| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
REVIEW |
1 Department of Biomineralization, The Forsyth Institute, 140 The Fenway, Boston, MA 02115, USA; and
2 GlaxoSmithKline, Weybridge, Surrey, UK
* corresponding author, hmargolis{at}forsyth.org
 |
ABSTRACT |
|---|
 TOP ABSTRACT (I) INTRODUCTION(II) AMELOGENESIS(III) BIOMIMETIC APPROACHES TO...(IV) CONCLUDING REMARKSREFERENCES |
|---|
KEY WORDS: amelogenin • amelogenesis • enamel • matrix proteins • self-assembly
|
(I) INTRODUCTION |
|---|
|
TOPABSTRACT(I) INTRODUCTION(II) AMELOGENESIS(III) BIOMIMETIC APPROACHES TO...(IV) CONCLUDING REMARKSREFERENCES |
|---|
The chemical compositions, structural organization, and mechanical properties of the various mineralized tissues are a result of highly orchestrated extracellular processes involving matrix molecules, proteases, and mineral ion fluxes that collectively regulate the nucleation, growth, and organization of forming mineral crystals. The formation of biominerals in general (including those of mollusks, algae, and other organisms) appears to be regulated by the same fundamental processes. In particular, it has been suggested (Weiner, 1986; Addadi and Weiner, 1992) that biological mineralization is regulated by an interplay between hydrophobic and hydrophilic molecules, where hydrophobic molecules provide a skeletal or space-filling structure (like collagen in bone) and the hydrophilic (acidic) molecules (like phosphophoryn in dentin (Veis et al., 1991; He et al., 2005) are involved in the regulation of crystal nucleation and growth. Given the well-established ability of type I collagen to self-assemble to form repetitive structures and extended networks of (aligned) fibers (Veis and George, 1994; Kadler et al., 1996), the envisioned role and importance of collagen self-assembly in regulating the structural organization of collagenous mineralized tissues are clear and supported by considerable amounts of experimental evidence (Weiner and Wagner, 1998). Since the collagen matrix persists in the mature tissue, direct observations of collagen organization (Weiner and Traub, 1992; Weiner and Addadi, 1997; Weiner et al., 1999), its interactions with non-collagenous proteins (Beniash et al., 2000; Keene et al., 2000) and its positional relationship with mineral crystals within mature tissues (Traub et al., 1989; Landis et al., 1993, 1996) provide convincing evidence that the pre-assembled collagen matrix serves as a template for organized mineralization in dentin and bone.
Enamel differs markedly from dentin and bone, in that mature enamel contains little to no matrix protein. Mature enamel is mainly composed of carbonated hydroxyapatite and is the hardest vertebrate tissue. Enamel is also uniquely composed of extremely long and narrow crystals, packed into parallel arrays, called enamel rods, which can form intricate interwoven patterns. Its high degree of structural organization strongly suggests that extracellular enamel matrix proteins secreted by ameloblasts during early stages of amelogenesis must regulate this organization, as collagen does in bone and dentin. However, the exploration of this mechanism has been hindered, in part, by the fact that key mineral-matrix associations cannot be easily established in vivo, since the organic matrix does not persist in the mature enamel tissue. Nevertheless, in recent years, considerable progress has been made in advancing an understanding of the mechanism by which enamel matrix proteins regulate enamel mineral formation. Such progress has led investigators to suggest that the predominant enamel matrix protein, amelogenin, self-assembles to form organized supramolecular structures that facilitate crystal organization (Fincham et al., 1994, 1995; Moradian-Oldak et al., 1998a, 2000), prior to its subsequent removal during tissue maturation. This suggestion was based, in part, on the detection of chains of nanometer-sized spheres in TEM studies of both dehydrated resin-embedded (Fincham et al., 1995) and non-dehydrated freeze-fractured sections of forming dental enamel (Robinson et al., 1981). It was further suggested that such aggregates, alone or in combination with other proteins, facilitate the nucleation and organization of mineral phases (Fincham et al., 1999). Although this conceptualization is very appealing, especially given its similarity to mineralization mechanisms in collagen-based tissues, the precise mechanism by which enamel matrix proteins regulate such processes is not well-understood. Nevertheless, as will become apparent below, the preponderance of evidence suggests that both protein-protein interactions and protein-mineral interactions play crucial roles in the regulation of enamel mineral formation and organization.
The purpose of this review is a critical evaluation of the present state of knowledge regarding protein-mediated mineralization and structural organization in amelogenesis. More specifically, this paper will emphasize protein-mineral interactions and the potential role of the assembly of enamel matrix proteins in the regulation of mineral crystal size and shape, and the structural organization of the resulting enamel tissue. Although several reviews have appeared within the last ten years (Simmer and Fincham, 1995; Fincham et al., 1999; Moradian-Oldak, 2001) that address these and other important aspects of dental enamel formation, this review is prompted by recent significant advances in amelogenesis research and in our understanding of macromolecular control of both biological and synthetic mineral formation. Thus, this paper will also provide a brief review of novel in vitro approaches that have used synthetic molecules (i.e., surfactants and polymers) to regulate the formation of hierarchical inorganic (composite) structures in a fashion analogous to that believed to take place in biological systems such as enamel. Accordingly, this paper will illustrate the potential for the development of biomimetic or bio-inspired approaches to mineralized tissue repair and regeneration. There is indeed a significant need for the development of such repair procedures, including those designed to treat damaged (e.g., dental caries) or diseased (amelogenesis imperfecta) enamel tissues. Although this review focuses on the mechanism of enamel formation, the broader concepts considered here should be generally applicable to other mineralizing systems.
|
(II) AMELOGENESIS |
|---|
|
TOPABSTRACT(I) INTRODUCTION(II) AMELOGENESIS(III) BIOMIMETIC APPROACHES TO...(IV) CONCLUDING REMARKSREFERENCES |
|---|
). In developing enamel (Nylen et al., 1963; Travis and Glimcher, 1964; Glimcher et al., 1965b), the c-axes (long axes) of these crystals are co-aligned and generally run in parallel to the overall direction of the enamel rod. The orientation of the enamel crystals within the rod in embryonic bovine enamel has been shown to vary only slightly, by ± 5–10°, along the c-axes, with more mature rods having an even higher degree of alignment (Glimcher et al., 1965b). However, enamel crystals are oriented randomly with respect to other dimensions (i.e., thickness and width) (Nylen et al., 1963; Travis and Glimcher, 1964; Glimcher et al., 1965b). Subsequently, during the maturation stage, coincident with the almost complete removal of enamel proteins by resident proteases, these mineral ribbons grow rapidly in thickness and width, resulting in a mineralized tissue that is > 95% mineral (by weight), with only 1–2% remaining protein (the missing percentage is mostly water), in a fashion that maintains the structural organization of the enamel crystals established during the secretory stage (Robinson and Kirkham, 1985). These findings suggest a functional relationship between organic matrix removal and subsequent mineral growth. The transient nature of the enamel matrix is unique and distinguishes enamel formation from other biomineralization systems.
