Noncollagenous proteins of rat compact bone.

In order to obtain a comprehensive overview of the noncollagenous proteins (NCPs) of bone matrix, the NCPs were extracted from rat compact bone and fractionated using methods aiming to prevent artifactual degradation and losses of protein. The NCP content of rat bone was found to be similar to that of rat dentin in several respects but different in others. The soluble NCPs of bone fell into four categories: acidic glycoproteins, gamma-carboxyglutamate-containing proteins, phosphoproteins, and proteoglycans. With the exception of the gamma-carboxyglutamate-containing proteins, the majority of NCPs had apparent molecular weights exceeding 50,000. As in rat dentin, several gamma-carboxyglutamate-containing proteins could be demonstrated in rat bone. Earlier studies have only taken one molecular species into consideration. No highly phosphorylated phosphoprotein could be demonstrated in bone. However, at least two phosphoproteins with a low degree of phosphorylation were found to be present. No plasma proteins could be demonstrated in any of the chromatographic fractions from the EDTA extracted NCPs by means of double diffusion. The NCPs, remaining firmly associated with the collagenous matrix after thorough demineralization and extraction, were analyzed after CNBr and collagenase degradation of the matrix. Much smaller amounts of phosphoprotein were recovered after CNBr digestion than reported earlier. Collagenase digestion released small amounts of acidic glycoprotein, phosphoprotein, and proteoglycan. The results give additional evidence that this small remainder might be explained, not by any covalent linkage to collagen, but by an inefficient extraction.


Goteborg, Sweden
In order to obtain a comprehensive overview of the noncollagenous proteins (NCPs) of bone matrix, the NCPs were extracted from rat compact bone and fractionated using methods aiming to prevent artifactual degradation and losses of protein.
The NCP content of rat bone was found to be similar to that of rat dentin in several respects but different in others. The soluble NCPs of bone fell into four categories: acidic glycoproteins, y-carboxyglutamate-containing proteins, phosphoproteins, and proteoglycans. With the exception of the y-carboxyglutamate-containing proteins, the majority of NCPs had apparent molecular weights exceeding 50,000. As in rat dentin, several y-carboxyglutamate-containing proteins could be demonstrated in rat bone. Earlier studies have only taken one molecular species into consideration. No highly phosphorylated phosphoprotein could be demonstrated in bone. However, at least two phosphoproteins with a low degree of phosphorylation were found to be present. No plasma proteins could be demonstrated in any of the chromatographic fractions from the EDTA extracted NCPs by means of double diffusion.
The NCPs, remaining firmly associated with the collagenous matrix after thorough demineralization and extraction, were analyzed after CNBr and collagenase degradation of the matrix. Much smaller amounts of phosphoprotein were recovered after CNBr digestion than reported earlier. Collagenase digestion released small amounts of acidic glycoprotein, phosphoprotein, and proteoglycan. The results give additional evidence that this small remainder might be explained, not by any covalent linkage to collagen, but by an inefficient extraction.
Type I collagen constitutes about 90% of the protein matrix in bone and dentin. The remaining part consists of different groups of NCPs': phosphoproteins, Gla-containing proteins, acidic glycoproteins, proteoglycans, and, possihly, serum proteins. Due to its higher proportion many investigators have studied the function of collagen in the process of tissue mineralization and several authors are of the opinion that collagen might have a nucleating function for initial crystal nuclei.
On the other hand, the noncollagenous components possess such chemical characteristics that they have been suggested to influence calcification in a number of ways (1, 2). Thus NCPs, including proteoglycans, have been proposed to inhibit or nucleate mineralization, to govern the type of calciumphosphate mineral that is formed, and to act as rate and size limiters for mineral crystal formation. Some of these functions may be performed in intimate functional association with collagen. The NCPs may also influence extracellular collagen fiber formation in osteoid and predentin. Changes in the characteristics and properties of NCPs may thus be causative factors in different metabolic bone disorders.
