| Sign In to gain access to subscriptions and/or personal tools. |
Sphingomyelin Degradation is a Key Factor in Dentin and Bone Mineralization: Lessons from the fro/fro Mouse
1 Laboratoire "Réparation et Remodelage des Tissus Orofaciaux", EA 2496, Faculté de Chirurgie Dentaire, Université Paris 5, 1, rue Maurice Arnoux, Montrouge 92120, France; Correspondence: * corresponding author, michel.goldberg{at}odontologie.univ-paris5.fr or mgoldod{at}aol.com
Key Words: phospholipids • bone • dentin • sphingomyelin • mineralization INTRODUCTION We have identified (by chemical analysis) and mapped (at the ultrastructural level) the phospholipids present in enamel and dentin. Analysis of the data obtained suggests that, as matrix extracellular (MEC) components, phospholipids play important functions not only during the formation of dental tissues, but also in mineralization processes (Goldberg et al., 1983; Goldberg and Boskey, 1996; Goldberg and Septier, 2002). Nevertheless, we were still looking to substantiate what was perhaps a matter of faith regarding the role of lipids, when finally we met the fro/fro mouse. Lipids play major biological roles as components of cell membranes. As components of the extracellular matrix (ECM) of mineralized tissues, lipids have been infrequently studied. Their presence is now well-recognized, but their role in biological and pathological mineralization is far from clear. In dental tissues, the total lipid content is 0.26–0.36% in dentin, 0.014% of which is phospholipids, and 0.60% in enamel, among which 12.5% is phospholipids (0.075% of total enamel) (Goldberg and Boskey, 1996). Compared with the major dentin ECM components, namely, the phosphorylated proteins of the SIBLING family, lipids appear as minor components. Due to the low ratio between lipids and the total ECM molecules, and because of the fact that the major part of the lipid fraction is lost after routine fixation and processing for light and electron microscopy, lipids have been investigated less than have other matrix components easier to preserve and analyze. THE CHEMISTRY AND HISTOCHEMISTRY OF DENTIN LIPIDS Our interest in lipids was stimulated by the finding that about 66% of the total lipids in dentin and enamel are extracted prior to demineralization, and only 33% after demineralization (Shapiro et al., 1966). The lipid composition of the non-demineralized extract suggested that its composition is close to that of biological membranes, namely, sphingomyelin (SM), phosphatidylcholine (PC), and phosphatidylethanolamine (PE). In dentin, some of these lipids may originate from odontoblast processes located inside dentin tubules. The lipid fraction extracted after demineralization has a different composition. Indeed, the acidic phospholipids (PLs) associated with the mineral phase include phosphatidyl inositol (PI), phosphatidylcholine, phosphatidylethanolamine, phosphatidyl serine (PS), phosphatidic acid (PA), and cardiolipin (CL). Presumably, there is a close association between these acidic phospholipids and the surfaces of the crystallites, and they are released only after mineral dissolution. This finding points to a group of acidic phospholipids interacting with the mineral phase, and therefore possibly being involved in dental mineralization. Historically, histochemical reports aiming to detect and localize lipids in bone and dental tissues were obtained after formol-saline fixation, hot pyridine or ethanol-benzene extraction, demineralization, and gelatin embedding. After such treatments, a Sudan black B stainable material was detected at the junction between predentin and dentin, osteoid, and bone, and at the interface between the forming and mineralizing enamel (Irving, 1958). In many articles, this publication is quoted as a reference providing evidence of a role for lipids in biomineralization. However, the method used was questionable, because after such treatment, most, if not all, lipids would have been spontaneously lost or extracted from the tissue. As a matter of fact, it has been demonstrated that lipophilic proteins, and not lipids, are stainable by this histochemical method (Fincham et al., 1972). It has been well-established that most lipids are solubilized during fixation and dehydration, and that the remaining lipids translocate during histological processing and tissue embedding (Goldberg and Boskey, 1996). However, we found two methods allowing for preservation of phospholipids within dental tissues, while avoiding gross translocation (Teichman et al., 1974). Malachite green, in the presence of aldehyde (MGA), forms insoluble complexes with cellular and extracellular lipids, and post-fixation with osmium tetroxide reinforces the electron density of the complexes, therefore facilitating ultrastructural detection (Teichman et al., 1974). The specificity of the