| Sign In to gain access to subscriptions and/or personal tools. |
Biglycan is a Repressor of Amelogenin Expression and Enamel Formation: An Emerging Hypothesis
1 Laboratoire de Biologie et Physiopathologie Crânio-faciales, Groupe Matrices Extracellulaires et Minéralisations, Faculté de Chirurgie Dentaire, Université Paris V , 1, rue Maurice Arnoux, 92120 Montrouge, France; Correspondence: * corresponding author, mgoldod{at}aol.com
Key Words: biglycan • proteoglycan • amelogenin • ameloblasts • odontoblasts THIRTY YEARS OF RESEARCH ON GLYCOSAMINOGLYCANS AND PROTEOGLYCANS About 30 years ago, proteoglycans (PGs) were considered as minor components of mineralized tissues. Few studies were related to these molecules, for two reasons: First, the quantity of PGs is negligible compared with that of major matrix components such as collagens or phosphorylated proteins, which make up about half of the non-collagenic proteins. In human dentin, for example, the total glycosaminoglycan (GAG) content of the organic matrix is about 0.43% (Jones and Leaver, 1972), which represents less than 5% of the non-collagenous matrix components. Second, at that time, the general concept in mineralized tissues was that GAGs or PGs were not directly involved in the mineralization process but were merely negative factors preventing the cascade of events that occurs during the initiation of mineralization from happening. For example, the GAG content in the osteoid was reduced during mineralization (Baylink et al., 1972), and consequently the small part retained or entombed in the mineralized tissue was thought to be of little importance. From there and despite original conflicting reports, interest grew progressively as PGs were identified as actual components of the mineralized part of the bone (Hunter et al., 1984; Fisher et al., 1983; Prince et al., 1983). In vitro, PGs were shown to have an inhibitory effect on hydroxyapatite (HA) formation and growth (Blumenthal et al., 1979; Chen et al., 1984). The fact that no net loss of PGs occurs during endochondral ossification (Poole et al., 1982) and the evidence that PGs promote hydroxyapatite (HA) formation led to the re-interpretation of the data (for review, see Hunter, 1991), and from other studies, it was concluded that, in solution, PGs inhibit mineral nucleation and growth, whereas immobilized on a surface, the same molecules promote mineral formation (Linde et al., 1989). During the last 30 years, chemical methods facilitating the study of matrix components have improved (Steinfort et al., 1994) and advanced from conventional histochemistry to GAGs and PGs immunohistochemistry. The precise identification and localization of PGs are now in the process of being achieved, but their function still remains to be elucidated. With new tools allowing for identification of the genes regulating the expression of Small Leucine-rich PGs (SLRPs) involved in the formation of dental tissues, and with the creation of genetically engineered mice (Young and Xu, 2001), we may now expect to gain new insights into the role and function of these molecules. When the senior author of this article (MG) began his research, an assistant position was available quite by chance in the Department of Oral Biology at the Dental School, University Paris Vth. Was it a pure accident or "God’s fingerprint"? The department head was Robert Weill, who built his reputation by working on the histochemistry of "mucopolysaccharides", a group of molecules soon renamed GAGs. Using radiolabeled sulphate incorporation, he published pioneer results (Verne et al., 1952), two years before Bélanger (1954). He contributed to the characterization of the light microscope histochemistry of the sulfated GAGs implicated in dentinogenesis (for review, see Symons, 1967). Consequently, all the members of the lab had to work along these lines, trying to identify new extracellular matrix components. It turned out to be a productive direction. ELECTRON HISTOCHEMISTRY We first explored the histochemistry of GAGs with the electron microscope. Together with Dominique Septier, who joined the lab soon after the beginning of this research, we used alcian blue at pH 1.4 on conventionally fixed undemineralized ultrathin Epon sections (Goldberg et al., 1976, 1978). Ultrastructural data, and other pioneer publications (Nagai and Takuma, 1973; Nygren et al., 1976), reported that GAGs were massively present in the predentin and appeared as small dots or granules associated with collagen fibrils in dentin. Various histochemical methods (including cationic dyes or cationic detergents) were developed to preserve and stain the GAGs that otherwise would be lost during the