Search

Bone mineral: update on chemical composition and structure

Bone sialoprotein (BSP) is one of the major non-collagenous glycosylated phosphoproteins of the extracellular matrix in bone. In vitro studies suggest that BSP may play important roles in the initiation and/or growth of calcium-phosphate crystals. To investigate the potential role of BSP in more complex in vivo environments, we implanted purified bovine BSP with type-I collagen as a carrier into surgically created rat calvarial defects and thoracic subcutaneous pouches. The responses to the implants were assessed by histochemistry, immunohistochemistry, in situ hybridization, quantitative real-time PCR, and biochemical analyses. BSP-collagen, but not collagen alone, elicited mineral deposition in the matrix of proliferating cells near the dura at days 4-5 followed by osteoblast differentiation and synthesis of new bone in the mid-portion of the calvarial defects. In contrast, implantation of BSP-collagen into subcutaneous pouches did not induce calcification or osteogenesis over the same experimental period. We explored the underlying mechanisms for the site-specific responses to BSP-collagen implants and found that higher levels of calcium content and alkaline phosphatase activity at the cranial site at days 2-5 were associated with the BSP-mediated calcification. We also found that BSP stimulated osteoblast differentiation through up-regulation of cbfa1 and osterix, key transcription factors of osteoblast differentiation, which occurred in the calvarial defects but not in the subcutaneous tissue. These results demonstrate that BSP stimulates calcification and osteogenesis in a site-specific manner, and that local environment and the specificities of responding cells may play critical roles in the function of BSP in vivo.

Apatite crystals of the calcified zone of the subarticular cartilaginous growth plates of the long bones of young growing chickens and calves were isolated by low temperature reaction with hydrazine and plasma ashing and examined by electron microscopy, electron diffraction and microprobe analysis, and computer-generated deconvolution of the spectra obtained by Fourier transform infrared spectroscopy. The crystal habit was that of wide, very thin, relatively long rectangular plates, which tended to form small clusters of crystals, possibly because reaction with hydrazine alone did not remove all of the organic matrix constituents. Further reaction with low power plasma ashing released more of the isolated crystals although to a lesser extent than was possible with bone. Stereograms of the small clusters showed that many of the crystals in the small isolated aggregates of crystals were bent and/or curved. Together with the resultant overlap of individual adjacent crystals, they also produced images of sharp, very dense lines, reminiscent of the electron-dense needle or rod-like appearances frequently observed by transmission electron microscopy of thin sections of calcified cartilage and thought to represent the habit of the apatite crystals. No true rod or needle-like crystals were observed in the isolated crystals. Although the overall general apatitic structure of the apatite crystals was similar to that of the apatitic crystals of bone, the individual crystals were significantly larger than those of bone from the same specimen, and there were small but significant differences in the concentrations of acid phosphate and carbonate groups and in their short range order.

Activating transcription factor-2 (ATF-2) is a basic region leucine zipper protein whose DNA target sequence is the widely distributed cAMP response element (CRE). We report here that mice carrying a germline mutation in ATF-2 demonstrated unique actions of ATF-2 not duplicated by other ATF/CREB family members. Mutant mice had decreased postnatal viability and growth, with a defect in endochondral ossification at epiphyseal plates similar to human hypochondroplasia. The animals had ataxic gait, hyperactivity and decreased hearing. In the brain, there were reduced numbers of cerebellar Purkinje cells, atrophic vestibular sense organs and enlarged ventricles. Unlike CREB alpha/delta-deficient mice whose main defect is in long-term potentiation, the widespread abnormalities in ATF-2 mutant mice demonstrate its absolute requirement for skeletal and central nervous system development, and for maximal induction of select genes with CRE sites, such as E-selectin.

