Induction of Apatite by the Cooperative Effect of Amelogenin and the 32-kDa Enamelin

Extracellular matrix proteins are considered to play essential roles in controlling the nucleation, growth, and organization of hydroxyapatite crystals during enamel formation. The effects of amelogenin and the 32-kDa enamelin proteins on apatite nucleation were investigated by a steady-state gel diffusion device containing 10% gelatin gels loaded with 0, 0.75%, and 1.5% (w/w) native porcine amelogenins. It was found that the induction time for hydroxyapatite precipitation was strongly increased by the presence of amelogenins, suggesting an inhibitory effect of apatite nucleation. Addition of 18 μg/mL of 32-kDa enamelin to 10% gelatin also caused inhibition of nucleation. Remarkably, addition of 18 and 80 μg/mL of 32-kDa enamelin in gels containing 1.5% amelogenin accelerated the nucleation process in a dose-dependent manner. Our observations strongly suggest that the 32-kDa enamelin and amelogenins cooperate to promote nucleation of apatite crystals and propose a possible novel mechanism of mineral nucleation during enamel biomineralization.

INTRODUCTION

Fundamental concepts related to the mechanisms of dental enamel mineralization are common among many mineralized tissues (Lowenstam and Weiner, 1989). Dental enamel is the most highly mineralized tissue in the vertebrate body and is comprised of carbonated hydroxyapatite crystals that are unusually large and elongated compared with apatite crystals in other skeletal tissues (Daculsi et al., 1984). Enamel mineralization is an extracellular event that takes place in an amelogenin-rich matrix constituting more than 90% of the total protein components (Termine et al., 1980). The remaining matrix is composed of acidic enamelins, proteinases, and ameloblastins (Fincham et al., 1999). Numerous structural studies have supported the notion that amelogenins self-assemble to form nanosphere structures that further assemble to form a supramolecular framework serving as a scaffold for the initiation and oriented growth of enamel apatite crystals (Robinson et al., 1998; Fincham et al., 1999; Wen et al., 1999; Paine et al., 2000). While several authors have supported the concept of enamel crystal orientation and habit being controlled through an organized assembly of the constituent proteins of the extracellular matrix, the mechanisms for crystal nucleation remain unclear and controversial. Two different scenarios have been proposed: (1) Crystal nucleation occurs at the DEJ and is promoted by either one of the components of dentin extracellular matrix or dentin crystals (Arsenault and Robinson, 1989; MacDougall et al., 1997); and (2) crystal nucleation is independent of dentin mineral formation and is modulated by the components of the enamel extracellular matrix (Diekwisch et al., 1995). Multiple organized nucleation sites (nanoscale precursor subunits) composed of amelogenins, enamelins, ameloblastins, and calcium-phosphate ions have also been proposed to serve as nucleators of enamel apatite crystals (Robinson et al., 2003). Enamelins that were originally defined as non-amelogenin acidic proteins (Termine et al., 1980) are ameloblastin-specific (Hu et al., 2000). The localization of enamelin proteolytic products has suggested their close association to enamel apatite crystal faces and their potential involvement in controlling nucleation and possibly growth modulation of the crystallites (Uchida et al., 1991). The conservation of enamelin structural features—namely, glycosylation and phosphorylation sites on the 32-kDa enamelin in the human, mouse, and pig—suggests those structural domains to be important functional domains of the protein (Yamakoshi et al., 1998; Hu et al., 2000). The putative spatial relationship of enamelin protein with the apatite crystallographic axis (Jodaikin et al., 1986) and its high affinity to interact with apatite crystals support our hypothesis that, among other enamel proteins of the extracellular matrix, the 32-kDa enamelin is the most appropriate candidate to act as crystal nucleator of enamel apatite (Doi et al., 1984; Tanabe et al., 1990).