|
In addition to amelogenin, the enamel matrix contains other important matrix components. Two key proteinases have been identified within the enamel matrix. MMP-20, or enamelysin, is expressed during the secretory stage (Bartlett et al., 1998; Bègue-Kirn et al., 1998; Nagano et al., 2003) and is responsible for processing enamel matrix proteins (Bartlett and Simmer, 1999; Simmer and Hu, 2002). The MMP-20 null mouse (Caterina et al., 2002; Beniash et al., 2006), in which the full-length amelogenin is not proteolytically degraded, exhibits a severely abnormal tooth phenotype, with an altered rod pattern and hypoplastic enamel that delaminates from the dentin. Thus, proteolytic processing of the enamel matrix is essential for proper enamel formation. The associated cleavage products could directly affect matrix assembly and/or mineralization, although there is currently no evidence to support either possibility. Alternatively, these products may be involved in the cell signaling (see Veis, 2003) that is also essential for proper enamel formation. Recently, two mutations in the enamelysin gene leading to amelogenesis imperfecta were also identified (Kim et al., 2005; Ozdemir et al., 2005), again emphasizing the importance of this protease in enamel formation. Kallikrein 4, a serine proteinase also called ESMP1 (Simmer et al., 1998), is expressed in enamel during the maturation stage (Hu et al., 2002) and is believed to be responsible for the complete breakdown of enamel proteins (Simmer and Hu, 2002). Other proteinases have also been detected within the enamel matrix (Smith et al., 1996; Bartlett and Simmer, 1999).
Two other key non-amelogenin matrix proteins, enamelin and ameloblastin, may also play crucial roles in enamel formation, even though they represent a small proportion of the enamel matrix. Enamelin, a glycoprotein with a molecular mass of 125 kDa, has been characterized and cloned from porcine enamel (Hu et al., 1997b). This protein, hydrophilic and acidic (primarily due to post-translational modification), and rich in glycine, aspartic acid, and serine, has also been shown, with immunohistochemistry, to co-localize with growing enamel crystallites (Uchida et al., 1991a, 1991b; Hu et al., 1997b) and strongly associate with mineral surfaces (Tanabe et al., 1990). However, strongest support for a key role for enamelin in enamel formation comes from a recent study that showed that multiple unrelated families having the same ENAM mutation exhibited severe phenotypic autosomal-dominant amelogenesis imperfecta (Hart et al., 2003; Hu and Yamakoshi, 2003). Like amelogenin, enamelin undergoes gradual enzymatic degradation extracellularly (Dohi et al., 1998), possibly suggesting that various degradation products of enamelin have different roles in amelogenesis. It was initially reported that the 32-kDa fragment of enamelin has proteolytic activity (Moradian-Oldak et al., 1996). However, later studies demonstrated that these data were likely erroneous, and that this finding was probably due to kallikrein 4 that co-purified with the 32-kDa enamelin fragment (Simmer et al., 1998; Bartlett and Simmer, 1999).
Ameloblastin (also called sheathlin or amelin) has been identified in the rat and pig (Cerny et al., 1996; Krebsbach et al., 1996; Hu et al., 1997a) and was suggested to play a role in crystal growth. The suggested function of ameloblastin was based on its location relative to the Tomes process (Nanci et al., 1998), which is the secretory end of the ameloblast and the site of crystal growth initiation. Ameloblastin has a polar structure, with the 66-residue hydrophilic C-terminus, which has a pI of 4.5, and an N-terminal domain with a pI of 10.8 (Hu et al., 1997a). Importantly, it has been shown that the over-expression of ameloblastin in mice significantly alters enamel crystal habit and enamel rod morphology, in a fashion resembling amelogenesis imperfecta (Paine et al., 2003a). Inhibition of ameloblastin expression by a synthetic hammerhead ribozyme also resulted in severe enamel malformations in the mouse (Lyngstadaas, 2001). However, it was recently found that enamel is not formed in the ameloblastin knockout mouse, since the ameloblasts de-differentiate soon after the onset of secretory enamel deposition (Fukumoto et al., 2004). The authors of this paper have proposed that ameloblastin is a cell adhesion molecule that facilitates the attachment of ameloblasts to the enamel matrix. This factor may be essential for the maintenance of the ameloblasts in their differentiated state, which is ultimately required for proper enamel deposition.
It is important to keep in mind that amelogenin may function alone or in a concerted fashion (Weiner, 1986; Dohi et al., 1998) with other matrix molecules, although limited and conflicting information is available on this topic. It has been shown that murine amelogenin specifically binds to recombinant ameloblastin via its tyrosine-rich domain (amino acids 33–45), suggesting the possible formation of a heteromolecular assembly (Ravindranath et al., 2004). Similarly, enamelin may have the potential to bind with amelogenin (Ravindranath et al., 1999) or its smaller cleavage products (Yamakoshi et al., 2003a). In contrast, studies using yeast-two-hybridization have concluded (see Paine and Snead, 2005) that amelogenin self-assembles but does not interact directly with either ameloblastin or enamelin. With the same approach and ameloblast-like cells, however, it was recently found that amelogenin, ameloblastin, and enamelin can interact with several other proteins, including extracellular macromolecules like biglycan and collagen (Wang et al., 2005). Nevertheless, little is known about the potential role of any of the noted heteromolecular assemblies. Most functional studies related to the self-assembly of enamel matrix proteins and mineralization have utilized native and recombinant amelogenins. For this reason, and since amelogenin represents the major enamel matrix protein, this review will primarily focus on the structure and function of amelogenin.
