was assessed in bone sections of similar length. Note that at the 6 h time point most wild type osteoclasts border a DBM whereas most osteoclasts of the deficient bones border mineralized bone. At the 24 h time point the numbers are similar for both phenotypes. Data are expressed as mean number of osteoclast per section (±SEM, n = 6).
V. Everts et al. / Bone 43 (2008) 915–920
919
Cysteine proteinase activity in bone extracts Extracts of bones were analyzed for their cysteine proteinase activity using a fluorescent substrate selective for this class of enzymes. Bone extracts obtained from uPA/tPA−/− mice proved to contain a significant higher level of activity (Fig. 7). Osteoclasts generated in vitro In order to analyze whether the above described delay in resorption was due to the osteoclast itself or to other cells present in the bone microenvironment in the vicinity of the osteoclast (e.g. bone lining cells), osteoclasts were generated in vitro. In addition to the resorbing activity of these cells we analyzed the localization of the integrin αvβ3. Osteoclasts were readily generated by using bone marrow cells of both genotypes. After 6 days of culturing in the presence of M-CSF and RANKL numerous TRACP positive dentin-resorbing osteoclasts were found. By comparing the two populations of osteoclasts a difference in size was noted. Osteoclasts generated from uPA/tPA−/− marrow were almost 5-times larger than those generated from wild type cells (WT: 1347 ± 1203 μm2; uPA/tPA−/− 6210 ± 6440 μm2). Such a difference in size, however, was not seen between osteoclasts present in the bone explants in which the cells were formed in vivo. Analysis of resorption by osteoclasts of the two genotypes generated on dentin slices and cultured either with or without the cysteine proteinase inhibitor E64, did not reveal any difference. Immunolocalization of β3 integrin Since several data indicate that the receptor for uPA, uPAR, may bind to the vitronectin receptor (αvβ3 integrin, 27,28) we analyzed the presence and localization of the β3 integrin by in vitro generated osteoclasts. This integrin subunit was highly expressed by osteoclasts generated from both genotypes (Fig. 8). However, the distribution of β3 integrin was markedly different between WT osteoclasts and those generated from uPA/tPA−/− mice. WT osteoclasts showed a belt-like β3 integrin staining, whereas cells of the deficient mice were characterized by a more diffuse labeling (Fig. 8). Assessment of the percentage of cells with this difference in localization revealed that 69% of the wild type cells had a clear belt-like localization, whereas this was 42% for the uPA/tPA−/− cells. Discussion The data presented in this study are the first to demonstrate a role for plasminogen activators in the process of osteoclastic bone resorp-
Fig. 7. Cysteine proteinase activity in extracts of calvarial bone explants obtained from uPA/tPA deficient mice and wild type mice. Activity was assessed in the extracts using a fluorescent substrate. Data are expressed as mean fluorescence (arbitrary units) ± SD, n = 5.
Fig. 8. Immunolocalization of β3-integrin subunit in osteoclasts generated in vitro. Bone marrow cells of both genotypes were cultured in the presence of M-CSF and RANKL and after 6 days the presence and localization of β3-integrin subunit was visualized with an anti-β3 antibody. Confocal scanning microscopy showed differences in the localization of the integrin subunit between wild type (A, C) and uPA/tPA−/− (B, D) mice: in the cells of the latter genotype the integrin is more diffusely localized. Cells were counterstained with DAPI to stain the nuclei. Note the difference in size of osteoclasts derived from wild type precursors (A) and those from deficient mice (B). Bar: A,B 60 μm; C,D 15 μm.