staining is better after treatment of the section with lead citrate, whereas uranyl acetate also stains proteinaceous components, inducing additional non-specific staining. Using spot tests, we have shown that this method preferentially stains phosphatidyl serine and sphingomyelin and, to a lesser extent, phosphatidylcholine (Vermelin et al., 1994). This method allowed for the visualization of a fibrillar and granular material inside the predentin, located in the intercollagenous spaces. The dentin edge at the predentin/dentin junction was not stained, and, in dentin, an electron-lucent group of collagen fibers was surrounded by MGA-positive needle-like structures reminiscent of the "crystal ghosts" described in the mineralized cartilage and in bone, formed by phospho-glyco-lipoproteins (Bonucci, 1987). The structures persisted after demineralization when the sections were floated on EDTA or acetic acid. The demonstration that dentin crystallites are formed or are located inside lipoprotein envelopes sheds light on the potential role of some lipids in the mineralization processes (Goldberg and Septier, 1985). After tissue samples were rapid-frozen, followed by freeze-substitution, a method that presumably avoids the artifacts induced by chemical fixation and dehydration, the same features were observed in dental tissues (Goldberg and Escaig, 1987). In the predentin, rapidly frozen MGA-positive structures were thicker than after aqueous chemical fixation, forming a network of liposome-like structures. The MGA-positive structures were abolished by pre-treatment of the samples with methanol, but not by acetone, thus confirming that they are actual phospholipids (since acidic phospholipids precipitate in acetone). They resisted enzyme digestion by bovine testicular hyaluronidase, but were largely reduced and even abolished by phospholipases, and therefore are distinct from the proteoglycans located in the same intercollagenous areas. The iodoplatinate (IP) method indicates a similar localization of the phospholipids, although the molecules that were visualized were not identical to those of the MGA-stained components. The spot test assay indicated a major affinity of IP for sphingomyelin, and a strong staining was also seen for phosphatidylethanolamine and phosphatidylcholine, whereas the staining appeared weaker, as a narrow ring, around the phosphatidyl serine dot (Vermelin et al., 1994). In the predentin, IP-positive granules have been detected in the spaces between collagen fibrils, and again, in dentin, needle-like structures were associated with the crystallites. The transition between predentin and the dentin edge was unstained. In addition, large electron-dense amorphous areas were seen in dentin, disappearing when the sections were treated with chloroform/methanol or phospholipase, but still present after acetone or hyaluronidase treatment of the sections. From our histochemical investigations, we concluded that some phospholipids are actually present as extracellular matrix components. They appear as space-filling components between collagen fibrils in the predentin, sharing this location with proteoglycans. Although the reasons for such co-distribution have not yet been clarified, interaction among decorin, low-density lipoprotein (LDL), and type I collagen has been reported in another biological model (Pentikainen et al., 1997). Association between proteoglycans and matrix vesicles was shown in the extracellular matrix of growth plate cartilage (Wu et al., 1991). The same physico-chemical interaction may also apply to the predentin situation. Water desorption is due to the self-association of hydrophobic molecules that combine and form an appropriate volume where ions concentrate. These events contribute to the pre-patterning of an ordered precipitation. This cascade is suggested by Glimcher’s model of enamel formation and mineralization (Glimcher, 1979), and provides an acceptable theory for a potential role for phosphorylated proteins in enamel mineralization, a working hypothesis that may be expanded to acidic phospholipids (Vogel and Boyan-Salyers, 1976). THE ORIGIN OF DENTIN PHOSPHOLIPIDS Until recently, the origin of dentin ECM phospholipids was unclear. It was assumed that they were synthesized by the odontoblasts and further secreted as matrix components. To investigate the truth of this assumption, we injected (3H) choline intravenously into rats and studied how the radiolabeled precursor was incorporated into the cells, and released into the predentin and dentin by radioautography. To avoid loss and translocation, the rats were perfused with MGA, the incisors were removed by dissection, and segments were post-fixed with osmium tetroxide. Surprisingly, after 30 min, no radiolabeling was detectable within the odontoblasts, but a significant labeling over background