Effects of non-collagenous proteins on the formation of apatite in calcium β-glycerophosphate solutions

A&S oral Biol.Vol. 31, No. I, pp. 15-21, 1992 Printed in Great Britain.All lights reserved

0003-9969192 S5.00+ 0.00 Copyright 0 1992PergamonPressplc

EFFECTS OF NON-COLLAGENOUS PROTEINS ON THE FORMATION OF APATITE IN CALCIUM Q-GLYCEROPHOSPHATE SOLUTIONS YUTAI;A Dot,’ TAKASHIHOIUGUCHI,’SEUG-HYOKIM,’ YUTAKAMORIWAKI,’ NOBUKAZUWAKAMATXJ,’MASANORIADACHI,’KYOMIIBARAKI,*K~JI MORIYAMA,~ SATOSHISASAKI~and HITOYATASHIMOKAWA’ ‘Department of Dental Materials and Technology, School of Dentistry, Asahi University, Gifu, *Department of IBiochemistry and ‘Second Department of Orthodontics, Faculty of Dentistry, Tokyo Medical and Dental University, Tokyo, Japan (Received 16 January 1991; accepted 30 July 1991) Summary-The effects of the non-collagenous proteins; osteonectin, bone Gla protein and dentine phosphoprotein, on the formation of apatite were studied in calcium jl-glycerophosphate solutions containing catalytic amounts of alkaline phosphatase under physiological conditions. In the system used, calcium phosphate precipitates de nova at levels of supersaturation precisely determined through the enzymatic hydrolysis of p-glycerophosphate. At 1.7 mM of calcium b-glycerophosphate, calcium phosphate precipitated when inorganic phosphate accumulated to about 1.4mM. In the presence of the proteins, however, a greater accumulation of inorganic phosphate was needed for calcium phosphate to precipitate, suggesting that a higher degree of supersaturation, though still a slight undersaturation with respect to dicalcium phosphate dihydrate, is required for calcium phosphate to precipitate in the presence of the proteins. At the same protein @g/ml) concentration, dentine phosphoprotein was approximately four times as effective as bone Gla protein, which was about twice as effective as osteonectin in delaying precipitation. The proteins also retarded subsequent crystal growth, with apatite formed in the presence of the more inhibitory proteins having the smallest crystals, especially in width. Key words: osteonectin, bone Gla protein, dentine phosphoprotein,

INTN!ODUCITON

mineralization, glycerophosphate.

in employing solutions prepared solely from inorganic calcium and phosphate compounds. Also, in most of these systems, the initial degree of supersaturation was kept the same, regardless of the presence of proteins. We have now investigated the effects of Gla protein, osteonectin and dentine phosphoprotein on in vitro mineralization from solutions consisting of calcium /?-glycerophosphate and alkaline phosphatase. This system, which was used by Banks et al. (1977) to show that fibrous apatite can grow on modified collagen fibres, offers an advantage in that the degree of supersaturation and the onset of calcium phosphate precipitation can be precisely determined by the alkaline phosphatase hydrolysis of fi-glycerophosphate.

Non-collagenous protems have the potential to regulate the mineralization of bone and dentine matrices. Among these proteins, Gla protein (Price et al., 1976; Poser and Price, 1979; Price and Williamson, 1985), osteonectin (Termine c?t al., 1981a; Termine et al., 1984; Romberg et al., ‘1985, 1986; Engel et al., 1987; Fisher et al., 1987; Doi et al., 1989) and dentine phosphoprotein (Nawrot et al., 1976; Termine, Eanes and Conn, 1980a; Lee and Glimcher, 1981; Veis, 1985) have been studied extensively because of their abundance in bone and dentine matrix and their close association with mineralization. However, there is controversy for Gla protein (Poser and Price, 1979; Price and Williamson, 1985; Linde, Lussi and Crenshaw, 1989), osteonec:tin (Termine et al., 1981a; Romberg et al., 1985; Romberg et al., 1986; Doi et al., 1989) and dentine phosphoprotein (Nawrot et al., 1976; Termine et al., 1980) as to whether they inhibit or promote the formation of hydroxyapatite in vitro. Although it cannot be ruled out that some proteins may either inhibit or promote hydroxyapatite formation (Termine et of., 1981a; Linde, Lussi and Crenshaw, 1989), delpending upon experimental conditions, the in vitro systems used have been similar