(b) Primary Structure and Solubility of Amelogenins
The amelogenin amino acid sequence can be divided into 3 domains, based on differences in composition (Fig. 2). The 45-amino-acid N-terminal domain is rich in Tyr, and hence is called TRAP (tyrosine-rich amelogenin peptide). The central region of amelogenin is hydrophobic and primarily composed of Xxx-Yyy-Pro- repeat motifs (where Xxx and Yyy are primarily Gln), while the 11-amino-acid-long C-terminal domain is charged and hydrophilic. The primary structure of the N- and C-terminal regions of amelogenin is almost completely conserved across mammalian species, although variations occur in the central portion of the protein, primarily via deletions or insertions of Xxx-Yyy-Pro repeats (Toyosawa et al., 1998; Fincham et al., 1999; Wang et al., 2005). Such homology is illustrated in Fig. 2, where the deduced aligned amino acid sequences of amelogenins from the mouse (M180), pig (P173), and cow (B197) are given. Although amelogenins from different species are not identical, the conservation of the N- and C-terminal regions of amelogenin led to the suggestion that these segments play a conserved and important role in amelogenesis (Fincham et al., 1999). Hydrophilic portions of amelogenin are lost during proteolytic processing, for example, resulting in P148 (loss of the 25-amino-acid C-terminus) that represents a frequently studied degradation product of amelogenin found in vivo.
|
). Thus, rM179 is analogous to the full-length M180. The solubility of rM179 was lowest (0.71 ± 0.36 mg/mL) near its isoelectric point (pH 6.8) at 25°C at 0.15 M ionic strength, but increased sharply at both lower and higher pH values (Fig. 3a, from Tan et al., 1998). The solubility of rM166, which lacks the 13-amino-acid C-terminus, decreased similarly from pH 4.0 to pH 6.0, but stayed relatively constant from pH 6.0 to pH 9.0. The observed difference in the solubility of rM179 and rM166 above neutral pH was attributed to the absence of the hydrophilic C-terminus in rM166. The solubility of the native porcine amelogenins is also lowest near neutrality, although the porcine proteins appear to behave differently from the recombinant mouse proteins, particularly above pH 7, despite the significant amino acid homology between the mouse and pig proteins (Fig. 3b, from Tan et al., 1998). Surprisingly, the ’20K’ porcine amelogenin (P148) was found (Fig. 3b) to be substantially more soluble than its ’23K’ (P161) and ’25K’ (P173) precursors under acidic conditions, particularly between pH 6.0 and pH 7.0, which may be related to the absence of the extremely hydrophobic 12-amino-acid domain (Fig. 2) in P148 that is present in P161. These authors suggested that the enhanced solubility of the ’20K’ protein under such conditions may be important for its further enzymatic degradation. It has also been shown (Tan et al., 1998) that the solubility of rM179 is markedly affected by ionic strength. At pH 5.00 and 25°C, the solubility of rM179 at an ionic strength of 0.01 M was 19.08 mg/mL and decreased to 2.63 mg/mL at an ionic strength of 0.20 M. It was further reported that, at an ionic strength of 0.10 M at pH 5.00 and 25°C, the solubility of rM179 was affected only slightly by concentrations of magnesium and calcium up to 1.5 mM.
|
(c) Aggregation of Enamel Matrix Proteins and Amelogenins in vitro
Physiological conditions during the secretory stage of amelogenesis have been well-characterized with respect to ionic strength and pH. Enamel fluid from developing pig teeth has been found to have a relatively high ionic strength of 165 mM (Aoba and Moreno, 1987). In addition, the pH within secretory enamel has been found to be close to neutrality (from 7.2 to 7.26), as determined directly in enamel fluid from the pig (Aoba and Moreno, 1987) and by pH measurement of reconstituted enamel strips from the rat (Smith et al., 1996). It has been estimated that, in the secretory stage of amelogenesis, the concentration of enamel matrix (mostly amelogenin) from pig and cow teeth is approximately 20–30 weight % (i.e., from 200 to 300 mg/mL) (Robinson et al., 1988; Fukae, 2002). Thus, amelogenins are likely to be present in solid or gel form under physiological conditions. Nevertheless, it is important to recognize that such insoluble forms of amelogenin would be in some form of equilibrium with soluble forms of amelogenin (Aoba and Moreno, 1989). Soluble and insoluble forms of amelogenin could possibly play distinct roles in the regulation and control of early enamel mineralization. The apparent solubility and aggregation properties of amelogenin are clearly related and affected by primary protein structure. It is likely that the secondary and tertiary structures of amelogenin will also affect these important properties.
Early in the course of studies of enamel formation, enamel matrix proteins were observed to exhibit a tendency to aggregate (e.g., see Eastoe, 1979). Most notably, it was shown that a major component of solubilized bovine enamel protein formed translucent gels at 4°C that reversibly transformed to an opaque material at room temperature (18°C) (Nikiforuk and Simmons, 1965). Interestingly, these authors suggested that this temperature-dependent aggregation may be related to the high proline content of enamel proteins, based on reported findings (e.g., Noguchi et al., 1957) that synthetic poly-L-proline II also undergoes heat-dependent aggregation. As noted in the INTRODUCTION, the basis for suggestions that supramolecular assemblies of enamel matrix proteins play a role in enamel mineral formation comes from TEM observations suggesting that filamentous (Travis and Glimcher, 1964; Smales, 1975) and spherical (Fearnhead, 1965; Robinson et al., 1981; Fincham et al., 1995) nanometer-sized (from 5 to 100 nm) structures are associated with developing enamel. Several hypotheses have been proposed that center around spherical aggregate structures of amelogenin (nanospheres) and their control of enamel mineral organization (Fincham et al., 1999; Robinson et al., 2003). In support of these hypotheses, in vitro studies (Fincham et al., 1994) with atomic force microscopy (AFM) showed that rM179 readily forms nanometer-sized particles on mica at pH 8, with a distribution of sizes ranging from 5 to 35 nm, similar to structures observed in vivo. TEM images of rM179 aggregates in Tris-HCl at pH 8 were also interpreted as being similar in appearance to the stippled structures observed in sections of the extracellular matrix of developing murine molar enamel. Despite this similarity, the observed aggregation of the enamel matrix (mostly amelogenin) in vivo may reflect the interaction between native amelogenins (full-length and specific degradation products) and other proteins. The authors of this 1994 study (Fincham et al., 1994) concluded that "amelogenin function in biomineralization may be mediated through these supramolecular structures rather than directly through the action(s) of discrete amelogenin molecules..." and noted that "without more precise information on the molecular structure of the amelogenin protein, the manner in which such aggregates may form and function remains unknown". More recently, it was proposed that the nanoparticles observed in developing enamel may also contain mineral precursors (e.g., particles of amorphous calcium phosphate) that could represent stabilized sub-units that subsequently fuse to form enamel crystals (Robinson et al., 2003), although the presence of transient mineral phases in forming enamel has not been well-established. Given the potential importance of such organic structures in the regulation of mineralization, this section will focus on the advances that have been made in our understanding of amelogenin aggregation, structure, and function, with particular attention to recent findings and concepts.