A resolution-enhanced Fourier Transform Infrared (FTIR) Spectroscopic study of the CO3(2-) ion in pig enamel of increasing age and maturity has demonstrated the existence of four different, main carbonate locations. The major CO3(2-) site arises as a result of the substitution of CO3(2-) ions in the positions occupied by PO4(3-) ions in the apatitic lattice. In addition, two minor locations have been identified in positions in which the CO3(2-) ions substitute for OH- ions. The fourth carbonate group appears to be in an unstable location. Its concentration has been found to decrease with aging and maturation, during which there is a progressive increase in the amount of mineral deposited in the enamel. The distribution of the carbonate ions in the different apatitic sites varies randomly during the formation of the mineral phase in enamel and during its maturation. Although these changes have been shown to be related to changes in the composition of the mineral phase, a comparison of the parameters assessing the degree of crystallinity of the mineral phase from upsilon 2CO3(2-) and upsilon 4PO4(3-) infrared absorption data reveals a significant discrepancy related to the nonhomogeneous partition of the CO3(2-) ion in the mineral phase. After maximum mineralization is reached, the composition of the mature mineral phase is decidedly different than that of the initial mineral deposited; the changes affect principally the concentrations of Ca2+, OH-, and HPO4(2-) ions, but not the CO3(2-) ions.

Localization of a ∼66 kD glycosylated phosphoprotein during morphogenesis of the embryonic chick tibia has been accomplished using immunohistochemistry. Although initial expression of the tibial osteoblast phenotype is detected as early as stage 28.5, with the deposition of osteoid matrix beginning at stage 30, little or no immunoreactivity against the ∼66 kD glycosylated phosphoprotein is observed in pre-osteoblasts, osteoblasts, osteocytes, or in the uncalcified osteoid matrix during the early events of tibia development. Immunoreactivity was first observed at stage 32 when mineralization of the osteoid matrix is initiated. At this and all later stages, the phosphoprotein is located almost exclusively in the extracellular matrix at the mineralization front with essentially no detectable staining in the adjacent unmineralized osteoid matrix. Similarly, no cellular staining is observed when even the lightly mineralized extracellular matrix is strongly immunoreactive. Only scant immunostaining is present over the heavily mineralized regions, although demineralization of these areas with EDTA exposes a low intensity, punctate staining pattern. Additionally, cryosections of developing calvaria stained with this antiserum only display reactivity in regions of bone matrix undergoing mineralization. These localization studies support the hypothesis that this phosphoprotein is intimately associated with the process of bone matrix mineralization in the developing chick long bone.

In order to investigate the possible existence in biological and poorly crystalline synthetic apatites of local atomic organizations different from that of apatite, resolution-enhanced, Fourier transform infrared spectroscopy studies were carried out on chicken bone, pig enamel, and poorly crystalline synthetic apatites containing carbonate and HPO4 2− groups. The spectra obtained were compared to those of synthetic well crystallized apatites (stoichiometric hydroxyapatite, HPO4 2−-containing apatite, type B carbonate apatite) and nonapatitic calcium phosphates which have been suggested as precursors of the apatitic phase [octacalcium phosphate (OCP), brushite, and β tricalcium phosphate and whitlockite]. The spectra of bone and enamel, as well as poorly crystalline, synthetic apatite in thev 4 PO4 domain, exhibit, in addition to the three apatitic bands, three absorption bands that were shown to be independent of the organic matrix. Two low-wave number bands at 520–530 and 540–550 cm−1 are assigned to HPO4 2−. Reference to known calcium phosphates shows that bands in this domain also exist in HPO4 2−-containing apatite, brushite, and OCP. However, the lack of specific absorption bands prevents a clear identification of these HPO4 2− environments. The third absorption band (610–615 cm−1) is not related to HPO4 2− or OH− ions. It appears to be due to a labile PO4 3− environment which could not be identified with any phosphate environment existing in our reference samples, and thus seems specific of poorly crystalline apatites. Correlation of the variations in band intensities show that 610–615 cm−1 band is related to an absorption band at 560 cm−1 superimposed on an apatite band. All the nonapatitic phosphate environments were shown to decrease during aging of enamel, bone, and synthetic apatites. Moreover, EDTA etching show that the labile PO4 3− environment exhibited a heterogeneous distribution in the insoluble precipitate.

