The Roles of Annexins and Types II and X Collagen in Matrix Vesicle-mediated Mineralization of Growth Plate Cartilage*

Annexins II, V, and VI are major components of matrix vesicles (MV), i.e. particles that have the critical role of initiating the mineralization process in skeletal tissues. Furthermore, types II and X collagen are associated with MV, and these interactions mediated by annexin V stimulate Ca2+ uptake and mineralization of MV. However, the exact roles of annexin II, V, and VI and the interaction between annexin V and types II and X collagen in MV function and initiation of mineralization are not well understood. In this study, we demonstrate that annexin II, V, or VI mediate Ca2+ influx into phosphatidylserine (PS)-enriched liposomes, liposomes containing lipids extracted from authentic MV, and intact authentic MV. The annexin Ca2+ channel blocker, K-201, not only inhibited Ca2+ influx into fura-2-loaded PS-enriched liposomes mediated by annexin II, V, or VI, but also inhibited Ca2+ uptake by authentic MV. Types II and X collagen only bound to liposomes in the presence of annexin V but not in the presence of annexin II or VI. Binding of these collagens to annexin V stimulated its Ca2+ channel activities, leading to an increased Ca2+ influx into the liposomes. These findings indicate that the formation of annexin II, V, and VI Ca2+ channels in MV together with stimulation of annexin V channel activity by collagen (types II and X) binding can explain how MV are able to rapidly take up Ca2+ and initiate the formation of the first crystal phase.

Annexins II, V, and VI are major components of matrix vesicles (MV), 1 which are particles that, after being released from the plasma membrane of hypertrophic chondrocytes or osteoblasts, have the critical role of initiating the mineralization process in cartilage and perhaps in bone (3,5). Three independent lines of evidence indicate that annexin II, V, and VI exhibit distinct Ca 2ϩ ion channel properties. First, when inserted into artificial phosphatidylserine bilayers they form voltage-dependent Ca 2ϩ ion channels (6 -8). Second, the crystal structures of these annexins are largely ␣-helical with parallel barrels of ␣-helical domains forming a hydrophilic, charged pore through the center of the protein (6,8,9). Third, annexin II, V, and VI are able to mediate Ca 2ϩ influx into artificial liposomes (10;11). It was shown that annexin-mediated Ca 2ϩ influx into liposomes is rapid during the first 20 min and then reaches a plateau after 20 min (10,11).
The initial phase of MV-mediated mineralization is characterized by the uptake of mineral ions by these particles and the formation and growth of the first mineral phase inside the vesicles (5). Because MV are enclosed by a membrane, channel proteins are required to mediate the influx of mineral ions into these particles. Previous findings from our and other laboratories, showed that chymotrypsin treatment, which removes most of the annexins from MV, and zinc treatment, which inhibits annexin-mediated Ca 2ϩ influx into phosphatidylserine (PS)-enriched liposomes, diminished MV Ca 2ϩ uptake (12)(13)(14)(15), suggesting that annexins II, V, and VI serve as ion channels in MV, enabling Ca 2ϩ loading of the vesicles during the initial phase of mineralization.
Previous studies have revealed that collagen types II and X are associated with the outer surface of MV (16). We demonstrated that the selective removal of these collagens from the MV surface drastically reduced the ability of vesicles to take up Ca 2ϩ . The addition of purified type II or X collagen to these collagen-depleted MV restored their Ca 2ϩ uptake ability to levels similar to MV containing these collagens (12). Types II and X collagen were shown to bind directly to annexin V (17,18). Thus, it is possible that annexin II, V, and VI form Ca 2ϩ channels in MV. In addition, binding of types II and X collagen to annexin V would anchor MV to the extracellular matrix, and it might further activate the annexin V channel properties. This would explain the rapid influx of Ca 2ϩ into MV that is required for the formation and growth of the first crystal phase within the vesicles. To test this hypothesis, we isolated MV from growth plate cartilage and measured Ca 2ϩ uptake by these particles in the absence or presence of K-201 (JTV519), a specific annexin channel inhibitor (19,20). Furthermore, we tested whether annexin II, V, and VI are able to mediate Ca 2ϩ influx into MV and whether binding of type II or X collagen to annexin V is able to stimulate its Ca 2ϩ channel activity by measuring Ca 2ϩ influx into fura-2-loaded liposomes comprised of extracted MV lipids, or PS and phosphatidylethanolamine (PE) in a molar ratio of 9:1, in the absence or presence of annexin II, V, or VI, or complexes of annexin V and type II or X collagen.