tion. Our findings clearly reveal that bone resorption by osteoclasts is delayed in the absence of these serine proteinases. This effect was especially detected when the activity of cysteine proteinases was blocked. Culturing wild type (control) bone explants in the presence of a cysteine proteinase inhibitor for 6 h resulted in a high level of nondigested bone matrix adjacent to the osteoclasts. This effect was not observed in the explants obtained from the uPA/tPA deficient mice indicating that due to the lack of plasminogen activators the dissolution of mineralized bone is delayed. How exactly serine proteinases are involved in this process is not known yet. Data presented in the literature indicate a role for these serine proteinases in at least four processes that are related to the bone resorption sequence: (i) migration of osteoclasts and/or their precursors to the resorption site [11,13,24], (ii) resorption of non-mineralized collagen protruding from the bone surface prior to osteoclast resorption [12,13,24], (iii) activation of enzymes like several MMPs [25], and (iv) degradation of noncollagenous proteins [12,26]. Our data do not seem to support the first possibility. If migration of mature osteoclasts was affected by the absence of the PAs, one would expect to observe an unhindered resorption by the osteoclasts attached to the mineralized bone surface. Yet, at the 6 h time point the demineralized area per osteoclast was significantly lower in the uPA/tPA deficient bones (see Fig. 2). Since migration is limited during this relatively short incubation period [22], the findings rather indicate that resorption itself is affected. Moreover, if migration was indeed disturbed the difference in bone resorption would also have been apparent at the 24 h time point. This proved not to be the case; the number of resorption pits was similar for both genotypes, thus indicating that migration along the bone surface does not depend on the activity of PAs. Are our findings explained by the observed role of the PAs in collagen degradation by osteoblasts/bone lining cells? This seems very unlikely since, as outlined in the previous paragraph, the effect we noted in the absence of the PAs was seen adjacent to osteoclasts already
920
V. Everts et al. / Bone 43 (2008) 915–920
present at that site and not depending anymore on the activity of osteoblasts/bone lining cells. Another possible explanation implies that the PAs are, by an unknown mechanism, involved in activating cysteine proteinases, comparable to their role in MMP activation. However, our data on the activity of cysteine proteinases in the bone extracts do not support this notion. On the contrary, a higher activity was found in the PA deficient bone extracts, suggesting rather a compensatory mechanism when the PAs are deficient. Finally, the PAs might be involved in the initial steps of the osteoclast-mediated bone resorption process. This possibility is supported by our findings showing that the increase in non-mineralized bone areas induced by the cysteine proteinase inhibitor was absent at an early time point in PA deficient bone explants and became only apparent at later time points. A similar delay was seen in a previous study in which we interfered with MMP activity [1]. These observations were explained by assuming a sequence of proteolytic activities: first digestion by cysteine proteinases at a low pH, followed by a slight increase in pH sufficient for MMPs to exert their activity. The exact function of the PAs during the initial step of osteoclastic bone degradation remains still to be elucidated, but the findings strongly suggest a PA-mediated effect at the onset of this process likely occurring prior to the acidification of the resorption site. Since the latter process of acidification only happens when the osteoclast is firmly attached to the bone surface, we propose that the PAs are involved in this attachment/sealing process. It remains to be elucidated whether for this process the enzymatic activity is needed or whether it depends on the binding of the PAs to their receptor. Since it has been shown that the receptor for uPA, uPAR, may interact with the vitronectin receptor (αvβ3 integrin [27,28], an integrin essential for osteoclast adhesion to the bone surface [29], the sealing process may be hampered in the absence of this PA. Our present findings appear to support this assumption since the osteoclasts of the deficient mice showed a more diffuse labeling pattern of the β3 integrin subunit. Collectively, our data suggest a role for the plasminogen activators in integrin-mediated attachment to the bone surface; we propose that in the absence of the PAs osteoclasts are hampered in their attachment resulting in a somewhat delayed resorption. Acknowledgments G. Carmeliet was funded by grant G.0229.04 of FWO (Fund for Scientific Research Flanders), Vlaanderen, Belgium. References [1] Everts V, Delaissé JM, Korper W, Beertsen W. Cysteine proteinases and matrix metalloproteinases play distinct roles in the subosteoclastic resorption zone. J Bone Miner Res 1998;13:1420–30. [2] Bromme D, Okamoto K, Wang BB, Biroc S. Human cathepsin O2, a matrix proteindegrading cysteine protease expressed in osteoclasts. Functional expression of human cathepsin O2 in Spodoptera frugiperda and characterization of the enzyme. J Biol Chem 1996;271:2126–32. [3] Delaissé JM, Andersen TL, Engsig MT, Henriksen K, Troen T, Blavier L. Matrix metalloproteinases (MMP) and cathepsin K contribute differently to osteoclastic activities. Microsc Res Tech 2003;61:504–13. [4] Furuyama N, Fujisawa Y. Distinct roles of cathepsin K and cathepsin L in osteoclastic bone resorption. Endocr Res 2000;26:189–204. [5] Godat E, Lecaille F, Desmazes C, Duchene S, Weidauer E, Saftig P, et al. Cathepsin K: a cysteine protease with unique kinin-degrading properties. Biochem J 2004; 383: 501–6.