The development of reliable procedures to produce ample quantities of highly purified recombinant (Simmer et al., 1994) and native (Yamakoshi et al., 2003b) proteins has greatly facilitated both qualitative and quantitative studies of the aggregation and assembly of amelogenins in vitro. Initial characterizations of amelogenin assemblies, as noted above, were carried out by TEM and AFM. Subsequently, investigators used dynamic light-scattering (DLS) to provide information on the aggregate sizes of various amelogenins prepared under a variety of conditions (Table). DLS provides an indirect measure of particle size and distribution of sizes in solution, based on measurements of particle diffusion. The data presented in the Table generally represent results of mono-modal distribution analyses that yielded a measure of mean particle size, reported as hydrodynamic radii (RH), and the distribution of sizes (polydispersity) around that mean, represented by the standard deviation.
|
, collectively, shows that the RH of aggregate structures of rM179 in solution are significantly affected by pH. Under acidic conditions (pH < 6), the RH of rM179 were found to exhibit mean values of 3.9 to 8 nm over a wide temperature range of 5 to 40°C. Molecular-weight estimations (Moradian-Oldak et al., 1994) derived from the DLS data suggest that these aggregate structures of rM179 are comprised of 10 to 17 molecules. When similar measurements were carried out in a more hydrophobic solvent (60% acetonitrile), a RH of 2.2 nm was obtained, suggesting that the aggregation of rM179 involved hydrophobic interactions (Moradian-Oldak et al., 1994). However, the aggregation behavior of rM179 under neutral and basic conditions differed sharply from that observed under acidic conditions (Table). Despite the fact that rM179 was shown to have ’similar’ solubility behavior under acidic and basic conditions (Fig. 3a), aggregates of rM179 observed at pH 
8 had RH values from 2 to 5 times larger (i.e., from 15 to 22 nm) than those observed under acidic conditions in the temperature range of 5 to 34°C. At higher temperatures, the mean RH of rM179 was found (Moradian-Oldak et al., 1998a; Aichmayer et al., 2005) to increase significantly to around 60 nm (Table), although the temperature at which this transition took place differed in the noted studies (i.e., at 37°C and 45–46°C, respectively). Hence, under the latter conditions, particles of rM179 had RH values that were from 8 to 15 times larger than those observed under acidic conditions. Similarly, aggregate sizes of rM179 at neutral pH were found to be larger than those observed under acidic conditions (Moradian-Oldak et al., 1994, 1998a), although more disparate results were observed at pH 7.4 (Table). It was also reported (Moradian-Oldak et al., 1998a) that DLS data could not be obtained for rM179 at 1 mg/mL at pH 7.4, since the protein was found to precipitate at T > 18°C. Thus, authors of this latter study concluded that hydrophobic interactions resulting in the formation of amelogenin nanospheres were not optimal at pH 7.4. Apparently, for these reasons, fewer studies were carried out near neutrality, even though neutrality better represents the physiological condition. Analysis of the noted DLS data has provided insight into the influence of temperature and pH on the sizes of amelogenin aggregates, but little information on their structural features. However, DLS data have been used in an attempt to provide insight into the mechanism of amelogenin assembly. A bimodal analysis of DLS data (Fincham et al., 1998) suggested that rM179 in solution consisted of large and small particles at both pH 5.9 (i.e., 12.1 nm and 4.4 nm) and pH 8.0 (i.e., 20.3 nm and 5.6 nm) at 20°C. This led the investigators to speculate that the smaller nanospheres are sub-units of the larger nanospheres. However, this idea was not confirmed in a later AFM study (Wen et al., 2001), and it was suggested that nanosphere formation proceeds through the progressive accretion of amelogenin molecules. In subsequent studies, based on multimodal analyses of particle size distribution of native and recombinant amelogenins, authors from the same laboratory suggested the possibility that the larger particles result from the association of smaller particles, followed by their fusion (Moradian-Oldak et al., 2002, 2003). However, no direct evidence or mechanism for this fusion process was presented.
As shown in the lower portion of the Table, variable differences in RH were observed between native (P173) and recombinant amelogenin (rP172) counterparts, and between full-length amelogenins and their degradation products from both the mouse (rM179 vs. rM166) and the pig (P173 vs. P161 and P148) at pH 8. The full-length native pig protein (P173) was found to have a larger RH than its recombinant counterpart rP172 (Moradian-Oldak et al., 2002). The authors of this study attributed this difference to the presence of the phosphate group on serine-16 of the native amelogenin, within the N-terminal regions believed to play a critical role in protein-protein interactions (Paine and Snead, 1997), as discussed below. In addition, the removal of the 12- to 13-amino-acid hydrophilic C-terminus (see Fig. 2) from the parent molecules (rM179 
rM166; P173 P161) generally resulted in an increase in particle size that is likely due to an increase in protein hydrophobicity. The mean particle size of P148, however, was found to be substantially smaller than that of P161, which, like their difference in solubility, may again be related to the absence of the extremely hydrophobic 12-amino-acid domain (Fig. 2) in P148 that is present in P161.
Self-assembly is a common property of extracellular organic matrix macromolecules, facilitated by specific intermolecular interactions. Accordingly, several studies have focused on revealing the molecular recognition sites in the amelogenin molecule involved in amelogein-amelogenin interactions. Paine and Snead (1997), using a yeast-two-hybrid system, have determined that the 42-amino-acid N-terminal domain (domain "A") of murine amelogenin M180 and the 17-amino-acid C-terminal fragment from 156 to 173 (domain "B") are essential for amelogenin-amelogenin intermolecular interactions. Subsequent DLS and AFM studies (Moradian-Oldak et al., 2000) suggested that domain "A" was involved in molecular interactions for nanosphere formation, while domain "B" contributed to nanosphere stability, preventing the formation of larger assemblies. Later, the same authors demonstrated that proline in the "B" domain, located at the border of hydrophilic C-terminal teleopeptide (Fig. 2), plays a critical role in these intramolecular interactions (Paine et al., 2003a). The importance of "A" and "B" domains for enamel matrix assembly and mineralization was further confirmed in a series of elegant studies using transgenic mice with altered "A" and "B" domains (Paine et al., 2000, 2003b; Dunglas et al., 2002).