fixation, dehydration, and embedding steps. GAGs were also visualized with hyaluronidase-gold complex and poly-L-lysine-gold complexes (for review, see Goldberg et al., 1987; Goldberg and Takagi, 1993; Embery et al., 2001). The electron-dense aggregates were demonstrated to be composed of GAGs, because the aggregates were dissociated by calcium chloride and digested by bovine testicular hyaluronidase. Depending on the fixation/staining method used, the shape of the aggregates varied from round to a star-like or boomerang-like appearance, suggesting that the aggregates are artifacts due to the coiling of the molecules. Indeed, following rapid freezing and freeze substitution, a method which better preserves the native form of biomolecules, the GAGs or PGs were seen to be an amorphous expanded gel in predentin (Goldberg and Escaig, 1984). In dentin, with some variations related to the method used, tiny electron-dense granules were associated with collagen cross-striations, or needle-like structures were evidenced along collagen fibrils or in intercollagen spaces. These structures were reminiscent of what has been called, in bone and mineralized cartilage, "crystal ghosts" (Bonucci, 1987). This organic material, located at the surface of crystallite, wraps mineralized structures in intertubular dentin. Histochemically, the needle-like structures were demonstrated to be glycosylated proteins associated with phospholipids. RADIOAUTOGRAPHY Radioautographic investigations demonstrated an uptake of radiosulfate into dentin- and enamel-forming cells that led to dentin and enamel labeling. Shortly after incorporation, labeling appeared as a double band, the first band located in predentin and the second in dentin, close to the mineralization front. This finding was confirmed by radioautography with either [35S]sulphate or [3H] glucosamine as precursor, on tissues fixed either by rapid freezing and freeze substitution or conventionally (Goldberg and Escaig, 1985), strongly suggesting that the double band was not artifactual. In predentin, studying [35S]sulphate incorporation between 30 min and 120 hrs in molars of seven- to 15-day-old rats on sections that had not been demineralized, we observed, as early as 30 min after injection, a detectable labeling that reached a maximum at 4 hrs. A slow and regular decrease of radiolabeling was seen thereafter, and at 120 hrs, labeling was totally absent. In dentin, labeling appeared as early as 30 min, and increased at the dentin-predentin junction between 1 and 2 hrs, reaching a maximum at 4 hrs. The dentin layer formed subsequently was not labeled. The labeled band stayed at a constant distance from the dentino-enamel junction and was quantitatively stable during the time studied (Lormée et al., 1996). This suggests that two distinct groups of sulfated GAGs are secreted: The first is secreted in predentin, in the proximal zone, together with native collagen fibrils. This group of GAGs spreads throughout predentin and is not stable. Enzymes present in the predentin and not specifically involved in GAG or PG degradation may be implicated in their degradation. Cathepsin-D and stromelysin 1 (MMP-3) have been identified in predentin (Nygren et al., 1979; Hall et al., 1999). Immunogold labelings demonstrated that MMP-2 and MMP-9 were present in predentin and might contribute to GAG degradation in predentin (Goldberg et al., 2000). Combined with other data, this suggests that, in predentin, GAGs are mostly involved in collagen transport toward the mineralization front, where fibrils are packed and undergo mineralization. During this step, the horseshoe-shaped SLRPs may be implicated in collagen fibrillogenesis, linking subunits and orienting collagen fibrils (Scott, 1996). The possibility that some predentin GAGs are further incorporated into dentin cannot be ruled out, but it seems clear, from radioautographic data, that a second group of dentin PGs is actually present. In contrast to the predentin group, dentin GAGs are secreted distally by odontoblast processes, within dentin tubules, perhaps some distance from the predentin-dentin junction. Associated with the mineralized dentin, this group of GAGs is stable with time, at least up to 120 hrs after synthesis. In accordance with our observations, Rahemtulla et al. (1984) also suggested the presence of two distinct groups of PGs in the rat incisor. A SEARCH FOR SPECIFICITY: LIGHT AND ELECTRON MICROSCOPE IMMUNOHISTOCHEMISTRY Histochemical methods are hampered by a lack of specificity of what is labeled. Several sulfated, phosphorylated, and carboxylated extracellular matrix components can be stained by cationic dyes or cationic detergents. pH variations