• α-2-HS-glycoprotein;
  • ameloblast;
  • ameloblastin;
  • amelogenesis;
  • amelogenin;
  • biglycan;
  • calnexin;
  • CD63;
  • dentin;
  • dentin-enamel junction;
  • dentine;
  • enamel;
  • enamelin;
  • fetuin-A;
  • melanoma 1 antigen;
  • tetraspanin;
  • yeast two-hybrid assay


  1. Top of page
  2. Abstract
  7. Acknowledgements

The recognized structural proteins of the enamel matrix are amelogenin, ameloblastin, and enamelin. While a large volume of data exists showing that amelogenin self-assembles into multimeric units referred to as nanospheres, other reports of enamel matrix protein-protein interactions are scant. We believe that each of these enamel matrix proteins must interact with other organic components of ameloblasts and the enamel matrix. Likely protein partners would include integral membrane proteins and additional secreted proteins.

Introduction: The purpose of this study was to identify and catalog additional proteins that play a significant role in enamel formation.

Materials and Methods: We used the yeast two-hybrid assay to identify protein partners for amelogenin, ameloblastin, and enamelin. Once identified, RT-PCR was used to assess gene transcription of these newly identified and potential “enamel” proteins in ameloblast-like LS8 cells.

Results: In the context of this yeast assay, we identified a number of secreted proteins and integral membrane proteins that interact with amelogenin, ameloblastin, and enamelin. Additionally, proteins whose functions range from the inhibition of soft tissue mineralization, calcium ion transport, and phosphorylation events have been identified as protein partners to these enamel matrix proteins. For each protein identified using this screening strategy, future studies are planned to confirm this physiological relationship to biomineralization in vivo.

Conclusion: Identifying integral membrane proteins of the secretory surface of ameloblast cells (Tomes' processes) and additional enamel matrix proteins, based on their abilities to interact with the most abundant enamel matrix proteins, will better define the molecular mechanisms of enamel formation at its most rudimentary level.


  1. Top of page
  2. Abstract
  7. Acknowledgements

Since the discovery of a cDNA sequence for amelogenin in 1983,(1) our understanding of enamel formation has been significantly aided by the subsequent discoveries of additional structural organic components of the enamel extracellular matrix, including ameloblastin(2,3) and enamelin.(4) In addition, data from an animal model null for the biglycan gene(5,6) indicate that the biglycan protein, whereas not unique to the enamel matrix environment, plays a role in amelogenesis.(7) Two enamel-specific proteases have also been recently characterized and discussed, and these are kallikrein-4(8) and matrix metalloproteinase-20 (MMP20).(9,10) The spatiotemporal expression of each of these enamel proteins has been, and continues to be, defined, but what remains to be studied is how each of these enamel matrix components interacts with one another to form a self-assembled matrix competent to initiate and orchestrate the events of mineralization.(11–13) While amelogenin self-assembles into multimeric units known as nanospheres,(13) reports of other enamel matrix protein-protein interactions are scarce.(11,12) Events of mineralization ultimately result in mature enamel that is almost completely absent of any history of its protein origins. The important role that each individual protein plays toward the creation of enamel can be appreciated from the well-ordered hierarchical structure seen in mature enamel,(13) but their individual roles in creating this elegant architecture has yet to be fully illuminated.

It is apparent that the timing and amount by which these enamel gene products are presented to the extracellular enamel matrix must be finely controlled and regulated. Understanding their gene transcription and the various transcriptional factors that govern RNA expression are of major significance in our quest to understand amelogenesis. In the laboratory, this requires careful dissection of genepromoter regions. This work has just begun for amelogenin,(14–18) ameloblastin,(19,20) and biglycan,(21,22) but has yet to be initiated for enamelin. Post-transcriptional modifications such as alternative splicing and post-translational modifications such as phosphorylation (e.g., in amelogenin(23)) ensure that, even within the four secreted proteins discussed (amelogenin, ameloblastin, enamelin, and biglycan), functional diversity from a single gene is possible. Because of the unique properties of these enamel proteins, it seems that their removal from the enamel matrix during enamel maturation has required specific proteases whose spatiotemporal expression must also be exquisitely regulated. This need is met partially or fully with the serine protease kallikrein-4(8,24,25) and the matrix metalloproteinase MMP20.(10)

The yeast two-hybrid (Y2H) assay was developed in 1989(26) and is a yeast-based genetic assay to detect protein-protein interactions in vivo. The Y2H assay system reflects the ability of two proteins to interact under physiologic conditions that approximate those observed experimentally in the enamel organic matrix milieu.(27) Moreover, yeast cells will perform post-translational modifications on transgene-derived proteins, thus allowing, for example, glycosylation and phosphorylation reactions. The Y2H assay is based on the fact that the GAL4 enhancer-binding protein consist of two physically dividable molecular domains: one is acting as a DNA-binding domain and the other is acting as the transcription-activating domain. Normally, the two functions are located on the same molecule, but it has been shown that a functional molecule can be assembled in vivo from the separated parts. In this case, the DNA-binding domain and the activating domain for the GAL4 transcription factor have been cloned separately into different expression vectors, each of which can accept a target gene (gene A or gene B) to form fusion proteins. Both hybrid proteins are targeted to the nucleus by nuclear localization signals. If protein A interacts with protein B, the DNA binding domain of GAL4 will be tethered to its transcription-activating domain and the activity of the transcription complex reconstituted. Expression of reporter genes, including HIS3 and lacZ, downstream from the GAL4 binding site are used as selective and colorimetric assays indicating the avidity of the interaction between the two proteins.

