Differential effects of transforming growth factor-beta on the synthesis of extracellular matrix proteins by normal fetal rat calvarial bone cell populations.

To determine the effects of transforming growth factor-beta (TGF-beta) on the different cell types that exist in bone, cell populations (I-IV), progressively enriched in osteoblastic cells relative to fibroblastic cells, were prepared from fetal rat calvaria using timed collagenase digestions. TGF-beta did not induce anchorage-independent growth of these cells, nor was anchorage-dependent growth stimulated in most populations studied, despite a two- to threefold increase in the synthesis of cellular proteins. In all populations the synthesis of secreted proteins increased 2-3.5-fold. In particular, collagen, fibronectin, and plasminogen activator inhibitor synthesis was stimulated. However, different degrees of stimulation of individual proteins were observed both within and between cell populations. A marked preferential stimulation of plasminogen activator inhibitor was observed in each population, together with a slight preferential stimulation of collagen; the effect on collagen expression being directed primarily at type I collagen. In contrast, the synthesis of SPARC (secreted protein acidic rich in cysteine/osteonectin was stimulated approximately two-fold by TGF-beta, but only in fibroblastic populations. Collectively, these results demonstrate that TGF-beta stimulates matrix production by bone cells and, through differential effects on individual matrix components, may also influence the nature of the matrix formed by different bone cell populations. In the presence of TGF-beta, osteoblastic cells lost their polygonal morphology and alkaline phosphatase activity was decreased, reflecting a suppression of osteoblastic features. The differential effects of TGF-beta on bone cell populations are likely to be important in bone remodeling and fracture repair.

Specific immunohistochemical localization of osteonectin and collagen types I and III in fetal and adult porcine dental tissues.

Affinity-purified antibodies have been used in combination with the peroxidase-antiperoxidase technique to study the distribution of osteonectin and collagen types I and III in porcine dental tissues. Tissue sections (2 mm thick), including unerupted (fetal) or erupted (adult) teeth, were fixed in periodate-lysine-paraformaldehyde, demineralized in 12% w/v ethylenediaminetetraacetic acid, and after embedding, 6 micron sections were prepared for immunolocalization. Strong staining for osteonectin was observed in dentine of unerupted teeth and in the associated alveolar bone. Light to moderate staining was observed in the dental pulp, stratum intermedium, stellate reticulum, and the reticular elements in the endosteal spaces. In erupted teeth, osteonectin staining in dentine was concentrated around dentinal tubules and the associated alveolar bone stained with variable intensity. Cementum was poorly stained. However, the periodontal ligament and reticular material in the endosteal spaces showed moderate to strong staining. Weaker staining was apparent in the pulp and lamina propria of the gingiva. In comparison, type I collagen showed a similar distribution to osteonectin in both fetal and adult tissues, whereas type III collagen was generally restricted to the periodontal ligament, reticular elements of the endosteal spaces, and Sharpey's fibers in bone and cementum. Both odontoblast and ameloblast layers in fetal tissues stained for osteonectin and type III collagen.

Immunohistochemical localization of SPARC (osteonectin) and denatured collagen and their relationship to remodelling in rat dental tissues.

To study this relationship, specific antibodies were used to determine the distribution of these proteins in mature rat dental tissues. Staining for SPARC with affinity-purified polyclonal antibodies was prominent throughout molar and incisor ligaments, endosteal tissue, dental pulp and muscle. More moderate staining was observed in other soft tissues including the lamina propria of gingiva, whereas the staining of demineralized bone was weak and in dentine was barely detectable. A monoclonal antibody (MBP 322), raised against a denatured form of a small collagenous bone protein, reacted strongly with osteoblastic cells but more moderately with alveolar bone. A strong reaction, indicative of unfolded collagen, was also evident throughout the dental pulp and molar ligament, whereas in the incisor ligament staining was largely restricted to the tooth-related half. Moderate staining with this antibody was also observed in other soft tissues and in dentine. The monoclonal antibody also stained the nuclei of certain cells; notably, whereas most of the fibroblasts in the tooth-related half of the incisor ligament were stained strongly, only occasional nuclei of fibroblasts in the molar ligament and in the bone-related half of the incisor ligament showed immunoreactivity. The differential staining of nuclei provides evidence for phenotypic differences between fibroblast populations within these tissues. The prominence of SPARC in the ligament tissues is consistent with their embryonic characteristics, whereas unfolded collagen recognized by the MBP 322 antibody may indicate sites of rapid collagen remodelling.