Recently, a combined approach of small-angle x-ray scattering (SAXS) and DLS was used for the study of amelogenin assembly (Aichmayer et al., 2005). This study took advantage of the fact that these two techniques are sensitive to different structural elements. SAXS directly reflects structural features related to electron-density differences (Fratzl, 2003), while DLS provides an indirect estimation of particle size based on diffusion characteristics. Two important and surprising observations were made during this study. First, the hydrodynamic radii (RH) of rM179 particles at pH 8 (Tris-HCl, 2 mg/mL), determined by DLS over the temperature range of 4°C to ~ 44°C, were 1.5 to 2 times larger than the particle radii determined by SAXS (RS) (Fig 4a). Second, a dramatic size increase was observed by DLS at a higher temperature (46°C), but not by SAXS (Fig. 4a). Since SAXS responds only to inhomogeneities in electron density, whereas DLS is related to the diffusion characteristics of the particle, it was proposed that, at pH 8 and below 46°C, amelogenin nanospheres consist of a dense protein core (visible by SAXS) surrounded by a loose protein shell, consistent with a larger hydrodynamic radius (RH), but invisible by SAXS due to its low electron density. The calculated numbers of protein molecules per particle (N) from SAXS and DLS analyses were nearly identical (Aichmayer et al., 2005) at lower temperatures (Fig. 4b). Thus, the particle density (g/cm3) determined by SAXS is significantly greater than that determined by DLS (Fig. 4c). Based on these data and the protein sequence of the full-length amelogenin, it was suggested that observed nanospheres are best described by a dense, predominantly hydrophobic, protein core, surrounded by a loose shell comprised of the negatively charged hydrophilic C-terminus that is exposed to the aqueous environment. It was suggested that this negatively charged surface prevents the aggregation of nanospheres at room temperature and below. Similar hydrophobic-core/hydrophilic-shell models have been previously proposed (Fukae, 2002; Snead, 2003), although sufficient experimental evidence in support of such structures has not been obtained.
|
) to explain the onset of amelogenin nanosphere aggregation, based on these data. At pH 8 and low temperatures, nanospheres of rM179 are stabilized by an outer shell comprised of hydrophilic protein segments. As the temperature increases, these charged protein segments condense onto the core, leading to a slight decrease in RH (observed by DLS) and a slight increase in RS (observed by SAXS) (Fig. 4a). This results in a reduction in repulsion between the spheres and in the onset of agglomeration and the formation of chain-like structures, rather than the fusion of particles as was previously proposed (Moradian-Oldak et al., 2002, 2003). The results of the combined DLS/SAXS study provide direct evidence showing that the full-length amelogenin can aggregate, through a process mediated by its hydrophilic C-terminus, to form assemblages (e.g., chain-like structures). Chains of rM179 nanospheres of similar size were also recently observed in vitro by TEM (pH 7.5-8, 37°C) (Beniash et al., 2005), although their formation was found to depend on the presence of calcium ions. It has also been reported (Du et al., 2005) that the full-length recombinant pig amelogenin (rP172) forms chains of nanospheres under conditions designed to induce protein crystallization and in small amounts in water (pH ~ 6.5, 20°C). Collectively, these observations provide evidence to support the suggestion that the full-length amelogenin nanospheres can interact through hydrophilic domains, and possibly assemble into higher-order structures that could guide the formation of organized arrays of enamel crystals as seen in developing enamel, consistent with the yeast-two-hybrid findings (Paine and Snead, 1997). Clearly, the use of SAXS has provided new insight into the structure and mechanism of amelogenin aggregation.
|
Secondary Structure
CD Spectroscopy
In a pioneering structural study, Renugopalakrishnan et al.(1986) used CD and FTIR spectroscopy to monitor structural changes in bovine amelogenin in solution. The amino acid sequence reported in that paper was based on the direct amino acid sequencing by Takagi et al.(1984). This sequence lacks a few repeat domains in the midsection of the protein, compared with the deduced sequence of the bovine amelogenin from AMELX gene (Gibson et al., 1991) (Fig. 2). Based on a comparison of the reported and deduced sequences, it appears that the protein studied by Renugopalakrishnan et al.(1986) is the bovine amelogenin (B184) that lacks the 13 C-terminal amino acid fragment (Fig. 2). The CD spectra were taken at different pH and Ca2+ concentrations. Analysis of the CD spectra revealed that bovine amelogenin consists predominantly of ß-sheet, ß-turn, and random coil fractions. Under all conditions, the analyses determined that 
-helix was the minor fraction. Both pH changes and the addition of Ca2+ ions significantly affected amelogenin conformation. Later, Goto et al.(1993) performed CD spectroscopic studies of the full-length porcine amelogenin, its fragments, and synthetic mimics. Based on these CD data, the authors concluded that amelogenin consists of three structural modules, corresponding to the distinct regions of its amino acid sequence (see Fig. 2): the N-terminal TRAP-ß-sheet domain; the center section, adopting an extended polyproline II or ß-turn rich structures, such as ß-spirals; and the C-terminal domain as a random coil (Goto et al., 1993; Matsushima, et al., 1998). (Distinguishing between PPII and ß-turn structures in CD spectra is a challenging task, especially in proteins consisting of different structural domains [Wellman et al., 1992; Ladokhin et al., 1999].)
CD studies of other synthetic peptides mimicking fragments of amelogenin have also been reported. Renugopalakrishnan (2002) reported a CD spectroscopic study of a 27-amino-acid synthetic polypeptide, reproducing the center domain of bovine amelogenin. The author reported significant changes in CD spectra, with an increase in temperature that is characteristic of polyproline II and type I ß-turn structures (Noguchi et al., 1957). The analysis of the spectra suggested a high number (up to 70%) of ß-turns in the structure. Based on the spectroscopic data, the author proposed that the peptide adopts a ß-spiral structure. In another CD study, (Leu-Gln-Pro)n peptides, also mimicking the center section of bovine amelogenin, were investigated (Sogah et al., 1994). When solubilized in TFE, the peptide existed primarily in an extended unordered state; however, the addition of multivalent metal ions (e.g., Ca2+, Mg2+, Sr2+, and Ba2+) significantly increased the organization of the polypeptide, leading to the formation of repeating class C type I ß-turn structures. Interestingly, the addition of monovalent metal ions did not affect the structural organization of the peptide. XRD studies of the amelogenin mimic peptide support the notion that the midsection of amelogenin adopts an extended conformation with a 165 helix secondary structure and exists as isolated "rigid rods" (Sogah et al., 1994).
NMR Spectroscopy
A 1H-NMR CIDNP study of the full-length porcine amelogenin and its cleavage products (Aoba et al., 1990) suggested that C-terminal Trp161 is exposed to solvent as well as Tyr residues in the N-terminal TRAP domain. Studies of His residues demonstrate that their exposure to solvent is different in different cleavage products. Together, these results suggest that both the N- and C-terminal domains of amelogenin are exposed to the solvent, and that the molecular structure of amelogenin can change upon cleavage. In this paper, solutions of the full-length amelogenin (8.33 mg/mL) and its fragments (5 mg/mL) were analyzed at pH 5.2 in D2O.
