Two Related Low Molecular Mass Polypeptide Isoforms of Amelogenin Have Distinct Activities in Mouse Tooth Germ Differentiation In Vitro


  • Kevin Tompkins,

    1. Department of Cell and Molecular Biology, Northwestern University, Feinberg School of Medicine, Chicago, Illinois, USA
    2. Department of Oral Biology, University of Illinois at Chicago College of Dentistry, Chicago, Illinois, USA
    Search for more papers by this author
    • These authors are the principal contributors to this work.

  • Keith Alvares,

    1. Department of Cell and Molecular Biology, Northwestern University, Feinberg School of Medicine, Chicago, Illinois, USA
    Search for more papers by this author
  • Anne George,

    1. Department of Oral Biology, University of Illinois at Chicago College of Dentistry, Chicago, Illinois, USA
    Search for more papers by this author
  • Arthur Veis PhD

    Corresponding author
    1. Department of Cell and Molecular Biology, Northwestern University, Feinberg School of Medicine, Chicago, Illinois, USA
    • Department of Cell and Molecular Biology, Feinberg School of Medicine, Northwestern University, 303 E. Chicago Avenue, Chicago, IL 60611, USA
    Search for more papers by this author
    • These authors are the principal contributors to this work.

  • The authors have no conflict of interest.


Embryonic mouse tooth germs were cultured in vitro in the presence of two related amelogenin isoforms to determine their effects on tooth development. Our results show that these individual proteins have specific but quite different effects on epithelial-derived ameloblasts versus mesenchymal-derived odontoblasts.

Introduction: Amelogenins, the main protein components of enamel matrix, have been shown to have signaling activity. Amelogenin isoforms differing only by the presence or exclusion of exon 4, designated ‘A+4’ (composed of exons 2, 3, 4, 5, 6d, and 7) and ‘A-4’ (composed of exons 2, 3, 4, 5, 6d, and 7) and ‘A-4’ (composed of exons 2, 3, 5, 6d, and 7), showed similar, but different, effects both in vitro and in vivo on postnatal teeth.

Materials and Methods: Lower first molar tooth germs of E15/16 CD1 mice were microdissected and cultured in vitro in a semisolid media containing either 20% FBS, 2% FBS, or 2% FBS with either 1.5 nM ‘A+4’, ‘A-4’, or both for 6 days. Tooth germs were analyzed by H&E staining and immunohistochemistry for collagen I, dentin matrix protein 2, and DAPI nuclear staining.

Results: Teeth cultured in media containing 20% FBS showed normal development with polarized ameloblasts, and odontoblasts producing dentin matrix, and DMP2 expression in odontoblasts and pre-ameloblasts. Culture in 2% FBS media resulted in no ameloblast polarization and modest odontoblast differentiation with scant dentin matrix. Tooth germs cultured with ‘A+4’ in 2% FBS media had well-polarized odontoblasts with robust dentin production and concomitant ameloblast polarization. DMP2 expression was equal to or greater than seen in the 20% FBS culture condition. In cultures with ‘A-4’ in 2% FBS media, odontoblast polarization and dentin production was reduced compared with ‘A+4’. However, the pre-ameloblast layer was disorganized, with no ameloblast polarization occurring along the dentin surface. DMP2 expression was reduced in the odontoblasts compared with the 20% FBS and ‘A+4’ conditions and was almost completely abrogated in the pre-ameloblasts.

Conclusion: These data show different signaling activities of these closely related amelogenin isoforms on tooth development. Here we make the novel observation that ‘A−4’ has an inhibitory effect on ameloblast development, whereas ‘A+4’ strongly stimulates odontoblast development. We show for the first time that specific amelogenin isoforms have effects on embryonic tooth development in vitro and also hypothesize that DMP2 may play a role in the terminal differentiation of both ameloblasts and odontoblasts.