in the fixative solution and/or handling of the fixation process at a critical electrolyte concentration restricts the group of matrix components which can be detected. Nevertheless, many potential members may still be labeled in each group. Radioautographic approaches suffer from the same limitation. For example, the incorporation of radiolabeled sulfate into the forming enamel may provide evidence for PGs, as suggested by Blumen and Merzel (1976), but also for short-lived sulfated enamel proteins (Smith et al., 1995). In our attempt to visualize individual GAGs or PGs present in dental tissues, we decided to use more specific approaches. We first used the 2B6 antibody, which specifically recognizes chondroitin sulfate/dermatan sulfate (CS/DS). Immunolabeling demonstrated the presence of a gradient in predentin; the labeling was high in the proximal part and decreased in the distal third, near the predentin/dentin junction (Septier et al., 1998). A reverse situation was observed with the 5D4 antibody, which reveals keratan sulfate (KS) chains. KS labeling was increased in the distal predentin near the mineralization front compared with the proximal predentin. Labeling was not detectable with the light microscope in the undemineralized dentin, where the staining was limited to the lumen of the tubules. As a working hypothesis, we assumed that the reverse gradient seen with the two antibodies could be due to the degradation process of CS/DS. MMP-3 immunodetected in predentin at the junction between the inner third and the central part is a good candidate for such an event. CS/DS degradation may provide a renewed access to collagen-binding sites that consequently allow for a correlated increase of KS in the tissue (Hall et al., 1999). BIGLYCAN AND DECORIN IMMUNOGOLD LABELING
This series of investigations was further extended to immunogold labeling with antibodies directed against specific amino acid sequences of 2 small leucine-rich proteoglycans (SLRPs)—biglycan and decorin—expressed in the tooth (Septier et al., 2001). Biglycan and decorin are synthesized with a pro-peptide which is cleaved under tissue-specific and age-dependent conditions (Roughley et al. 1996). We studied the pro- and processed forms of decorin and biglycan (hereafter called, respectively, probiglycan/prodecorin and biglycan/decorin) in the rat incisor (antibodies were a kind gift from Larry Fisher; Fisher et al., 1995 [see Fig
A SEARCH FOR A FUNCTION: THE KNOCK-OUT MICE At the end of a seminar given at the NIH, and following a friendly informal discussion, we decided on a joint venture between the NIDCR group and the French group with knock-out mice being explored with the expectation that we would gain some insights into the role of SLRPs during odontogenesis. A first report on biglycan knock-out mice (BGN KO) had already been published (Xu et al., 1998), establishing that this mouse line displays an osteoporosis-like phenotype. Our results reinforced this observation. In the day-1 mandible, large nodules that have not merged are present in the BGN KO alveolar bone, leaving unmineralized interglobular spaces between them. The same was seen in dentin, with calcospherites surrounded by poorly mineralized interglobular dentin. Besides these defects, and compared with the wild-type situation, the diameter of the collagen fibrils in predentin was decreased in the proximal third, but significantly increased in the central and distal parts. Several studies now support that SLRPs control the initial collagen fibril assembly and induce changes in the fibril diameter (Ameye and Young, in press). Together, these results confirmed the role of BGN as a regulator of extracellular matrix organization. Another totally unexpected finding was that, at day 1 in the BGN KO mouse molar, the forming enamel was about three- to five-fold thicker compared with that of the wild-type. Interprismatic enamel alone was seen in the outer half of the forming enamel. Instead of the expected enamel rods, tubule-like structures filled with stippled material and protracted Tomes’ processes were seen in the outer-forming enamel. This suggests that secretory ameloblasts rapidly withdrew backward from the enamel surface, leaving membrane residues and Tomes’ processes in unfilled tubular spaces. The results presented here fit well with the data published by Matsuura et al.