Current Y2H expression/hybrid vectors, available cDNA expression libraries, screening protocols, and data analysis have improved significantly since the Y2H assay was first introduced. It is now feasible to do large-scale screenings of multiple commercially available Y2H libraries for protein partners to the known enamel and DEJ proteins (amelogenin and its isoforms, ameloblastin and enamelin). One improvement has been the inclusion of four reporter genes (ADE2, HIS3, lacZ, and MEL1) into the plasmids that have allowed for the automatic exclusion of the majority of false positives. A second major improvement has been the inclusion of epitope tags on both the DNA binding-domain vector and the DNA activating-domain vector to eliminate the need to generate specific antibodies to new proteins to identify fusion proteins. The use of specific epitopes has also allowed for the confirmation of physiological protein-protein interactions using the more classical co-immunoprecipitation methodologies without additional gene manipulations. A third improvement was to include the T7 promoter in the vector cassette that allows one to transcribe and translate epitope-tagged gene fusions in vitro. Co-immunoprecipitation studies can now be done on identified interacting proteins without the inclusions of the GAL4 binding domain and GAL4 activating domain. Many commercially available libraries are now available, including pretransformed libraries. The advantage of pretransformed libraries is that the majority of the “leg-work” has been done, from transformation into yeast to library amplification. For this system, the “bait”-containing cassette is transformed into a suitable yeast host strain and mated with a compatible yeast strain containing the library. Positive protein-protein interactions are selected for with appropriate selective media containing agar plates (i.e., histidine-deficient selection media).

Recently, the biggest change to screening any expression cDNA library has occurred because many of the genome projects are now complete, including mouse and human. With very little DNA sequence data, for example, as little as 100 bp, it is now possible to identify a specific gene and to confirm that its open-reading fame (ORF) is correct with respect to position in the hybrid vector. By knowing the identity of the gene/protein, the annotated databases make it is possible to then determine whether this Y2H identified protein has a signal peptide, transmembrane regions, nuclear transport signal motifs, or indeed any peptide region, that give clues as to function. Proteomic maps are now also including protein-interacting partners, further supporting the analysis of the function of a protein in physiologic relevant pathways.

Using the Y2H assay, we have identified a small number of potential enamel organic matrix-interacting proteins that were selected from approximately 4 × 107 cDNA expressed in mouse embryos. Further characterization of each of these clones should better define the molecular events of amelogenesis.


  1. Top of page
  2. Abstract
  7. Acknowledgements

Subcloning of amelogenin, ameloblastin, and enamelin into the GAL4 DNA-binding domain vector for use in the Y2H system

The GAL4 DNA-binding domain plasmid that was used for these studies is pGBKT7 (BD Biosciences Clontech, Palo Alto, CA, USA). Significant features of this vector related to the hybrid protein are the GAL4 DNA-binding domain region, followed by T7 promoter, c-Myc epitope tag, and the multicloning site. The cDNA sequences and ORF information for the enamel matrix proteins included in the GAL4 DNA-binding domain hybrid can be accessed from the National Center for Biotechnology Information web page ( These gene accession numbers are mouse amelogenin M180(28) (NM_009666), mouse enamelin (NM_017468), and rat ameloblastin (NM_012900). PCR DNA primers were designed and synthesized to coding regions of each of the cDNA (Table 1). For amelogenin and ameloblastin, PCR was used to generate the entire ORF, whereas the strategy for enamelin amplification and subcloning required a different approach, which is briefly discussed below. All PCR steps were performed per standard protocol.(29) The primer design also included selected restriction enzyme sites to allow for the efficient, subsequent, and “in-frame” subcloning into the Y2H vector (pGBKT7). All PCR-produced cDNA was initially subcloned into a TA cloning vector (pCR2.1-TOPO; Invitrogen, Life Technologies) and sequenced in their entirety to ensure that the PCR did not introduce errors into the amplified DNA sequence. In each case, the protein's signal peptide was excluded from the hybrid protein construction.

Table Table 1.. Primers Used to Produce PCR Products in the Construction of the Y2H Constructs
Thumbnail image of

The ORF for enamelin contains in excess of 3.8 kb of cDNA sequence. To prepare the Y2H enamelin vector, the 5′ region of the enamelin cDNA, excluding the signal peptide but including a unique EcoRV restriction site, was PCR amplified (product 520 bp), subcloned, and sequenced. The enamelin 3′ cDNA region from this EcoRV site through to a BglII site (which was part of the multicloning site of the plasmid containing the cDNA) was removed from the original vector. These two enamelin cDNA regions were ligated at their EcoRV sites, and the complete and intact enamelin cDNA was removed from its plasmid backbone with the SmaI and BglII restriction enzymes and subcloned into pGBKT7 at the SmaI/BamHI site.