Identification of lysyl oxidase and TRAMP as the major proteins in dissociative extracts of the demineralized collagen matrix of porcine dentine.

Carbonated apatite (dahllite) is formed within and between collagen fibrils in the mineralization of connective tissues. However, the mechanism of crystal nucleation at these sites has not been resolved. To identify non-collagenous proteins that may be involved in the nucleation process we have utilized a dissociative extraction procedure to isolate proteins associated non-covalently with the de-mineralized collagen matrix of dentine isolated from tooth roots of adult porcine incisors. Following extraction of dentine fragments with 4M GuHCl (G1-extract) and 0.5M EDTA (E-extract), de-mineralized collagen matrix-associated proteins were isolated with a second series of extractions with 4M GuHCl (G2-extract). Analysis of the G2-extracts on SDS-PAGE revealed two major 32 kDa and 24 kDa protein bands, comprising > 80% of the extracted non-collagenous proteins. The 32 kDa protein was purified by FPLC on hydroxyapatite and Mono Q resins, followed by HPLC reverse-phase chromatography. Small amounts of 26 kDa and 6 kDa proteins, which appear to represent proteolytically processed, disulphide-linked fragments of the 32 kDa protein, co-eluted with the major protein. The 32 kDa protein was identified as lysyl oxidase from amino acid sequence analysis of a 13 kDa CNBr peptide obtained from protein purified by preparative electrophoresis on SDS-PAGE. Fractionation of the 24 kDa protein on FPLC Mono Q resin generated < 5 closely eluting protein peaks. The proteins from these peaks were similar in size, staining properties, amino acid composition and CNBr digestion patterns. Each protein was immunoreactive with antibodies raised against a tyrosine-rich acidic matrix protein (TRAMP), reported previously to co-purify with lysyl oxidase. These studies, therefore, show that lysyl oxidase, which is important in collagen cross-link formation, and proteins with properties of TRAMP, a protein that can modulate collagen fibrillogenesis, are the major proteins in dissociative extracts of de-mineralized porcine dentine.

Biosynthesis of bone proteins by fetal porcine calvariae in vitro. Rapid association of sulfated sialoproteins (secreted phosphoprotein-1 and bone sialoprotein) and chondroitin sulfate proteoglycan (CS-PGIII) with bone mineral.

To study the biosynthesis of bone proteins, fragments of fetal porcine calvariae were cultured in the presence of 50 micrograms/ml ascorbate and 10 mM beta-glycerophosphate and individual cultures labeled for either 4 h or 48 h with [35S]-methionine, Na2[35SO4], Na3[32PO4] or [14C]-glycine plus [14C]-proline. The radiolabeled proteins in tissue extracts were obtained by sequential extraction with 4 M GuHCl (G1-extract), 0.5 M EDTA (E-extract), and again with 4 M GuHCl (G2-extract) and analyzed together with the radiolabeled proteins secreted into the medium. SPP-1 (secreted phosphoprotein 1, osteopontin) was the major non-collagenous protein deposited into the bone matrix, with lesser amounts of BSP (bone sialoprotein), osteocalcin and chondroitin sulfate proteoglycans (CS-PG II and CS-PG III). SPP-1 was also the major phosphorylated protein and was recovered, together with several fragmented forms, almost entirely in the demineralizing extracts. Moreover, approximately one-half of the [35SO4] incorporated into E-extract proteins was present in SPP-1, the remainder being incorporated into PGs with smaller amounts associated with BSP. Over 65% of the [35SO4] in the proteoglycans of the demineralizing extracts was recovered in the small CS-PG III with less than 35% in CS-PG II, the bone homologue of DS-PG II (decorin). In contrast, CS-PG II was the predominant small proteoglycan in culture media and in guanidine extracts. Some sulfated BSP was also observed in guanidine extracts and small amounts appeared to bind to collagen. Radiolabeled SPARC (osteonectin), a prominent protein of fetal porcine bone, was not detected in the mineralized bone tissues but was prominent in the culture medium. These results demonstrate that following secretion, the major proteins expressed by osteoblastic cells are initially incorporated into different tissue compartments, with most of the sulfated sialoproteins and CS-PG III associating rapidly with the hydroxyapatite crystals. The initial distribution of these proteins is of importance in the evaluation of their role in bone formation and mineralization.

