INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

OSTEOBLASTS INITIATE MINERALIZATION of the pericellular matrix by promoting formation of hydroxyapatite crystals in the sheltered interior of extracellular membrane-limited vesicles (3, 4, 9, 10, 23). These structures, called matrix vesicles (MVs), are shed from specialized areas of the osteoblast plasma membrane (2, 3). MVs contain factors that regulate the composition of the extracellular matrix (3, 9, 10, 20). By an unknown mechanism, MVs are markedly enriched in tissue-nonspecific alkaline phosphatase (TNAP) relative to both whole cells and the plasma membrane (3).

Physiological bone matrix mineralization is hypothesized to be dependent on the availability of P_i released from a variety of substrates by certain MV ectoenzymes (3, 23). For example, ATP is hypothesized to drive the initiation of calcification by MVs in vivo, and a specific bone and cartilage ATPase appears to be responsible for the ATP-dependent calcium and P_i-depositing activity of bone and cartilage-derived MVs in vitro (23).

TNAP is the only tissue-nonrestricted isozyme of a family of four homologous human alkaline phosphatase (AP) genes (EC. 3.1.3.1) (31). Expressed as an ectoenzyme transported to the osteoblast plasma membrane and anchored via a phosphatidylinositol glycan moiety, TNAP has been demonstrated to play an essential physiological role during osteoblastic bone matrix mineralization (31, 44, 46). Specifically, defective bone mineralization (osteomalacia) occurs in TNAP deficiency (hypophosphatasia) (44). The severity of hypophosphatasia is variable and modulated by the nature of the TNAP mutation (18, 21, 31, 40, 44, 46).

Skeletal TNAP can catalyze P_i release from ATP (23), and TNAP catalyzes several transphosphorylation reactions (44). Although TNAP does not appear to dephosphorylate membrane proteins (17), TNAP has been hypothesized to modulate bridging of MVs to matrix collagen (44). TNAP has been demonstrated to bind calcium (14). Moreover, TNAP degrades at least three phosphocompounds [phosphoethanolamine, pyridoxal 5'-phosphate, and inorganic pyrophosphate (PP_i)] that accumulate endogenously in hypophosphatasia (45).

The central function(s) of TNAP in conditioning mineralization has not been completely defined (31, 44). Importantly, aberrant localization of TNAP can occur, including defective transport of TNAP to the plasma membrane associated with hypophosphatasia (16, 18). This has suggested that TNAP might act at the level of plasma membrane-derived structures such as MVs. In subjects with perinatal hypophosphatasia, the most severe form of the disease, MVs were present in approximately normal numbers and distribution, and these MVs contained internal hydroxyapatite crystals (4). However, propagation of hydroxyapatite crystals outside of isolated MVs was impaired in perinatal hypophosphatasia, but by an undefined mechanism (4).

The ability of TNAP to hydrolyze PP_i to P_i (45) has been hypothesized to be central to the ability of TNAP to promote osteoblastic mineralization (4, 44). A major action of PP_i is to suppress both the deposition and propagation of hydroxyapatite crystals in vitro (27, 30). Thus critically timed removal or exclusion of PP_i at sites of mineralization appears to be necessary for active crystal deposition to proceed (27, 30).

PP_i is generated by multiple biochemical reactions (38), including hydrolysis of the phosphodiester I bond in purine and pyrimidine nucleoside triphosphates (nucleoside triphosphate pyrophosphohydrolase activity) by ectoenzymes of the phosphodiesterase I nucleotide pyrophosphatase (PDNP/NTPPPH) family (EC 3.6.1.8, EC 3.1.4.1) (19). Significantly, cultured fibroblasts from subjects with infantile hypophosphatasia retain an apparently normal capacity to generate extracellular PP_i through NTPPPH activity (13). It also should be noted that osteoblasts have particularly high NTPPPH-specific activity (12, 26, 33, 34).

Plasma cell membrane glycoprotein-1 (PC-1) is a NTPPPH isozyme expressed by cultured osteoblastic cells (19, 26, 41). Furthermore, similar to the case for TNAP and other ectoenzymes, PC-1 expression and the extent of PC-1 distribution to MVs are regulated by certain growth factors and calciotropic hormones, including transforming growth factor- , basic fibroblast growth factor, and 1,25 dihydroxyvitamin D₃ (7, 26, 33, 34, 41). Osteoblast-derived MV PC-1 appears to function directly to increase MV fraction PP_i content and to restrain mineralization by isolated MVs in vitro (26). In this regard, a two- to fourfold increase in osteoblast PC-1 expression decreases, by >80%, the amount of hydroxyapatite deposited in the pericellular matrix of osteoblasts in vitro (26). Moreover, the effects of PC-1 on mineralization have been demonstrated to be physiologically significant in ttw/ttw mice, which are homozygous for a naturally occurring PC-1 truncation mutation and develop osteoblast-mediated and chondrocyte-mediated hyperostosis and articular cartilage calcification in early life (32).