GAL4 DNA-binding domain hybrid vectors were prepared using cDNA for amelogenin M180, ameloblastin, and enamelin. Each of these hybrid constructs was co-transformed into a yeast host (PCY2)(30) with the unaltered GAL4-activating domain vector (pGADT7; BD Biosciences Clontech) and assayed for intrinsic β-galactosidase activity by the filter assay.(31) No β-galactosidase activity was noted for any of these GAL4 DNA-binding domain hybrid constructs; therefore, each of these enamel matrix “bait” constructs was deemed suitable to screen Y2H expression libraries.

Large-scale library screenings of a mouse 17-day embryo (E17) pretransformed Y2H expression library (BD Matchmaker Pretransformed cDNA Library; BD Biosciences Clontech) were performed on agar-containing plates deficient in histidine, leucine, and tryptophan using the manufacturer's recommended protocol and recommended yeast host strains. Positive interacting protein pairs identified in host yeast were replated on the same medium but including X-α-Gal, at 25 mg/liter. Double-transformed yeast clones that grew as blue colonies were selected, and the unknown library cDNA-containing plasmid was recovered and partially sequenced from the 5′ direction. Sequencing covered the splice region between the vector and the insert, such that the reading frame and amino-acid composition of the cloned insert was determined.

RT-PCR reaction conditions

Ameloblast-like LS8 cells(18,20,32) were cultured, and their mRNA was isolated using standard procedures.(29) RT-PCR was performed on LS8 mRNA using the RETROscript first-strand synthesis kit from Ambion (Austin, TX, USA), and PCR was performed(29) using gene-specific primers spanning intron regions. Both positive and negative controls were included during the RT-PCR reaction steps and gave predicted results.


  1. Top of page
  2. Abstract
  7. Acknowledgements

Identification of amelogenin-, ameloblastin-, and enamelin-interacting proteins

Our search criteria were that newly identified proteins are either integral membrane proteins, proteins with predicted transmembrane domains, secreted structural proteins, or novel proteolytic enzymes. The enamel matrix-interacting proteins listed (Table 2) are proteins that fit our search criteria. Each of these proteins/cDNA was in the correct ORF with respect to the GAL4-activating domain, and as summarized in Table 2, often the entire protein was recovered from the hybrid protein.

Table Table 2.. Amelogenin (AMEL), Ameloblastin (AMBN), and Enamelin (ENAM) Interacting Proteins
Thumbnail image of

At this stage, we screened 20 × 106 independent proteins from an E17-mouse Y2H library for amelogenin-interacting proteins, 7.5 × 106 independent proteins for ameloblastin-interacting proteins, and 9.6 × 106 independent proteins for enamelin-interacting proteins. These data are from two rounds of library screenings for ameloblastin, two rounds of screening for amelogenin, and a single round of screening for enamelin.

For each of the experimental screenings with amelogenin and ameloblastin, between 200 and 400 double-transformants grew on the selection media lacking leucine, tryptophan, and histidine. Approximately one-half of this number showed α-galactosidase activity when grown on X-α-Gal-containing media, whereas the library construct alone showed no endogenous β-galactosidase activity. Between 50 and 150 GAL4-activating domain hybrid-protein plasmids containing the potential enamel matrix-interacting proteins were recovered for sequencing and further characterization.

The screening of 9.6 × 106 library clones with enamelin as bait yielded close to 103 double-transformants able to grow on selection media. Again, after the appropriate selections, approximately one-half of these were identified as clones of interest. To date, ∼200 of these library clones have been sequenced.

Confirming enamel matrix protein interactions with human CD63 and human biglycan

A number of proteins listed in Table 2, including melanoma 1 antigen (Cd63) and biglycan, were identified more than once. The mouse and human proteins for melanoma 1 antigen and biglycan are 87% and 97% similar, respectively, and it is likely that the physiological contribution to enamel formation for both of these proteins is similar across species. The protein-protein-interacting abilities of the rodent enamel matrix proteins and human biglycan (BGN) and melanoma 1 antigen (CD63) were examined. cDNA to human CD63 (NM_001780) and human biglycan (NM_001711) was purchased from Origene Technologies (Rockville, MD, USA). The entire ORFs for both cDNA sequences, not including the signal peptide region of biglycan, were amplified using primers containing appropriate restriction sites (Table 1), to allow for the efficient in-frame subcloning of the cDNA into the vector pGADT7 (BD Biosciences Clontech). PCR-generated DNA was sequenced to ensure no errors were introduced during the manipulations. All possible protein-protein combinations between the “bait” enamel matrix proteins; mouse amelogenin, enamelin, and rat ameloblastin with human CD63 and human biglycan were tested using the Y2H assay (Fig. 1). Also tested were the leucine-rich amelogenin peptides (LRAP) versus CD63 and biglycan. Control interactions consisted of the positive interacting combination of tumor suppressor protein p53 and SV40 large T antigen,(30) and the negative controls of pGBKT7 versus pGADT7 (empty vectors), pGBKT7 (empty vector) versus CD63, and pGBKT7 (empty vector) versus biglycan. Protein-protein interactions were confirmed on histidine-negative plates and confirmed and documented on X-α-Gal-containing plates after 6 and 20 h (Fig. 1). The strength of the interaction was assessed relative to the positive combination of p53 versus SV40 large T antigen (Table 3).