Amelogenin and enamelin have been shown to be critical for normal enamel formation, as documented by recent transgenic and null mice model studies (Paine et al., 2000; Gibson et al., 2001), as well as by linkage analysis of human pedigrees with defective tooth enamel formation, called amelogenesis imperfecta (Kindelan et al., 2000; Kida et al., 2002). While such studies prove that these extracellular matrix proteins play critical physiological functions in the formation of enamel formation, clear insight into the mechanism of their action is still lacking.

Considering the scenario that nucleation of enamel crystals is independent of apatite crystal nucleation in dentin, we have implemented a steady-state gel-diffusion in vitro experimental system that was originally developed by Hunter et al.(1986) to assess the effect of the 32-kDa enamelin on hydroxyapatite nucleation in 10% gelatin gel. The present study was aimed at examining the hypothesis that the 32-kDa, the most stable proteolytically cleaved enamelin, has the potential to serve as a nucleator of enamel apatite crystallites. The use of gelatin gel allowed amelogenin and enamelin to be applied at different concentrations, and therefore dose-dependency of nucleation by these proteins could be evaluated. It was observed that while incorporation of amelogenin into the gel caused delay in apatite nucleation, addition of enamelin to the amelogenin/gel mixture system enhanced the induction, suggesting a cooperative effect between the two proteins in promoting the nucleation of apatite crystals.

MATERIALS & METHODS

Protein Extraction

Enamel scrapings were collected from unerupted fourth and fifth mandibular molars of fresh six-month-old pig jaws. The pig mandibles were obtained fresh from a local slaughterhouse (Farmer John Clougherty Co., Los Angeles, CA, USA) through Sierra For Medical Science (Santa Fe Springs, CA, USA). The United States Department of Agriculture has inspected and approved the process of sample harvesting. Proteins were isolated from porcine developing enamel by a slight modification of the dissociative extraction technique described by Termine et al.(1980).

Amelogenins were extracted in guanidinium hydrochloride (4M)-Tris (50 mM) at pH 7.4, de-salted, and concentrated by Amicon ultrafiltration, with a 10-kDa cut-off Amicon YM-10 membrane against 0.5% formic acid. The protein solution was lyophilized and stored in −20°C. The extract was characterized by SDS-PAGE to be a mixture of secreted amelogenin (25 K, 7.4%) and its processed products (23 K, 10.7%; 20 K, 49.5%; < 18 kDa, 32.4%) (Wen et al., 1999) (Fig. 1A, lane 2).

The 32-kDa enamelin was extracted from the pellet after the addition of 4 M guanidine-HCl for the isolation of amelogenins as described by Termine et al.(1980). Further purification was performed by a combination of ammonium sulfate precipitation and reverse-phase HPLC, with slight modification of the procedure developed by Yamakoshi (1995). The pellet was suspended in 4 M guanidine-HCl, 0.05 M Tris, pH 7.4, 0.5 M EDTA for two days. The soluble fraction was de-salted against 0.5% formic acid, and the enamelins were precipitated at 65% ammonium sulfate saturation after precipitation of amelogenin residues at 40% saturation. The 32-kDa enamelin was further purified by reverse-phase HPLC (C18, Vydac) with a gradient of 30–95% B (A, 0.1% TFA; B, 60% acetonitrile in 0.1% TFA) in 60 min. The protein which was eluted at 55–60% B appeared as one major band around 32 kDa on the SDS-PAGE gel (Fig. 1A, lane 3).

Apatite Nucleation

The experiments were carried out according to the gel-diffusion technique developed by Hunter et al.(1986) with slight modifications (Fig. 1B). Stock solutions of 2 M NaCl, 1 M Tris-HCl, pH = 8, 2 M CaCl₂, and 2 M NaHPO₄ were prepared in de-ionized water with reagent-grade crystalline reagents. Three types of working solutions containing 0.01% sodium azide were prepared with the following compositions: (Solution A) 10 mM Ca, 50 mM NaCl, 100 mM Tris-HCl, pH = 7.4; (Solution B) 6 mM P, 50 mM NaCl, 100 mM Tris-HCl, pH = 7.4; and (Solution C) 50 mM NaCl, 100 mM Tris-HCl, pH = 7.4.

