Sequencing of Bovine Enamelin (“Tuftelin”) a Novel Acidic Enamel Protein*

Enamelins are a major group of 28-70-kDa acidic proteins rich in aspartic acid, glutamic acid, serine, and glycine found in developing and mature extracellular enamel; a unique and highly mineralized ectoder- mal tissue covering vertebrate teeth. They have been associated with the mineralization and structural or- ganization of this tissue. In an attempt to elucidate the primary structure of enamelin, a 2674-base pair cDNA isolated from a bovine ameloblast-enriched, lambda Zap 2 expression library, was sequenced. The identity and localization of the deduced protein was confirmed by amino acid composition, enzyme-linked immunosor- bent assay, Western blotting, indirect immunohistochemistry, and high resolution protein-A gold immu- nocytochemistry. The immunological techniques were employed using antibodies directed against synthetic peptides corresponding to the protein sequence deduced from the cloned cDNA sequence. The results reveal the deduced protein to be a novel acidic enamel protein. It contains 389 amino acids and has a calcu- lated molecular weight of 43,814. Its amino acid composition is similar to that of “tuft” proteins (enamel matrix protein fragments remaining in the mature tis-sue). It contains one potential N-glycosylation site and 6 cysteine residues. Southern hybridization of the cloned cDNA with genomic bovine DNA indicated the existence of a single gene with one enamelin clones were identified, one containing a 2.8-kb insert with an internal EcoRI site and the other an insert of 1.8 kb with no internal EcoRI site (5). The present paper describes the sequencing of the 2.8-kb cDNA enamelin clone, tuftelin. cDNA Sequencing-The cloned 2.8-kb cDNA was sequenced using both subcloned fragments in M13 (31) and double-stranded sequenc- ing of the denatured plasmid DNA, with T4, T7, and synthetic oligo-primers using the di-deoxy chain termination method (33). For primer walking (32) synthetic oligonucleotides, 19-21 bases long, were de-signed every 250 bases in both directions of the cDNA. DNA sequences were analyzed by the program of the University of Wisconsin, Genetic Computer Group (WGCG). to nitrocellulose of (34). DNA nitrocellulose anhydrous HF 10% anisole -5 "C. Purity of the peptides was examined by HPLC using 0.1% trifluoroacetic acid/water/acetonitrile gra- dients. Conjugation of the Peptides to Chick Serum Albumin and Synthetic Peptide Antisera Production-The conjugation of peptides with the carrier protein chick serum albumin was performed by first bromo-acetylating the available amino groups of the carrier by the reaction with bromoacetyl-N-hydroxysuccinimide ester (37) followed by ad-dition of a cysteine-containing peptide. Chick serum albumin shows no immunological cross-reactivity with bovine or human serum albumin (52). Reaction conditions for conjugation were the following: to 15 mg of carrier protein dissolved in 2.5 ml of 0.1 M NaHC03 was added 4.7 mg of bromoacetyl-N-hydroxysuccinimide ester dissolved in 100 pl of acetonitrile. The mixture was stirred for 1 h at room temperature. Then 15 mg of the peptide was added, and the reaction mixture was stirred for a further 1 h. The peptide-protein conjugates were obtained following exhaustive dialysis against 0.1 M NaHCO:? and lyophilization. Antisera were produced by injection of 1 mg of peptide conjugate in Freund's complete adjuvant into 10 intradermal and two intramuscular sites followed by two (at 2 and 4 weeks) 1-mg booster injections into multiple sites in incomplete adjuvant. The rabbits were bled at 2-week intervals.

The functions of these two major matrix protein groups are still not clear, but they are thought to be involved in the mineralization and structural organization of enamel. The acidic enamelins are secreted at a very early stage of enamel formation (3,13,(17)(18)(19), and are tightly bound to the surface of the growing crystallites. They have been reported to possess @-pleated sheet structures (potential nucleating structures for hydroxylapatite (20)), and under certain in vitro conditions can inhibit crystal growth (21). For these reasons, enamelins have been suggested to be involved in the nucleation and regulation of enamel crystal growth. Unlike the amelogenins, however, which have been cloned and were sequenced first in the mouse (22) and later in the bovine (23), no amino acid sequence typical of enamelins has been published. This is probably one reason for the uncertainty of its identity (24)(25)(26). Enamelins have been shown to possess common immunologically reactive epitopes with enameloid of aquatic species spanning 450 million years of vertebrate evolution (i.e., shark and hagfish), suggesting that they are highly conserved, and again indicating their importance in the development and mineralization of enamel (27,28). The present work describes the sequencing of bovine enamelin protein (tuftelin), a novel acidic protein.