EXPERIMENTAL PROCEDURES
Isolation of Matrix Vesicles-MV were isolated from growth plate cartilage of 6-to 8-week-old broiler strain chickens and from cultures of non-mineralizing hypertrophic and mineralizing post-hypertrophic chondrocytes as described previously (12) . To harvest MV, the partially digested tissue was vortexed and the suspension was subjected to differential centrifugation at 13,000 ϫ g for 20 min and 100,000 ϫ g for 1 h. Protein concentration of the vesicles was determined using the BCA assay (Pierce, Rockford, IL), and alkaline phosphatase activity was assayed as described previously (12).
Ca 2ϩ Uptake by MV Isolated from Tissue-Ca 2ϩ uptake was assayed by incubating MV aliquots (100 g of protein) in 1 ml of 45 Ca 2ϩ -labeled SCL at 37°C under mild shaking. After 24-h incubation, 100-l aliquots of the SCL incubations were sampled by microfiltration through Millipore HA filters (0.45-m pore size). After washing the filters twice, radioactivity associated with the filters was determined by scintillation counting. Ca 2ϩ uptake was measured in the presence of various concentrations of K-201 (JTV519). This compound was generously provided by Drs. N. Kaneko and T. Tanaka.
Ca 2ϩ Uptake by MV Isolated from Chondrocyte Cultures-After 24-h incubation of MV aliquots (100 g of protein) in 1 ml of SCL in the absence or presence of 200 nM of annexin II, V, or VI at 37°C, MV were pelleted, washed twice in 150 mM NaCl, 10 mM TES (pH 7.4), and 200 M EDTA (buffer 1), and resuspended in 1 ml of buffer 1 containing 1 M fura-2. MV suspensions were then incubated with 1% Triton X-100 to burst matrix vesicles and to release intraluminal Ca 2ϩ (blast method). Changes in the fluorescence ratio, 340:380 nm, were measured.
Isolation of Lipids from Authentic MV-To isolate lipids from authentic MV, lyophilized MV pellets isolated from growth plate cartilage were extracted three times with a chloroform/methanol mixture of 2:1. Insoluble material was removed by centrifugation at 15,000 ϫ g. The chloroform/methanol solution was evaporated and the lipid film was redissolved in chloroform.
Preparation of Liposomes-Phospholipids were obtained from Avanti Polar Lipids (Birmingham, AL). Fura-2 was obtained from Molecular Probes (Eugene, OR). Large thin-walled liposomes were prepared using a dehydration/rehydration method and were loaded with fura-2 as described previously (11,21). The liposomes contained either PS and PE in a molar ratio of 9:1 or lipids extracted from MV (see above). After the lipids were dried down from chloroform, the lipid film was rehydrated for 30 min in a stream of nitrogen saturated with water. The rehydrated phospholipids were overlaid with a solution containing 150 mM NaCl, 10 mM TES, pH 7.4, 200 M EDTA, and 100 M fura-2. After incubation at 37°C for 2 h, the suspension was washed four times and used for Ca 2ϩ influx measurements within the next 24 h. In addition, aliquots of the liposome suspensions were extruded through membranes with various pore sizes (20, 2, and 0.2 m) using the Extruder (Lipex Biomembranes).