MATERIALS AND METHODS Preparation

of the proteins

Osteonectin and bone Gla protein. The diaphyses of

young calf bone were scraped clean of soft tissues and washed in cold water. The washed bone was frozen in liquid nitrogen, and ground in a Wiley mill to a particle size of 1 mm3. The particles were demineralized in 0.6 M HCl at 4°C for 48 h and washed in cold water. Then the particles were extracted with 4 M guanidine HCl. The proteins soluble in this were precipitated by dilution with seven volumes of deionized cold water and allowed to stand overnight at

Abbreviations: Gla, ycarboxyglutamate-containing; SDSPAGE, sodium dodecyl sulphate-polyacrylamide gel electrophoresis. 15

16

YUTAKA Dot et al.

4°C. The precipitate was removed by centrifugation at 40,OOOg for 20 min, and the supernatant was dialysed against distilled water. The water-soluble fraction (Gla protein-rich fraction) and the waterinsoluble fraction (osteonectin-rich fraction) were purified by two passes of DEAE-Sephacel ionexchange chromatography, as described by Termine et al. (1981b). Osteonectin was further purified by gel-filtration, high-pressure liquid chromatography (Waters Protein Pak 125), as described by Doi et al. (1989). Bone Gla protein and osteonectin were electrophoresed as single bands with apparent molecular weights of 14 and 40 kDa, respectively, in the presence of /I-mercaptoethanol. These proteins were further identified by amino acid composition and immunoblot using anti-osteonectin and anti-bone Gla protein monoclonal antibody (data not shown). The molecular weight of osteonectin estimated from SDS gel was essentially the same as the reported value (Termine et al., 1981a; Romberg et al., 1985) whereas that of bone Gla protein was approximately twice the value (about 6000) estimated, based on the amino acid sequence of calf bone Gla protein (Hauschka, 1985). It is generally accepted that apparent molecular weight of bone Gla protein calculated from SDS gels ranges from 7000 to as high as 17,000. This is considerably greater than the true molecular weight established by sequencing, and results from polyelectrolyte effects rather than aggregation of bone Gla protein (Gundberg et al., 1984). Dentine phosphoprotein. Guanidine-EDTA extracts of dentine were obtained from incisor tooth germs of approx. I-yr-old calves and applied to a DEAE-Sephacel column (2.5 x 10 cm) equilibrated with 7 M urea in tris-HCl (pH 7.5). Dentine phosphoprotein was eluted with a linear gradient of O-O.7 M NaCl at flow rate of 30 ml/h. The dentine phosphoprotein fraction was collected and further purified on a Sepharose CL-6B column (1.6 x 120 cm) under dissociative conditions (4 M guanidine), as described by Termine et al. (1980b). The apparent molecular weight for the dentine phosphoprotein was approx. 120 kDa by SDSPAGE. Egg-yolk phosvitin purchased from Sigma Chemical Co. (St Louis, MO) and purified by dialysis had an apparent molecular weight of 40 kDa (Allerton and Perlmann, 1965). In presenting the molar concentrations of the proteins, molecular weights of 40, 6, 120 and 40 kDa were used for osteonectin, bone Gla protein, dentine phosphoprotein and phosvitin, respectively. Precipitation of calcium phosphate. Calcium phosphates were precipitated from calcium /I-glycerophosphate solutions in the presence of alkaline phosphatase. Calcium /I-glycerophosphate was purified by recrystallization of the commercially purchased chemical (Nacalai Tesque Inc., Kyoto, Japan) and dissolved in 100 mM NaCl and 50 mM tris solution buffered at pH 7.40, at 37°C. Alkaline phosphatase (EC3.1.3.1) was purchased from Sigma Chemical Co. (St Louis, MO) and used without further purification. All the proteins were dissolved in the solution of calcium /3-glycerophosphate before adding the alkaline phosphatase solution, which was prepared by dissolving alkaline phosphatase in the buffered solution (0.87 mg/ml) before each series of experiments.