Poorly crystalline apatites (PCA) are the major mineral component of mineralized tissues in vertebrates. Their physical-chemical properties are, however, not very well known due to their relative instability and the difficulties to characterize nanocrystalline compounds. Several studies using spectroscopic techniques (Fourier transform infrared [FTIR]; 31P nuclear magnetic resonance [NMR]) have demonstrated the existence, both in precipitated and biological PCA, of labile non-apatitic environments of the mineral ions. These environments are involved in the high surface reactivity and evolution ability of PCA and they are believed to form a hydrated layer at the surface of the nanocrystals in aqueous media. The extent of the hydrated layer may vary considerably depending on the conditions of precipitation and maturation time. As PCA age, the decrease of the non-apatitic environments proportion is associated with a decrease of intracrystalline disorder and an increase of stable apatitic domains. For synthetic and biological apatites, the carbonation rate of the mineral and the uptake of essential or toxic trace elements can be related to the maturation processes. The mineral ions of the hydrated layer can be easily and reversibly substituted by other ions which can either be included in the growing stable apatite lattice during maturation or remain in the hydrated layer. In addition, the non-apatitic environments seem to be involved in the binding of soluble non-collagenic proteins. This phenomenon could be related to calcium phosphate formation; we showed that, at an albumin concentration close to that in human serum, this protein has an inhibitory effect on octacalcium phosphate crystallization on collagen in vitro.

environments very similar to those ofdicalcium phosphate dihydrate (DCPD) [56].”2. There were four errors in the section headed “Precursorphase(s) of apatitic nanocrystals”:(a) The last sentence of the second paragraph (left column,p. 1019) should have been: “We do not feel that theconclusion that OCP or “OCP-like” crystals exist in boneasdistinctcrystallinephasescanaccountfortheverylargenumber of non-apatite environments found in newlyformedbonemineralatanystageofnewboneformation ”.(b) The second sentence of the sixth paragraph (rightcolumn, p. 1019) should have been: “In addition, acareful examination of spectroscopic data obtained bydifferent techniques and an exploration of all spectro-scopic characteristics (not only special features) doesnot support the existence of separate crystalline phasesdifferent from those of apatite, even in the samples ofbone from the youngest animals.”(c) Thesentencebeginningonline24ofthesixthparagraph(right column, p. 1019) should have been: “As alreadynoted, these ions are not compatible with the formationof OCP crystals, and to date, no carbonate-containingOCP crystals or other non-apatitic phases have beendetected.”(d) The first sentence of the penultimate paragraph (leftcolumn, p. 1020) should have been: “We also notethe following reservations we have about the con-clusions reached by Mahamid et al. [69]: The FTIRband at 961 cm

High performance liquid chromatography was used to resolve O-phosphoserine, O-phosphothreonine, and O-phosphotyrosine as their fluorescent o-phthalaldehyde derivatives. By adjusting the buffer system, very small amounts of O-phosphothreonine could be detected and quantitated in the presence of very large amounts of O-phosphoserine. In addition, gamma-carboxyglutamic acid and glutamic acid were also separated and quantitated. Depending on the buffer used, various combinations of these amino acids could be resolved in a single run.

The amount of non-collagenous proteins is increased greatly during the pathological calcification of rat skin experimentally induced by dihydrotachysterol (DHT) and Ovalbumin (topical cutaneous calciphylaxis). This is accompanied by an increase in the total amount and concentrations of protein-bound serine phosphate [Ser(P)], threonine phosphate [Thr(P)] and gamma-carboxyglutamic acid (Gla), almost all of which can be extracted from the tissue and can be dissociated from collagen in 0.5M EDTA. The EDTA-soluble, non-collagenous proteins are rich in aspartic and glutamic acids, similar to the non-collagenous, EDTA-soluble proteins of bone, cementum and calcified cartilage, and quite distinct from those of dentin and enamel.

The crystallinity of bone mineral at different stages of maturation has been measured by quantitative X-ray diffraction methods. Crystallinity measurements were made on tibial mid-diaphyses from 17-day embryonic chicks, newly-formed periosteal bone from embryonic chicks, and density-fractionated bone from post-hatch chickens from 5 weeks to 2 years of age. For a given animal age and degree of mineralization, crystallinity increases with animal age, indicating that changes in bone mineral occur even after mineralization is complete or nearly complete.