PC-1 does not account for all the NTPPPH activity of osteoblastic cells (41). Moreover, several tissues that actively synthesize a collagenous extracellular matrix, including articular chondrocytes and hepatocytes, express not only PC-1 but also the closely related isozyme B10 (also called PDNP3) (19, 27).

In this study, we investigated and compared the potential interactions, in osteoblastic cells and MV fractions derived from these cells, among TNAP, PC-1, and B10/PDNP3 (principally referred to as B10 in the text). To do so, we employed the clonal immortalized murine calvarial osteoblast line MC3T3-El (abbreviated here as MC3T3) (26) and also studied primary calvarial osteoblasts from TNAP-knockout (ko or /) mice (31), which serve as a model of infantile hypophosphatasia.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Reagents

Mouse Genotyping and Isolation and Culture of Primary Calvarial Osteoblasts

Primary cultures of osteoblasts were isolated from calvariae of 1- to 4-day-old pups that were hybrids of C57Bl/6 × 129/J mouse strains with wild-type, heterozygote, and homozygote TNAP-null genotypes. Isolation was performed by a slightly modified version of the time sequential collagenase technique described by Boonekamp et al. (8). In brief, calvariae of the same genotype were pooled and three 10-min incubations in 4 mM EDTA, 137 mM NaCl, 2.7 mM KCl, 3 mM NaH₂PO₄, pH 7.2, were performed, followed by seven 10-min digestions in EDTA-free buffer, containing 180 U/ml collagenase type II (Worthington Biochemical, Lakewood, NJ). All isolations were performed at 37°C in a shaking water bath. The last five collagenase digestions, containing an enriched cell population of osteoblastic phenotype, were pooled and seeded at 4 × 10⁴ cells/cm² in MEM- (GIBCO-BRL, Grand Island, NY), containing 10% heat-inactivated FCS, penicillin (50 U/ml) and streptomycin (0.5 mg/ml).

Culture of Osteoblastic Cells Under Mineralizing Conditions

Immunofluoresence/Confocal Microscopy

To detect AP activity, fixed cells were stained with a 1:1 solution of 0.2 mg/ml naphthol AS-MX and 1.2 mg/ml fast red TR salt dissolved in 0.2 M Tris · HCl for 30 min in the dark at 25°C. This was followed by blocking (in PBS with 2% goat serum and 0.02% thiomersal) for 1 h. Where indicated, cells were permeablized by using 0.1% Triton X-100 in the blocking buffer for 15 min, followed by continued incubation in blocking buffer for another 45 min.

To detect PC-1 and B10, the murine anti-PC-1 alloantibody IR518 (42), or a rabbit antibody to the COOH terminus (amino acids 580-875) of rat B10 (5), was diluted in blocking buffer and added to the cells for 18 h at 4°C, washed three times in PBS, and incubated with Alexa 488-conjugated goat anti-mouse IgG (Molecular Probes, Eugene, OR) or goat anti-rabbit FITC (Sigma) at 1:400 in blocking buffer for 1 h at 22°C. Coverslips were mounted with Slowfade media (Molecular Probes), and cells were studied by using a Zeiss Axiovert 100M laser scanning microscope using the FITC channel to detect PC-1 or B10 staining, and the Texas red channel to detect AP staining.

Western Blotting, Densitometry, and Immunoprecipitation

NTPPPH-specific, polyclonal antibodies to PC-1 (R1769) and B10 (26, 27) and rabbit antibody to tubulin (Sigma) served as primary antibodies in Western blotting, performed as described (26, 27). Washed nitrocellulose membranes were incubated with horseradish peroxidase-conjugated secondary antibody in blocking buffer for 1 h and washed again, and immunoreactive products were detected by using the enhanced chemiluminesence system (Amersham, Arlington Heights, IL). Where indicated, semiquantitative analyses (of Western blots) were performed, using a previously described densitometry protocol (26).

To specifically immunoprecipitate TNAP from osteoblast cell lysates, we used a previously described method (41). We used a rabbit polyclonal antibody against rat TNAP that cross-reacts with mouse TNAP (22) (a gift from Prof. Yukio Ikehara, Fukuoka, Japan).