Reaction kinetics Reaction kinetics were followed either by monitoring calcium activity with a calcium electrode (Microprocessor Ionalyser/90 1, Orion Research, Nagoya, Japan) or by analysing dissolved calcium and inorganic phosphate. The withdrawn solutions were filtered through 0.2 pm Millipore filters and an equal volume of 0.05 M HCI solution was added to each filtered solution. The final pH of the filtered, acidified solution was about 1.7 at room temperature. The addition of 0.05 M HCl was found to suppress almost completely the enzymatic action of alkaline phosphatase and also to inhibit the formation of calcium phosphate precipitates before analysis of the solution. Calcium concentrations were determined by ionexchange chromatography (Shimadzu Ion Chromatography System, HIC-6A) using a solution of 4 mM tartaric acid and 2 mM ethylenediamine as the carrier. Inorganic phosphate concentrations were determined by the method of Murphy and Riley (1962). The composition thus obtained was further analysed by chemical potential plots, as described by Doi et ul. (1989), with one further assumption that small amounts of glycerine formed as a by-product during the enzymatic hydrolysis of /I-glycerophosphate did not contribute significantly to the dielectric constant of the reacting solution. Identification of solid At regular intervals, samples of the reaction solution were withdrawn and suspended solids were prepared for transmission electron microscopy (Doi and Eanes, 1984; Doi et al., 1984a) (JEOL1200EX analytical electron microscope operated at 80 kV with an objective aperture of 50 pm). Precipitated solids at the end of reactions were identified by X-ray diffraction (Rigaku Rotaflex) at 50 kV and 15OmA, as described in detail by Doi et al. (1989). RESULTS

The time at which calcium phosphate precipitated depended on the initial concentration of calcium /I-glycerophosphate and alkaline phosphatase. Text Fig. 1 shows time plots of electrode-measured cal-

0

5

10

15

20

time/h Fig. 1. Time plots of solution calcium measured with a calcium electrode at four concentrations of calcium B-glycerophosphate in the presence of 0.05 pg/ml alkaline phosphatase. Calcium jI-glycerophosphate at each concentration [6.8 mM(a), 5.1 mM(b), 3.4mM(c) and 1.7 mM(d)J was dissolved in 100 mM NaCl buffered at pH 7.40 at 37°C with 50mM tris.

Non-collagenous proteins and apatite formation cium in solutions where the initial concentration of calcium fl-glycerophosphate varied from about 2-7 mM while keeping the concentration of alkaline phosphatase constant alt 0.05 mg/ml. It is evident that the higher the initial c,oncentration of the glycerophosphate the earlier the precipitation, as measured by the steep decrease in the dissolved calcium. At higher concentrations, however, a less prominent but noticeable decrease preceded the steep decrease in the calcium, indicating that at least two reactions might contribute to the observed precipitation kinetics. As the biological fluid responsible for matrix mineralization is known to have less than 2 mM calcium under comparable ion strength and pH (Howell et al., 1968), the least concentrated (about 1.7mM) solution of calcium B-glycerophosphate was used hereafter. The concentration of alkaline phosphatase was set at 0.05 mg/ml to initiate precipitation about 8 h after reaction. As shown in Text Fig. 2, in this reaction the dissolved calcium remained constant at the initial value of 1.7 mM, while the dissolved inorganic phosphate increased monotonically with time until both suddenly decreased coincidentally. The point at which this last event occurred was taken as the induction time for the precipitation reaction. Text Figs 3(A) and (B) show time plots of the dissolved calcium and inorganic phosphate, respectively, in the presence of bone Gla protein and dentine phosphoprotein. For each protein, the induction time lengthened with increiases in the protein concentration. At 5 pg/ml (0.042 PM) of dentine phosphoprotein or 14 pg/ml(2.?1 PM) of bone Gla protein, the reaction was completely inhibited, and no precipitation of calcium phosphate observed even after one week. Text Fig. 4 compares the inhibitory activity of osteonectin (0.088 PM), bone Gla protein (0.58 PM), egg-yolk phosvitin (0.088 p M) and dentine phosphoprotein (0.029 PM) at a protein concentration of 3.5 pg/ml by plotting the dissolved inorganic phosphate as a function of time. Osteonectin appeared to be least effective in inhibiting the precipitation of calcium phosphate and is about half as effective as bone Gla protein in lengthening the induction time. The inhibitory activity of egg-yolk phosvitin was