(2001), establishing that, in the newborn mouse, in situ hybridization reveals intense biglycan expression in pre-secretory ameloblasts and weak labeling in the secretory odontoblasts. (In contrast, decorin mRNA expression is very high in odontoblasts and weak in pre-secretory ameloblasts). Additional immunolabeling confirmed that the BGN KO did not stain with BGN antibody, whereas normal staining was seen in predentin with decorin and fibromodulin antibodies (antibodies kindly provided by L.W. Fisher; Fisher et al., 1995). Immunostaining was further carried out on this material with two different antibodies, one raised against the whole amelogenin (a gift from Dr. Ivan Slaby, Biora, Malmö, Sweden) and the other raised against the amelogenin C-terminal end (gift from Dr. Carolyn Gibson, University of Pennsylvania, Dental Medicine, Philadelphia, PA, USA). In the wild-type mouse, the two antibodies stained exclusively thin half-moon structures located at the tips of the cusps. The stained material included only secretory ameloblasts and forming enamel. In BGN KO mice, the staining was enhanced, because the enamel was thicker. In addition, pre-secretory and secretory ameloblasts were intensely stained. In the molar, the odontoblasts located near the tips of the cusps were also densely stained. In the incisor, not only the crown-like labial odontoblasts but also the root-like lingual odontoblasts were positively immunostained. For the labial odontoblasts, the staining may result from diffusion of amelogenins produced by secretory ameloblasts. This diffusion would be very unlikely for the lingual odontoblasts far away from ameloblasts, strongly suggesting that the immunostaining is due to amelogenin synthesis by the odontoblasts. Expression of amelogenin by odontoblasts is now recognized, but is usually not detectable by immunohistochemical methods in wild-type mice. Only its over-expression in the BGN KO mice allows it to be detected. Spliced forms of 5- to 7-kDa amelogenins (A+4 and A-4) have been reported to be dentin matrix components (Nebgen et al., 1999; Veis et al., 2000). Structural and immunohistochemical studies have also been carried out on single decorin (DCN) and fibromodulin (FM) KO mice. Although this is still under investigation, at this stage of our research it is clear that the DCN and FM KOs do not develop the same enamel phenotype as the BGN KO. From the close homology among BGN, DCN, and FM, common features could have been expected among the three KOs, yet this is not the case. On the contrary, in the DCN KO, enamel formation is repressed in the molar, while enamel appears to be mostly aprismatic in the incisor. In the FM KO, enamel formation is apparently normal. Each SLRP therefore bears its own biological specificity. To conclude the discussion on an emerging hypothesis, we suggest that biglycan acts as a repressor of the expression of amelogenin in the two unique groups of cells that are involved in amelogenin synthesis, namely, the secretory ameloblasts and odontoblasts. The mechanisms that are involved are not yet clarified but highlight that BGN is an extracellular matrix molecule able to influence cellular activities. It is now admitted that some molecules are multifunctional. The new properties evidenced here for BGN suggest that two tracks should be explored: One, the events reported here are due to direct cellular activities targeted by BGN. According to this line of thought, BGN is recognized to induce gap junction expression (Fujita et al., 1987). BGN also controls some of the uptake pathway for lipoproteins (Halvorsen et al., 1998), and, due to the interaction of cationic lipid-DNA complexes with glycosaminoglycans, may play a role in cationic lipid-mediated gene delivery (Wiethoff et al., 2001). As a second alternative possibility, the effects of BGN may be indirectly mediated. SLRPs are biological ligands for growth factor receptors in an integrin-independent manner (Iozzo et al., 1999). This may also be the case for growth factors such as TGF-â (Iozzo, 1999) and FGF2 (Kinsella et al., 1997), which are key factors in the stimulation or inhibition of extracellular molecule synthesis. Because BGN occupies some cell membrane or pericellular sites, the activity of receptors may be repressed by its presence. The binding properties of some amelogenin motifs with N-acetyl-D-glucosamine, reported by Ravindranath et al. (1999), may contribute to the BGN-amelogenin interrelationship reported here, even if acetylglucosamine is present mostly in HSPG or KSPG, whereas it is acetylgalactosamine which is found in CS/DS. In any case, the previously unsuspected interaction between biglycan and enamel formation now opens new avenues of research into the exploration of dental tissue formation.
REFERENCES
Journal of Dental Research, Vol. 81, No. 8,
520-524 (2002) This article has been cited by other articles:
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Sign In to gain access to subscriptions and/or personal tools.