Detailed analyses of some specific NCPs have been performed, e.g. the a2HS-glycoprotein, Gla-containing protein, by some authors referred to as osteocalcin, the sialoprotein of bone, and the PP-Hz of dentin, and these studies have given increased insight into the possible mechanisms whereby such components may influence the mineralization process. However, reviews of the NCP content in bone (3) reveal a complicated picture of the total NCP composition of bone with numerous ill-defined components whose qualitative and quantitative interrelationships are not well understood. It is difficult to obtain such an overview, since the literature is remarkably devoid of any studies using a straightforward chromatographic approach subsequent to demineralization and extraction, the exception being the recent study of Termine and coworkers of NCPs of fetal calf bone (4). One of the aims of this study was thus to give a comprehensive picture of the total NCP content of compact bone from rat in order to provide a basis for further biochemical characterization of specific NCP components and their metabolism during diverse physiological and pathological conditions. One problem with extracting and analyzing connective tissue proteins is the risk caused by the existence of protease activities intrinsic to the tissue. Such problems would be expected to be especially pronounced with the extended decalcification periods usually necessary for hard tissues. Another factor that may cause artifactual degradation of tissue proteins is a too thorough disintegration of the tissue by mechanical means, e.g. by milling into a fine powder. Earlier findings in bone of numerous proteinaceous components of small molecular size and the presence of hydroxyproline in a number of fractions, where its existence would be unexpected, may be taken as support for this.
In fact, in an earlier study of dentin NCPs strong evidence was obtained that the use of certain precautionary measures is advantageous in preventing artifactual degradation and losses of protein (5). No grinding of the tissue fragments was performed. An initial extraction, with 4 M GdmCl was made to remove NCPs not associated with the tissue apatite, such as soluble proteases and contaminating proteins from blood and some soft tissues. Both the GdmCl extraction and the subsequent EDTA decalcification were performed at 4 "C and with several protease inhibitors present in the solutions. All chromatography was run in the cold. Similar extraction conditions have also been used by Termine and co-workers (4,6). The conclusion was drawn that such procedures provided an improvement over earlier studies, since a limited number of major dentin NCPs were obtained and these had, with two exceptions, apparent molecular weights higher than 50,000. It is thus evident that similar precautionary measures should be taken when attempting to re-examine bone NCPs.
A controversial question is whether some NCPs occur in bone and dentin matrix as "collagen adjuncts" (3), i e . covalently, or at least very strongly, bound to the collagen matrix. It has been argued that the NCPs of bone and dentin exist in two pools, one set extractable at neutral pH during or following decalcification, and the other set closely associated with collagen and released only after degradation of the collagen. In bovine bone the collagen-associated acid-insoluble residue, after CNBr degradation, has been reported to account for 2.5% of the EDTA-insoluble matrix, i e . about one-fifth of the total NCP (7).
The existence of such a pool of collagen adjuncts has been criticized. Cohen-Sold et al. (8) found that 98% or more of the phosphoproteins containing 0-phosphoserine or O-phosphothreonine were solubilized by extraction of chicken bone. Collagen itself was postulated to contain significant amounts of bound phosphate, presumably in the form of y-glutamyl phosphate. Smith and Leaver (9) were unable to demonstrate any phosphoprotein remaining after extensive demineralization and extraction of human dentin matrix, and the same result was found with rat incisor dentin matrix (10). After solubilization with CNBr or with bacterial collagenase, unusually small amounts of phosphoprotein were detected, and evidence was found that even this small quantity of phosphoprotein was not covalently bound to collagen (10). Even though the situation in dentin matrix thus seems to be clarified, the existence of a separate pool of NCPs covalently linked to bone collagen deserves further study, since such a pool has been implied as having a special function in mineralization (1).
In the present investigation the content of NCPs in rat compact bone has been reassessed using methods aiming at the prevention of artifactual degradation and losses of protein.
By these means an inclusive outline of the NCPs was obtained and may serve as a basis for further studies in this area. It is shown that, except for the Gla-containing proteins, the majority of NCPs in rat bone have an apparent M, greater than 50,000. The results reported here show that, like in rat dentin (ll), several Gla-containing proteins are present. In contrast to dentin, however, no PP-H could be demonstrated, but the presence of at least two phosphoproteins with a low degree of phosphorylation (PP-L) was shown. Results obtained also show that the amount of NCPs strongly associated with the collagen is far less than has been reported earlier.