TEETH FORM AS the result of a series of specific, reciprocal, and sequential epithelial-mesenchymal signals between the oral epithelium and the underlying neural crest derived ectomesenchyme.(1, 2) This signal-regulated process is initially directed by the epithelium, but control subsequently shifts to the mesenchymal cells.(3) This series of signaling events results in a progressive restriction in developmental potential that culminates in the differentiation of the mesenchymal-derived odontoblasts and epithelial-derived ameloblasts, the cells responsible for producing the extracellular matrices that form dentin and enamel, respectively.(4) In the mouse, tooth formation is initiated at E10, with epithelial signals to the mesenchyme directing the process. This control shifts to the mesenchyme at E12 and continues through the bud (E12-E13), cap (E14-E15), and bell (E16-E18) stages.(3) The signals follow a pattern common to all organs such as tooth, hair, lung, and kidney that develop by epithelial-mesenchymal interactions.(5) The major signaling molecules involved include those of the fibroblast growth factor (FGF),(6, 7) bone morphogenetic protein (BMP),(8) hedgehog,(9) and Wnt families.(10) However, there are specific, as yet unknown, signals that cause the terminal differentiation of the odontoblasts and ameloblasts, which culminate in the formation of dentin and enamel.

Amelogenins are the predominant proteins found in developing enamel matrix of the tooth.(11) They are specifically degraded by certain proteases, such as enamelysin,(12) during mineralization so that amelogenin present in the developing tooth is a heterogeneous mixture of related peptides. Another source of heterogeneity arises from the fact that multiple isoforms are present as the result of alternative splicing of pre-mRNA.(13) Moreover, differences in the expression patterns of individual splice products have been shown at various developmental stages.(14) Thus, the amelogenins extracted from enamel at any developmental point are a complex mixture of specific gene products and degraded polypeptides. The functions of the amelogenins have been primarily investigated relative to their proposed structural roles in creating the space and milieu for promoting enamel mineralization in the developing tooth, although the mechanisms of these activities are currently unknown.(15-17) The intrinsic heterogeneity of the amelogenins may indicate that different products perform different functions.(13) A quite different function, cell signaling related to cementogenesis, has been recently proposed for amelogenin.(18) In fact, an impure mixture of porcine enamel proteins has been used clinically to induce cementogenesis along the tooth root surface.(19) That activity has been attributed to amelogenin, although neither enamel nor ameloblasts are in the root domain.

It has also been shown that low molecular mass amelogenins (6-10 kDa), isolated from rat and bovine dentin matrix, were able to induce embryonic muscle fibroblasts in vitro to switch to a chondrogenic phenotype.(20, 21) A rat pulp-odontoblast cDNA library was probed using primers derived from the protein sequences identified.(22) This identified the presence of two specific amelogenin gene splice products composed of rat exons 2, 3, 5, 6d, and 7, with a translated protein mass of 6736 Da ([A-4]), and rat exons 2, 3, 4, 5, 6d, and 7, with a translated protein mass of 8135 Da ([A+4]). The presence of mRNA specific for amelogenins in odontoblasts has since been confirmed by others.(23, 24) These two specific gene product peptides fell exactly into the molecular mass range that had been identified as the chondrogenic factors in the original tissue isolates. We designated these two proteins as [A-4] and [A+4], respectively, to emphasize that they differ in sequence only by the absence or presence of 14 amino acids comprising the polar, hydrophilic amelogenin sequence encoded by exon 4. These isoforms were shown to have cell-signaling activity. The peptides induced embryonic muscle fibroblasts (EMFs) in culture in vitro to switch to the chondrogenic/osteogenic phenotype.(22) However, [A-4] and [A+4] induced upregulation of different phenotypic markers, CBFA1 and SOX9, respectively.(22) Similarly, matrix supported implants into muscle in vivo in 100-g Long-Evans rats induced the formation of mineralized tissue containing typical bone matrix proteins, although to different degrees. Interestingly, differences in implant vascularization were also seen. This was the initial demonstration of biological activity by specific amelogenin isoforms.(22)