(doy10.5 I

0 IJ IIO

17

102 D"

I 104

fime/min (doy)o.5 I 3 57 # , Illmll

0

LJ I JO

_ I

I02

1

c

10s

IO

time/min Fig. 3. Effectsof bone Gla protein (A) and dentine phosphoprotein (B) on the formation of apatite. In each figure, dotted lines represent time plots of solution calcium and inorganic phosphate in the protein-free reaction. The solid and broken lines represent, respectively, time plots of solution calcium (solid symbol) and solution inorganic phosphate (open symbol) in the protein-added reaction. The numbers in each figure represent the protein concentration &g/ml).

almost the same as that of dentine phosphoprotein, which was approximately four times as effective as bone Gla protein in delaying the onset of precipitation. It is of interest that in the presence of the more inhibitory proteins the dissolved inorganic phosphate

35

r”----Irl-mr

te- time/min

O

loo

Kloo

10000

Fig. 2. Time plots of solution calcium and inorganic phosphate at initial concentration of 1.7mM calcium /3-glycerophosphate.

Fig. 4. Comparison of effects of osteonectin (ost), bone Gla protein (BGP), dentine phosphoprotein (DPP) and egg-yolk phosvitin (phosv.) on the formation of apatite by showing the time plot of the solution inorganic phosphate at protein concentration of 3.5 pg/ml.

Yuruu

18

Do1 et al.

4450 n-in 5887 min

3\ 530.5

31.0

J degree/26

PtH’fCPO:-1

Fig. 6. X-ray diffraction patterns of apatite formed in the absence and the presence of the protein. From top, solid at 4450 min of the protein-free, at 5887min of osteonectin (ost)-added, at 6845 min of phosvitin (phosv)-added, and at 7175 min of dentine phosphoprotein (DPP)-added reactions.

16.5

16.0

P(H’ftPO:-) Fig. 5. Chemical potential plots of the solution composition in the protein-added reactions. (A) In the presence of 3.5 pg/ml osteonectin (ost) (A) or bone Gla protein (BGP) (0). (B) In the presence of 3.5,ug/ml dentine phosphoprotein (DPP) (A) or egg-yolk phosvitin (phosv) (0). In each figure the dotted line represents solution movement in the protein-free reaction. The numbers in each figure represent the reaction time in min. TCP, j-tricalcium phosphate; OCP, octacalcium phosphate; DCPD, dicalcium phosphate dihydrate.

became more concentrated before the onset of precipitation. Text Fig.5 shows the chemical potential plots of the solution compositions shown in Text Fig. 4. In the absence of the protein [dotted line in each figure of Text Figs 5(A) and (B)] the solution composition first moved toward the solubility line of dicalcium phosphate dihydrate, parallel with the abscissa representing negative log(Hf),(P043-). Because the system was well buffered, this movement simply meant an increase in activity of phosphate. However, before it became saturated with respect to this phase, the solution composition moved in a different direction away from dicalcium phosphate dihydrate and toward the solubility line of octacalcium phosphate.