Vibrational Spectroscopy (FTIR, FT-Raman)
Vibrational spectroscopy has been widely used for the assessment of structural characteristics of amelogenins. Renugopalakrishnan et al. (1986), using FTIR, carried out a comprehensive structural analysis of bovine amelogenin that lacks the 13-amino-acid C-terminal domain (B184). Proteins were examined dry as well as in an aqueous solution with pH values from 1.8 to 9.23. Specific absorbance bands were identified by deconvolution. Analysis of data obtained in this study suggests that the amelogenin contains primarily ß-sheet and extended repetitive ß-turn structures. The authors also showed that the spectroscopic data were sensitive to changes in pH, which might reflect both pH-induced intramolecular conformational changes, as well as changes in structure that are induced by intermolecular interactions associated with aggregation. In a later report, the same group of investigators (Zheng et al., 1987), using Raman spectroscopy, confirmed a mixed ß-sheet/ß-turn structure of bovine amelogenin. Jodaikin et al. (1987) also studied guanidine-purified and -lyophilized rat amelogenins, prepared as previously described (Termine et al., 1980) and concluded that the protein has no regularly ordered secondary structure (i.e., random coil). The authors contributed this finding to the fact that the forming enamel matrix contains proteolytically degraded amelogenins. More recently, Aoba et al. (2001) carried out FTIR spectroscopic studies of porcine amelogenin (P148) in solution as a function of temperature. This protein lacks the 25-amino-acid C-terminal domain. The results of this study suggested that amelogenin contains multiple structural motifs, including ß-sheet, ß-turn, helices, and random coil.
X-ray Crystallography
To date, no one has been able to discern an x-ray crystallographic structure of pure amelogenin that could provide insight into both its solution and aggregation behavior. In a recent study (Du et al., 2005), authors reported an x-ray diffraction (XRD) pattern of recombinant amelogenin birefringent microribbons. However, upon further analysis, this pattern was later identified as diffraction from a cellulose fibril contaminant (Du et al., 2005).
Tertiary Structure
Matsushima et al. (1998) performed a synchrotron SAXS study of the ’20K’ porcine amelogenin that lacks the 25-amino-acid C-terminus (P148) in acidic solution at 5°C. Based on the scattering properties of the protein at different concentrations, the authors concluded that individual amelogenin molecules in solution adopt asymmetric rod-like or ellipsoid shapes. Consistent with this report, it was also found, by SAXS (Aichmayer et al., 2005), that rM179 at pH < 3.5 and at pH 4.5 (20°C), as well as rM166 at pH < 6, gave very weak x-ray scattering, suggesting the presence of monomers or dimers. The elongated shapes of the amelogenin molecules were later confirmed by filtration studies (Fukae, 2002). In these experiments, 20-kDa porcine amelogenin could pass through a 3-kDa cutoff membrane, suggesting that this molecule adopts a highly anisotropic form.
Molecular Modeling
Chou-Fasman calculations of the bovine amelogenin sequence (Renugopalakrishnan, 1986) strongly suggest that amelogenin is a modular protein with an N-terminal TRAP domain adopting largely ß-sheet conformation, whereas the midsection of the protein represents an extended structure which consists of ß-turns and polyproline-type structures. At the same time, the Chou-Fasman analysis of the amelogenin sequence has demonstrated a low probability of -helical conformation in this protein. In a later study from the same lab, molecular mechanics calculations and molecular dynamics simulation were used to model structural organization of a 27-amino-acid segment from the center section of bovine amelogenin. The results of these in silico studies suggested that the central domain, comprised of 9 Gln-Pro-Xxx repetitive motifs, form a series of ß-turns adopting ß-spiral conformation. The models also strongly favored the formation of hydrogen bonds between side-chain carboxyl-amides of glutamines and the backbone carbonyls of prolines, which stabilize the structure. These models are consistent with CD and vibrational spectra data. Interestingly, later independent studies of poly-L-proline type II (PPII) helical structures in a variety of proteins have demonstrated that Gln is the most common residue in these structures after Pro, due to the ability of Gln to form hydrogen bonds between side-chain carbonylamides of Gln with the backbone carbonyls of Pro, which helps to stabilize the PPII structure (Stapley and Creamer, 1999; Kelly et al., 2001).
Matsushima et al.(1998) have developed a computer model of the 20-kDa porcine amelogenin that lacks the 25-amino-acid C-terminal domain. Initial parameters of the model were based on the SAXS and CD data (Goto et al., 1993; Matsushima et al., 1998). The model predicted that the molecule adopts an extended conformation, referred to by the authors as a ’folded bundle’ structure, comprised of the TRAP fragment adopting an extended ß-strand structure, and Pro- and Gln-rich midsection in an extended PPII conformation. They proposed that the molecule folded on itself in two places, with amino acid sequences strongly favoring the formation of ß-turns. The resulting molecular model of the 20-kDa amelogenin consisted of an extended ß-strand and two PPII helical motifs, co-aligned with the long axis of the molecule and connected by two ß-turns.
Summary of Analyses of Amelogenin Structures in Solution
Collectively, the results of structural analyses reported here suggest that amelogenins are modular, moderately hydrophobic proteins, with their primary features conserved across different classes of Tetrapoda. In general, large portions of amelogenin adopt extended secondary structures such as ß-sheet/ß-strand (TRAP) and ß-spiral/PPII structures (midsection). This conclusion is strengthened by the SAXS analysis of the tertiary structure of amelogenin, demonstrating that individual molecules in a moderately acidic solution exist in a form of anisotropic rods (Matsushima et al., 1998). Such extended conformations would enhance both protein-protein (Williamson, 1994) and protein-mineral (Xu and Evans, 1999; Zhang et al., 2000; Evans, 2003) interactions, consistent with the functional role of amelogenin. In particular, the midsection of amelogenin, rich in Pro and Glu ß-spiral/PPII repetitive motifs, is very similar to structures found in other self-assembling proteins that form viscous gels (Matsushima et al., 1990; Urry, 1993). Extended ß-spiral/PPII Pro-rich motifs are also widespread in the binding sites of signaling proteins (Williamson, 1994; Kay et al., 2000), which may explain the suggested signaling effect of amelogenin breakdown and splice products, described by several authors (e.g., Veis, 2003). Another general characteristic of amelogenin, which likely has important functional implications, is that its secondary structures are affected by pH, temperature, and the presence of multivalent metal ions. However, it is not yet clear if or how these structural changes are related to the aggregation process (and the regulation of crystal growth), although amelogenin aggregation is similarly influenced by the same parameters.
(e) Structural Characteristics of the Enamel Matrix and Amelogenin Gels
Attempts have been made to characterize the structure of the enamel matrix and amelogenins in the gel state, since, as noted, this may best represent the physiological condition. The question remains the same: Do these entities have structural characteristics that would be consistent with the formation of highly oriented arrays of enamel crystals within the developing enamel matrix? As discussed here, several approaches have been used to answer this critical question.