Construction and Identification of the Bovine Enamelin cDNA
Clone-The construction and preliminary identification of the enamelin cDNA clones have been previously reported (5). In brief, tooth organs (mandibular permanent molar) were removed from 3-month old Holstein calves. The teeth generally contained only the forming stage of enamel (first stage).
Enamel organ (ameloblast-enriched tissue) was separated from the adjacent enamel, pooled, and then extracted for total RNA. Poly(A) mRNA was selected by affinity chromatography on oligo(dT)-cellulose and 20 fig of poly(A) mRNA was used to make cDNA (29). The cDNA was size-selected to be enriched in molecules 2 1.0 kb' and inserted into the EcoRI site of a lambda Zap 2 expression library (Stratagene). The amplified cDNA expression library was screened using affinity-purified polyclonal antibodies against the enamelin 66-kDa protein (30). Two types of The abbreviations used are: kb, kilobase(s); ELISA, enzymelinked immunoabsorbent assay; PBS, phosphate-buffered saline; SDS, sodium dodecyl sulfate; BSA, bovine serum albumin; HPLC, high performance liquid chromatography; FITC, fluorescein isothiocyanate.

cDNA-derived Enamelin Protein
Sequence and Localization enamelin clones were identified, one containing a 2.8-kb insert with an internal EcoRI site and the other an insert of 1.8 kb with no internal EcoRI site (5). The present paper describes the sequencing of the 2.8-kb cDNA enamelin clone, tuftelin.
cDNA Sequencing-The cloned 2.8-kb cDNA was sequenced using both subcloned fragments in M13 (31) and double-stranded sequencing of the denatured plasmid DNA, with T4, T7, and synthetic oligoprimers using the di-deoxy chain termination method (33). For primer walking (32) synthetic oligonucleotides, 19-21 bases long, were designed every 250 bases in both directions of the cDNA. DNA se-quences were analyzed by the program of the University of Wisconsin, Genetic Computer Group (WGCG).
Genomic DNA Extraction and Southern Analysis-High molecular weight bovine genomic DNA was extracted from a fetal bovine liver and aliquots of 10 pg digested with excess amounts of various restriction endonucleases for 3 h at 37 "C using buffers recommended by the manufacturer. After enzyme treatment, fragmented DNA was resolved through 0.8% agarose gels and transferred to nitrocellulose using the method of Southern (34). DNA bound nitrocellulose was hybridized with full length tuftelin cDNA which was previously FIG. 1. Nucleotide sequence of bovine enamelin (tuftelin) cDNA and deduced amino acid sequence. Nucleotide residues are numbered on the right. A Kozak consensus sequence for eukaryotic initiation sites is underlined. A possible polyadenylation signal in the 3'-untranslated sequence is underlined. A termination signal (TAG), potential N-glycosylation site (Asn-Lys-Ser), and tripeptide (Glu-Ser-Leu), the phosphorylated form of which appears in all mineralizing tissues, are also marked. Cysteine residues are marked with arrows. The deduced protein sequences to which synthetic peptides (73,74,75) were produced, and to which anti-sera LF-73, LF-74, LF-75 were made, are underlined.  (41). labeled with ' ' P to a specific acitivity of -10' per pg of DNA.
Hybridization was performed in a solution of 50% formamide, 5 X SSC (1 X is 15 mM sodium citrate and 150 mM sodium chloride), 50 mM Tris-HCI, 5 X Denhart's (1 X is 0.2% bovine serum albumin, 0.2% Ficoll, 0.2% polyvinyl pyrrolindone), and 0.33 pg/ml denatured salmon sperm DNA at 41 "C for 16 h. Nonspecifically bound probe was removed by washing three times in 2 X SSC/O.l% sodium dodecyl sulphate (SDS) at 25 "C followed by three additional washes in 0.1 X SSC/O.l% SDS a t 68 "C. Specifically bound probe was detected by exposure of the nitrocellulose to XAR film for 1 week a t -70 "C.
Computer Analysis-The National Biochemical Research Foundation protein data base and GenBank nucleic acid data base were searched for sequence homology. Hydrophilicity was predicted by the method of Kyte and Doolittle (35). Isoelectric point analysis was carried out by the WGCG program.