EXPERIMENTAL PROCEDURES AND RESULTS~
In preliminary experiments it was found that GdmCl extracts of rat compact bone contained several protein components, although present in small amounts. It was reasoned, however, that this primarily represented contaminating material, such as blood and remaining cellular elements, and material not closely associated with the tissue mineral. This notion is supported by earlier findings that little NCP, notably some proteoglycan, is extracted from rat dentin without prior demineralization (5). The GdmCl extracts were thus discarded. T o study the content of soluble NCPs in rat compact bone matrix, this was totally demineralized in the cold by several changes of EDTA solution containing protease inhibitors (Fig. 3). The NCPs are extracted simultaneously and can be recovered from the combined EDTA solutions after desalting.
On an average the EDTA-extracted NCPs accounted for 6% of the organic matrix that was accounted for, i e . the EDTAextracted NCPs plus the extensively purified collagen matrix. DEAE-Sepharose Chromatography of Soluble NCPs-Several UV-absorbing fractions were detected when the total NCP extract was chromatographed on DEAE-Sepharose ( Fig.   1). No protein could be recovered from peak A . Peak B gave essentially the same analyses aspeak C, and it was concluded that the high peak B was due to remaining EDTA. EDTA elutes a t this ionic strength and is difficult to remove from acidic connective tissue protein (5).
Fraction C, eluting around 0.15 M NaCl in the linear gradient, was found by SDS-PAGE to contain one major component, with an apparent M, of about 60,000 in addition to a few other components (Fig. 4). The amino acid analysis of fraction C demonstrated the presence of acidic glycoprotein with fairly high amounts of aspartate, glutamate, alanine, and glucosamine ( Table I). Fraction C thus primarily contained one major and one minor component that belong to the group of acidic glycoproteins.
Fractions D, E, and F, eluting between 0.20-0.25 M NaC1, all contained a major component running with the front on 10% SDS-PAGE (Fig. 4). Fraction D contained in addition some material with the same electrophoretic properties as fraction C. Little material, in addition to the front band, was found in fractions E and F. In the amino acid analyses, small amounts of hydroxyproline were found in all three fractions ( Table I).
The only organic phosphate detectable in the effluent from the DEAE chromatography of rat bone NCPs was found in the biphasic fraction G, eluting at 0.30-0.35 M NaCl (Fig. 1). Also the phosphate curve had a biphasic appearance but, in contrast to the UV absorbance curve, with the second peak invariably higher than the fist. Fraction G contained several Coomassie-staining components as seen on 10% SDS-PAGE (Fig. 4). When the material in the phosphoprotein-containing fraction G was subjected to electrophoresis on 15% polyacrylamide gels and stained with Alcian blue, most of the stainable material migrated with the front. In addition, one band was seen in the middle of the gel and some Alcian blue staining material obviously did not enter the gel (Fig. 4)

Sephadex G-50 Chromatography of Soluble NCPs followed by DEAE-Sepharose Chromatography of 10K Fraction-
When the EDTA-extracted NCPs from rat compact bone were chromatographed on Sephadex G-50, four well separated peaks were obtained (not shown). The one eluting in the void volume ( Vo fraction) and the one eluting at the same position as globular proteins with an M, = 10,000 both contained protein. The other two UV-absorbing peaks, containing EDTA and salt, were devoid of any protein.
Since fractions D, E, and F of the total bone EDTA extract ( Fig. 1) were absent from a DEAE-Sepharose chromatogram of the VO fraction ( Fig. 6 in Miniprint), the components constituting these three peaks should be recovered in the 10K fraction. All protein in the 10K fraction migrated with the front on 10% SDS-PAGE, thus corresponding to the strongly stainable front bands in the gel runs of fractions D-F (Fig. 4).
No other Coomassie-stainable components were seen in the 10% SDS-PAGE of the 10K fraction. When increasing the gel concentration to 17.5%, the SDS-PAGE revealed one major and two minor bands (Fig. 4). The amino acid analysis of the 10K fraction showed the presence of hydroxyproline and high amounts of aspartate, glutamate, proline, and leucine ( Table  11).