Transfections of PC-1, B10 and TNAP

MV Fraction Isolation and ⁴⁵Ca Precipitation Assay: Electron Microscopy of MV Fractions

MV fractions (0.04 mg protein in 0.025 ml) were added in triplicate to 0.5 ml of "calcifying medium" [(in mM) 2.2 CaCl₂ (1 µCi/ml ⁴⁵Ca), 1.6 KH₂PO₄, 1 MgCl₂, 85 NaCl, 15 KCl, 10 NaHCO₃, 50 mM N Tris, and, where indicated, 1 ATP disodium salt, and/or 0.1% Triton X-100, pH 7.6] and vortexed and incubated at 37°C for 24 h (26). Samples were then centrifuged at 14,000 g for 10 min at 4°C. The pellet was washed twice with cold calcifying medium without ATP. The ⁴⁵Ca in the mineral phase was solubilized in HCl and counted in 5 ml scintillation fluid.

The MV fraction of MC3T3 cells at 10 days in culture was assessed by conventional EM, using previously described methods (11). For specimen processing, we carried out double fixation (using 2% glutaraldehyde in cacodylate buffer, followed by 1% OsO₄ in H₂O), which was followed by dehydration in 2% uranyl acetate in 70% ethanol, and then in 100% ethanol, which was then followed by infiltration using acetonitrile/Epon (11).

RT-PCR Analysis

PP_i, Enzyme, and Cellular DNA Assays

To measure transglutaminase activity, we added 30 µg of protein in 0.08 ml of 5 mM Tris · HCl (pH 7.5), 0.25 M sucrose, 0.2 mM MgSO₄, 2 mM DTT, 0.4 mM phenylmethylsulfonyl fluoride, and 0.4% Triton X-100 to an equal volume of 300 mM Tris · HCl, pH 7.5, 10 mM CaCl₂, 20 mM DTT, 10 mg/ml N,N' dimethylcasein and 10 µCi/ml [³H]putrescine (15) for 30 min at 37°C. The reaction was stopped by adding cold 10% TCA and 0.5% tannic acid. The precipitate was washed with 10% TCA and 0.25% tannic acid. The pellet was resuspended in 0.1 ml 1 M NaOH. Samples were then treated with 2% H₂O₂ for 4 h at 22°C, followed by 4 h of treatment at 40°C, and then counted for 5 min in 5 ml of scintillation fluid. Transglutaminase activity was designated as nanomoles of putrescine incorporated into casein (per mg protein in each sample).

Statistical Analysis

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Impaired ATP-Dependent Calcium Precipitation by Detergent-Treated Osteoblast-Derived MV Fractions From TNAP-Deficient Mice

View larger version (199K): [in this window] [in a new window]   Fig. 1.   Morphology of matrix vesicles (MVs) isolated from mineralizing osteoblastic cells. The MV fractions from the conditioned media of mineralizing MC3T3 cells were collected at 10 days in culture, as described in the MATERIALS AND METHODS. After double fixation, dehydration, and infiltration with acetonitrile/Epon, the MVs were studied by conventional electron microscopy as described in MATERIALS AND METHODS. Representative membrane-limited MVs from 4 separate studies are shown. Nearly all the MVs had intact membranes. Many MVs contained electron-dense internal deposits. The size range of the osteoblast-derived MVs was mainly 100-200 nm in diameter. Bar = 100 nm.

In the absence of detergent treatment, TNAP-deficient osteoblasts and control osteoblasts released MV fractions that were comparably able to precipitate calcium in an ATP-dependent manner (Fig. 2). However, treatment of the MV fractions with Triton X-100 unmasked a significant defect in the ATP-dependent calcium-precipitating ability of TNAP/ osteoblast-derived MV fractions relative to TNAP+/+ MV fractions (Fig. 2). The defect directly correlated with the extent of TNAP deficiency, because it was greater in MV fractions derived from TNAP/ mice than from TNAP+/ mice (Fig. 2). The defect was exposed by using ATP as a substrate and was active at the level of the MV fractions. Because NTPPPH activity hydrolyzes ATP and is associated with MVs, we assessed NTPPPH expression in osteoblasts and osteoblast-derived MV fractions.

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Impaired ATP-dependent calcium precipitation by detergent-treated MV fractions from tissue-nonspecific alkaline phosphatase (TNAP)-deficient osteoblasts. MV fractions were collected from pooled conditioned media at the time periods in culture indicated, from primary calvarial osteoblasts of different (n = 12 each) wild-type (+/+), TNAP heterozygous (+/), and homozygous knockout (/) mice, cultured as described in MATERIALS AND METHODS in medium supplemented with -glycerophosphate and ascorbate. MV fractions (0.04 mg protein in 0.025 ml Hanks' balanced salt solution) were added to 0.5 ml of the calcifying medium described in MATERIALS AND METHODS, with or without the presence of 1 mM ATP, and/or 0.1% Triton X-100, at pH 7.6, vortexed, and incubated at 37°C for 24 h. After centrifugation, the pellet was washed and counted for 45Ca in the mineral phase, as indicated in MATERIALS AND METHODS. Background ATP-independent 45Ca precipitation for detergent-treated MV fractions was not significantly different from the results without detergent treatment (not shown). n = 5, * P < 0.001 for TNAP/ vs. TNAP+/+ at 3-6 days and 6-10 days in Triton-treated MV fractions with ATP present.