In retrospect, the observation that the amelogenin peptides had the ability to interact with cells of nonoral, non-neural crest origin such as embryonic muscle-derived fibroblasts and induce chondrogenesis and osteogenesis, while very interesting, was not germane to the function of the peptides in the tooth environment. These considerations led us to study directly the effects of these amelogenins in tooth development. Because the amelogenin peptides were shown to be synthesized by the odontoblasts, a primary question was the autocrine-like effect of the peptides on odontoblast function. Thus, we initially elected to assay for expression of type I collagen (COL1), as a marker of dentin collagen production, dentin matrix protein 2 (DMP2), as a marker of an odontoblast specific protein and odontoblast maturation,(25) and cementum attachment protein (CAP), as a marker for cementum formation.(26) In the first experiment, lower first molar tooth germs from postnatal day 1 and 2 mice were cultured in the Begue-Kirn organ culture system(27) under serum-free conditions. This organ culture technique, using a 0.5% agar media, lends structural support to the developing tooth germ to maintain the 3-D architecture. This technique had been used to show the effects of various growth factors such as FGFs, transforming growth factor (TGF)βs, BMPs, and insulin-like growth factors (IGFs) in tooth germ, dental papilla, and dental pulp development.(28-31) Our first experiment showed the effects of these amelogenin peptides on tooth germs that had undergone overt differentiation to secretory cells. We found that exogenous [A-4] induced the mesenchymal cells of the dental follicle to express CAP, whereas [A+4] induced the expression of DMP2 in the odontoblast layer in the dental papilla.(32) Unfortunately, the serum-free medium did not support the development of the epithelial component of the tooth germ. However, 2% FBS-containing medium was found to be sufficient to sustain the development of the enamel organ. These experiments used tooth germs from E15/16 mice. This time-point was used because, at this developmental stage, the pre-odontoblasts and pre-ameloblasts had not yet withdrawn from the cell cycle or begun to polarize and express phenotypic markers,(33) but had undergone most of the sequential signaling required for terminal differentiation. The mRNA for the amelogenin splice products in developing tooth germ was not detected until E18.(34) In the work reported here, we have examined the effects of the exogenous addition of [A+4] and [A-4] on tooth germ development in cultures supported by 2% FBS. The two peptides have strikingly different effects on odontoblast and ameloblast development.


Production of recombinant peptides

[A-4] and [A+4] were expressed in BL21 E. coli as glutathione-S-transferase (GST) fusion proteins. The expressed proteins were collected on GST affinity columns, cleaved from GST with thrombin, and purified by HPLC on a C18 reverse-phase column as previously described.(22) These were lyophilized and resuspended in PBS at a concentration of 10 ng/ml (∼1.5 nM).

Organ culture

Lower first molars of CD-1 (Charles River) mice were obtained by microdissection, using a stereoscopic microscope, from E15 and E16 mice. All surgical explant protocols and animal care procedures were reviewed and approved by the Northwestern University Animal Care and Use Committee. Each tooth germ was cultured in semisolid media composed of RPMI-1640, 180 μg/ml ascorbic acid, 2 mM l-glutamine, 1% penicillin/streptomycin, 1.5 × 10−7 M retinoic acid (All-trans; Sigma), and 0.5% agar.(27) Cultures were carried out with FBS and different proteins added to the semisolid medium: (1) 20% FBS as positive control; (2) 2% FBS as negative control; (3) 10 ng/ml [A-4] + 2% FBS; (4) 10 ng/ml [A+4] + 2% FBS; and (5) 10 ng/ml of both [A+4] and [A-4] + 2% FBS as the experimental conditions. The tooth germs were carefully positioned on their buccal or lingual sides on Millipore filters and covered with the semisolid media at 37°C. The filters were placed at 4° for 10 minutes and subsequently cultured in a humidified incubator (5% CO2, 37°C) for 6 days in Trowell-type cultures in 3 ml of medium with the same components as the semisolid media, absent the agar. The media was changed every 48 h. Six explants were used per each condition.


The embedded tooth germs on the Millipore filters were fixed in 4% formalin, dehydrated in graded ethanol, cleared in xylene, and embedded in paraffin. Careful placement using the Millipore filter as a guide was particularly important. This allowed the mesial-distal axis of the tooth germs to be positioned in sectioning parallel to the plane of section. Thus, the tooth germ in each section could be viewed in the same orientation. Serial sections (5 μm) were cut, and every seventh section was stained with standard H&E. The remaining sections were used for immunohistochemistry.

Immunodetection of COL1 and DMP2

Immunoperoxidase detection of COL1 and DMP2 was performed, and images were captured as previously described.(28) Anti-rat DMP2 antibody was purified from serum from rabbits by Bethyl Laboratory (Montgomery, TX, USA) from recombinant DMP2 prepared by our laboratory. COL1 antibody was from Calbiochem. Antibody binding was visualized using the Vector Elite ABC detection kit (Vector Laboratories) per the manufacturer's directions. Nuclei were labeled with DAPI reagent (Pierce) as previously described.(28)