Table Table 3.. Relative X-α-Gal Activity after 6 and 20 h
Thumbnail image of
thumbnail image

Figure FIG. 1.. Protein-protein interactions as measured by the yeast two-hybrid assay. Possible protein-protein combinations between the “bait” enamel matrix proteins (mouse amelogenin and enamelin and rat ameloblastin) with human CD63 and human biglycan were assayed. Also included and tested were the leucine-rich amelogenin peptide (LRAP) vs. CD63 and biglycan; the positive interacting combination of tumor suppressor protein p53 and SV40 large T antigen (positive control); and the negative controls of pGBKT7 vs. pGADT7 (empty vectors), pGBKT7 vs. CD63, and pGBKT7 vs. biglycan. Protein-protein interactions were documented on X-α-Gal containing plates after 6 (left) and 20 h (right). Strength of interaction (−, +, or ++) was assessed relative to the positive combination of p53 vs. SV40 large T antigen, which was arbitrarily given a (+) interacting strength.

Download figure to PowerPoint

Ameloblast-like LS8 cells express some of the enamel matrix-interacting proteins

LS8 cells have previously been used to study ameloblastin expression(19,20) and amelogenin expression(18,32,33); thus, LS8 cells seem to be a good isolated cell model system to study enamel-specific gene activities. Using gene-specific primers spanning intron regions (Table 4), a number of the genes identified in Table 2 such as biglycan, Cd63, and calnexin, along with amelogenin, ameloblastin, and enamelin, were shown to be expressed in ameloblast-like (LS8) cells as determined by transcript identification based on RT-PCR (Table 5). The resulting amplified cDNA from mRNA for biglycan, Cd63, and calnexin was sequenced to confirm their identity. Expression of the mRNA in LS8 cells suggests that the resulting protein is available for assembly into the enamel organic matrix.

Table Table 4.. Primers Used in RT-PCR to Show Gene Transcription in Ameloblast-Like LS8 Cells
Thumbnail image of
Table Table 5.. Relative Levels of mRNA in LS8 Cells Determined by RT-PCR
Thumbnail image of

α-2-HS-glycoprotein (Ahsg or fetuin-A) interacts with amelogenin and enamelin by the Y2H assay (Table 2). mRNA for Ahsg could not be shown in LS8 cells by RT-PCR (Table 5), which is not surprising, because Ahsg is a circulating serum protein produced by hepatocytes,(34–36) and this suggests that the enamel matrix may specifically take up and use circulating proteins.

Two amelogenin-interacting proteins listed in Table 2 are designated “unknown ESTs,” and these have gene accession numbers NM_026325 and NM_026808. Both these cDNA sequences were identified as having the correct reading frame within the Y2H vector, and both contain the entire secreted protein portion produced from these genes (Fig. 2). Both these unknown ESTs have a predicted signal peptide, as determined by the software SignalP 3.0 Server ( (Fig. 2). However, neither predicted protein contains any other identifiable functional peptide domains. Gene NM_026325 contains five exons and gene NM_026808 contains two exons. To determine if LS8 cells transcribe either gene, RT-PCR using unique primer sets spanning intron regions (Table 4) was used to assay for the “unknown ESTs.” mRNA for NM_026325 was detected in ameloblast-like LS8 cells (Table 5), and its identity was confirmed by sequencing the PCR product. The mRNA from NM_026808 could not be shown in LS8 cells using similar conditions (Table 4).

thumbnail image

Figure FIG. 2.. Predicted amino-acid sequence for two “unknown ESTs” that interact with amelogenin by the yeast two-hybrid assay. Arrow indicates the cleavage site of the N-terminal signal peptide as predicted by SignalP 3.0 server (

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  7. Acknowledgements

The proteins listed in Table 2 are expected to impact our understanding of enamel formation. The difficulty in discussing each is that, for most of these proteins, little if anything has been published relating protein to function, and the information presented most frequently relates to a possible function based on protein domain predictions. For example, there are two amelogenin-interacting proteins listed that are designated “unknown ESTs.” These have gene accession numbers NM_026325 and NM_026808. Both unknown ESTs have a conceptual signal peptide; however, neither predicted protein contains an identifiable functional peptide domain. Gene NM_026325 contains five exons and gene NM_026808 contains two exons. Ameloblast-like LS8 cells transcribe NM_026325, and the mRNA from this gene has also been identified in ESTs from libraries derived from whole body, testis, thymus, kidney, mammary gland, brain, spleen, skin, placenta, eye, pituitary gland, lymph node, stomach, liver, pancreas, uterus, heart, lung, and colon. Thus, the protein produced from this message is not tooth-specific nor is it specifically expressed in mineralized tissues. The mRNA from NM_026808 could not be shown in LS8 cells by RT-PCR but has been identified in ESTs from libraries derived from small intestine, testis, kidney, liver, pancreas, muscle, and lung. Despite these findings, it remains plausible that both unknown ESTs may play a functional role in amelogenesis, but this possibility must be studied further.