About one day after reaction the solution composition closely paralleled the solubility line of octacalcium phosphate for about 1 h and then became undersaturated with respect to this phase. The movement of the solution composition in the presence of protein resembled that of the protein-free reaction except for more closely approaching the solubility line of dicalcium phosphate dihydrate before the onset of precipitation and then moving away from this line more slowly after the onset of precipitation. Text Fig 6 shows the X-ray diffraction patterns of the precipitates obtained at the end of reaction. These patterns were characteristic of a poorly crystallized hydroxyapatite. In the presence of the more inhibitory proteins, egg-yolk phosvitin and dentine phosphoprotein, however, the diffraction peaks, especially in the range of 30”-35” 28, were less resolved in spite of the fact that the reaction time was much longer in these cases. This finding suggests that apatite crystals formed in the presence of protein are smaller in size or are more defective in lattice structure, or both. Plate Fig. 7 shows typical electron micrographs prepared from solution samples taken at intervals throughout the course of the reaction in the presence of the protein. At approximately the same reaction times, the osteonectin micrographs were similar to the protein-free micrographs, and the dentine phosphoprotein micrographs were almost the same as the phosvitin micrographs. Before the induction time all the grids investigated were free of precipitates of calcium phosphates. The first precipitates observed just after the induction time were rounded in appearance [Plate Figs 7(A) and (D)]. As the reaction proceeded a crystalline-like phase became the dominant feature [Plate Figs 7(B) and (C)l. Although the size of these crystals depended on the reaction time,

Plate 1 Fig. 7. Transmission electron micrographs of solid specimens 1015 (A) and 2620 min (C) after initiating the Gla protein-added reactions; 3960 min (B) after initiating osteonectin-added reactions; and 4077 (D) and 7175 min (E) and (F) after initiating dentine phosphoprotein-added reactions. Note that in the presence of dentine phosphoprotein, individual apatite crystal becomes smaller [Plate Fig. 7(E)], especially in width, indicating that crystal growth in this direction was considerably slowed by the protein, although very few crystals [Plate Fig. 7(F)] grew in a way similar to those seen in the protein-free reaction.

Non-collagenous

proteins and apatite formation

Plate 1

19

YUTAKA DOI et

20

it was more greatly affected by the type of protein present. Osteonectin was found to be the least effective [Plate Fig. 7(B)], the Gla protein somewhat more so [Plate Fig. 7(C)], and the dentine phosphoprotein and the phosvitin the most effective in reducing the crystal size. Although all the crystalline precipitates were identified by X-ray diffraction as apatite, in the presence of dentine phosphoprotein or phosvitin [Plate Fig. 7(E)] most of the crystals were needle-like with only a few showing the plate-like features of crystals observed in protein-free reactions. DISCUSSION

We here demonstrate that the matrix proteins, bone Gla protein, osteonectin and dentine phosphoprotein, when present free in solutions, delay the onset of calcium phosphate formation from solutions of calcium /I-glycerophosphate and alkaline phosphatase. When compared at a protein concentration of 3.5 pg/ml, the inhibitory activity increased in the order of osteonectin, bone Gla protein and dentine phosphoprotein. The last was approx. four times as effective as bone Gla protein, which was about twice as effective as osteonectin in delaying the onset of precipitation. The protein concentration of 3.5 pg/ml corresponds to a molar concentration of 0.088, 0.58 and 0.029 PM, respectively, for osteonectin, bone Gla protein and dentine phosphoprotein. As shown in Text Fig. 4, crystal growth after the onset of precipitation is also more protracted in the presence of the more inhibitory proteins. This finding clearly suggests that the proteins can inhibit not only the formation of the initial phase but also its transformation to apatite. Inorganic phosphate here increased in concentration through the enzymatic hydrolysis of calcium /I-glycerophosphate (Banks et al., 1977) until calcium phosphate formed de nouo. In mineralization in cultures grown in the presence of /I-glycerophosphate (Tenenbaum and Heersche, 1982; Ecarot-Charrier et al., 1983; Gerstenfeld et al., 1987) organic phosphate as a source of phosphate ions has been postulated to be more important than circulating inorganic phosphate in initiation of cellular-controlled mineralization (Tenebaum and Heersche, 1982). Although Robinson’s mechanism (Robinson, 1923) for mineralization is no longer accepted, our system has the advantage that the degree of supersaturation can be raised without perturbing the reaction solution and that the supersaturation at the onset of precipitation can be determined precisely. At an initial concentration of 1.7 mM calcium /3-glycerophosphate, no precipitation of calcium phosphate appeared in the protein-free reaction until the concentrations of inorganic phosphate was raised to about 1.4 mM. These concentrations of calcium and phosphate are similar to those found in the biological fluid responsible for matrix mineralization (Howell et al., 1968). In the presence of the protein, however, the concentration of inorganic phosphate at the onset of precipitation was higher than that in the corresponding proteinfree reaction. This increase was in proportion to the protein’s inhibitory activity. The finding that the inorganic phosphate in the protein-added reactions increased essentially in the same way as that of the protein-free reaction for the first 8 h (Text Fig. 4) is