Peptide Synthesis-The sequences used for the synthetic peptides (73, 74, 75; see Fig. 1) were derived from the deduced protein sequences of the sequenced cDNA tuftelin clone. The three peptides were synthesized using an automated solid-phase peptide synthesizer (Model 430A, Applied Biosystems). The procedure was based on the original solid-phase peptide synthesis procedure described by Merrifield (36). The protection and release of the peptides from the PAM resin were accomplished by treating the resin with anhydrous H F containing 10% anisole at -5 "C. Purity of the peptides was examined by HPLC using 0.1% trifluoroacetic acid/water/acetonitrile gradients.
Conjugation of the Peptides to Chick Serum Albumin and Synthetic Peptide Antisera Production-The conjugation of peptides with the carrier protein chick serum albumin was performed by first bromoacetylating the available amino groups of the carrier by the reaction with bromoacetyl-N-hydroxysuccinimide ester (37) followed by addition of a cysteine-containing peptide. Chick serum albumin shows no immunological cross-reactivity with bovine or human serum albumin (52). Reaction conditions for conjugation were the following: t o 15 mg of carrier protein dissolved in 2.5 ml of 0.1 M NaHC03 was added 4.7 mg of bromoacetyl-N-hydroxysuccinimide ester dissolved in 100 pl of acetonitrile. The mixture was stirred for 1 h at room temperature. Then 15 mg of the peptide was added, and the reaction mixture was stirred for a further 1 h. The peptide-protein conjugates 0 loo were obtained following exhaustive dialysis against 0.1 M NaHCO:? and lyophilization. Antisera were produced by injection of 1 mg of peptide conjugate in Freund's complete adjuvant into 10 intradermal and two intramuscular sites followed by two (at 2 and 4 weeks) 1-mg booster injections into multiple sites in incomplete adjuvant. The rabbits were bled at 2-week intervals.
Dynatech microtiter plates were coated overnight at 4 "C with either 1.0 pg of purified enamelin and 73, 74, and 75 peptides (not conjugated to chick serum albumin) or with dentin matrix proteins. The plates were then blocked with 125 pl of 2% bovine serum alhumin (BSA) in phosphate-buffered saline (PBS) for 1 h at room temperature. 100 pl of antisera (LF-73, LF-74, LF-75, and affinity-purified polyclonal antibody against enamelin) at increasing dilutions (150 to 1:6400) (in 2% BSA solution) were then added. Incubation with the respective preimmune sera served as controls. The plates were then incubated for 1 h at room temperature, washed for 3 X 10 min (PBS with 0.05% Tween 20) and incubated in a 1:2,000 dilution of peroxidase-conjugatedgoat anti-rabbit IgG in 2% BSA for 1 h. After washing 3 X 10 min in PBS-Tween, the wells were incubated in 100 p1 of 2.2'azino-di-3-ethylbenzthiazoline sulfate) for 30 min, then monitored a t 405 nm in a Dynatech microplate spectrophotometer.
Indirect Immunohistochemistry and Immunofluorescence-Indirect immunohistochemistry using synthetic peptide antisera as first antibody (LF-73, LF-74, LF-75) and goat antirabbit IgG coupled to horseradish peroxidase (second antibody) was carried out on 5-pm sections of undecalcified paraffin-embedded developing bovine teeth. Indirect and immunofluorescence using synthetic peptide antisera as first antibody, and fluorescein-conjugated antirabbit (FITC) as second antibody, was performed on Cryostat sections (5 pm) of nondecalcified developing bovine teeth and on nonetched and HC1-etched (5 s) 30-40-pm thick longitudinal ground sections of hovine and human mature erupted teeth. Dilutions of 300-1000 were used for each first antibody. The second antibody solution was generally twice as concentrated as that used for the first antibody. PBS was used as a washing solution and 1% fetal calf serum to block unspecific sites. The tissue specimens were incubated with the first antibody for 1 h and the second antibody for 7 min. Longitudinal tooth sections treated with corresponding preimmune sera served as a control.