Of the other proteins identified by the Y2H assay, based either on their known function or the repeated isolation of these proteins from the screenings, four of these will be discussed in relation to their possible role in amelogenesis. Their potential importance and novelty to amelogenesis may relate to calcium ion transport and binding (calnexin), inhibition and control of hydroxyapatite (HAP) mineralization (α-2-HS-glycoprotein),(34–36) secreted protein-to-protein interactions (biglycan versus the enamel matrix proteins), and docking and orientation properties of Tomes' processes to the enamel matrix proteins (melanoma 1 antigen).

The human calnexin (CANX) gene is located on chromosome 5q35. Calnexin contains a C-terminal endoplasmic reticulum (ER) retention motif “RKPRRE” and is implicated to play a role in a number of cell activities, including ER membrane-related protein folding, calcium ion binding, calcium ion storage, and protein secretion through secretory vesicles.(38–41) Using software for the Signal P3.0 server (, the N-terminal 20 amino acids of calnexin seem to code a signal peptide (with a 100% probability figure for both human and mouse proteins). This signal peptide domain for calnexin has not previously been recognized or discussed in the literature, but it is an intriguing finding and suggests yet another role for this multifunctional protein.

The human CD63 antigen (melanoma 1 antigen; CD63) gene is located on chromosome 12q12. CD63 is a member of the transmembrane-4 glycoprotein superfamily, which is also known as the tetraspanin family.(42,43) Most of these family members are cell surface proteins that are characterized by the presence of four hydrophobic (transmembrane) domains.(42,43) These proteins mediate signal transduction events that play a role in the regulation of cell development, activation, growth, and motility.(43,44) In particular, cell surface glycoprotein CD63 is known to complex with integrins.(43,45) A little more deductive information can be extracted from Table 2 concerning the ability of Cd63 to interact with amelogenin. As stated, Cd63 has four transmembrane domains.(42) Both the N- and C-terminals are cytoplasmic; therefore, there are two extracellular domains. The two extracellular domains of Cd63 are well defined in mice from amino acids 35-51 and 103-205.(42,43,46) These domains are mirrored in the human CD63 protein. Whereas ameloblastin showed that it interacted with a complete and intact Cd63 protein, amelogenin interacted with the C-terminal 73 amino acids (amino acids 166-238). If we assume that amelogenin interacts with Cd63 in the extracellular matrix, the logical domain of Cd63 to posses this activity is present within amino acids 166-205. This hypothesis is testable and is the focus of current study.

The human α-2-HS-glycoprotein (AHSG or fetuin-A) gene is located on chromosome 3q27. AHSG is a glycoprotein present in the serum and is synthesized by hepatocytes.(34,35,47) The AHSG molecule consists of two polypeptide chains, which are both cleaved from a proprotein encoded from a single mRNA.(48,49) AHSG is involved in several functions, such as endocytosis, brain development, and the formation of bone tissue.(34,35,47) The protein is commonly present in the cortical plate of the immature cerebral cortex and bone marrow hemopoietic matrix, and it has therefore been postulated that it participates in the development of these tissues.(34,36) It is also quite possible that AHSG thus has a regulatory role in the mineralization process of amelogenesis. We hypothesis that such a unique mineralized tissue like enamel is likely to have multiple regulatory protein controls that contribute to the inhibition, initiation, and growth of individual crystallites. Ahsg-null animals have been engineered,(34–36) and the most notable feature of these animals is that there are ectopic soft tissue calcifications that ultimately result in death. Based on these observations and the suggested protein function, it would be of interest to study the enamel phenotype of these Ahsg-null animals.

The human biglycan (BGN) gene is located on chromosome Xq28. Biglycan is a small cellular or pericellular matrix proteoglycan that is closely related in structure to two other small proteoglycans: decorin and fibromodulin.(5,50,51) Decorin contains one attached glycosaminoglycan chain, whereas the biglycan protein probably contains two such chains.(51) For this reason, the protein is called biglycan and is thought to function in connective tissue metabolism by binding to collagen fibrils and TGF-β.(5,51) Biglycan-null animals have been generated to study the role of biglycan (Bgn) in vivo.(5) These Bgn-null animals appear normal at birth, but as these mice age, they display a phenotype characterized by reduced growth rate and decreased bone mass. Whereas this type of phenotype is commonly observed in specific collagen-deficient animals, it is rarely observed in skeletal abnormalities in animals lacking noncollagenous proteins. Biglycan is also expressed in dentin.(52) Goldberg et al.(7) studied these biglycan-null animals, focusing on how dentin is affected. In addition to changes to the dentin, significant changes were noted in the enamel of these animals. It was shown that biglycan is expressed in ameloblast cells during normal tooth development and that biglycan is present in the enamel extracellular matrix.(7) In biglycan-null mice, the forming enamel was between 3- and 5-fold thicker than in normal control animals, and this difference was explained by an enhanced level of amelogenin synthesis and secretion from secretory ameloblasts.(7) From their observations, Goldberg et al.(7) concluded that biglycan either directly or indirectly act as a repressor of amelogenin expression. However, the mature enamel of these biglycan-null animals appears to be of relatively normal thickness (M Young and M Goldberg, personal communication, 2004), suggesting that biglycan may alter the temporal events of amelogenesis (including the forming or immature enamel thickness), but biglycan plays little or no role in determining the final enamel thickness.