Measurement of Ca 2ϩ Influx into Fura-2-loaded Liposomes-Ca 2ϩ influx into fura-2-loaded liposomes was measured in the absence or presence of annexin II, annexin V, or annexin VI, or complexes of annexin V/type II collagen, annexin V/type X collagen, annexin V/type IX collagen in a fluorescence cuvette with a 1-cm path length using a method previously described (11). Briefly, liposomes were resuspended in 150 mM NaCl, 10 mM TES, pH 7.4, and 200 M EDTA, and Ca 2ϩ was added to a final concentration of 400 M. The suspensions were prewarmed to 37°C. Ca 2ϩ influx was initiated by adding annexin II, V, or VI (200 nM) or complexes containing annexin V and collagen types II, X, or IX. In addition, Ca 2ϩ influx was measured in the presence of annexin II, V, or VI and various concentrations of K-201 (JTV519). The rate of Ca 2ϩ influx into fura-2-loaded liposomes was determined by measuring the fluorescence ratio at the two excitation wavelengths of 340 and 380 nm and the emission wavelength of fura-2 of 510 nm as a function of time. The excitation wavelength of the Ca 2ϩ -bound form of fura-2 was 340 nm, whereas the excitation wavelength of the Ca 2ϩ -free form of fura-2 was 380 nm.
Annexin V/Collagen/Liposome Binding Studies-To test the binding of type II or X collagen to liposomes in the absence or presence of annexin II, V, or VI, a 200 nM concentration of these annexins was first incubated with liposomes in the presence of 400 M Ca 2ϩ . Liposomes containing annexins or liposomes without annexins were then incubated with type II or X collagen (10 g) for 1 h at room temperature. Liposomes were quantitatively pelleted by centrifugation at 200,000 ϫ g for 15 min. The pellets were then washed twice and resuspended in 20 l of TES buffer. Aliquots of these suspensions (5 l) were dotted onto nitrocellulose membranes. After blocking with low-fat milk protein, the membranes were immunostained with antibodies specific for annexin II, annexin V, annexin VI, collagen type II, or collagen type X. The optical density of the color reaction was determined using a densitometer. To determine the amount of annexin II, V, or VI bound to liposomes, various defined concentrations of purified recombinant annexin II, V, or VI protein were applied onto the nitrocellulose membranes and immunostained with the appropriate primary antibodies. Measurement of the optical density of the color reaction for each annexin standard revealed a linear standard curve, which was used to calculate the amount of annexin II, V, or VI bound to liposomes.
Proteins and Antibodies-Recombinant annexin II, V, or VI were prepared using the pGEX expression vector (Pharmacia, Piscataway, NJ) as described previously (11). Full-length annexin II (22), annexin V (23), and annexin VI (24) cDNA clones were cloned into the pGEX expression vector. Recombinant annexin-glutathione S-transferase (GST) fusion protein was expressed in Escherichia coli DH5␣FЈ and purified. The recombinant annexin-GST fusion proteins were subjected to thrombin cleavage to release annexin molecules from the GST moiety. The reversible Ca 2ϩ -dependent binding of annexins to PS-containing liposomes was used as an affinity purification step to remove bacterial contaminants and thrombin (11). Types II, IX, and X collagen were isolated from chicken sternal and growth plate cartilage and purified as described previously (25,26). Aliquots were treated with pepsin as described previously (26). Denatured types II and X collagen were obtained after heat denaturation of native type II or X collagen at 100°C for 10 min and chilling the samples immediately on ice.
The preparation and specificity of antibodies specific for annexin II, V, and VI, type II collagen, and type X collagen were described previously (17,27,28).
SDS-Polyacrylamide Gel Electrophoresis and Immunoblotting-Samples were dissolved in 3% SDS sample buffer (29) with dithiothreitol, denatured at 100°C for 3 min, and analyzed by electrophoresis in 10% or 12% (w:v) SDS-polyacrylamide gels as described previously (26). Samples were electroblotted onto nitrocellulose filters after electrophoresis. After blocking with a solution of low fat milk protein, blotted proteins were immunostained with primary antibodies followed by peroxidase-conjugated secondary antibodies.
Electron Microscopy-MV were fixed in 2.5% glutaraldehyde in 0.1 M sodium cacodylate (pH 7.4) containing 2% tannin and postfixed in 1% osmium tetroxide. MV were dehydrated in a graded series of ethanol and embedded in epoxy resin. Ultrathin sections were contrasted with uranyl acetate and lead citrate. Specimens were examined in a transmission electron microscope (100CX II, JEOL, Peabody, MA) operated at 80 kV.