al.

evidence that the proteins did not affect the enzymatic hydrolysis of b-glycerophosphate, and also that they were not themselves enzymatically hydrolysed enough to produce measurable additional inorganic phosphate during this period. However, this was found not to be the case for the highest concentration of dentine phosphoprotein toward the end of the precipitation period. Here the total concentration of inorganic phosphate became slightly higher than that expected from the initial concentration of /?-glycerophosphate alone, owing to some evaporation of the solution or partial hydrolysis of the protein phosphate (Nawrot et al., 1976) although the preliminary experiment, in which the proteins were incubated in 100 mM NaCl and 50 mM tris solution buffered at 7.40 at 37°C in the presence of 0.05 mg/ml alkaline phosphate, but without /I-glycerophosphate, suggested that no measurable inorganic phosphate was produced from the proteins (except egg-yolk phosvitin) even after 1 week of incubation. For eggyolk phosvitin at a molar concentration of 1.25 PM (50 pgg/ml), about 0.01 mM of inorganic phosphate were detected after 3 days of the incubation. If eggyolk phosvitin has approx. 10.4% phosphorus available for the formation of inorganic phosphate by alkaline phosphatase hydrolysis (Allerton and Pertmann, 1965) about 50% of the phosvitin must have been dephosphorylated. Because the dissolved calcium was always at the same concentration, the fact that the inorganic phosphate concentration at the onset of precipitation was invariably higher in the protein-added reactions suggests that a higher degree of supersaturation was needed for the initial calcium phosphate phase to form de nouo from these solutions. The thermodynamic analysis of the solution compositions also confirms this finding, showing that the protein-added solution composition was closer to the solubility line of dicalcium phosphate dihydrate at the onset of precipitation. However, because the initial precipitation took place when the solution was still undersaturated with respect to dicalcium phosphate dihydrate, neither the dihydrate nor amorphous calcium phosphate are likely to be involved in the formation of the initial phase. The additional finding that the solution composition closely paralleled the solubility line of octacalcium phosphate during part of reaction (Text Fig. 5) suggests, on the other hand, that octacalcium phosphate was possibly involved in the formation of initial phase (Sobel and Laurence, 1960; Meyer and Eanes, 1978; Driessens and Verbeeck, 1986). The plate-like crystals seen by transmission electron microscopy, which were shown to be apatite by X-ray diffraction, also suggest an origin from the octacalcium phosphate phase (Eanes and Meyer, 1977). Of possible biological significance is the finding that the apatite crystals formed in the presence of the protein became smaller, especially in width (Plate Fig. 7). This reduction in width causes the individual crystals of apatite to be more needle-like. Although a quantitative measure was not made, this decrease in size in the presence of the more inhibitory proteins, especially of dentine phosphoprotein, appeared to be accompanied by the formation of a greater number of crystals. It is possible that the strong inhibitory action

Non-collagenous proteins and apatite formation of the protein toward growth of octacalcium phosphate crystals (Doi et al., 1984b, 1989) may have favoured the direct formation of apatite at higher supersaturations. Generally, crystals formed at higher supersaturations are smaller in size. In this regard, the proteins can play an important role in the early stage of the matrix mineralization. At this stage, mineralization proceeds by an increase in the number of apatite crystals, rather than by growth of individual crystals (Glimcher, 198 1). Acknowledgement-This work was supported in part by research Grants No. 6070862.63570821and 01571005from the Ministry of Education of Japan.