Several studies have attempted to determine the structural characteristics of the developing enamel matrix by x-ray crystallography. Attempts to determine structural characteristics of demineralized and fixed enamel matrices by x-ray diffraction were first performed in the early 1960s (Glimcher et al., 1961; Pautard, 1961), followed by several studies over the next decade. A common feature of the diffraction patterns collected from the developing enamel matrix and reconstituted enamel matrix gels was the presence of strong reflections, with d-spacings of 4.65–4.7 Å and 9.6–10 Å, characteristic of ß-sheet structures (Glimcher et al., 1961, 1965a; Pautard, 1961; Bonar et al., 1965; Höhling, 1965; Angmar-Månsson, 1971; Jodaikin et al., 1986).
Interestingly, several authors reported oriented diffraction patterns, while others reported homogeneous diffraction with no preferred orientation. This controversy resulted in a fired debate and led to the development of two hypotheses regarding orientation of enamel proteins in the developing matrix. One hypothesis, based on the characteristic oriented x-ray diffraction patterns from demineralized enamel matrices and filaments prepared from reconstituted enamel matrix protein gels, proposed that proteins in the enamel matrix form filaments and adopt a cross-ß structure (the same as those found in ß-amyloid fibrils) (Glimcher et al., 1961, 1965a; Pautard, 1961; Bonar et al., 1965). This hypothesis was further supported by TEM studies (Travis and Glimcher, 1964) that described networks of 5-nm-thick filaments surrounding forming enamel crystals. Other researchers, however, have suggested that the enamel matrix is an isotropic thixotropic gel (a consistency that is gel-like at rest, but fluid when agitated, having, simultaneously, high static shear strength and low dynamic shear strength, and losing viscosity under stress) (Fearnhead, 1965; Ångmar-Mansson, 1971), based on their own x-ray diffraction (Fearnhead, 1965; Ångmar-Mansson, 1971), TEM (Fearnhead, 1965), and optical birefringence (Sundström, 1966; Ångmar-Mansson, 1971) studies. Reverberations of these earlier disputes continue, due, in part, to these conflicting data. Such studies are complicated by the fact that the alignment of enamel matrix proteins is likely susceptible to mechanical forces (both tensile and shear). It has been suggested that matrix orientation can be induced by stretching the samples or by other factors, such as demineralization, drying, etc. (Bonar, 1965; Pautard, 1965; Ångmar-Mansson, 1971; Jodaikin et al., 1986). Thus, the question of the organization of the enamel matrix remains open.
More recently, to improve our understanding of the structure of the enamel matrix, a group of investigators studied the microstructure of fixed translucent and opaque gels of amelogenin prepared from developing porcine enamel extracts (primarily P148) at 4°C and 24°C, respectively, using AFM, TEM, and SEM (Wen et al., 1999b). These gels were similar to those originally described decades ago (Nikiforuk and Simmons, 1965). In general, the porcine amelogenin gels were shown to be comprised of assemblies of quasi-spherical nanospheres of amelogenin. It was reported that assemblies of nanospheres with diameters of 8–20 nm further assemble to form spherical structures 40-70 nm in diameter, and that these structures assembled to form larger spherical structures of 70–300 nm in diameter. The clear gel at 4°C appeared dense and was comprised of uniformly dispersed particles (< 150 nm), while the opaque gel at 24°C was comprised mostly of larger spherical assemblies (150–200 nm). The opaque gel was also completely porous, with the presence of numerous separate cavities ranging from 0.3 to 0.7 µm. It was presumed that these cavities were fluid-filled, thus giving rise to their opaque appearance. Similar clear to opaque transitions were observed (Wen et al., 2001) in gels of purified rM179; however, both clear and opaque gels were comprised of 15- to 20-nm nanospheres, and the opaque gels had smaller cavities (~ 0.1 µm). These differences were attributed to differences in amelogenin composition used in the two studies. Although these collective observations have provided insight into the subunit structure of fixed amelogenin gels, no evidence for hierarchical organization can be gleaned from these data.
Thus, it has not been shown convincingly that the enamel matrix or gels of amelogenin adopt extended higher-order or oriented assemblies. The present data may, therefore, be more consistent with the formation of an amorphous gel with properties that could potentially allow matrix molecules to reassemble and organize upon some form of mechanical stimulation (e.g., mineral growth or tension applied by ameloblasts), in a fashion that would help regulate the oriented growth of enamel mineral crystals. The basis for this idea was presented over 40 years ago (Eastoe, 1963), when Eastoe suggested that the enamel matrix "must either be capable of diffusional movement or alternatively unstable" and that "possession of thixotropic properties" would cause protein flow due to "local increase in pressure" caused by the rapid growth of apatite crystallites. This suggested critical interplay between crystal growth and matrix assembly is also consistent with a recently developed model (Colfen and Mann, 2003) based on in vitro studies. These studies have elegantly demonstrated that interplay between self-assembly of organic molecules and mineralization, and the subsequent cooperative re-organization of hybrid inorganic-organic nanostructures, can give rise to the formation of hierarchical structures. In fact, these investigators have suggested that such interplay may have a direct bearing on the mechanism of dental enamel formation. Support for this mechanism was recently obtained (Beniash et al., 2005) with the use of recombinant amelogenins in vitro, as discussed below. As a further test of this hypothesis, however, microstructural studies of amelogenin gels are needed under conditions that support in vitro mineralization, as was previously suggested (Wen et al., 1999b).