The 10K fraction gave one major (yz), one intermediate (y3), and two minor (yl and y4) fairly well separated peaks, eluting in the interval 0.18-0.23 M NaC1, when chromatographed on DEAE-Sepharose using a shallow gradient (Fig. 2). These obviously corresponded to peaks D, E, and F in Fig. 1. Although no determinations of Gla were made in these experiments, the amino acid analyses of yl-y4 with hydroxyproline present and high amounts of aspartate, glutamate, and proline (Table 11) clearly show the close relationship to the Glacontaining proteins of bone (23, 24), by some referred to as "osteocalcin," and those of dentin (11). We have earlier demonstrated the presence of several Gla-containing proteins in dentin (ll), and it is clear that several components can be separated also from rat bone. Certain differences could be seen between the y-fractions. yl and y4 clearly contained serine, whereas at least yz was devoid of this amino acid. y4 had a higher content of aspartate, proline, and leucine, whereas the other three y-fractions were similar in this respect. When the contamination between the different y-fractions is taken into account, this difference seems to be compensated for by the lower amount of lysine, threonine, isoleucine, and possibly phenylanine in y4.

DISCUSSION
As in rat dentin (5) the main classes of NCPs in rat compact bone tissue were found to be acidic glycoproteins, Gla-containing protein, phosphoprotein, and proteoglycan. On the other hand, in spite of the many similarities between these tissues in general composition, e.g. the mineral phase and the collagen type, certain differences were found. It is thus obvious that a comparison between these related tissues might elucidate some of the fundamental processes involved in the calcification mechanism.
Although several earlier studies have been made of NCPs of bone these have either concentrated on specific components of the tissue or utilized techniques that can now be regarded as not adequate. It is known e.g. that dentin (26) and bone (27) tissues contain protease activities and thus every effort should be taken to minimize protein loss and degradation.
Both rat incisor dentin (5,ll) and rat compact bone contain large quantities of Gla protein. Although no specific determination of the amino acid Gla was made in the fractions, the general amino acid composition and the chromatographic behavior on Sephadex G-50 and DEAE-Sepharose (Fig. 2) clearly show the intimate relationship of the fractions 71-74 with the Gla-containing proteins of bone from other species (23, 24) and rat dentin (11). In addition, in current studies Noncollagensus Proteins of Bone (data not shown) we have demonstrated the virtually complete immunological cross-reactivity between the total Gla protein fractions of rat bone and rat incisor dentin using a radioimmunoassay technique developed for rat bone Gla protein (generously provided by Dr. P. Hauschka, Harvard University) and an immunosorbent assay with antiserum against the total Gla protein fraction of rat incisor dentin.
We have earlier shown that several Gla-containing proteins exist in rat incisor dentin (11). The present study shows that this is the case also in bone. The amino acid analyses and the chromatographic behavior of the four Gla protein fractions yl-y4 (Fig. 2, Table 11) indicate, as in dentin, that these are not simply degradation products of each other. Whether they represent different gene products cannot be settled at the present stage. Extensive studies of the Gla-containing protein of bone, by some authors referred to as osteocalcin, have been performed, but the existence of several such proteins has not been taken into account hitherto. Careful chromatography should be able to resolve several Gla proteins from other species as well.
The presence in bone of phosphoprotein, with a fairly low degree of phosphorylation, and containing both O-phosphoserine and 0-phosphothreonine, was fist demonstrated by Spector and Glimcher (28,29). Several PP-L occur as minor NCP constituents of rat incisor dentin Analysis data for a phosphoprotein from chicken bone (30) and three phosphate-containing proteins from fetal calf bone (4), all belonging to the PP-L class, have been published. The present study shows that rat compact bone contains at least two phosphoproteins with a low degree of phosphorylation.
One striking difference between bone and dentin is the absence in the former of significant amounts of PP-H, the major NCP in dentin from rat incisor (5) and calf (7).4 This was shown in the DEAE-chromatography (Figs. 1 and 6) but also by the absence of any precipitate in the initial step of the sequential precipitation procedure (Fig. 1). The presence of EDTA in extraction and demineralization solutions should inhibit any alkaline phosphatase activity (31), so it is unlikely that the absence of any component with a higher degree of phosphorylation than that of PP-L could be explained by the high nonspecific alkaline phosphatase activity present in bone. In addition, the alkaline phosphatase inhibitor l-p-bromotetramisole (0.1 mM; Ref. 32) was added to the solutions in one series of preparation without any significant difference in the results (data not shown). The absence of PP-H in bone, when present in the closely related dentin, is paradoxical and is not understood at present.