E15 tooth germs

Figure 1 compares development of E15 stage molar tooth germs after culture for 6 days in 2% FBS alone or as supported by the addition of 1.5 nM [A+4] or [A-4]. The tooth germs are viable in 2% FBS alone (Fig. 1A), but development is not robust. As shown by H&E staining, the most advanced odontoblasts at the cusp tips are polarized, but their production of dentin matrix is minimal. It is evident by the fact that the stratum intermedium is one or two cells thick that the tooth germs were sectioned in the plane parallel to the mesial-distal axis of the tooth germ. The layer of ameloblasts is also poorly developed. In fact, the layer of pre-ameloblasts is quite thick but, as emphasized by the nuclear DAPI staining, there is little evidence for ameloblast polarization and no enamel matrix secretion. In 2% FBS alone, immunostaining with anti-COL1 antibody (Fig. 2A) showed that some dentin collagen was secreted in the odontoblast layer, but staining with anti-DMP2 was essentially negative (Fig. 2D), showing only trace amounts of DMP2 in the dentin. Surprisingly, the undeveloped pre-ameloblast layer was faintly more positive for the presence of DMP2. DSPP mRNA has been seen transiently expressed in developing pre-ameloblasts.(35) The immunostaining shows that DMP2 protein is also expressed.

Figure FIG. 1..

Cultured E15 mandibular first molar mouse tooth germs. (A-C) H&E stained. (D-F) DAPI stained to show cell nuclei. Note that in both the H&E- and DAPI-stained sections, the stratum intermedium is one or two cells thick. This indicates that the tooth germs were sectioned approximately parallel to the mesial-distal axis. (A and D) Negative control, tooth germs cultured in medium containing 2% FBS, no peptide additives. The odontoblasts are mature and polarized at the cusp tip with some dentin production. The ameloblasts are completely unorganized and show no polarization. (B and E) Tooth germ cultured in media containing 1.5 nM [A-4] and 2% FBS. The odontoblasts have matured and polarized down the length of the tooth and have secreted a prominent layer of dentin. The ameloblasts have elongated, and the ameloblast layer has thickened into a double layer, but the ameloblast nuclei have not retreated from the dentino-enamel junction. (C and F) Tooth germ cultured in medium containing 1.5 nM [A+4] and 2% FBS. The odontoblasts are polarized and dentin robustly produced. The ameloblasts are polarized at the cusp and continue to polarize in the apical direction (bars = 25 μM; A-D, black bar; E and F, yellow bar).

Figure FIG. 2..

Cultured E15 mandibular first molar mouse tooth germs. (A-C) Immunostaining for COL1. (D-F) Immunostaining for DMP2. (A and D) Negative control, cultured in medium containing 2% FBS. Scant staining is seen for COL1 or DMP2. (B and E) Tooth germ cultured in medium containing 1.5 nM [A-4] and 2% FBS. Staining is seen for COL1 at the odontoblast-dentin layer; DMP2 staining is also seen in the odontoblast layer, in ameloblast layer faint staining seen. (C and F) Tooth germ cultured in media containing 1.5 nM A+4 and 2% FBS. Extensive staining is seen for both COL1 and DMP2 in the odontoblast layer. DMP2 in the odontoblast layer seems most intense in the regions of the most intense collagen staining. In F, an intense appearance of DMP2 is seen in the ameloblast layer, especially in the zone where the ameloblasts are becoming most polarized (bar = 25 μM).

Tooth development was markedly enhanced by the addition of either [A-4] or [A+4] to the 2% FBS culture medium. The H&E and DAPI stains showed the polarization of the odontoblasts along the entire length of the cusp and strong production of the dentin collagen (Fig. 1). Similarly, the ameloblast polarization was also enhanced. The degree of odontoblast function was shown as well in the anti-COL1 staining (Fig. 2), which was greater in both intensity of staining and involvement of the odontoblasts in secretory activity compared with the 2% FBS control. The staining with anti-DMP2 was even more striking. The most active collagen-secreting odontoblasts were also engaged in the robust secretion of DMP2 in the odontoblast layer. A difference in activities of [A+4] and [A-4] was very clear. The presence of DMP2 protein in the ameloblasts was much more intense in the adjacent polarizing ameloblast layer in the tooth germs cultured in the presence of [A+4] than in the presence of exogenous [A-4], both qualitatively and quantitatively. COL1 production by the odontoblasts was enhanced by [A-4] (Figs. 1 and 2) compared with the 2% FBS control but not as much as with [A+4], and DMP2 was localized to the odontoblast layer, again less robustly than the case with [A+4] (Fig. 2).