Finally, a number of integral membrane proteins, a protein phosphatase, and a number of collagens are identified as interacting either with amelogenin, ameloblastin, and enamelin. Little is known about the proteins that comprise the dentin-enamel junction (DEJ), but as shown by our data (Table 2), the discovery that enamel matrix proteins directly interact with selected collagens opens the possibility that the DEJ is a protein-based union between two dissimilar structures rather that a biological interface.

In conclusion, our understanding of ameloblast differentiation and physiology and the enamel matrix assembly and biomineralization is in its infancy. The Y2H assay in this study has served as a proteomic tool that has shown a biochemical relationship between the known enamel matrix proteins and many additional proteins not previously implicated in tooth development. Additionally, RT-PCR of selected mRNA from LS8 cells suggests physiological roles in amelogenesis for some of these newly identified proteins. Further studies are planned to confirm these relationships.


  1. Top of page
  2. Abstract
  7. Acknowledgements

This work was supported by NIH Grants DE13045 and DE13404. The cDNA was a kind gift from James Simmer (enamelin) and Paul Krebsbach (ameloblastin). The authors thank Caroline Paine and Benton Yoshida for critical reading of the manuscript, and we appreciate the constructive and critical comments provided by the two reviewers of this manuscript, whose assistance helped to improve this work.