Gelatin-amelogenin gels

Using the following procedure, we prepared 10% gelatin gels containing 0, 0.75, and 1.5% amelogenins. Samples (containing phosphate ions) and control (without phosphate ions) were prepared in triplicate. An appropriate amount (30 or 15 mg) of the lyophilized amelogenins was mixed with 0.150 mL of solution B, and the resulting mixture was vortexed for 15 sec. The same procedure was repeated with solution C. Gelatin powder (BioRad, Hercules, CA, USA) was mixed with solution B or C to produce a 10.9% w/v mixture. The pH was adjusted to 7.4 by the addition of NaOH via a combination glass electrode. Finally, 1.85 mL of the 10.9% w/v gelatin gel was mixed with 0.15 mL matrix (solution B with amelogenin), and 0.4 mL of the produced mixture was transferred into plastic wells. The gels were stored overnight at 4°C and used the next day for the nucleation experiments.

The effect of amelogenin

The gel slabs were equilibrated at room temperature for 3 hrs and overlaid with 2.5 mL of solution A containing 10 mM Ca plus 1.25 μCi ⁴⁵CaCl₂ (Perkin Elmer Life Sciences, Boston, MA, USA). Aliquots of 50 μL of the upper part of each well were transferred into scintillation vials containing 4 mL of scintillation liquid (Bio-Safe RPI, Chicago, IL, USA), and the radioactivity was measured in a Beckman liquid scintillation counter (Beckman LS-5801, Beckman Instruments, Fullerton, CA, USA). Samples were collected every 1–2 hrs during the first day and then after 12 and 24 hrs. All experiments were carried out at room temperature for 5 days. The results are expressed as the ratio of ⁴⁵Ca radioactivity (counts per min) measured at time t to that measured at time zero and represent the mean ± standard deviation of 3 measurements. Comparisons between the sample and the control for each measurement were made by Student’s t test, and statistically significant differences were defined at p < 0.05. Induction time was defined as the time when the comparison between the sample (phosphate-containing gel) and the control for each measurement was statistically significant.

The effect of the 32-kDa enamelin

Enamelin was adsorbed onto the gelatin gel by 35 μL of enamelin solution (18 or 80 μg/mL) being spread on top of the gel. The gel was then covered with calcium solution containing ⁴⁵Ca, and the calcium uptake was measured as described above. In parallel experiments, phosphovitin was used as a positive control for nucleation.

Transmission electron microscopy

Apatite precipitates were recovered from the gel-solution interface, heated at 40°C, and centrifuged immediately. The supernatant was discarded, and a small piece of the pellet was re-suspended in ethanol, placed on carbon-backed 300-mesh Parlodion-coated copper grids, and air-dried as previously described (Moradian-Oldak et al., 1991). The specimens were examined by a JEOL TEM (JEM1200-EX, Tokyo, Japan) operated at 80 kV. Selected-area electron diffraction was performed with an aperture size of 20 μm.

RESULTS

A thin layer of apatite crystals was formed at the interface between the gel and the liquid overlaid on the surface of the gel (Fig. 1B). A representative TEM image of crystals formed in 10% gelatin gel containing 1.5% amelogenin and the corresponding electron diffraction pattern is presented in Fig. 2 . Morphological analysis of the images revealed elongated thin crystals about 100 nm long. The arrowhead reflections 1 and 2 correspond to lattice spacing of 3.44 and 2.74 Å, respectively, which are characteristic of hydroxyapatite (Fig. 2 ).