The H&E and DAPI staining (Fig. 1) brought out another prominent and important difference in the effects of [A+4] and [A-4] on tooth germ development. In 2% FBS alone, the pre-ameloblasts do not progress along the maturation pathway, and the inner enamel epithelium cells remain as a more or less disorganized layer wherein the cell nuclei occupy relatively central positions. In the normal course of development, as exemplified by the addition of [A+4], the dentin layer forms, and the opposing ameloblasts begin to elongate and polarize, so that their nuclei move distally from the dentino-enamel junction (DEJ). Although dentin forms in the presence of [A-4], the ameloblasts still remain as a thick layer of nonpolarized or less well polarized cells, and it is likely that they would not become secretory cells in the presence of [A-4]. It thus seems that the exogenous [A-4] inhibits ameloblast development and maturation to the secretory stage.

E16 tooth germs

Similar 6-day culture of the initially more advanced E16 tooth germs accentuated the differences in effect between [A+4] and [A-4]. As a positive control of the best culture conditions, Figs. 3 and 4 show the effect of culture in 20% FBS. Culture in 2% FBS yielded sections similar to that of the 2% culture of E15 tooth germs (Fig. 1), but in 20% FBS, odontoblast and ameloblast polarization to elongated form with the cell nuclei moving in opposite directions from the DEJ was evident (H&E, Fig. 3; DAPI, Fig. 4). Antibody staining with anti-COL1 and anti-DMP2 were intense, and DMP2 was clearly labeled in the ameloblasts as well as in the odontoblasts (Figs. 5 and 6). Addition of 1.5 nM [A+4] in the 2% FBS had the same effect as the 20% FBS. In fact, dentin formation seemed to be stimulated, the dentin layer was wider, the polarization of both ameloblasts and odontoblasts was more complete (Fig. 3), and the COL1 and DMP2 antibody stainings were equally intense (Figs. 5 and 6). The DMP2 was in both odontoblast and ameloblast layers (Fig. 6).

Figure FIG. 3..

Cultured E16 mandibular first molar mouse tooth germs, H&E stained. Note that the stratum intermedium is one or two cells thick. This indicates that the tooth germs were sectioned approximately parallel to the mesial-distal axis. (A) Tooth germ cultured in medium containing 2% FBS (negative control). Polarized odontoblasts and a thin layer of dentin are present. The ameloblasts at the cusp tip have begun to lengthen, but the ameloblasts are otherwise disorganized. (B) Tooth germ cultured in medium containing 20% FBS (positive control). Odontoblasts are seen polarized along the cusp length, producing a thick dentin layer. The ameloblasts lying against the dentin are polarized at the cusp tip, but become less polarized down the cusp in a smooth transition. (C) Tooth germ cultured in media containing 1.5 nM [A+4] in 2% FBS. Odontoblast polarization and dentin production are greater over both cusps than that seen in positive control tooth germs. Polarized ameloblasts are observed along much of the dentin layer. Tooth germ development is clearly accelerated in comparison with the 20% FBS positive control. (D) Tooth germ cultured in medium containing 1.5 nM [A-4] in 2% FBS. Odontoblast polarization and collagen production observed is similar to that in the E15 tooth germs, less robust than with [A+4]. Ameloblast show an abrupt transition midway down the major cusp to multilayered disordered cuboidal cells (arrowhead) along the dentin layer (A and D, same magnification; bars = 25 μM).

Figure FIG. 4..

Cultured E16 lower first molar tooth germs, DAPI stained. Note that the stratum intermedium is one or two cells thick. This indicates that the tooth germs were sectioned approximately parallel to the mesial-distal axis. (A) Tooth germ cultured in medium with 2% FBS, showing ameloblast elongation restricted to outer cusp tip area. (B) Tooth germ cultured in medium containing 20%FBS. Extensive ameloblast polarization is seen at the cusp tip, gradually diminishing apically with ameloblast and odontoblast nuclei approaching each other. (C) Tooth germ cultured in media containing 1.5 nM [A+4] in 2% FBS. At the cusp tip, ameloblasts along dentin are polarized, but at midcusp, show a disorganization not seen in the 20% FBS culture. (D) Tooth germ cultured in medium containing 1.5 nM [A-4] in 2% FBS. Odontoblast nuclei seen well aligned, but ameloblasts along the dentin layer are not polarized at the cusp tip, and at midcusp, become abruptly disorganized (arrowhead; A and D, same magnification; bars = 25 μM).

Figure FIG. 5..