Differential Localization of PC-1 Relative to B10 in Osteoblasts and Osteoblast-Derived MV Fractions

View larger version (31K): [in this window] [in a new window]   Fig. 3.   Plasma cell membrane glycoprotein-1 (PC-1) and B10 mRNA are expressed in MC3T3 cells and primary calvarial osteoblasts in culture. Total RNA isolation and RT-PCR analysis were performed to determine whether PC-1 and B10 mRNA were expressed in cultured MC3T3 cells and primary calvarial osteoblasts from TNAP wild-type, heterozygotic, and homozygous knockout mice (designated as in Fig. 2). Cells were cultured for the number of days indicated and as described in MATERIALS AND METHODS. Assay of the ribosomal "housekeeping gene" L30 served as the control for RNA loading.

PC-1 protein expression also was readily detected in cell lysates and in the MV fractions of primary calvarial osteoblasts of TNAP+/+ mice, and of TNAP+/ and TNAP/ mice (Fig. 4A). However, PC-1 was more readily detected in equal amounts of protein from MV fractions of the osteoblasts of TNAP+/+ mice than TNAP/ mice (Fig. 4A). PC-1 and B10 both were detected in cell lysates of primary calvarial osteoblasts and MC3T3 cells (Fig. 4). In contrast, in MV fractions from calvarial osteoblasts of all mice tested, B10 was below the limits of detection in Western blotting (Fig. 4A), and B10 could not be detected in MV fractions of MC3T3 cells (Fig. 4B). Semiquantitative densitometric analyses of these Western blot findings suggested a greater than eightfold increase in MV fraction PC-1 in MV fractions from TNAP+/+ compared with TNAP/ mice, and levels of cell lysate PC-1 were more than 50% greater in TNAP+/+ cells than in TNAP/ cells, under conditions where tubulin was not greater in cells of TNAP+/+ mice compared with TNAP/ mice (not shown).

View larger version (42K): [in this window] [in a new window]   Fig. 4.   Preferential localization of PC-1 relative to B10 in MV fractions of primary calvarial osteoblasts (A) and MC3T3 cells (B). A: primary calvarial osteoblasts from TNAP +/+, +/ and / mice were cultured, and MV fractions isolated as described above. Then, 30 µg of protein from cell lysates (at days 6 and 10 in culture) and MV fractions (at day 10) were analyzed by SDS-PAGE and Western blotting as described in MATERIALS AND METHODS. Detection of immunoreactive products, performed as described in MATERIALS AND METHODS, is illustrated for PC-1 and B10 (each ~130 kDa) and for tubulin (~70 kDa) as the housekeeping gene control. B: C3T3 cells were cultured with or without transfection with empty plasmid, or PC-1 or B10 in pcDNA3.1, as indicated, using the procedure described in MATERIALS AND METHODS. Cell lysates and MV fractions (30 µg of protein) were studied by Western blotting after isolation at day 7 in culture (all lanes indicated were taken from the same gel).

Because we detected PC-1 but not B10 expression in MV fractions of primary calvarial osteoblasts and MC3T3 cells, we assessed whether direct upregulation of B10 expression would be associated with localization of B10 in MV fractions. To do so, we transfected MC3T3 cells with human PC-1 and B10 cDNAs, which are ~80% identical to their respective rodent homologs (16). We cultured the MC3T3 cells under mineralizing conditions as used for primary calvarial osteoblasts.

Transfection of PC-1 increased immunoreactive PC-1 in MC3T3 cells (Fig. 4B) and elevated both cell-associated and MV fraction NTPPPH activity in MC3T3 cells (Fig. 5A). PC-1 also significantly elevated MV fraction PP_i, but not extracellular PP_i, suggesting enrichment of PC-1 in the MV fractions (Fig. 5C). In contrast, transfection of B10 elevated cell-associated NTPPPH activity but did not augment MV fraction NTPPPH activity in MC3T3 cells (Fig. 5A). B10 transfection also did not elevate MV fraction PP_i (Fig. 5C). Transfected MC3T3 cells demonstrated an increase in PC-1 or B10 protein (Fig. 4B), but only PC-1 was detectable by Western blotting in MV fractions derived from the cells transfected with either NTPPPH (Fig. 4B).