Induction time was defined as the time when the difference between the sample (the gel containing phosphate ions) and the control (without phosphate) for each measurement was statistically significant (p < 0.05) (the differences in the curves in Fig. 3 ). This difference represents the difference between the quantity of calcium ions diffused into the gel and the calcium ions that participated in apatite induction. This induction time represents the kinetics of a new phase formation and is considered to be inversely proportional to the nucleation rate (Mullin, 1997). The induction times of apatite formation in the absence and presence of amelogenin and enamelin were determined based on the percentage ratios of ⁴⁵Ca measured at time t to that measured at time zero as the function of time (Fig. 3).

We found that the addition of 0.75% amelogenin into 10% gelatin doubled the induction time, while increasing the amelogenin concentration to 1.5% caused an almost four-fold delay in the induction time (Table ).

The addition of 18 μg/mL of 32-kDa enamelin to 10% gelatin had an effect similar to that of 0.75% amelogenin. Remarkably, addition of the 32-kDa enamelin at 18 μg/mL to the gel containing 1.5% amelogenin resulted in a faster induction time, indicating promotion of apatite nucleation (Table ). This acceleration effect appeared to be dose-dependent, as documented by the significant decrease in induction time as the result of increasing the concentration of the 32-kDa enamelin to 80 μg/mL. An equivalent concentration of phosvitin, used as a positive control (18 μg/mL), caused a delay in induction; however, the acceleration effect was found to be weaker than that of the 32-kDa enamelin (Moradian-Oldak et al., 1998) (Table).

DISCUSSION

Our present study was aimed at exploring the possibility of enamel apatite nucleation being independent of the presence of dentin apatite crystals (Diekwisch et al., 1995). We hypothesized that the process of crystal nucleation during enamel biomineralization is highly controlled and mediated through molecular interactions, both protein-protein (amelogenin self-assembly and amelogenin-enamelin interactions) and protein-mineral interactions. The best-studied, and apparently the most stable, enamelin cleavage product is the 32-kDa (Yamakoshi, 1995).

We systematically examined the effects of amelogenin and the 32-kDa enamelin on the induction of synthetic apatite crystals precipitated in 10% gelatin gel. Analysis of the data showed that incorporation of amelogenins into gelatin matrix at concentrations of 0.75 and 1.5 w/w% (equivalent to 7.5–15 mg/mL) resulted in inhibition of hydroxyapatite nucleation in a dose-dependent manner (Table). Using the same gel-diffusion system without the presence of phosphate ions, we have previously shown that the presence of amelogenin in gelatin gel did not affect the rate of calcium diffusion into the gel (Moradian-Oldak et al., 2003). We therefore propose the inhibition by amelogenin to be a direct effect of the protein on the process of crystal initiation. This inhibitory effect may result from structural re-organization of the amelogenin/gelatin-gel that may block the clusters of ions needed for the formation of nuclei (Blumenthal et al., 1991). It is noteworthy that the same level of delay was achieved by the 32-kDa enamelin, an acidic glycoprotein, at much lower concentration (18 μg/mL or 0.0018%).

Remarkably, addition of the 32-kDa enamelin (18 μg/mL) to the gelatin/amelogenin gel caused promotion of apatite nucleation almost six-fold, reducing the induction time from 20–48 hrs to 5–6 hrs (Table). It has been shown that many acidic glycoproteins act as strong inhibitors of crystal nucleation and growth in solution but can serve as effective nucleators once they have been adsorbed onto surfaces and adopted defined structures (Lussi et al., 1988). Indeed, enamelin showed a strong affinity to apatite crystals in vitro and was the most potent inhibitor of apatite crystal growth in solution (Doi et al., 1984; Tanabe et al., 1990).