An investigation of the molecular structure of bone mineral and synthetic calcium phosphates was carried out using radial distribution function (RDF) techniques. The X-ray data were collected using CuKα and MoKα radiation to insure the validity of the RDFs. Synthetic preparations of hydroxyapatite (HA) varying in their crystal size and crystallinity, and amorphous calcium phosphate (ACP), were studied, as well as bone samples from a 1-year-old chicken and 16-day embryonic chicks. Mixtures of embryonic bone and synthetic ACP were also investigated. The RDFs of bone and crystalline HA samples are similar in peak position, and show evidence of an atomic order extending to 2.5 nm and beyond. The RDF of ACP differs from that of HA, showing only short range order up to 0.9 nm, as well as small differences in peak shape. The decrease in intensity of the RDF function with increasing distance (r), observed with both HA and bone samples can be related to a decrease in crystallinity and crystal size. The RDF data indicate there is no significant amount of ACP in either very young or mature bone. The RDFs of the embryonic bone + synthetic ACP mixtures showed that a small amount of ACP can be readily detected in a sample of bone with a poorly crystalline mineral phase; from this we estimate the threshold for detection of ACP in bone to be 12% or less.

Ethylenediaminetetraacetate and hydrochloric acid (EDTA) (HCl) extracts of cementum were fractionated by molecular sieving, ion exchange chromotography, and reverse phase high precision liquid chromatography (HPLC). Nine fractions were isolated, all of which contained serine phosphate, threonine phosphate, and high concentrations of aspartic acid (asp) and glutamic acid (glu). Five of the fractions obtained by repeated HPLC consisted of a single band by SDS-PAGE; the others contained at least one other minor component. All of the protein bands stained with both Rhodamine B and alcian blue, the latter consistent with analytical determinations that demonstrated that the phosphoprotein component contained a significant amount of carbohydrate, including neuraminic acid.

Amino acid analyses of carefully dissected samples of the articular cartilage of postnatal chickens indicates the presence of significant amounts of a collagen with a low hydroxylysine content. It was found that 9 M LiCl solutions preferentially extract the components of a collagen whose components when identified by ion exchange chromatography and amino-acid analyses, established that they were derived from Type I collagen. Type II collagen was essentially insoluble in solutions of 9 M LiCl, but could be solubilized in 4 M CaCl2 after prior treatment of the cartilage with pepsin. The αl (II) components were then isolated and characterized by their behavior during ion exchange chromatography and by their amino acid composition. The findings establish that the articular cartilage of postnatal chickens contains substantial amounts of Type I collagen and suggest that the nature and stability of the crosslinkages in Type I collagen of cartilage differs significantly from those of Type II collagen of cartilage. Moreover, it is also postulated that the nature of the major intermolecular cross-linkages in the Type I collagen of cartilage are similar to those present in Type I collagen of bone rather than those present in the Type I collabens of skin and other connective tissues.

O-phosphoserine [Ser(P)], O-phosphothreonine [Thr(P)], and gamma-carboxyglutamic acid (Gla) have been identified in native, calcified cementum and in non-collagenous proteins which can be extracted from the tissue in EDTA at neutral pH. The concentrations of Ser(P) and Thr(P) and the amino acid composition of the EDTA extractable proteins are more similar to those found in bone than in dentin or enamel. The concentration of Gla in cementum is lower than it is in bone and higher than it is in enamel, which contains essentially no Gla. Based on the contents of Gla in these mineralized tissues and the distribution of alkaline and acid phosphatases in these tissues, it is speculated that Gla may be part of these or other proenzymes rather than being involved directly and structurally with the deposition of the mineral phase.