RESULTS
Ultrastructural analysis revealed that MV freshly isolated from growth plate cartilage of 6-to 8-week-old chickens were, in general, round to oval with diameters ranging from 100 to 250 nm (Fig. 1A). Immunoblot analysis demonstrated that these vesicle preparations contained annexin II, V, and VI and that types II and X collagen were associated with the vesicles (Fig. 1B). These findings are in agreement with previous studies (12,16,30). Furthermore, these vesicle fractions contained a high content of alkaline phosphatase activity (data not shown) and showed significant Ca 2ϩ uptake when incubated in SCL in vitro (Fig. 3).
To address the question of whether MV annexins mediate Ca 2ϩ influx into these particles, we first determined whether annexin II, V, or VI mediates Ca 2ϩ influx into fura-2-loaded PS-enriched liposomes. Annexin II, V, or VI was able to mediate Ca 2ϩ influx into liposomes containing PS and PE in a molar ratio of 9:1 (Table I and Fig. 2). These annexins not only mediated Ca 2ϩ influx into large liposomes but also into liposomes, which were extruded through various pore size membranes, including liposomes extruded through a 0.2-m pore size membrane (Table I). These unilamellar liposomes were similar in size to authentic matrix vesicles isolated from growth plate cartilage. There was a ϳ2-fold difference in the rate of annexin-mediated Ca 2ϩ influx between large-sized versus small-sized liposomes. It is likely that this difference reflects the amount of annexin molecules bound per unit amount of phospholipid surface area or curvature. Based on these findings, that the annexins not only mediate Ca 2ϩ influx into mixed-sized, multi-, and unilamellar liposomes but also into unilamellar liposomes with similar size as MV, and evidence presented below, that annexin II, V, or VI mediates Ca 2ϩ influx into intact MV, it is reasonable to assume that the mixed-sized, multi-, and unilamellar liposomes provide an accurate and simple model to study annexin Ca 2ϩ channel formation and activities in MV. Thus, mixed-sized, multi-, and unilamellar liposomes were used in the subsequent experiments.
The 1,4-benzothiazepine derivative K-201, a specific annexin Ca 2ϩ channel blocker (19,20,31), inhibited Ca 2ϩ influx into liposomes mediated by annexin II, V, or VI in a dose-dependent manner (Fig. 2). Buffer containing only K-201 had no effect on Ca 2ϩ flux across the membrane or leakage of fura-2 from the liposomes (Fig. 2). Interestingly, K-201 also inhibited Ca 2ϩ uptake by authentic MV isolated from growth plate cartilage in a dose-dependent manner (Fig. 3), revealing that annexin II, V, and VI mediate Ca 2ϩ influx into authentic MV.
To further test whether the lipid composition of the MV membrane allows annexins to form channels and mediate Ca 2ϩ influx, we extracted lipids from MV using chloroform/methanol in a ratio of 2:1. The extracted lipids were then reconstituted into liposomes and loaded with fura-2. Table II shows that annexin II, V, or VI mediated Ca 2ϩ influx into liposomes containing lipids extracted from MV. The rate of Ca 2ϩ influx mediated by these three annexins was similar (Table II). The rate of Ca 2ϩ influx mediated by annexin II, V, or VI into liposomes containing lipids extracted from authentic MV (340: 380 nm ratio of ϳ0.69, see Table II) was slightly higher than the rate of influx mediated by these annexins into liposomes containing PS and PE in a molar ratio of 9:1 (340:380 nm ratio of ϳ0.55, see Table I), suggesting that the MV membrane contains, besides PS, other lipids that might influence annexin binding and/or Ca 2ϩ channel formation.