REVERENCES

AlIerton S. E. and Perlmann G. E. (1965) Chemical characterization of the phosphoprotein phosvitin. J. biof. Chem. 240,3892-3908.

Banks E., Nakajima S., Shapiro L. C., Tilevitz 0.. AIonzo J. R. and Chianell R. P. (1977) Fibrous apatite grown on modified collage. Science 198, 1164-I 166. Doi Y. and Eanes E. D. (1984) Transmission electron microscopic study of calcium phosphate formation in supersaturated

solution seeded with apatite. Calc. Tiss.

Int. 36, 39-47.

Doi Y., Eanes E. D., Shimokawa H. and Termine J. D. (1984a) Inhibition of seeded growth of enamel apatite crystals by amelogenin and enamelin proteins in vitro. J. dent. Res. 63, 98-105. Doi Y., Eanes E. D., Shimokawa H. and Termine J. D. (1984b) Modulation of seeded enamel apatite crystal growth in vitro by enamel matrix amelogenin and enamelin proteins. In Tooth A%zumelIV (Us Feamhead R. W. and Suga S.), pp. 19-23. Elsevier Science, Amsterdam. Doi Y., Okuda R., Takezawa Y., Shibata S., Moriwaki Y., Wakamatsu N., Shimitu N., Moriyama K. and Shimokawa H. (1989) Osteonectin inhibiting de novo formation of apatite in the presence of collagen. Calc. Tiss. Int. 44, 200-208.

Driessens F. C. M. and Verbeeck R. M. H. (1986) The dynamics of bone mineral in some vertebrates. Z Nuturf. 41C, 46847 1. Eanes E. D. and Meyer J. C. (1977) The maturation of crystalline calcium phosphates in aqueous suspensions at physiologic pH. Cajc. 7’iss. Rex 23,.259-269. Ecarot-Charrier B.. Glorieux F. H.. Van Der Best M. and Fereira G. (198j) Osteoblasts isolated from mouse calvaria initiate matrix mineralization in culture. J. Cell Biol. %,63M3.

Engel J., Taylor W., Paulsson M., Sage H. and Hogan B. (1987) Calcium bindin;% domains and calcium-induced conformational transition of SPARC/BM-40/osteonectin, an extracellular glycoprotein expressed in mineralized and nonmineralized tissues. Biochemie 26, 695-965. Fisher L. W., Eanes E. D., Denholm L. J., Heywood B. R. and Termine J. D. (1987) Two bovine models of osteogenesis imperfecta exhibit decreased apatite crystal size. Calc. Tiss. Inf. 40, 282-285.

Gerstenfeld L., Chipman 13.D., Glowacki J. and Lian J. B. (1987) Expression of differentiated function by mineralizing cultures of chicken osteoblasts. Devf Biol. 122,4=. Glimcher M. J. (1981) On the form and function of bone: from molecules to organs. Woltl’s law revisited, 1981. In The Chemistry and Biology of Mineralized Connective Tissues (Ed. Veis A.). Elsevier/North Holland, NY.

Gundberg C. M., Hauschka P. W., Lian J. B. and Gallop P. M. (1984) Osteocalcin: isolation, characterization, and detection. Meth. Enzym. 107, 516544 Hauschka P. W. (1985) Osteocalcin and its functional do-

21

mains. In The Chemistry

and Biology of Mineral Tissue (Ed. Butler W. T.), pp. 149-158. Ebsco Media Inc.