View larger version (21K): [in this window] [in a new window]   Fig. 5.   Effects of transfection of PC-1, B10,and TNAP on nucleotide pyrophosphatase (TPPPH) and alkaline phosphatase (AP) specific activities and inorganic pyrophosphate (PPi) concentration in MC3T3 cells and MV fractions. MC3T3 cells were transfected with 5 µg of empty plasmid DNA or plasmid DNA containing PC-1, B10, or TNAP cDNA. Alternatively, cells were cotransfected with 2.5 µg each of PC-1 and TNAP and 2.5 µg each of B10 and TNAP cDNA constructs. Cell lysates, conditioned media, and MV fractions (n = 5) were collected at day 7, and, as indicated (A-C), were assessed for specific activity of NTPPPH, AP, and PPi, respectively, relative to cell or MV fraction protein. Units of NTPPPH and of AP were designated as 1 µmol of substrate hydrolyzed/h. Transglutaminase specific activity (D) was measured as described in MATERIALS AND METHODS and served as a control. A: * P < 0.01 for NTPPPH activity relative to control for all conditions except MV fraction NTPPPH after B10 transfection. B: * P < 0.0001 for increase in MV fraction AP after transfection of TNAP, PC-1+TNAP, B10+TNAP. C: * P < 0.05 for change in MV fraction PPi relative to control after transfection of PC-1 and TNAP, and for decrease in MV fraction PPi after transfection of PC-1+TNAP relative to PC-1 alone.

Because PC-1 but not B10 appeared to localize in isolated osteoblast-derived MV fractions, we assessed and compared the distribution of PC-1 and B10, relative to TNAP, in mineralizing MC3T3 cells, using confocal microscopy (Fig. 6). Before matrix mineralization, at 48 h in culture, TNAP staining was readily detected on the surface of the subconfluent MC3T3 cells. Surface PC-1 and B10 immunostaining were weak at this point in time (Fig. 6A). Permeabilization of MC3T3 cells revealed that B10 staining was predominantly intracellular at 48 h (Fig. 6A).

View larger version (69K): [in this window] [in a new window]   Fig. 6.   Confocal microscopy: PC-1 but not B10 colocalizes with TNAP on the surface and in the pericellular matrix of mineralizing MC3T3 cells. MC3T3 cells were cultured with ascorbate and -glycerophosphate as described above and stained for AP activity by using naphthol AS-MX and fast red TR salt and immune stained for PC-1 and B10 as described in MATERIALS AND METHODS. Where indicated, cells were permeabilized with 0.1% Triton X-100, as described in MATERIALS AND METHODS, in blocking buffer for 10 min, and then again treated with blocking buffer for 45 min. Cells were studied by using a Zeiss Axiovert 100M laser scanning microscope using the FITC channel to detect PC-1 or B10 staining, and the Texas red channel to detect AP staining. Results are shown before onset of mineralization (at 48 h in culture) (A), and during mineralization by confluent cells (at 7 days in culture) (B-D). In D, all cells were permeablized. Colocalization of PC-1 or B10 with TNAP was detected by yellow or orange staining. LM, light microscopy control.

At 7 days in culture, when mineralization of the matrix by confluent MC3T3 cells was detectable, PC-1 and TNAP staining colocalized in broad areas of the pericellular matrix (Fig. 6B). Osteoblast surface and pericellular matrix B10 staining remained weak at this time (Fig. 6C). Moreover, permeabilization of the MC3T3 cells revealed that B10 staining remained predominantly intracellular in MC3T3 cells in a distribution distinct from that shared by PC-1 and TNAP (Fig. 6D).

Functional Relationship Between TNAP and PC-1 in MV Fractions

Transfection of wild-type TNAP significantly elevated cell-associated and MV fraction-associated AP activity in MC3T3 cells and decreased the PP_i associated with MV fractions derived from MC3T3 cells transfected with PC-1 (Fig. 5, B and C). Paradoxically, transfection of wild-type TNAP was associated with a significant increase in cell-associated NTPPPH, and an even greater increase in MV fraction-associated NTPPPH activity in MC3T3 cells (Fig. 5A) but no significant change in activity of transglutaminase activity, which served as a control enzyme for these experiments. We established that this result was not attributable to intrinsic NTPPPH activity of TNAP. Specifically, immunoprecipitation of TNAP from osteoblastic cell lysates isolated >85% of the TNAP activity but was not associated with any coimmunoprecipitation of NTPPPH activity (starting mean cell lysate NTPPPH of 1.7 U and AP of 4.3 U; beads after immunoprecipitation with antibody to TNAP had mean of 0.09 U NTPPPH and 4.1 U alkaline phosphatase; beads after immunoprecipitation with antibody to PC-1 had mean of 1.3 U NTPPPH and 0.11 U AP, n = 5).