Cultured E16 mandibular first molar mouse tooth germs immunostained for COL1. (A) Tooth germ cultured in medium containing 2% FBS. COL1 staining is seen in odontoblast layer of the major cusp. (B) Positive control cultured in medium containing 20% FBS. Strong staining is seen for COL1 along the odontoblast layer. (C) Tooth germ cultured in media containing 1.5 nM [A+4] in 2% FBS. Extensive staining is observed for COL1 in the odontoblast layer similar to that seen in the 20% FBS condition. (D) Tooth germ cultured in medium containing 1.5 nM [A-4] in 2% FBS. Preimmune serum staining is shown as a negative control (A and D, same magnification; bars = 25 μM).

Figure FIG. 6..

Cultured E16 mandibular first molar mouse tooth germs immunostained for DMP2. (A) Tooth germ cultured in medium containing 2% FBS. DMP2 staining is seen along the odontoblast layer in the major cusp and in localized areas of developing ameloblasts in the minor cusp. (B) Tooth germ cultured in medium containing 20% FBS. DMP2 staining is seen in both odontoblast and pre-ameloblast layers. (C) Tooth germ cultured in media containing 1.5 nM [A+4] in 2% FBS. Extensive staining observed in the odontoblast layer and in the pre-ameloblast layer, similar to that seen in the 20% FBS condition. (D) Tooth germ cultured in media containing 1.5 nM [A-4] in 2% FBS. Similar to the observation noted in Fig. 2E with the E15 tooth germs; DMP2 staining is seen in the odontoblast layer, but very weakly in the pre-ameloblast layer (A and D, same magnification; bars = 25 μM).

The H&E sections of [A-4]-treated cultured E16 tooth germs were striking compared with [A+4]-treated tooth germs (Fig. 3). Collagen production was less, and cuspal ameloblast polarization was inhibited. The anti-DMP2 staining was much less, and the DMP2 was essentially restricted to the odontoblast layer. The ameloblasts were labeled to a minor extent (Fig. 6). The progression of ameloblast terminal differentiation and maturation, and the effects of [A+4] and [A-4] on the cell elongation and polarization are particularly well shown in the E16 cultured tooth germs. After the additional in vitro culture for 6 days, the different culture effects are seen on the initially less well-developed cells. As shown in Figs. 7C and 7D, E16 tooth germs cultured in 20% FBS form a well-developed seam of dentin matrix, both the odontoblasts and ameloblasts are polarized, and the ameloblasts are particularly elongated. The odontoblast and ameloblast DAPI-stained nuclei are well separated. In contrast, E16 tooth germs cultured in 2% FBS (Figs. 7A and 7B), the odontoblasts are poorly polarized and organized and form very little dentin matrix. The ameloblast layer is entirely unorganized. The addition of [A+4] to the 2% FBS drives the development of the odontoblasts and dentin formation almost to the same extent as the 20% FBS, (Figs. 7E and 7F). Whereas ameloblast maturation is not as robust as in the 20% FBS, the DAPI staining shows considerable polarization of the ameloblast nuclei. When [A-4] is present, odontoblast maturation and dentin matrix formation proceeds, but ameloblast maturation is inhibited, and the ameloblasts, while elongated, are of indeterminate polarization (Figs. 7G and 7H). When [A+4] and [A-4] are added together (Fig. 8), the inhibitory effects of [A-4] are evident. The odontoblast maturation proceeds as usual in the presence of the [A+4], but the disorganization of the ameloblast layer is prominent even in the most cuspal zone.

Figure FIG. 7..

Higher magnification view of H&E and DAPI stains for culture E16 tooth germs. The different culture conditions are specified. Compare the odontoblast polarization between the (A and B) 2% FBS condition and the experimental conditions. Also note the difference in ameloblast polarization seen with the (C and D) 20% FBS condition and the experimental conditions. The polarized ameloblasts in 20% FBS smoothly transition to less polarized down the cusp, whereas the (G and H) [A-4] ameloblasts abruptly change to multiple cuboidal cells midway down the cusp. The (E and F) [A+4] tooth germs, in contrast, show a double layer of ameloblasts, suggesting individual effects of these two peptides on ameloblast development (bar = 5 μM).

Figure FIG. 8..