[6] Gowen M, Lazner F, Dodds R, Kapadia R, Feild J, Tavaria M, et al. Cathepsin K knockout mice develop osteopetrosis due to a deficit in matrix degradation but not demineralization. J Bone Miner Res 1999;14:1654–63. [7] Saftig P, Hunziker E, Wehmeyer O, Jones S, Boyde A, Rommerskirch W, et al. Impaired osteoclastic bone resorption leads to osteopetrosis in cathepsin-Kdeficient mice. Proc Natl Acad Sci U S A 1998;95:13453–8. [8] Everts V, Korper W, Hoeben KA, Jansen ID, Bromme D, Cleutjens KB, et al. Osteoclastic bone degradation and the role of different cysteine proteinases and matrix metalloproteinases: differences between calvaria and long bone. J Bone Miner Res 2006;21:1399–408. [9] Everts V, Korper W, Jansen DC, Steinfort J, Lammerse I, Heera S, et al. Functional heterogeneity of osteoclasts: matrix metalloproteinases participate in osteoclastic resorption of calvarial bone but not in resorption of long bone. FASEB J 1999; 13:1219–30. [10] Grills BL, Gallagher JA, Allan EH, Yumita S, Martin TJ. Identification of plasminogen activator in osteoclasts. J Bone Miner Res 1990;5:499–505. [11] Hoekman K, Lowik CW, van de RM, Bijvoet OL, Verheijen JH, Papapoulos SE. The effect of tissue type plasminogen activator (tPA) on osteoclastic resorption in embryonic mouse long bone explants: a possible role for the growth factor domain of tPA. Bone Miner 1992;17:1–13. [12] Daci E, Udagawa N, Martin TJ, Bouillon R, Carmeliet G. The role of the plasminogen system in bone resorption in vitro. J Bone Miner Res 1999;14:946–52. [13] Tumber A, Papaioannou S, Breckon J, Meikle MC, Reynolds JJ, Hill PA. The effects of serine proteinase inhibitors on bone resorption in vitro. J Endocrinol 2003;178:437–47. [14] Webber MM, Waghray A. Urokinase-mediated extracellular matrix degradation by human prostatic carcinoma cells and its inhibition by retinoic acid. Clin Cancer Res 1995;1:755–61. [15] Jennings JC, Mohan S. Heterogeneity of latent transforming growth factor-beta isolated from bone matrix proteins. Endocrinology 1990;126:1014–21. [16] Murphy G, Ward R, Gavrilovic J, Atkinson S. Physiological mechanisms for metalloproteinase activation. Matrix Suppl 1992;1:224–30. [17] Everts V, Delaissé JM, Korper W, Niehof A, Vaes G, Beertsen W. Degradation of collagen in the bone-resorbing compartment underlying the osteoclast involves both cysteine-proteinases and matrix metalloproteinases. J Cell Physiol 1992;150:221–31. [18] Kerkvliet EH, Jansen ID, Schoenmaker TA, Docherty AJ, Beertsen W, Everts V. Low molecular weight inhibitors of matrix metalloproteinases can enhance the expression of matrix metalloproteinase-2 (gelatinase A) without inhibiting its activation. Cancer 2003;97:1582–8. [19] Creemers LB, Jansen ID, Docherty AJ, Reynolds JJ, Beertsen W, Everts V. Gelatinase A (MMP-2) and cysteine proteinases are essential for the degradation of collagen in soft connective tissue. Matrix Biol 1998;17:35–46. [20] Everts V, Beertsen W, Schroder R. Effects of the proteinase inhibitors leupeptin and E-64 on osteoclastic bone resorption. Calcif Tissue Int 1988;43:172–8. [21] Everts V, Korper W, Docherty AJ, Beertsen W. Matrix metalloproteinase inhibitors block osteoclastic resorption of calvarial bone but not the resorption of long bone. Ann N Y Acad Sci 1999;878:603–6. [22] Everts V, Delaissé JM, Korper W, Jansen DC, Tigchelaar-Gutter W, Saftig P, et al. The bone lining cell: its role in cleaning Howship's lacunae and initiating bone formation. J Bone Miner Res 2002;17:77–90. [23] Daci E, Everts V, Torrekens S, Van Herck E, Tigchelaar-Gutterr W, Bouillon R, et al. Increased bone formation in mice lacking plasminogen activators. J Bone Miner Res 2003;18:1167–76. [24] Leloup G, Lemoine P, Carmeliet P, Vaes G. Bone resorption and response to calcium-regulating hormones in the absence of tissue or urokinase plasminogen activator or of their type 1 inhibitor. J Bone Miner Res 1996;11: 1146–57. [25] Murphy G, Stanton H, Cowell S, Butler G, Knauper V, Atkinson S, et al. Mechanisms for pro matrix metalloproteinase activation. APMIS Acta Pahol Microbiol Immunol Scand 1999;107:38–44. [26] Dallas SL, Rosser JL, Mundy GR, Bonewald LF. Proteolysis of latent transforming growth factor-beta (TGF-beta)-binding protein-1 by osteoclasts. A cellular mechanism for release of TGF-beta from bone matrix. J Biol Chem 2002;277: 21352–60. [27] Wei Y, Waltz DA, Rao N, Drummond RJ, Rosenberg S, Chapman HA. Identification of the urokinase receptor as an adhesion receptor for vitronectin. J Biol Chem 1994;269:32380–8. [28] Nip J, Rabbani SA, Shibata HR, Brodt P. Coordinated expression of the vitronectin receptor and the urokinase-type plasminogen activator receptor in metastatic melanoma cells. J Clin Invest 1995;95:2096–103. [29] Helfrich MH, Nesbitt SA, Horton MA. Integrins on rat osteoclasts: characterization of two monoclonal antibodies (F4 and F11) to rat beta 3. J Bone Miner Res 1992;7:345–51.