Osteonectin is a minor component of mineralized connective tissues in rat.

Osteonectin is a major glycoprotein of porcine and bovine bones and teeth that is found associated with hydroxylapatite crystal surfaces. From the ability of osteonectin to bind calcium ions, it has been proposed as a possible nucleator of hydroxylapatite crystal formation. Analysis of hydroxylapatite-bound proteins of rat bone and dentine, however, has revealed that osteonectin represents only 2.5 +/- 1.5% of the hydroxylapatite-bound protein in long bones, 0.9 +/- 0.5% in calvariae, and less than 0.1% in incisor dentine of animals of different ages. Further, in vivo pulse-chase studies carried out in young adult rats have shown osteonectin to be synthesized at low levels in these tissues. Similarly, low levels of osteonectin were synthesized by rat calvarial cells in vitro. In contrast, fibroblastic cells from periodontal ligament and gingiva synthesized significantly greater amounts of osteonectin. These studies indicate that the low quantities of osteonectin in rat mineralized tissues are a consequence of low rates of formation rather than being due to rapid turnover. The virtual absence of osteonectin in incisor dentine correlates with the lack of peritubular dentine in rat, whereas the low osteonectin content of rat bones may reflect differences in their structure and biophysical properties compared with bones of larger mammals.

Mineral-binding proteoglycans of fetal porcine calvarial bone.

To provide a more definitive characterization of the hydroxylapatite-associated proteoglycans (HAPG) of bone, proteins were extracted from the mineralized matrix of fetal porcine calvaria with 0.5 M EDTA in the absence of guanidine HCl. The small proteoglycans obtained in the extract were fractionated by gel filtration on Sepharose CL-6B, purified by ion-exchange chromatography on Polyanion matrix (fast protein liquid chromatography), and then separated into three major populations of chondroitin sulfate proteoglycans by chromatography on hydroxylapatite, all in the presence of 7 M urea. Based on immunological and chemical properties, two classes of bone proteoglycan were resolved. In one class (HAPG1), the proteoglycan and specific CNBr-derived peptides cross-reacted with three monoclonal antibodies that recognize different epitopes of the protein core of bovine skin proteodermatan sulfate. The other class of proteoglycan included two species (HAPG2, HAPG3) which were not recognized by these antibodies. In addition, these proteoglycans did not stain with Coomassie Blue R-250 nor with silver stain nor did they bind to nitrocellulose membranes used in Western blots. However, the cationic dye Stains-all stained both HAPG2 and HAPG3; the protein cores of these proteoglycans were stained a characteristic turquoise blue, whereas the protein core of HAPG1 was stained pink. The average Mr values of the bone proteoglycans, from gradient sodium dodecyl sulfate-polyacrylamide gel electrophoresis were: HAPG1, 120,000, with a protein core (chondroitinase AC-digested) of 45,000; HAPG2 and HAPG3, 110,000, with protein cores of 37,000-38,000. On 15% polyacrylamide gel electrophoresis, the protein cores of HAPG2 and HAPG3 migrated with an Mr 30,000, while HAPG1 protein core was unchanged (Mr 45,000). Based on amino acid analysis, the protein chains of HAPG2 and HAPG3 appear to be identical, although minor differences in the relative amount of glucosamine were evident. In contrast, the composition of HAPG1 was quite different, with higher relative amounts of hydrophobic and aromatic residues and lower amounts of Asx and Glx. The presence of 360 residues/1,000 of Asx and Glx in HAPG2 and HAPG3 may in part explain the characteristic staining and immunotransfer properties of these proteoglycans. The unique amino-terminal sequence of HAPG2 (Asn-Pro-Val-Ala-Arg-Tyr-Gln), together with the immunological and chemical properties, would indicate that HAPG2 and HAPG3 are novel proteoglycans and, unlike HAPG1, could be unique to mineralized tissues.

Characterization of fetal porcine bone sialoproteins, secreted phosphoprotein I (SPPI, osteopontin), bone sialoprotein, and a 23-kDa glycoprotein. Demonstration that the 23-kDa glycoprotein is derived from the carboxyl terminus of SPPI.