We next determined whether exogenous annexin II, V, and VI are able to mediate Ca 2ϩ influx into intact vesicles isolated from non-mineralizing hypertrophic chondrocytes. These vesicles did not contain annexins II, V, and VI (4; see also Fig. 4A) and were not able to take up Ca 2ϩ and to mineralize (4) (Fig.  4B, bar b). In contrast, MV isolated from mineralizing posthypertrophic chondrocytes contained annexins II, V, and VI (Fig. 4A), and they took up significant amounts of Ca 2ϩ when incubated in SCL for 24 h (Fig. 4B, bar a). The addition of annexin II, V, or VI to MV isolated from non-mineralizing hypertrophic chondrocytes led to influx of Ca 2ϩ into these particles comparable to influx of Ca 2ϩ into annexin-containing MV isolated from mineralizing post-hypertrophic chondrocytes (Fig. 4B, bars c-e). Thus, annexin II, V, and VI were not only able to mediate Ca 2ϩ influx into liposomes containing PS and PE or lipids from authentic MV but also in intact authentic MV.
Previously, we and others have demonstrated that types II and X collagen bind to annexin V (16 -18). In addition, removal of types II and X collagen from MV led to a drastic decrease in Ca 2ϩ uptake by these particles (12). Thus, it is possible that binding of type II or X collagen to annexin V might stimulate its channel activities. To test this hypothesis, we first investigated whether type II or X collagen can bind to liposomes in the presence of annexin II, V, or VI. Liposomes containing PS and PE in a molar ratio of 9:1 were first incubated with 5 g of annexin II, V, or VI in the presence of Ca 2ϩ , to allow the annexins to bind to the liposomes. Then, these annexin-containing liposomes were incubated with type II or X collagen. After washing the liposomes, aliquots of the liposome fractions were dotted onto nitrocellulose membranes, which were immunostained with antibodies specific for the annexins, type II and X collagen. All three annexins (II, V, and VI) bound to the liposomes in presence of Ca 2ϩ in equal amounts. Approximately 0.7 g of annexin II, 0.7 g of annexin V, and 0.5 g of annexin VI bound to liposomes containing 300 g of total lipids. Type II or X collagen bound to liposomes that contained annexin V (Fig. 5, bars b), but not to liposomes that contained annexin II (Fig. 5, bars a) or annexin VI (Fig. 5, bars c). Type II FIG. 1. Ultrastructural and immunoblot analysis of MV. MV were isolated from growth plate cartilage of 6-day-to 8-week-old chickens as described under "Experimental Procedures." A, electron micrograph of MV. Bar, 300 nm. B, aliquots of the MV fractions were analyzed by SDS-polyacrylamide gel electrophoresis and transferred electrophoretically to nitrocellulose membranes. Membranes were immunostained with antibodies specific for annexin II (AnII), V (AnV), and VI (AnVI) and types II (␣1(II)) and X collagen (␣1(X)).

TABLE I Annexin II-, V-or VI-mediated Ca 2ϩ influx into liposomes containing PS and PE in a molar ratio of 9:1 and extruded through membranes with
various pore sizes Ca 2ϩ influx into fura-2-loaded liposomes in the presence of annexin II, V, or VI was measured over a 30-min time period as described under "Experimental Procedures." Liposomes containing PS and PE in a molar ratio of 9:1 were extruded through membranes with various pore sizes (a, no extrusion; b, 20 m; c, 2 m; d, 0.2 m). The data express the difference between the fluorescence ratios (340:380 nm) before and after addition of annexin. Data  or X collagen were not able to bind to liposomes in the absence of annexin V (Fig. 5, bars d).
Next, we determined whether binding of type II or X collagen to annexin V modulates its Ca 2ϩ channel activities. Ca 2ϩ influx into fura-2-loaded liposomes was measured in the presence of annexin V or complexes of annexin V and type II or X collagen. A significantly increased Ca 2ϩ influx into liposomes was measured in the presence of complexes containing annexin V and type II (Fig. 6, bar b) or X collagen (bar e) compared with annexin V alone (bar a). Pepsin-treated type II (bar c) or X collagen (bar f) was less effective in stimulating channel activities than the non-pepsin-treated collagens containing telopeptide and globular regions. Denatured type II (bar d) or X collagen (bar g) did not stimulate annexin V channel activities. Type IX collagen, which did not bind to annexin V (data not shown), was not able to stimulate its Ca 2ϩ channel activities (bar h). Type II or X collagen alone did not mediate significant Ca 2ϩ influx into liposomes (bars i and j).