  1. Top of page
  2. Abstract
  7. Acknowledgements
  • 1
    Snead ML, Zeichner-David M, Chandra T, Robson KJ, Woo SL, Slavkin HC 1983 Construction and identification of mouse amelogenin cDNA clones. Proc Natl Acad Sci USA 80: 72547258.
  • 2
    Fong CD, Slaby I, Hammarstrom L 1996 Amelin, an enamel related protein, transcribed in cells of the epithelial root sheath. J Bone Miner Res 11: 892898.
  • 3
    Lee SK, Krebsbach P, Matsuki Y, Nanci A, Yamada KM, Yamada Y 1996 Ameloblastin expression in rat incisors and human tooth germs. Int J Dev Biol 40: 11411150.
  • 4
    Hu C-C, Fukae M, Uchida T, Qian Q, Zhang CH, Ryu OH, Tanabe T, Yamakoshi Y, Murakami C, Dohi N, Shimizu M, Simmer JP 1997 Cloning and characterization of porcine enamelin mRNAs. J Dent Res 76: 17201729.
  • 5
    Xu T, Bianco P, Fisher LW, Longenecker G, Smith E, Goldstein S, Bonadio J, Boskey A, Heegaard AM, Sommer B, Satomura K, Dominguez P, Zhao C, Kulkarni AB, Robey PG, Young MF 1998 Targeted disruption of the biglycan gene leads to an osteoporosis-like phenotype in mice. Nat Genet 20: 7882.
  • 6
    Young MF, Bi Y, Ameye L, Chen XD 2002 Biglycan knockout mice: New models for musculoskeletal diseases. Glycoconj J 19: 257262.
  • 7
    Goldberg M, Septier D, Rapoport O, Young M, Ameye L 2002 Biglycan is a repressor of amelogenin expression and enamel formation: An emerging hypothesis. J Dent Res 81: 520524.
  • 8
    Hu JC, Zhang C, Sun X, Yang Y, Cao X, Ryu O, Simmer JP 2000 Characterization of the mouse and human PRSS17 genes, their relationship to other serine proteases, and the expression of PRSS17 in developing mouse incisors. Gene 251: 18.
  • 9
    Bartlett JD, Simmer JP, Xue J, Margolis HC, Moreno EC 1996 Molecular cloning and mRNA tissue distribution of a novel matrix metalloproteinase isolated from porcine enamel organ. Gene 183: 123128.
  • 10
    Caterina JJ, Skobe Z, Shi J, Ding Y, Simmer JP, Birkedal-Hansen H, Bartlett JD 2002 Enamelysin (MMP-20) deficient mice display an amelogenesis imperfecta phenotype. J Biol Chem 277: 4959849604.
  • 11
    Yamakoshi Y, Hu JCC, Fukae M, Tanabe T, Oida S, Simmer JP 2003 Amelogenin and 32kDa enamelin protein-protein interactions. In: KobayashiI, OzawaH (eds.) Biomineralization (BIOM2001): Formation, Diversity, Evolution and Application. Tokai University Press, Kanagawa, Japan, pp. 338342.
  • 12
    Ravindranath HH, Chen LS, Zeichner-David M, Ishima R, Ravindranath RM 2004 Interaction between the enamel matrix proteins amelogenin and ameloblastin. Biochem Biophys Res Commun 323: 10751083.
  • 13
    Paine ML, White SN, Luo W, Fong H, Sarikaya M, Snead ML 2001 Regulated gene expression dictates enamel structure and tooth function. Matrix Biol 20: 273292.
  • 14
    Chen E, Piddington R, Decker S, Park J, Yuan Z, Abrams W, Rosenbloom J, Feldman G, Gibson C 1994 Regulation of amelogenin gene expression during tooth development. Dev Dyn 199: 189198.
  • 15
    Adeleke-Stainback P, Chen E, Collier P, Yuan ZA, Piddington R, Decker S, Rosenbloom J, Gibson CW 1995 Analysis of the regulatory region of the bovine X-chromosomal amelogenin gene. Connect Tissue Res 32: 115118.
  • 16
    Snead ML, Paine ML, Chen LS, Yoshida B, Luo W, Zhu D-H, Lei Y-P, Liu Y-H, Maxson REJ 1996 The murine amelogenin promoter: Developmentally regulated expression in transgenic animals. Connect Tissue Res 35: 4147.
  • 17
    Snead ML, Paine ML, Luo W, Zhu D-H, Yoshida B, Lei Y-P, Chen LS, Paine CT, Burstein JM, Jitpukdeebudintra S, White SN, Bringas PJ 1998 Transgene animal model for protein expression and accumulation into forming enamel. Connect Tissue Res 38: 279286.
  • 18
    Zhou YL, Snead ML 2000 Identification of CCAAT/enhancer-binding protein alpha as a transactivator of the mouse amelogenin gene. J Biol Chem 275: 1227312280.
  • 19
    Dhamija S, Liu Y, Yamada Y, Snead ML, Krebsbach PH 1999 Cloning and characterization of the murine ameloblastin promoter. J Biol Chem 274: 2073820743.
  • 20
    Dhamija S, Krebsbach PH 2001 Role of Cbfa1 in ameloblastin gene transcription. J Biol Chem 276: 3515935164.
  • 21
    Ungefroren H, Gellersen B, Krull NB, Kalthoff H 1998 Biglycan gene expression in the human leiomyosarcoma cell line SK-UT-1. Basal and protein kinase A-induced transcription involves binding of Sp1-like/Sp3 proteins in the proximal promoter region. J Biol Chem 273: 2923029240.
  • 22
    Ungefroren H, Lenschow W, Chen WB, Faendrich F, Kalthoff H 2003 Regulation of biglycan gene expression by transforming growth factor-beta requires MKK6-p38 mitogen-activated protein Kinase signaling downstream of Smad signaling. J Biol Chem 278: 1104111049.
  • 23
    Fincham AG, Moradian-Oldak J, Sarte PE 1994 Mass-spectrographic analysis of a porcine amelogenin identifies a single phosphorylated locus. Calcif Tissue Int 55: 398400.
  • 24
    Nelson PS, Gan L, Ferguson C, Moss P, Gelinas R, Hood L, Wang K 1999 Molecular cloning and characterization of prostase, an androgen-regulated serine protease with prostate-restricted expression. Proc Natl Acad Sci USA 96: 31143119.
  • 25
    Ryu O, Hu JC, Yamakoshi Y, Villemain JL, Cao X, Zhang C, Bartlett JD, Simmer JP 2002 Porcine kallikrein-4 activation, glycosylation, activity, and expression in prokaryotic and eukaryotic hosts. Eur J Oral Sci 110: 358365.
  • 26
    Fields S, Song O 1989 A novel genetic system to detect protein-protein interactions. Nature 340: 245247.