Demineralizing extracts of porcine bone contain two large 66-80-kDa sialoproteins and smaller 20- and 23-kDa glycoproteins with similar chemical properties. Each protein was characterized following extraction from fetal calvariae and purification under dissociative conditions using Sepharose CL-6B, followed by fast protein liquid chromatography fractionation on hydroxyapatite and Mono Q resins. Unlike the large sialoproteins, the 20- and 23-kDa glycoproteins did not contain sialic acid. Nevertheless, affinity-purified antibodies raised against the 23-kDa protein recognized both the 20-kDa protein and a 67-kDa sialoprotein on immunoblots. These antibodies also immunoprecipitated a 60-kDa [35S]methionine-labeled protein produced by cell-free synthesis of calvarial bone mRNA, indicating that the smaller proteins were derived from the 67-kDa protein. The two sialoproteins were shown by primary sequence analysis to be secreted phosphoprotein I (SPPI, osteopontin, bone sialoprotein I) and bone sialoprotein (BSP, bone sialoprotein II). The SPPI was also characterized by its susceptibility to thrombin which produced a 23-kDa fragment, similar to the glycoprotein isolated, and a 30-kDa fragment. Amino-terminal sequence analysis of the 23- and 20-kDa proteins revealed that these proteins were derived from the carboxyl-terminal half of the SPPI molecule, the proteins showing 58% identity with human and rat, and 50% identity with mouse, SPPI sequences. Both the 23- and 20-kDa proteins appeared to be generated by the activity of an endogenous trypsin-like protease that cleaves at Arg-Ser (residues 155-156) and Lys-Ala (residues 182-183) bonds. Radiolabeling studies performed in vitro showed that the 23-kDa fragment was detectable in mineralized tissue within 4 h. The fragment was phosphorylated but, unlike SPPI, was not sulfated. The rapid generation of the 23-kDa glycoprotein and its presence in different bone tissues at different developmental stages indicate that the fragmentation of SPPI is important in bone formation and remodeling.

Characterization of porcine osteonectin extracted from foetal calvariae.

Osteonectin, extracted from foetal porcine calvariae with 0.5 M-EDTA, was purified to homogeneity by using gel filtration and polyanion anion-exchange fast protein liquid chromatography under dissociative conditions without the need of reducing agents. The purified protein migrated with an Mr of 40,300 on SDS/polyacrylamide gels and was similar to bovine osteonectin in both amino acid composition and in its ability to bind to hydroxyapatite in the presence of 4 M-guanidinium hydrochloride (GdmCl). However, unlike the bovine protein, porcine osteonectin did not bind selectively to hydroxyapatite when EDTA tissue extracts were used. In addition, purified porcine osteonectin did not show any apparent affinity for either native or denatured type I collagen, but did bind to serum albumin. Primary sequence analysis revealed an N-terminal alanine residue, with approximately one-half of the subsequent 35 residues identified as small hydrophobic amino acids and one-quarter as acidic amino acids. The only significant difference between the N-terminal sequences of the bovine and porcine proteins was the deletion of the tripeptide Val-Ala-Glu in porcine osteonectin. In contrast with bovine osteonectin, far-u.v.c.d. of porcine osteonectin revealed considerable secondary structure, of which 27% was alpha-helix and 39% was beta-sheet. Cleavage of the molecule with CNBr under non-reducing conditions generated five fragments, of which two major fragments (Mr 27,900 and 12,400) stained blue with Stains All, a reagent that stains sialic-acid-rich proteins/phosphate-containing proteins and/or Ca2+-binding proteins blue while staining other proteins pink. The 12,400-Mr fragment bound 45Ca2+ selectively, indicating a Ca2+-binding site in this part of the molecule. The 27,900-Mr fragment did not bind Ca2+, and since biosynthetic studies with 32PO4(3-) did not show phosphorylation of porcine osteonectin, this fragment is likely to be highly acidic. The incomplete cleavage of the molecule with CNBr and the ability of the molecule to regain its secondary structure after exposure to 7 M-urea are features consistent with the molecule having a compact structure that is stabilized by numerous disulphide bridges. The chemical and binding properties of porcine osteonectin are closely similar to the recently described 'culture shock', SPARC and BM-40 proteins, indicating that these are homologous proteins.

Identification of small collagenous proteins with properties of procollagen alpha 1 (I) pN-propeptide in fetal porcine calvarial bone.