was already located over the dentin compartment. After 1 hr, the situation was identical, and it took 2 hrs before the cells began to be significantly radiolabeled (Goldberg et al., 1999). Hence, dentin phosphatidylcholine could not arise from synthesis by the odontoblasts, followed by exocytosis, as is the case for proteins. As an alternative pathway, we hypothesized that isotopic exchanges between labeled serum lipoproteins and the outer membrane leaflet of erythrocytes take place within 5–20 min, and these events immediately precede the diffusion of radiolabeled serum albumin between the odontoblasts. Within 1 hr, radiolabeled albumin crosses the leaky intercellular junctional complex between odontoblasts, and diffuses into predentin and dentin (Kinoshita, 1979). Albumin has been recognized to be a potent lipid carrier (Veerkamp and Maatman, 1995). These results suggest that dentin phospholipids may originate from the blood serum, possibly as low-density lipoproteins (LDL). Rapid phospholipid interconversion and the chemical plasticity of the different forms of phospholipids may regulate the final phospholipid composition of the ECM. A SEARCH FOR EXPERIMENTAL EVIDENCE Analysis of these data led us to search for examples that would substantiate the role of lipids (phospholipids) in dental mineralization. Along this line, we investigated a veritable ’zoo’ of experimental animal models involving chemically induced lipidosis (the suramin-induced mucopolysaccharidosis and lipidosis, chloroquine-induced lipidosis), diseases induced by deficient food intake (zinc deficiency and/or essential fatty acid deficiency), and human genetic mutations (Krabbe galactoceramide lipidosis) (Goldberg and Boskey, 1996), or targeted mutations affecting dentin and bone (Boskey et al., 2006). Although slight abnormalities were detected, we did not obtain strong evidence for defective mineralization when the lipid composition was altered, and consequently the role of phospholipids in dental mineralization was still not substantially demonstrated. At that point, I received a call from Jean-Louis Guenet. We had never met, but he told me—and I don’t remember if this was the right order—that he was the father, first, of a mouse that displays an osteogenesis imperfecta and tooth color alteration, and, second, of a son who by chance happened to have been one of my past students. Apparently, the son told his father that I was one of the very few crazy persons who could be interested in such an investigation. After this fascinating introduction, we decided to meet, and immediately after, we started a wonderful collaboration, studying the teeth of the fragilitas ossium (fro/fro) mouse within the enriching environment of researchers in genetics and chemists who were the other partners in this game. THE fro/fro MOUSE The chemically induced autosomal mutation named "fragilitas ossium" is not linked to any of the collagen defects shown to be responsible for most forms of osteogenesis imperfecta (OI). Types I–IV, constituting about 90% of OI forms, are due to the mutation of genes encoding for type I procollagen, whereas types V–VII (10%) are OI lacking collagen mutations. The mutation was induced into the mouse germ line by the chemical mutagen tris (1-aziridinyl) phosphine sulphide (Thiotepa®) 2 wks before the mice were mated (the targeted germ cells were then spermatids). The mutant mice displayed a severe form of OI (Guenet et al., 1981), and therefore the mutation was designated as fragilitas ossium (fro/fro).
Compared with control heterozygous mice (+/fro), newborn homozygous mice (fro/fro) were smaller, with brittle long bones and ribs (Muriel et al., 1991). Longitudinal sections of the mandible showed shorter incisors in the fro/fro, about half the length of those in the heterozygotes (Figs, 1a, 1b
The bone of the skeleton of the adult fro/fro mouse recovers an almost "normal" phenotype as it ages, but not the teeth, where the abnormal distribution of ECM proteins seen in newborn mice is not corrected. The closure of the lingual part in the forming zone of the incisor is delayed in the adult fro/fro mouse, and a fissure persists between the 2 lingual arched portions. Why the 2 arches do not merge rapidly, as is the case in control teeth, has not yet been explored. In the rodent, the molar is not a continuously erupting tooth, in contrast to the incisor, and since there is no turnover in dental tissues, it may be expected to reflect unequivocally the effects of the mutation. In the fro/fro mouse, a thick layer of cellular cementum gives the root a bulbous appearance. In addition, the pulp chamber is filled with a homogeneous well-mineralized dentin-like structure (Figs. 1e, 1f  , 2a, 2b ). The general appearance of the tooth is characteristic of that of a human type II dentin dysplasia (Opsahl et al., 2005).