The secondary structures of two phosphoproteins from chicken bone matrix of Mr approximately 15kDa and approximately 28kDa, rich in Asx, Glx, and Ser, and containing Ser(P) and Thr(P) residues, have been investigated in solution by Circular Dichroism (CD) and Fourier Transform-Infrared Spectroscopy (FT-IR). CD spectroscopy, which yields useful information on the backbone conformation of polypeptides and proteins, suggests a predominantly beta-sheet structure for the two phosphoproteins. The FT-IR spectra of the approximately 15kDa protein, which is sensitive to secondary structure and hence provides complimentary information to CD spectroscopy, are consistent with the results obtained by CD studies.

Mouse bone collagenase, a specific tissue collagenase, was isolated from the tissue culture media of 5 day old mouse tibiae. By a combination of (NH4)2SO4 precipitation, molecular sieve filtration and ion exchange chromatography a fraction was obtained with a yield of approximately 11% which had a specific enzyme activity 120-fold greater than the original crude extract. The molecular weight of the fraction containing the enzyme activity was approximately 41000 as determined by calibrated molecular sieve filtration studies. There were at least two distinct major fractions and one minor fraction possessing collagenase activity which had distinct isoelectric points. It is not clear whether these fractions represent isoenzymes from bone or from bone and cartilage, or are derived from a single enzyme which has been minimally degraded by the extraction and purification techniques used. Collagenase activity was inhibited by EDTA, cysteine and horse serum, but not by phenylmethylsulfonylfluoride, epsilon amino caproic acid or soybean trypsin inhibitor.

Horse serum prepared free of trypsin inhibitors by passing it through a column of insolubilized trypsin (trypsin covalently bound to Sepharose 4B) no longer inhibits mouse bone collagenase activity. Of the various serum fractions tested, only two (beta-lipoprotein, which is a rich source of alpha-2-macroglobulin, and alpha-globulin) significantly inhibited mouse bone collagenase activity. Only the alpha-2-macroglobulin fraction of horse serum chromatographed on Bio-gel P-150 inhibited both mouse bone collagenase activity and trypsin activity. The alpha-1-antitrypsin fraction did not inhibit mouse bone collagenase activity, but did inhibit trypsin activity. Affinity chromatography of horse serum on Sepharose 4B to which trypsin had been covalently bound revealed inhibitory activity towards both trypsin and mouse bone collagenase in the fraction containing alpha-2-macroglobulin. The results suggest that the component in serum which inhibits mouse bone collagenase activity may be identical with the trypsin inhibitor present in the alpha-2-macroglobulin fraction. Serum and mouse bone collagenase chromatographed on the Bio-gel column did not bind together and the collagenase emerged at its own elution volume without loss of enzymatic activity. This suggests that an enzyme inhibitor complex (if it exists) is easily dissociable. Similar results to those above were obtained using mouse serum.

Monoclonal and polyclonal antibodies were raised against 150k, 60k, 32k, and 15k phosphoprotein components of 14-week chicken bone. Hybridomas were prepared by immunizing Balb/c mice with 150k, 60k, or 32k phosphoproteins followed by fusion of their spleen cells with X63-Ag8.653 myeloma cells. Polyclonal antibodies to the 60k, 32k, and 15k phosphoprotein components were produced in immunized rabbits. Immunological cross-reactivity between antigen and antibody was examined by enzyme-linked immunosorbent assay (ELISA) and dot-blotting. There was cross-reactivity between the 150k, 32k, and 15k phosphoprotein components and between the 150k, 90k, and 60k components. The antibodies raised against 60k component do not bind 32k and 15k antigens.

EARLIER we suggested a role for matrix-bound organic phosphorus in the initiation of mineralization. The evidence supporting our hypothesis was provided by observations on the interaction of collagen and inorganic orthophosphate in vitro1,2 and the finding of small amounts of protein-bound organic phosphorus in purified collagens and gelatines and the single stranded α chains from a number of different soft tissues from different animals3. The content of organic phosphorus was found to be considerably higher in the organic matrix proteins of embryonic and adult dental enamel, in which serine phosphate bound to protein was identified4. Enzymes which catalysed a phosphoryl transfer from adenosine triphosphate5 to enamel proteins, in addition to collagens and gelatines6, were also characterized.