E15 mandibular first molar tooth germs cultured with both 1.5 nM [A+4] and 1.5 nM [A-4]) in 2% FBS containing media. (A) H&E staining, polarized odontoblasts with abundant dentin production are seen along both the inner and outer length of the cusp. However, there are only cuboidal pre-ameloblasts seen along the dentin surface. (B) DAPI nuclear staining shows the cuboidal pre-ameloblasts and polarized odontoblasts. (C) Immunostaining for DMP2, odontoblasts well stained, whereas pre-ameloblasts show weak staining along the outer cusp length (bar = 25 μm).


The presence of various splice products of the amelogenin gene in enamel has been known for some time,(13) but the function of the smaller amelogenins has not been understood. We were surprised to find that the chondrogenic/osteogenic activity of dentin matrix could be ascribed to amelogenin protein, but confirmed that observation by showing that the amelogenin peptides found to be active were likely to be primary biosynthetic products, not degradation products of amelogenin produced in the course of enamel formation.(21) It was equally surprising to find evidence for amelogenin mRNA in a rat incisor pulp-dentin cDNA library, and two isoforms of small splice product were present.(22) We showed that, in a fibroblast culture system, these peptides differently stimulated the production of two transcription factors, SOX9 and Cbfa1/Runx2, both of which have been shown to be involved in chondrogenesis, osteogenesis, and odontogenesis.(36, 37) On the basis of those data, we postulated that these specific amelogenin peptides might have a role in tooth development, and in particular, in the signaling between epithelium and mesenchyme that is so important during tooth morphogenesis.(38) If that were the case, we further hypothesized that the specific amelogenins should have effects on the development of tooth germs in culture. These experiments were directed to that question. E15/E16 mouse lower first molar tooth germs were chosen to take advantage of their undifferentiated state. At this developmental point, the pre-odontoblasts and pre-ameloblasts are still proliferative, not yet having undergone their final mitotic division before terminal differentiation.(33) These respective cell types have not yet begun to express phenotypic markers characteristic of their secretory stages, and it has been shown that various amelogenin alternative splice product mRNA is not present before E18.(34) It has also been shown that, under in vitro culture conditions, 1 day in vitro is equal to 0.5 days in vivo.(39) Thus, in our 6-day culture condition, endogenous expression of the smaller splice product messages would be present only at the very end of the period, if at all.

Culture of the tooth germs in 20% FBS resulted in normal development of both the enamel and dentin matrix, as expected from the abundant tooth culture literature.(27, 28) The tooth culture system used here, 2% FBS supplemented with ascorbate, glutamine, and retinoic acid, provided minimal support for growth and thus allowed us to study the effects of the additions of specific peptides. As shown in Figs. 1, 2, 3, 5 and 6, both [A+4] and [A-4] supported and enhanced dentin collagen (COL1) production, but they had very different effects on DMP2 production and ameloblast differentiation. Several studies have shown antibody staining for amelogenin in the odontoblast layer after the breakdown of the basal lamina between the pre-ameloblasts and pre-odontoblasts.(40) This amelogenin was assumed to have originated by diffusion from the ameloblast layer.(41) Our work and that of Oida et al.(42) indicated that the amelogenin in the odontoblast layer was at least partly synthesized in the odontoblasts.

During tooth morphogenesis, the inner enamel epithelium cells of the enamel organ, destined to become secretory ameloblasts, sit on a basement membrane, which they have synthesized, separated from the underlying mesenchymal cells of the dental papilla, destined to become odontoblasts. The developing tooth enlarges by cell division in each layer, adding new cells in the apical direction. The more coronal terminally divided cells enter a unidirectional maturation phase. The relative rates of dentin and enamel formation and the time of initiation of formation of the two mineralized tissues are crucial for development of the proper mechanical properties. The hard but brittle enamel must be deposited on the softer preformed dentin, not the other way around. The terminal differentiation programs begin when the basement membrane separating the two cell layers degrades. The pre-odontoblasts are the first to begin their maturation program. Stimulated by signals from the pre-ameloblasts, the pre-odontoblasts polarize and begin to secrete the matrix that forms the pre-dentin, and subsequently, the dentin. During this period, the pre-ameloblasts are delayed in their maturation. They enter their polarization and secretory phase only after the dentin matrix thickens and begins to mineralize. The entire process of tooth formation is regulated by a series of reciprocal signals.(4) The initiation of the cell maturation may be the last of these signaling steps. The data presented above suggest that these last signaling steps may be related to the production of tissue specific peptides rather than the more ubiquitous TGFβ-BMP family members.