In addition to the Gla protein fraction, the PP-L components, and the proteoglycan fraction, the major NCP of rat compact bone was found to be an acidic glycoprotein eluting from the DEAE-Sepharose at about 0.2 M NaCl (fraction C in Figs. 1 and 6). AS seen on SDS-PAGE, this had an apparent M, of around 60,000.
As with dentin NCPs (12) it was possible to prefractionate bone NCPs into pools by a sequential precipitation procedure. However, it is unclear to what extent this technique is of value for the analysis of bone NCPs, since a major component with an apparent M , around 60,000 and eluting at about 0.2 M NaCl from DEAE-Sepharose was found in both the HCOOH precipitate and the HCOOH supernatant. The absence of any PP-H in bone, which should have been recovered in the Ca2+precipitation step, decreases the applicability of this technique.
Several authors have demonstrated the presence of serum proteins in bone such as the azHS-glycoprotein (33, 34) and albumin (35). According to Termine et al. (4) some ~z H Sglycoprotein may still remain in fetal calf bone matrix after an initial extraction with 4 M GdmC1. In contrast, no serum proteins, including the azHS-glycoprotein, could be demonstrated in the EDTA extract from rat compact bone by means of double diffusion against antiserum to rat serum proteins.
Nor was any protein found with the expected M, of mHSglycoprotein (4, 36). Whether this was due to a tissue difference, caused by a difference in species and/or degree of maturation (37), or to a more efficient initial-extraction in the present study is not known, but it must be concluded that no serum protein seems to be strongly associated with the mineral phase in rat compact bone. Earlier studies of the organic matrix of bone and dentin, performed in order to understand the local mechanisms in the tissue at the site of mineral formation, concentrated on the major component in the matrix, collagen. In effect, demineralization solutions were usually rejected and thus the easily solubilized portion of the NCPs was discarded. When the collagen portion was studied, it was found that a certain amount of NCPs, especially phosphoprotein, was present and it was suggested that this portion represented NCPs covalently attached to the collagen. When the major NCP portion, the soluble NCPs, later attracted considerable attention it was postulated that in mineralized tissues two pools of NCPs in fact exist (7) and that the "collagen adjuncts" (3) would be of a major importance for mineral formation.
The strategy for studying collagen-associated NCPs has been to degrade the collagen by bacterial collagenase, periodate, or CNBr. With collagenase all remaining NCPs of the collagen matrix would be recovered, while in the latter two procedures precipitation with weak acid has been used in order to isolate phosphoprotein.
In rat bone the acid-insoluble remainder after CNBr degradation was found to be much less (0.19%) than the 2.5% of the organic matrix reported for bovine bone (7). In addition, the hydroxyproline content, taken as representing collagen CNBr fragments present in the acid precipitate, was also lower, 10 compared to 30 residues per lo3 amino acid residues (Table IV and Ref. 7). Basically the same results were obtained in a similar study of rat incisor dentin (10) when comparing with data published earlier for bovine dentin. The finding that not only phosphoprotein but also acidic glycoprotein and proteoglycan were released by the collagenase digestion of rat bone matrix (Fig. lo), in spite of the previous extensive demineralization and extraction procedures, strongly supports the notion that the findings of collagen adjuncts in bone and dentin may be explained by an incomplete extraction of NCPs due to electrostatic interaction and steric hindrance (10).
The analysis of bone of different degree of maturity may yield different results. Thus e.g. the levels of anHS-glycoproteins have been reported to be about 15 times higher in fetal human bone than in adult cortical bone (37). Immature bone, or woven bone, differs in morphology from mature compact bone and the mechanism for its formation may, in fact, be different, since it has been postulated that the initial mineral crystals are formed inside extracellular mineralization vesicles (38). Such vesicles are not found when the immature bone is rebuilt. A parallel to this situation is the difference between mantle and circumpulpal dentin formation (2). Data for NCPs in fetal bone thus, presumably, pertain to calcification of the initial bone tissue formed during fetal life, while the proteins studied here may be of importance for the formation of the continuously remodelling compact bone that is present at later stages.