  • 27
    Aoba T, Moreno EC, Kresak M, Tanabe T 1989 Possible roles of partial sequences at N- and C-termini of amelogenin in protein-enamel mineral interaction. J Dent Res 68: 13311336.
  • 28
    Lau EC, Simmer JP, Bringas P, Hsu DD, Hu C, Zeichner-David M, Thiemann F, Snead ML, Slavkin HC, Fincham AG 1992 Alternative splicing of the mouse amelogenin primary RNA transcript contributes to amelogenin heterogeneity. Biochem Biophys Res Commun 188: 12531260.
  • 29
    Sambrook J, Russell DW 2001 Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, New York, NY, USA.
  • 30
    Paine ML, Snead ML 1997 Protein interactions during assembly of the enamel organic extracellular matrix. J Bone Miner Res 12: 221227.
  • 31
    Paine ML, Krebsbach PH, Chen LS, Paine CT, Yamada Y, Deutsch D, Snead ML 1998 Protein-to-protein interactions: Criteria defining the assembly of the enamel organic matrix. J Dent Res 77: 496502.
  • 32
    Zhou YL, Lei Y, Snead ML 2000 Functional antagonism between Msx2 and CCAAT/enhancer-binding protein alpha in regulating the mouse amelogenin gene expression is mediated by protein-protein interaction. J Biol Chem 275: 2906629075.
  • 33
    Chen LS, Couwenhoven D, Hsu D, Luo W, Snead ML 1992 Maintainance of amelogenin gene expression by transformed epithelial cells of mouse enamel organ. Arch Oral Biol 37: 771778.
  • 34
    Jahnen-Dechent W, Schinke T, Trindl A, Muller-Esterl W, Sablitzky F, Kaiser S, Blessing M 1997 Cloning and targeted deletion of the mouse fetuin gene. J Biol Chem 272: 3149631503.
  • 35
    Denecke B, Graber S, Schafer C, Heiss A, Woltje M, Jahnen-Dechent W 2003 Tissue distribution and activity testing suggest a similar but not identical function of fetuin-B and fetuin-A. Biochem J 376: 135145.
  • 36
    Schafer C, Heiss A, Schwarz A, Westenfeld R, Ketteler M, Floege J, Muller-Esterl W, Schinke T, Jahnen-Dechent W 2003 The serum protein alpha 2-Heremans-Schmid glycoprotein/fetuin-A is a systemically acting inhibitor of ectopic calcification. J Clin Invest 112: 357366.
  • 37
    Bendtsen JD, Nielsen H, von Heijne G, Brunak S 2004 Improved prediction of signal peptides: SignalP 3.0. J Mol Biol 340: 783795.
  • 38
    Rajagopalan S, Brenner MB 1994 Calnexin retains unassembled major histocompatibility complex class I free heavy chains in the endoplasmic reticulum. J Exp Med 180: 407412.
  • 39
    Okazaki Y, Ohno H, Takase K, Ochiai T, Saito T 2000 Cell surface expression of calnexin, a molecular chaperone in the endoplasmic reticulum. J Biol Chem 275: 3575135758.
  • 40
    Michalak M, Robert Parker J, Opas M 2002 Ca2+ signaling and calcium binding chaperones of the endoplasmic reticulum. Cell Calcium 32: 269278.
  • 41
    Schrag JD, Bergeron JJ, Li Y, Borisova S, Hahn M, Thomas DY, Cygler M 2001 The Structure of calnexin, an ER chaperone involved in quality control of protein folding. Mol Cell 8: 633644.
  • 42
    Stipp CS, Kolesnikova TV, Hemler ME 2003 Functional domains in tetraspanin proteins. Trends Biochem Sci 28: 106112.
  • 43
    Yunta M, Lazo PA 2003 Tetraspanin proteins as organisers of membrane microdomains and signalling complexes. Cell Signal 15: 559564.
  • 44
    Mantegazza AR, Barrio MM, Moutel S, Bover L, Weck M, Brossart P, Teillaud JL, Mordoh J 2004 CD63 tetraspanin slows down cell migration and translocates to the endosomal-lysosomal-MIICs route after extracellular stimuli in human immature dendritic cells. Blood 104: 11831190.
  • 45
    Berditchevski F 2001 Complexes of tetraspanins with integrins: More than meets the eye. J Cell Sci 114: 41434151.
  • 46
    Seigneuret M, Delaguillaumie A, Lagaudriere-Gesbert C, Conjeaud H 2001 Structure of the tetraspanin main extracellular domain. A partially conserved fold with a structurally variable domain insertion. J Biol Chem 276: 4005540064.
  • 47
    Kalabay L, Cseh K, Pajor A, Baranyi E, Csakany GM, Melczer Z, Speer G, Kovacs M, Siller G, Karadi I, Winkler G 2002 Correlation of maternal serum fetuin/alpha2-HS-glycoprotein concentration with maternal insulin resistance and anthropometric parameters of neonates in normal pregnancy and gestational diabetes. Eur J Endocrinol 147: 243248.
  • 48
    Magnuson VL, McCombs JL, Lee CC, Yang F, Bowman BH, McGill JR 1998 Human alpha 2-HS-glycoprotein localized to 3q27-q29 by in situ hybridization. Cytogenet Cell Genet 47: 7274.
  • 49
    Szweras M, Liu D, Partridge EA, Pawling J, Sukhu B, Clokie C, Jahnen-Dechent W, Tenenbaum HC, Swallow CJ, Grynpas MD, Dennis JW 2002 alpha 2-HS glycoprotein/fetuin, a transforming growth factor-beta/bone morphogenetic protein antagonist, regulates postnatal bone growth and remodeling. J Biol Chem 277: 1999119997.
  • 50
    Waddington RJ, Roberts HC, Sugars RV, Schonherr E 2003 Differential roles for small leucine-rich proteoglycans in bone formation. Eur Cells Mater 6: 1221.
  • 51
    Kinsella MG, Bressler SL, Wight TN 2004 The regulated synthesis of versican, decorin, and biglycan: Extracellular matrix proteoglycans that influence cellular phenotype. Crit Rev Eukaryot Gene Expr 14: 203234.
  • 52
    Septier D, Hall RC, Embery G, Goldberg M 2001 Immunoelectron microscopic visualization of pro- and secreted forms of decorin and biglycan in the predentin and during dentin formation in the rat incisor. Calcif Tissue Int 69: 3845.