In a previous report, it was suggested that the fro/fro mutation might be linked to an osteonectin (or SPARC protein) defect (Muriel et al., 1991), a gene mapping to chromosome 11. Isabelle Aubin and Christophe Poirier, under the leadership of Jean-Louis Guenet, established that the mutation in fro/fro is localized on mouse chromosome 8. A deletion was found, encompassing part of intron 8 and a major part of exon 9, including the coding sequence of this gene (Aubin et al., 2005). We were gambling on a restricted list of 23 potential genes of causality, none of them coding a protein implicated in the structure of the extracellular matrix. The cause of the mutation was finally identified as a sphingomyelin phosphodiesterase neutral 2 enzyme (Smpd3) deficiency. At this point, Salvayre’s group (Aubin et al., 2005) in Toulouse established that, in the cultured fibroblasts of the fro/fro mouse, Smpd3, activated by tumor necrosis factor-  (TNF-), has an enzyme activity reduced to 12 ± 4%, compared with +/fro mice.
Neutral sphingomyelinases cleave sphingomyelin to produce ceramide and phosphocholine and are activated by TNF- In 2005, Stoffel et al. also reported a knock-out mutation at the same Smpd3 locus as the chemically induced fro/fro mutation (Stoffel et al., 2005). However, they did not report any effect on teeth and mandibular bone. They emphasized the intracellular effects of the mutation on insoluble domains of the Golgi membrane. In their Smpd3 knock-out model, substantial evidence was provided for a severe growth hormone/insulin-like growth factor-1 (GH/IGF-1) deficiency. Consequently, the defective post-natal skeletal growth retardation was attributed to the perturbed secretion of GH/IGF-1 (Lupu et al., 2001). The phenotype of the knock-out allele mimicks human chondrodysplasia and dwarfism, in contrast to the fragilitas ossium (fro) mice described as a model of a recessively inherited, non-collagen-dependent type of osteogenesis imperfecta. Differences in the methods used to obtain the mutation, and in the characterization of the phenotype of the mutant mice, may account for the interpretation of the results. Glycosphingolipids have been identified in bone, cartilage, and dentin (Fukaya et al., 1989; Goldberg and Boskey, 1996; Goldberg and Septier, 2002). Sphingomyelinase is involved in sphingomyelin hydrolysis in osteoblast-like cells. Extracellular Smpd amplifies BMP-4-induced osteocalcin synthesis in osteoblast-like MC3T3-E1 cells (Kozawa et al., 2002). Finally, during cartilage matrix vesicle-induced mineralization, there is a progressive disappearance of sphingomyelin, presumably due to hydrolysis by sphingomyelinases (Wu et al., 2002). Together, adding these data to what we have observed in the fro/fro mutation in dental tissues, and in mandibular and long bones as well, we suggest that proper sphingomyelin metabolism is essential for mineralization in bone and dentin. Although an indirect role for lipids in dentin and bone mineralization cannot be ruled out, a series of in vitro experiments on the nucleation of hydroxyapatite by the calcium-phospholipid-phosphate complex, and in vivo data on normal and pathologic mineralization (e.g., salivary calculi, kidney stones, vascular calcification, atherosclerotic plaque), supports direct involvement. The fro/fro mice provide the very first in vivo experimental evidence that sphingomyelin and, moreover, sphingomyelin degradation play a crucial role in bone and dentin mineralization. In addition, this new finding suggests that it is necessary to revisit the classification of the non-collagenous forms of osteogenesis imperfecta and dentin dysplasia (Witkop, 1988).
ACKNOWLEDGMENTS The authors are grateful for support from the Ministry of Education and University Paris 5 (EA2496), and from NIH grants DE04141 and P30 AR046121 (ALB). FOOTNOTES Received for publication May 25, 2006. Revision received October 27, 2007. Accepted for publication October 29, 2007. REFERENCES
Journal of Dental Research, Vol. 87, No. 1,
9-13 (2008)
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Sign In to gain access to subscriptions and/or personal tools.