Annexins II, V, and VI Mediate Ca 2ϩ Influx into MV-Pre-
vious studies, including findings from our laboratories, have clearly demonstrated that annexins II, V, and VI form Ca 2ϩ channels in planar lipid bilayers and artificial liposomes leading to Ca 2ϩ influx into these particles (6,8,10,11,32). In addition, a previous report showed the similarity in Ca 2ϩ channel activity of annexin V and MV in planar lipid bilayers (14). These findings suggest that annexins II, V, and VI form channels in MV and mediate rapid Ca 2ϩ influx into these particles (6 -8, 14). In this study we provide several lines of direct evidence for this possibility. First, annexin II, V, and VI medi- were incubated in 1 ml of SCL for 24 h at 37°C under shaking. Intraluminal Ca 2ϩ was determined using the "blast method" as described under "Experimental Procedures." Although MV isolated from mineralizing post-hypertrophic chondrocytes were able to take up Ca 2ϩ (a), MV isolated from non-mineralizing hypertrophic chondrocytes showed no significant Ca 2ϩ uptake (b). However, the addition of exogenous annexin II (c), V (d), or VI (e) to these non-mineralizing MV could restore their ability to take up Ca 2ϩ .  (19,20,31) inhibited annexin II-, V-, or VI-mediated Ca 2ϩ influx into these liposomes. Second, K-201 drastically decreased Ca 2ϩ uptake and mineralization of authentic MV isolated from growth plate cartilage. Third, the addition of annexin II, V, or VI to vesicles isolated from non-mineralizing hypertrophic chondrocytes, which do not contain these annexins and do not take up Ca 2ϩ naturally, led to an influx of Ca 2ϩ into these particles. These findings clearly indicate that all three MV annexins are able to form Ca 2ϩ channels in the membrane of MV and mediate Ca 2ϩ influx into these particles, a necessary step to facilitate the formation of the first crystal phase inside the vesicle lumen. K-201 (JTV519) binds to the central region of annexin V formed by domains II, III, and IV (20). The compound was shown to inhibit Ca 2ϩ channel activities of annexin V without affecting its Ca 2ϩ -dependent binding to membranes (31). In this study we show that K-201 not only inhibits annexin Vmediated influx into liposomes, but also Ca 2ϩ influx mediated by annexin II or VI. This finding confirms previous studies showing similarities in structure, Ca 2ϩ channel characteristics, and activities among these three annexins (6 -8, 10). More interesting is our finding that K-201 also drastically reduced Ca 2ϩ uptake and mineralization of authentic MV isolated from growth plate cartilage, clearly establishing the role of these three annexins in mediating Ca 2ϩ loading of MV.
MV-mediated mineralization of cartilage is restricted to a few cell layers close to the chondro-osseous border. Thus, cartilage mineralization is, and must be, under cellular control. Only post-hypertrophic chondrocytes release mineralizationcompetent MV, which contain annexin II, V, and VI and are able to initiate matrix mineralization. On the other hand, hypertrophic chondrocytes release vesicles that do not contain these annexins and cannot mineralize (4). However, the addition of exogenous annexin II, V, or VI to these non-mineralizing vesicles leads to Ca 2ϩ influx into these particles, demonstrating that channel activities of annexins II, V, and VI enable MV to take up Ca 2ϩ . Thus, because of the absence of annexins, vesicles from non-mineralizing chondrocytes are not able to take up Ca 2ϩ and to form the first intraluminal crystal phase. Most of the Ca 2ϩ inside mineralization-competent vesicles is in a bound form (initial crystal phase) (33), thus permitting perpetual Ca 2ϩ influx into the vesicles mediated by annexin II, V, and VI.