We interpret our data to suggest that amelogenin and enamelin cooperate to promote nucleation of apatite crystals during enamel formation. We propose that the 32-kDa enamelin bound to amelogenin serves as a potential nucleator for apatite crystals through its oriented and structured ion-binding motifs, such as phosphoserines (Saito et al., 2000). We speculate that such structured orientation of the 32-kDa enamelin would not be achieved with gelatin gel only. The pig 32-kDa enamelin has 106 amino acids (residues 174–279), which include two phosphoserines and three glycosylated asparagines (Yamakoshi et al., 1998). The exact structural binding relationship between enamelin and amelogenin at the molecular level has not yet been firmly established. However, enamelin has the potential to assemble with the amelogenin matrix through the tri-tyrosine motif on amelogenin N-terminus, interacting with the N-acetylglucosamine on the 32-kDa enamelin (Ravindranath et al., 1999; Yamakoshi et al., in press). Our in vitro model system was designed in such a way that, after the gel was overlaid with enamelin-containing solution, the molecules of the 32-kDa enamelin will bind with the gel and, more specifically, with amelogenin. In this case, nucleation occurred on the surface of the gel. The observation that apatite formation took place in the outer surface of the gel supports the idea that induction of apatite was promoted by the 32-kDa enamelin bound on the surface of amelogenin/gelatin gel. Further detailed study on the amelogenin/enamelin complex is needed to support the proposed cooperative mechanism.

Table.

Influence of Amelogenin, the 32-kDa Enamelin, and Phosvitin (PSV) on the Nucleation of Hydroxyapatite in 10% Gelatin Gels

Protein (in 10% gelatin)	Induction Time (hrs)
0% amelogenin	4 - 5
0% amelogenin+18 μg/mL enamelin	8 - 10
0.75% amelogenin	8 - 10
1.5% amelogenin	20 - 48
1.5% amelogenin + 18 μg/mL enamelin	5 - 6
1.5% amelogenin + 80 μg/mL enamelin	3 - 4
1.5% amelogenin + 18 μg/mL PSV	6 - 8

Figure 1.

In vitro experimental set-up for the nucleation of apatite by amelogenins and the 32-kDa enamelin. (A) Sodium dodecyl sulfate-polyacrylamide gel (15% acrylamide) electrophoresis (SDS-PAGE) patterns of extracted amelogenins and the 32-kDa enamelin. Lane 1: Gibco BRL protein molecular-weight standard containing proteins with 43-k, 29-k, 18-k, 14-k, and 5-k molecular weights. Lane 2: amelogenins extracted from the extracellular enamel matrix of developing pig mandibular molars by the dissociative technique as described in MATERIALS & METHODS. The extract was characterized to be a mixture of secreted amelogenin (25 K, 7.4%) and its processed products (23 K, 10.7%; 20 K [the major band], 49.5%; 14–18 kDa, 32.4%) (Wen et al., 1999). Lane 3: The 32-KDa enamelin with an apparent molecular weight of 32 kDa. (B) Schematic representation for the experimental set-up used for monitoring nucleation of apatite crystals. Induction time was determined based on the difference in calcium uptake by the gel between the control (without phosphate) and the samples (with phosphate) as described in Fig. 3.

Figure 2.

Transmission electron micrograph of crystals grown in 10% gelatin-1.5% amelogenin gel, indicating the formation of apatite crystals. The insert shows the corresponding electron diffraction pattern. The arrowhead reflections 1 and 2 correspond to lattice spacings of 3.44 and 2.74 Å, respectively, which are characteristic of hydroxyapatite.

Figure 3.

Uptake of calcium ions from the upper solution by gelatin gel containing 1.5% amelogenin with (□, sample) and without (▪, control) phosphate ions. Comparisons between the sample and the control for each measurement were made by Student’s t test, and statistically significant differences were defined at p < 0.05. Induction time was defined as the time when the comparison between the sample (phosphate-containing gel) and the control for each measurement was statistically significant. Values are based on ratio percentages of ⁴⁵Ca radioactivity (counts per min) at time t to that measured at time zero as the function of time. The data (13 points) shown are the mean ± SD of 3 separate experiments. The variability ranged between SD = 0.4–1.9 for control and SD = 0.3–2.3 for sample.

Induction of Apatite by the Cooperative Effect of Amelogenin and the 32-kDa Enamelin

Abstract