Birmingham, AL. Howell D. S., Pita J. C., Marquez J. F. and Madruga J. E. (1968) Partition of calcium, phosphate, and protein in the &id -phase aspirated by &lcif$ng sites ih epiphyseal cartilage. J. clin. Invest. 47. 1121-l 132. Lee S. L: and Glimcher M. j. (1981) Purification, composition, and “P-NMR spectroscopic properties of a noncollageneous phosphoprotein isolated from chicken bone matrix. Calc. Tiss. Int. 33, 385-394. Linde A., Lussi A. and Crenshaw M. A. (1989) Mineral induction by immobilized polyanionic proteins. Calc. Tiss. Znt. 44,28&295.

Meyer J. L. and Eanes E. D. (1978) A thermodynamic analysis of the secondary transition in the spontaneous precipitationofcalciumphosphate. Calc. Tiss. Res.25,20%216. Murphy J. and Riley J. P. (1962) A modified single solution method for the determination of phosphate in waters. Analytica

chim. Acta 27, 31-36.

Nawrot C. F., Cambell D. J., Schroeder J. K. and Valkenburg M. V. (1976) Dental phosphoprotein-induced formation of hydroxyapatite during in vitro synthesis of amorphous calcium phosphate. Biochemie 15,3445-3449. Poser J. W. and Price P. A. (1979) A method for decarboxylation of y-carboxyglutamic acid in proteins. J. biol. Chem. 254,431436. Price P. A. and Williamson M. K. (1985) Primary structure of bovine matrix Gla protein, a new vitamin K-dependent bone protein. J. biol. Chem. 260, 14,971-14,975. Price P. A., Otsuka A. S., Poser J. W., Kristaponis J. and Raman N. (1976) Characterization of a y-carboxyglutamic acid-containing protein from bone. Proc. natn. Acad. Sci. U.S.A. 73, 1447-1459.

Robinson R. (1923) The possible significance of hexosephosphate esters in ossification. Biochem. J. 17, 283-293. Rimberg R. W., Werness P. G., Lollar P., Riggs B. L. and Mann K. G. (1985) Isolation and characterization of native adult osieon&tin. J. biol. Chem. 260, 2728-2736. Romberg R. W., Wemess P. G., Riggs B. L. and Mann K. G. (1986) Inhibition of hydroxyapatite crystal growth by bone-specific and other calcium-binding proteins. Biochemie 25, 117&l 180.

Sobel A. T. and Laurence P. A. (1960) Crystal growth in mineralizing tissues. Biochim. biophys. Acta 41, l-8. Tenenbaum H. C. and Heersche J. N. M. (1982) Differentiation of osteoblasts and formation of mineralized bone in vitro. Calc. Tiss. Int. 34, 7679.

Termine J. D., Eanes E. D. and Conn K. M. (1980a) Phosphoprotein modulation of apatite crystallization. Calc. Tiss. Int. 31, 247-251. Termine J. D., Belcourt A. B., Miyamoto M. S. and Konn K. M. (1980b) Properties of dissociatively extracted fetal tooth matrix proteins. II. Separation and purification of fetal bovine dentin phosphoprotein. J. biol. Chem. 255, 9769-9772.

Tennine J. D., Kleinman H. K., Whitson S. W., Conn K. M., McGarvey M. L. and Martin G. R. (1981a) Osteonectin, a bone-specific protein linking mineral to collagen. Cell 26, 99-105. Termine J. D., Belcourt A. B., Conn K. M. and Kleinman H. K. (198lb) Mineral- and collagen-binding proteins of fetal calf bone. J. biol. Chem. 256, 10,403-10,408. Termine J. D., Robey P. G., Fisher L. W., Shimokawa H., Drum M. A., Conn K. M., Hawkins G. R., Cruz J. B. and Thompson K. G. (1984) Osteonectin, bone proteoglycan, and phosphophoryn defects in a form of bovine osteogenesis imperfect. Proc. natn. Acad. Sci. U.S.A. 81, 2213-2217. Veis A. (1985) Phosphoproteins of dentin and bone. Do they have a role in matrix mineralization? In The Chemistry and Biology of Mineralized Tissue (Ed. Butler W. T.), pp. 170-176. Ebsco Media Inc., Birmingham, AL.