Because TNAP appeared to regulate the NTPPPH activity of osteoblastic MC3T3 cells, we assessed whether this effect was dependent on TNAP enzymatic activity. Wild-type TNAP, but not the enzyme-inactive mutant of TNAP, induced an increase in both MV fraction NTPPPH and AP activity (Fig. 7, A and B). In contrast to wild-type TNAP, the mutant TNAP failed to decrease MV fraction PP_i (Fig. 7C).

View larger version (13K): [in this window] [in a new window]   Fig. 7.   The ability of TNAP to augment MV fraction NTPPPH requires TNAP enzyme activity. MC3T3 cells were transfected with 5 µg of empty plasmid, or plasmids encoding TNAP or a mutant enzyme-inactive TNAP. Cell lysates, conditioned media and MV fractions (n = 5) were collected at day 7, and, as indicated in panels A-C, were assessed for specific activity of NTPPPH, AP, and PPi, respectively, relative to cell or MV fraction protein. A: *P < 0.0001 for increased NTPPPH after TNAP transfection. B: *P < 0.001 for increased MV fraction TNAP after TNAP transfection. C: *P < 0.05 for decreased MV fraction PPi after TNAP transfection.

The results presented in Fig. 4A had indicated that immunoreactive PC-1 decreased progressively over time in culture in TNAP/ osteoblasts relative to primary osteoblasts from TNAP+/ and TNAP+/+ mice. Thus we also assessed the relationship among TNAP deficiency and MV fraction PP_i-generating NTPPPH activity and PP_i concentration in primary osteoblasts.

Cell-associated NTPPPH but not transglutaminase activity progressively decreased over time in culture in osteoblasts from TNAP/ mice, relative to cells from TNAP+/ and TNAP+/+ mice, and was significantly less in TNAP/ cells than in TNAP+/+ cells at 14 days (Fig. 8). Moreover, NTPPPH activity but not transglutaminase activity was significantly lower in MV fractions derived at days 10-13 from TNAP/ mice compared with MV fractions from osteoblasts of TNAP+/+ animals, with the values for MV fraction NTPPPH activity being intermediate for TNAP+/ osteoblasts (Fig. 9). Despite the presence of the lowest MV fraction NTPPPH specific activity, it was in the TNAP/ state that the highest MV fraction-associated concentration of the mineralization inhibitor PP_i was observed (Fig. 9).

View larger version (34K): [in this window] [in a new window]   Fig. 8.   Cell-associated NTPPPH activity decreases over time in TNAP-deficient primary calvarial osteoblasts relative to cells expressing TNAP. Primary calvarial osteoblasts from TNAP+/+, and TNAP+/ and TNAP/ mice (n = 12 each) were cultured as described in MATERIALS AND METHODS, and cell lysate AP (A), NTPPPH (B), and transglutaminase activity (C) were sampled at the days in culture indicated. n = 5, * P < 0.01 for AP in TNAP / cells at all time points and for NTPPPH at 14 days.

View larger version (30K): [in this window] [in a new window]   Fig. 9.   PPi concentration and enzymatic activities of MV fractions from primary calvarial osteoblasts from wild-type and TNAP heterozygous and homozygous knockout mice. Osteoblasts were cultured for up to 13 days, and media were changed every fourth day as indicated. MV fraction-associated PPi (A), as well as the MV fraction specific activities of AP (B), NTPPPH (C), and transglutaminase (D) were studied in samples pooled from conditioned media of 12 mice of each genotype, between the days in culture indicated. A: *P < 0.01 for increased MV fraction PPi for TNAP/ cells relative to TNAP+/+ cells at days 6 and 13.B: *P < 0.01 for MV fraction AP for TNAP/ cells relative to TNAP+/+ and TNAP +/ cells at all days. C: *P < 0.01 for MV fraction NTPPPH for TNAP/ cells relative to TNAP+/+ cells at days 6 and 13.

Last, we directly assessed the effects of TNAP, PC-1, and B10 on MV fraction-mediated mineralization (Fig. 10). Transfection of PC-1, but not B10, significantly inhibited ATP-dependent calcium precipitation by MV fractions (detergent-treated and untreated) isolated from the transfected MC3T3 cells (Fig. 10). Wild-type but not mutant TNAP significantly antagonized the inhibitory effect of PC-1 on calcium precipitation by MV fractions derived from the MC3T3 cells (Fig. 10). Thus TNAP functioned to directly antagonize the inhibitory effect of PC-1 on the mineralizing activity of osteoblast-derived MV fractions.