The annexins are cytosolic proteins that, in the presence of Ca 2ϩ , bind to acidic phospholipids. However, as we have recently demonstrated, binding of annexins to a membrane is not sufficient for channel formation, but a specialized lipid composition is required for annexin II, V, and VI to form Ca 2ϩ channels (11). As shown in this study, the lipids extracted from MV provide an environment in which these annexins are able to form channels. By definition, a channel must span the membrane to allow flux of ions through a membrane. Annexin II and V monomers, however, are too small to span the membrane lipid bilayer. A recent crystal structure analysis of an annexin XII hexamer has shown not only that the spatial dimension of the hexamer is sufficient to span the membrane but that it also has a central pore lined with charged residues (34). On the basis of this structure, the authors proposed that an annexin V hexamer would also have a central pore lined with negatively charged residues (34). We have previously demonstrated that annexin II and V form hexameric structures in PS-rich liposomes and MV (11,35). Thus, it is plausible that multimeric annexin units are needed to form functional ion channels. In addition, although each annexin by itself is capable of mediating Ca 2ϩ influx into liposomes, it is also possible that interactions among annexin II, V, and VI molecules further optimize the channel activities.
Binding of Type II and X Collagen to Annexin V Stimulates Its Ca 2ϩ Channel Activities-It has been previously shown that types II and X collagen are associated with MV and that these interactions are mediated by annexin V (16,18). In addition, FIG. 5. Binding of type II or X collagen to liposomes in the presence of annexin II, V, or VI. Liposomes containing PS and PE in a molar ratio of 9:1 were first incubated with annexin II, V, or VI in the presence of Ca 2ϩ followed by incubation with type II or X collagen. In addition, liposomes were incubated with type II or X collagen only. Aliquots of the liposome fractions were dotted onto nitrocellulose membranes and immunostained with antibodies specific for type II or X collagen. The intensity of the bands was analyzed by densitometry. The optical density obtained for staining of annexin V-containing liposomes incubated with type II or X collagen followed by antibodies for type II collagen or type X collagen was set as 1. Data were obtained from three different experiments, values are means Ϯ S.E. a, liposomes containing annexin II incubated with type II or X collagen; b, liposomes containing annexin V incubated with type II or X collagen; c, liposomes containing annexin VI incubated with type II or X collagen; d, liposomes incubated with type II or X collagen. Note that type II or X collagen bound to liposomes containing annexin V, whereas no significant binding was observed to liposomes containing annexin II or VI or to annexin-free liposomes.
FIG. 6. Ca 2؉ influx into fura-2-loaded liposomes in the presence of annexin V or complexes of annexin V and type II or X collagen. Ca 2ϩ influx into fura-2-loaded liposomes containing PS and PE in a molar ratio of 9:1 was measured in the presence of annexin V (a), or complexes of annexin V/native type II collagen (b), annexin V/pepsin-treated type II collagen (c), annexin V/denatured type II collagen (d), annexin V/native type X collagen (e), annexin V/pepsintreated type X collagen (f), annexin V/denatured type X collagen (g), annexin V/type IX collagen (h), native type II collagen (i), or native type X collagen (j). Data were obtained from five different experiments; values are means Ϯ S.E. Note the increased influx of Ca 2ϩ into liposomes in the presence of annexin V/native type II or X collagen complexes compared with annexin V alone. selective removal of surface-attached types II and X collagen from MV leads to a reduction in Ca 2ϩ uptake by these particles, rendering the possibility that binding of type II and X collagen activates the Ca 2ϩ channel activities of annexin V (12). In this study, we show direct evidence that binding of type II or X collagen to annexin V stimulates its Ca 2ϩ channel activities, leading to an increased Ca 2ϩ influx into liposomes. Pepsintreated type II or X collagen is less effective in stimulating annexin V channel activities, which is in agreement with our previous findings that pepsin-treated type II or X collagen is less effective in binding to annexin V and stimulating Ca 2ϩ uptake by MV (12,18). Thus, the telopeptide regions of type II collagen and the globular domains of type X collagen seem to be required for optimal binding and stimulation of annexin V channel activities.