Several small collagenous apatite binding (SCAB) proteins have been extracted from the mineralized matrix of fetal porcine calvarial bone. One protein (SCAB 3), released on demineralization of bone with 0.5 M EDTA, appears to represent the alpha 1 pN-propeptide that is normally released during proteolytic processing of type I procollagen. The 28 Kd protein, which stains blue with "Stains-all", is reduced to a 19 Kd fragment by bacterial collagenase digestion, but is not susceptible to cyanogen bromide. The amino acid composition, blocked amino-terminus and immunological properties are all consistent with properties of alpha 1 (I) pN-propeptide. Fractionation on hydroxylapatite in the presence of urea has revealed a nonbinding (SCAB 3a) and a binding (SCAB 3b) form. Extraction of the demineralized matrix of bone with 4 M GuHCl revealed a third form (G2-28) which was similar to SCAB 3a on hydroxylapatite chromatography but showed differences on FPLC "Mono Q" resin. The occurrence of these different forms of pN-propeptide in bone may be of significance in collagen fibril-associated hydroxylapatite formation and in the regulation of osteoblastic function during bone resorption.

Characterization of multiple forms of small collagenous apatite-binding proteins in bone.

A number of small (Mrs 25-28 kDa) collagenous apatite-binding (SCAB) proteins that stain blue with 'Stains-All' have been isolated from fetal porcine bone by sequential extractions with 4M GuHCl (G1), followed by 0.5M EDTA (E), and again with 4M GuHCl (G2). Following purification under dissociative conditions, two types of SCAB proteins both with approximately one-third of their structure being collagenous, were identified in the EDTA extract. One type, which appears to be a novel protein, was revealed in two forms (SCABs 1 and 2, Mrs 25 and 28 kDa) that were recognized by a monoclonal antibody (MBP-322). The second type, SCAB 3, was also present in two forms; one form (SCAB 3a) having a lower affinity for hydroxyapatite than the other (SCAB 3b). These proteins were resistant to CNBr and displayed the chemical and immunochemical properties of the alpha 1 pN-propeptide of type I collagen. A third form of the propeptide (G2-28K) was a prominent component of the second 4M GuHCl extract. The chromatographic properties of serum alpha 1 (I) pN-propeptide were similar to SCAB 3a, indicating that SCAB 3b and G2-28K are post-translationally modified forms of the propeptide produced by bone cells. These propeptides may provide a link between the hydroxyapatite and collagen fibrils, and also have the potential to suppress collagen synthesis during bone resorption.

Characterization of a monoclonal antibody recognizing small collagenous proteins in fetal bone.

A monoclonal antibody (MBP-322) that recognizes two small collagenous, apatite-binding (SCAB) proteins associated with the mineral phase of fetal bone, has been prepared. The SCAB proteins, which are quantitatively extracted from bone with EDTA, have been shown by immunotransfer analyses to have MrS of 28K and 25K and both were selectively degraded by bacterial collagenase. Amino acid analysis of the collagenase-digested protein revealed a hypro:pro:gly ratio of approximately 0.5:1:0.7 for both proteins and indicated that one-third of the protein could have a collagen-like sequence. The SCAB proteins, unlike other collagens and collagen fragments tested bound quantitatively to hydroxylapatite in the presence of 4M guanidine hydrochloride and appear to be unique to bone. The antibody, however, was not specific for the SCAB proteins and showed comparable immunoreactivity against denatured alpha 1 chains of types I, II and III collagens and the alpha 2 chains of types I and V collagens but not type IV collagen nor native collagens I-V. The epitope was further localized to the CB6 fragment in the alpha 1(I) chain and the CB5 fragment of alpha 1(III) chain, and was present in both the TCA and TCB fragments of alpha 2(I). Despite the immunological reactivity, the properties of the SCAB proteins were not consistent with their being derived from known collagen types. Immunocytochemical staining of permeabilized bone cells with MPB-322 showed a perinuclear, punctate staining pattern in most cells with some cells showing specific nuclear staining. In non-permeabilized cells, the antibody stained various sized spherical particles, many of which were closely associated with the cell surface. Immunoblots of cell proteins revealed a number of immunoreactive proteins sensitive to collagenase digestion including two proteins with MrS similar to the SCAB proteins. The MBP-322 antibody appears useful for identifying sequence homology in various collagens, and for recognizing denatured collagen and specific collagen fragments in tissues, as well as being important for the further characterization of the SCAB proteins.