View larger version (24K): [in this window] [in a new window]   Fig. 10.   Effects of increased TNAP, PC-1, and B10 expression in MC3T3 cells on the ability of cell-derived MV fractions to precipitate calcium. MC3T3 cells were transfected with a total of 5 µg of plasmid DNA encoding the indicated enzymes, as described in MATERIALS AND METHODS. Values for NTPPPH, AP, PPi, and transglutaminase correspond to cells in Fig. 5. Seven days after transfection, MV fractions were isolated from conditioned media. The ability of the MV fractions to carry out ATP-dependent precipitation of 45Ca, with or without detergent treatment, was measured as described above (n = 5). Background ATP-independent 45Ca precipitation is shown in the absence of treatment with 0.1% Triton X-100 and was not different from background results with detergent treatment (not shown). * P < 0.001 and * P < 0.05 for inhibition by PC-1 of ATP-dependent precipitation with Triton treatment, and without Triton, respectively (effect of B10 not statistically significant). P < 0.01 and P < 0.05 for removal by TNAP transfection of the inhibitory effect of PC-1 (on ATP-dependent precipitation with and without Triton treatment, respectively).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

A definition of the function of TNAP is pivotal to understanding how the skeleton mineralizes. We previously demonstrated that PC-1 suppressed osteoblast MV-mediated mineralization under conditions where sodium phosphate was provided as the major phosphate source for hydroxyapatite formation (26). The present study employed -glycerophosphate, which, to provide free phosphate, requires hydrolysis by inorganic phosphatase activity such as that exerted by TNAP. PC-1 markedly inhibited ATP-dependent calcium precipitation by MV fractions under these conditions in this study. We also demonstrated that TNAP and PC-1 were concomitantly expressed and TNAP and PC-1 colocalized in both cultured osteoblasts that formed mineralized nodules and the MV fractions derived from the mineralizing cells.

TNAP functioned to directly antagonize the inhibitory effects of PC-1 on ATP-dependent precipitation of calcium by MV fractions. In this regard, a forced increase in TNAP localization in MV fractions diminished basal MV fraction-associated PP_i and prevented PC-1 from augmenting MV fraction PP_i. Our results suggest that TNAP promotes MV-mediated mineralization, in large part, by hydrolyzing the MV-associated mineralization inhibitor PP_i generated by PC-1 through catabolism of exogenous ATP.

A selective impairment of extravesicular propagation of hydroxyapatite by ATP-treated MVs has been demonstrated in subjects with perinatal hypophosphatasia (4). This suggested the possibility of defective metabolism of PP_i at the exterior face of MVs. In previous studies the majority of both NTPPPH activity (38) and TNAP activity (3, 25) have been found to be active on the external face of MVs. We did not directly assess surfaces where the enzyme activities were localized in the MVs obtained in the present study. The observation that detergent-induced perturbation of MV fractions was required to reveal a defect in ATP-dependent MV calcium precipitation associated with TNAP deficiency could be consistent with external-face orientation of TNAP enzyme activity and/or release of a mineralization inhibitor such as PP_i from MVs.

Significantly, a mineralizing defect correlated directly with the extent of TNAP deficiency measured in osteoblast-derived MV fractions. In contrast, the total concentration of MV fraction-associated PP_i was less clearly related to the mineralizing defect observed in TNAP+/ and TNAP/ osteoblasts. The finding that MV fractions from TNAP+/ and TNAP/ osteoblasts both demonstrated a mineralizing defect may help explain in vitro differences in mineralization between osteoblasts from TNAP+/ and TNAP+/+ mice (43a). More specifically, when TNAP+/ and TNAP+/+ mice at day 2-3 after birth are compared, no morphological differences could be detected in the bones in vivo. However, initiation of mineralization in vitro was delayed in the TNAP+/ osteoblast cultures. In this regard, in vitro cultures evaluated at different time points (days 4, 6, 8) revealed a delayed increase in TNAP activity in the TNAP+/ cultures compared with TNAP+/+ cultures, accompanied by a later onset of mineralization, as demonstrated by von Kossa staining and measurement of deposited calcium (43a).

We speculate that the foci in which PP_i might be concentrated at MVs may be more critical in regulating mineralization than the total PP_i concentration associated with MVs. Alternatively, our results suggest that PP_i is not likely to be the sole potent inhibitor of MV mineralization in normal or TNAP-deficient states. For example, the ability to precipitate calcium dramatically declined for MV fractions of normal primary calvarial osteoblasts recovered at days 10-13 in culture, when much of the mineralization of nodules had already been completed. In contrast, the concentration of PP_i associated with MV fractions changed little over 13 days in culture of the normal calvarial osteoblasts.