(f) Modulation of Calcium Phosphate Crystal Growth and Organization by Enamel Proteins
Although the present review is focused primarily on the role of amelogenin self-assembly in the structural organization of the developing enamel tissue, it is important to note that several in vitro studies have been carried out in attempts to assess the functional role of enamel matrix proteins as regulators of crystal nucleation and growth. As noted earlier, it has been suggested (Weiner, 1986) that, in biomineralization systems, hydrophobic molecules (like amelogenin) provide a skeletal or space-filling structure, while (acidic) hydrophilic molecules (like enamelin and ameloblastin) are involved in the regulation of crystal nucleation and growth. Nevertheless, it has been reported (Aoba et al., 1987; Aoba and Moreno, 1991) that the 25-kDa amelogenin from the pig adsorbs onto HA surfaces and partially inhibits crystal growth of HA seed crystals when added to a solution supersaturated with respect to HA. Importantly, the 20-kDa and 13-kDa protein degradation products of the parent 25-kDa molecule had greatly reduced capacities to adsorb onto HA and inhibit crystal growth. Similar results showing a modest effect of recombinant amelogenins on seeded crystal growth have been reported more recently (Moradian-Oldak et al., 1998b). In the latter study, it was also shown that the full-length recombinant amelogenin promoted the adhesion of HA crystals. However, it has been reported (Hunter et al., 1999) that the full-length recombinant amelogenin from the mouse (rM179, analogous to the 25-kDa protein) has no effect on de novo (spontaneous) HA formation, although these studies were carried out up to a protein concentration of only 30 µg/mL. It remains unclear at this time if the 25-kDa amelogenin functions as an inhibitor during the growth of the initially formed enamel ribbons, or if other enamel proteins are more likely to fulfill this function. With deproteinized bovine enamel crystals used as seed material, it was found (Doi et al., 1984) that isolated fetal bovine amelogenins (27 kDa, 22 kDa, and 16 kDa) and "enamelins" (a 40- to 70-kDa complex of proteins) were both effective in inhibiting seeded crystal growth. Smaller fragments of amelogenin (5 and 10 kDa) showed no significant inhibitory activity. It has also been reported (Hunter et al., 1999) that rM179 in an agarose gel has no effect on HA nucleation, although, in a recent study, it was reported (Bouropoulos and Moradian-Oldak, 2004) that the 32-kDa fragment of enamelin promotes the induction of apatite formation on a gelatin gel, but only in association with amelogenins. A similar group of 32-kDa protein fragments isolated from the pig were previously shown to adsorb strongly onto HA and to be a potent inhibitor of HA-seeded crystal growth (Tanabe et al., 1990). Hence, analysis of the data at hand suggests that soluble portions of the full-length and partially degraded amelogenins (as inhibitors) or fragments of enamelin (as a nucleator or inhibitor) may serve to regulate enamel crystal growth. Nevertheless, based on these findings, and the fact that a poorly organized enamel mineral layer forms in the amelogenin knockout mouse (Gibson et al., 2001), it is clear that a major function of amelogenin is to guide the organization and shape of the enamel crystallites as they develop in the matrix.
Several in vitro studies have also been carried out to examine the influence of amelogenins on calcium phosphate crystal formation under conditions of spontaneous precipitation. Such studies are deemed relevant, since there is no evidence that mineralization within the enamel matrix is triggered by the heterogenous nucleation of a metastable (calcium phosphate) aqueous phase of the extracellular matrix. Accordingly, it has been shown that a mixture of native amelogenins from the pig can modulate the crystal morphology of spontaneously formed octacalcium phosphate (a prototype for initial enamel mineral and a precursor of hydroxyapatite), when contained within gelatin gels, resulting in longer crystals with an increase in aspect (length/width) ratio (Wen et al., 2000a). These results suggested that such shape modification was brought about by selective adsorption of amelogenins on the growing crystal faces normal to the (long) c-axis. In addition, the same general effect has been observed with the use of gels of mixtures of bovine amelogenin (Iijima et al., 2001) and purified recombinant mouse amelogenins (Iijima et al., 2002). Such effects were subsequently shown to be dose-dependent (Iijima and Moradian-Oldak, 2004). A similar increase in aspect ratio of apatitic crystals was observed in supersaturated solutions that contained either rM179 or rM166 (Beniash et al., 2005). Despite the significant influence of the hydrophilic C-terminus in promoting amelogenin interactions with pre-formed crystals of HA (e.g., Doi et al., 1984; Aoba et al., 1987; Aoba and Moreno, 1991), it is apparent from these latter results that the noted shape regulation of growing HA crystals is controlled by hydrophobic portions of the amelogenin molecule. Consistent with these results, it has recently been shown (Habelitz et al., 2004) that nanoparticles from recombinant human amelogenin gels specifically adsorb onto (hk0) planes (crystal faces parallel to the c-axes) of large rod-like fluoroapatite crystals contained within a glass ceramic. This selective adsorption similarly resulted in a significant reduction in crystal growth at these surfaces, and in an increased rate of growth in the direction of the c-axes, when this material was exposed to a supersaturated calcium phosphate solution. Interestingly, it was shown, by AFM, that the selectively adsorbed amelogenin consisted of short strings of nanoparticles (from 4 to 8 nanoparticles, 30 to 60 nm in diameter) that were co-aligned parallel to the c-axes of the fluoroapatite crystals (Habelitz et al., 2004).
Importantly, recent in vitro studies (Beniash et al., 2005), under conditions that again support the spontaneous formation of calcium phosphate crystals, have provided unique insight into the potential ability of amelogenin to regulate their organization. When monomeric forms of rM179 were allowed to assemble simultaneously with the induction of mineral formation (by simultaneously increasing solution pH [from pH 4 to 7.5–8.0] and temperature [from 4° to 37°C]), bundles of crystals with their c-axes preferentially oriented were formed (Fig. 6a). A mean angular spread of 29 ± 14° was obtained, implying a strong preferred orientation of the crystals inside the bundles, which is comparable with, although not quite as good as, that in native enamel (Nylen et al., 1963; Glimcher et al., 1965b). However, pre-assembled rM179, as well as rM166, although capable of regulating crystal shape as described above, had no influence on crystal organization (Figs. 6b, 6c). These results strongly suggest that parallel arrays of calcium phosphate crystals, as seen in this study and in the early stages of enamel formation, result from a cooperative interaction between growing mineral crystals and assembling enamel proteins, by a process similar to that proposed for the formation of artificial nanostructured materials (Colfen and Mann, 2003). These results further highlight the unique capability of the full-length amelogenin and suggest that the hydrophilic C-terminus plays a critical role in regulating the organic-mineral assembly and the formation of parallel arrays of forming crystals. It has also been reported that recombinant amelogenins (rM179 and rP172) modulate the (epitaxial) formation of apatite crystals grown on bioactive Bioglass®, resulting in the formation of bundles of long apatitic crystals oriented in a parallel fashion (Wen et al., 1999a, 2000b).
|
|
(III) BIOMIMETIC APPROACHES TO PRODUCING HIERARCHICALLY ORGANIZED INORGANIC STRUCTURES |
|---|
|
TOPABSTRACT(I) INTRODUCTION(II) AMELOGENESIS(III) BIOMIMETIC APPROACHES TO...(IV) CONCLUDING REMARKSREFERENCES |
|---|
Biological materials such as enamel exhibit intricate architecture and outstanding physical properties, unobtainable by traditional methods of materials synthesis. Such biominerals bear little resemblance to the unit cell from which they are constituted, and patterning on the nanometer to macrometer length scale is achieved under mild physiological conditions. Although attempts to replicate the detailed structural and functional characteristics of biominerals have failed thus far, significant progress has been made in recent years. Motivation for biomimetic synthesis is largely driven by the desire to reproduce exceptional physical properties, such as the hardness and fracture resistance displayed by some mineralized tissues. Success in this regard may be expected to lead to the availability of better synthetic materials for the repair of dental hard tissues and bone, for example.
As described above for dental enamel, nature engineers biological hard tissues through interactions with macromol