As shown in this study, type II or X collagen stimulates annexin V-mediated Ca 2ϩ influx into liposomes by 2-to 3-fold. We have previously shown that the selective removal of types II and X collagen from MV reduced Ca 2ϩ uptake by these particles from 30 -35% to 5-7% of total 45 Ca 2ϩ in the incubation solution (12). These findings indicate that the interactions between types II and X collagen and annexin V play a crucial role for Ca 2ϩ loading and the formation of the first crystal phase inside MV and that the disruption of these interactions has severe consequences on mineralization of MV. In addition, these interactions may function to anchor MV to the extracellular matrix and might provide a bridge to facilitate the transfer of mineral from MV to extracellular matrix macromolecules.
Based on its restricted localization in hypertrophic and mineralizing growth plate cartilage, type X collagen is thought to play a role in the mineralization process. However, its exact functions in this process remain controversial. Two studies showed no association between type X collagen and focal calcification sites (36,37). Furthermore, previous findings revealed that the non-helical domain of type X collagen must be removed to facilitate cell-mediated mineralization of the egg shell membrane, suggesting that type X collagen functions as an inhibitor of mineralization (38). In contrast, based on the findings that type X collagen expression precedes mineral formation (39), and its capacity to bind Ca 2ϩ and MV (16,40), it was proposed that type X collagen might prepare the matrix for subsequent mineralization. In addition, calcified cartilage of transgenic mice, which express truncated type X collagen protein, contained less mineral, and the crystal phase was of poorer quality than the crystal phase found in calcifying cartilage of normal mice (41). In this study, we now demonstrate that binding of type X or II collagen to annexin V stimulates its channel activities, leading to an increased influx of Ca 2ϩ into liposomes. These findings are consistent with our previous studies showing that removal of type X collagen together with type II collagen inhibited Ca 2ϩ uptake by MV. Such inhibition can be reversed by restoring type X or II collagen to the MV surface (12). These findings together indicate that type X collagen has a stimulatory rather than inhibitory role during the initial phases of mineralization. How type X collagen affects the spreading and outgrowth of crystals from MV into the extracellular matrix remains to be established.
Conclusion-This study provides evidence that all three MV annexins (II, V, and VI) form Ca 2ϩ channels in these particles. In addition, once these vesicles are attached to the extracellular matrix via binding to types II and X collagen, the interaction between these collagens and annexin V further stimulates its Ca 2ϩ channel activities. These activated annexin V Ca 2ϩ channels plus the Ca 2ϩ channel activities of annexin II and VI enable MV to rapidly take up Ca 2ϩ as a likely initial step in the mineralization process.
MV are not only released from mineralizing post-hypertrophic chondrocytes, but also from osteoblasts and odontoblasts. We have isolated MV from fetal calvarial bone. These vesicle fractions also contained annexin II, V, and VI, and interestingly, types II and X collagen, 2 which is in agreement with our previous study showing the transient expression of types II and X collagen in fetal calvarial bone (42). In addition, we and others have shown the presence of mineralized areas, MV, and the expression of annexin II and V and types II and X collagen in the upper zone of osteoarthritic cartilage (43)(44)(45)(46). In that case, mineralization of articular cartilage is associated with this tissue's progressive destruction. Matrix metalloproteinases, such as collagenases, are activated during cartilage destruction in osteoarthritis (47). Thus, it would be interesting in future studies to determine how these metalloproteinases affect annexin V-mediated and types II and X collagen-stimulated Ca 2ϩ influx into MV. MV were not only found in skeletal tissues, but also in mineralizing arteriosclerotic plaques of arteries (48,49). As shown in this study, the MV annexins alone are sufficient to mediate Ca 2ϩ influx into MV. In tissues, where types II and X collagen are present, Ca 2ϩ uptake is greatly enhanced, probably leading to a more rapid initiation and progression of mineralization than in the absence of these collagens.
In conclusion, our study provides new and important insights into the regulatory mechanisms involved in the initiation of mineralization of tissues during normal development and under pathological conditions and suggests new strategies for the prevention of pathological mineralization.