The relative stability of PP_i in MV fractions produced at various time points may reflect parallel changes in MV AP and NTPPPH activity observed in normal osteoblasts. Significantly, TNAP expression directly promoted an increase in MV fraction NTPPPH specific activity in MC3T3 cells in this study. Moreover, TNAP-expressing primary calvarial osteoblasts ultimately produced MV fractions with significantly more NTPPPH activity than MV fractions produced by TNAP-deficient osteoblasts. Our results suggested that TNAP, by an enzyme activity-dependent mechanism, acts to control mineralization in part by modulating the content of its own inhibitor in osteoblasts and osteoblast-derived MVs. It will be of interest to determine whether modulation of PP_i concentration at a specific location is a regulatory signal for PC-1 gene expression or the distribution of PC-1 to the plasma membrane and MVs. Alternatively, transphosphorylation reactions catalyzed by TNAP (44) might be responsible for modulating MV composition in osteoblasts (3).

MV constituents include matrix proteins and proteoglycans, calcium binding proteins and phospholipids, metalloproteinases, and transglutaminase activity (3, 9, 10, 23, 37). Transglutaminase activity can promote mineralization by promoting activation of latent transforming growth factor- (28) and modifying extracellular matrix proteins including osteonectin (1). However, osteoblast MV fraction transglutaminase activity did not change in response to direct TNAP transfection of MC3T3 cells, and there was only a nonstatistically significant trend toward higher transglutaminase activity in MV fractions of TNAP-deficient osteoblasts. It will be of interest to determine what other constituents of MVs that regulate mineralization are modulated by TNAP activity in osteoblasts.

In this study, we demonstrated that PC-1 preferentially distributed to osteoblast MV fractions, compared with the closely related NTPPPH B10. MV fraction-associated NTPPPH and PP_i were increased by a direct elevation in expression of PC-1 but not of B10. The ability of PC-1 to suppress MV-mediated mineralization was not shared by B10. We observed that B10 remained in a predominantly intracellular distribution distinct from that of PC-1 and TNAP in mineralizing osteoblastic cells. These findings help explain how osteoblast-mediated hyperossification occurs in PC-1-deficient mice, despite the fact that mouse osteoblasts also contain abundant B10 and despite the physiological expression (as a component of BMP-2-mediated chondroosseous differentiation) of a distinct PDNP/NTPPPH isozyme called autotaxin, which is a predominantly secreted species (6).

The genes for PC-1 and B10 are located in close proximity on chromosome 6q21-23, presumably reflecting an antecedent gene duplication event (19). Each gene encodes a class II (intracellular NH₂ terminus) transmembrane glycoprotein of 120-130 kDa that shares a highly homologous extracellular domain containing two somatomedin B-like regions and a conserved catalytic site (19). However, B10 has a unique extracellular RGD cell adhesion motif, and the cytosolic tails of PC-1 and B10 share no significant homology (19). Moreover, differential localization and function of PC-1 have been observed in other tissues. For example, PC-1 translocates to the basolateral surface and B10 to the apical surface in a polarized cell type (hepatocytes), an effect attributable to differences in the cytosolic tail (39). Furthermore, PC-1 but not B10 preferentially localizes to the plasma membrane in human articular chondrocytes and only PC-1 causes an increase in extracellular PP_i in these cells (32).

In conclusion, TNAP promotes mineralization, in part, by removing the profound inhibitory effect of PC-1-generated PP_i on ATP-dependent MV-mediated mineralization. It will be of interest to determine whether TNAP does so by not only hydrolyzing PP_i but also by dephosphorylating ATP that would otherwise be used by PC-1 to generate PP_i. This study established that PC-1 preferentially distributed to osteoblast MV fractions compared with another NTPPPH isozyme, B10 (also known as PDNP3). It will be of interest to determine whether the signals responsible for PDNP/NTPPPH-selective basolateral membrane localization of PC-1 in polarized epithelia and PC-1 distribution to MVs are the same or different. Results of such studies would help elucidate how selective concentration of specific ectoenzymes occurs in MVs. We also conclude that TNAP upregulates the amount of NTPPPH expressed by osteoblasts, which influences the amount of NTPPPH activity distributing to osteoblast MVs. Once their bone-making duties in the basic multicellular unit of bone are completed, the majority of osteoblasts must become less active in bone formation (29). We speculate that the control of PC-1 expression by TNAP may be one of the physiological means by which normal osteoblasts regulate their bone-making activity.

Perspectives