The key observations were that [A+4] and [A-4] have quite different effects. [A+4] added to culture media clearly leads to the stimulation of COL1 and DMP2 in dentin production and to the transient appearance of DMP2 at lower levels in the developing ameloblast layer. In the [A+4]-stimulated cultures, the ameloblasts begin to elongate and polarize normally after the dentin thickens. In sharp contrast, exogenous [A-4] addition has inhibitory effects, the production of dentin proceeds, although somewhat less robustly than in [A+4], but the maturation of the ameloblast layer is disrupted. This is shown in the H&E sections in Fig. 3, with the comparison of the E16 [A+4] and [A-4] additions. This is also seen with the DAPI staining (Fig. 4). In both the 20% FBS and [A+4] conditions, the separation of the ameloblast and odontoblast nuclei is greatest at the cusp tip and decreases in the apical direction, reflecting the gradient of differentiation of these cells. However, with [A-4], the separation between these two cell types actually widens from the cusp apically. The DMP2 protein localization in these tooth germs correlates well with the abrupt change in ameloblast morphology seen in the DAPI-stained tooth germs treated with [A-4] (Figs. 4 and 6). Thus, we conclude that [A-4] in the 1- to 2-nM range of concentration has inhibitory effects on the ameloblast maturation, and we can hypothesize that [A-4] produced in the developing odontoblasts may delay pre-ameloblast maturation until a sufficiently thick layer of dentin is produced, blocking diffusion and further signaling interactions between the two cell layers. The ameloblast maturation program then ensues. Thus, in vivo, the brief period in which the odontoblasts can specifically provide signals to inhibit ameloblast maturation, possibly by the transitory synthesis and export of [A-4], may be the final epithelial-mesenchymal signaling event in tooth morphogenesis. Interestingly, whereas [A-4] has an inhibitory effect on DMP2 expression in pre-ameloblasts, there is not an inhibition in the pre-odontoblasts. Thus, both [A+4] and [A-4] could be expressed at the same time by developing odontoblasts, acting synergistically on odontoblast differentiation, whereas [A-4] inhibits ameloblast development until the dentin layer is adequately formed.

The addition of exogenous [A-4] to the tooth germ cultures does increase the thickness of the ameloblast layer to an almost double layer of elongated, but disorganized, ameloblasts. This may indicate that the disorganization and ameloblast polarization defect is related to both internal and external [A-4] signaling. That is, the diffusion of the exogenous [A-4] into the epithelial cell layer may provide a gradient opposite to that in the normal situation.

Another very interesting observation was the [A+4] enhancement of DMP2 production within the layer of elongating ameloblasts just opposite to the odontoblasts most actively secreting DMP2 and collagen matrix (Fig. 6). In situ hybridization studies have shown the transient expression of DSPP mRNA in developing ameloblasts.(35) Recent studies have proposed that dentin matrix proteins, particularly DMP1, are expressed in alveolar bone as well as dentin and may act as intracellular signaling molecules.(43) The DMP2 may likewise be an active, although transient, participant in ameloblast maturation. The mixture of [A+4] and [A-4] (Fig. 8) shows that the two amelogenin splice product proteins have distinct activities in culture and possibly during in vivo development.

The data and various derivative hypotheses discussed here raise several questions requiring further study. One necessary direction for future work is to determine the pattern of expression of the two amelogenins in the course of development at early stages of tooth development. Preliminary in situ hybridization studies suggest that there is expression of the amelogenins by the pre-odontoblasts at certain stages, as well as in the pre-ameloblasts. Such studies require careful design of the probes. It is evident that we need to include the DMPs in the in situ study. Another direction of our work is to examine more specifically, at the mRNA level using quantitative PCR, the changes in expression of other related transcription factors and matrix protein components. An important line of work that is currently underway is the elucidation of the mechanisms by which the signaling effects seen in this work occur, such as receptor identification, and the signaling pathways involved. Despite all of these complications, however, it seems to be clear that what were previously thought of as dentin- and enamel-specific peptides and proteins involved in the mineralization processes are neither dentin- nor enamel-specific nor restricted to the mineralization function.


We thank Dr Thomas Diekwisch (Brodie Laboratory for Craniofacial Research, University of Illinois College of Dentistry) for suggestions on orientation of tooth germs during culturing and sectioning. This work has been supported by National Institute for Dental and Craniofacial Research Grants DE-01374 and DE-08525 (AV).