High Resolution Protein A-Gold Immunohistochemistry-High resolution protein A-gold immunohistochemistry was carried out according to the methods of Bendayan (48) and Nanci et al. (49). Briefly, nondecalcified developing bovine incisor teeth were fixed with 1 % glutaraldehyde in 0.1 M phosphate buffer, pH 7.3, and postfixed with potassium ferrocyanide-reduced osmium. The tissues were dehydrated and then embedded in Epon. Sections were then prepared and processed for immunohistochemistry (48). The grids were floated on a drop of a saturated solution of sodium metaperiodate for 1 h, rinsed, and then incubated on a drop of antiserum (synthetic peptide antisera) for 1 h at room temperature. Following several washes with PBS, the grids were floated on a drop of protein A-gold for 30 min at room temperatures to reveal the antigen-antibody complex. The grids were then washed and conventionally stained. Reaction of the tissue with preimmune sera served as control.  I  I  I  I  I  l  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  I

RESULTS
cDNA Sequence Analysis-In a preliminary report we described the isolation of cDNA encoding bovine enamelin (5). The cDNA clone, 2674 base pairs long, was isolated by antibody screening (affinity-purified antibody against enamelin, (30) from a bovine ameloblast-enriched cDNA library in expression vector or lambda Zap 2. Northern blot analysis showed that it hybridized to an mRNA species of 2.7 kb and a shorter mRNA of 1.0 kb (5).
The deoxynucleotide sequence for the 2.7-kb enamelin cDNA clone and the predicted amino acid sequence of the protein it encodes is illustrated in Fig. 1. A methionine codon at base 182 surrounded by a Kozak (39) consensus sequence (GCCATGG) for eukaryotic initiation sites is followed by an open reading frame of 389 amino acids extending to a stop codon (TAG) at base 1349. A polyadenylation signal (AA-TAAA) begins at position 2637. The sequence contains 5 cysteine residues and a potential N-glycosylation site (Asn-Lys-Ser). It also contains a Glu-Ser-Leu peptide, the phosphorylated form of which has been shown to exist in all mineralizing tissues (40). 181 bases (1-181) from the 5' and 1322 (1352-2674) bases at the 3' ends are untranslated sequences.
Deduced Protein Characteristics-A search in the National Biochemical Research Foundation protein data base and GenBank nucleic acid data base revealed that the cDNA encodes a novel protein. Its composition resembles tuft proteins (protein remaining in the mature enamel, the origin of which is mainly the enamelins secreted in the forming stage (41,42) (Fig. 2). This figure shows the comparison of amino acid composition of the deduced bovine enamelin protein ("tuftelin") with one of the tuft protein sequences reported by Robinson et al. (41). The amino acid compositions are represented as rose diagrams according to the method of Robinson et a1. (41). The calculated molecular weight of the deduced protein is 43,814.
Computer analysis of the deduced amino acid sequence revealed the protein to be highly hydrophilic (Fig. 3) and acidic with an isoelectric point of 5.2. This isoelectric point is similar to that already reported for this protein (18, 44).
Analysis of Tuftelin Genomic DNA-In order to determine the nature and approximate copy number of the tuftelin gene, bovine genomic DNA was fragmented with enzymes that either cut the cDNA, or not at all. EcoRI, which cuts the cDNA into two asymmetric fragments of 1.8 and 1.0 kb, also generated two hybridizable genomic fragments that were distinct in size from the cDNA (Fig. 4). The restriction enzymes BamHI and HindIII, which do not cleave the cDNA, resulted in multiple hybridizable genomic fragments ranging in size from -2.5 to 1.6 kb (Fig. 4). These data imply the probable existence of only one gene copy per haploid genome containing one or more introns. Immunological Analysis of the Deduced Tuftelin Protein-Since no amino acid or cDNA sequences typical of enamelin were available in the literature, to confirm the identity of the deduced protein, antisera against three chick serum albuminconjugated synthetic peptides (corresponding to 73, 74, 75, respectively, Fig. 1) were reacted against forming and mature enamel (indirect immunohistochemistry and immunofluorescence) and denatured enamelin (ELISA, Western blotting). Fig. 5 are ELISA analyses of these synthetic peptide antisera. Fig. 5a shows the ELISA reaction of synthetic peptide antisera (LF-73, LF-74, LF-75) made to deduced tuftelin protein sequences (73, 74, 75) (see Fig. 1) with (i) nonconjugated peptides (73, 74, 75), (ii) dentin noncollagenous extracellular matrix proteins. Reaction with pre-immune serum served as control. The results revealed (i) that the antisera had indeed been produced against the peptides of concern, and (ii) that they did not react with dentin noncollagenous extracellular matrix protein. Fig. 5b shows the ELISA reaction of synthetic peptide antisera (LF-73, LF-74, LF-75) made to the deduced tuftelin protein sequences with sequentially extracted (6) and purified (15) enamelin protein. Reaction with preimmune sera served as controls. The results reveal strong reaction with the enamelin protein extracted from developing extracellular bovine enamel. Fig. 5c shows the ELISA reaction of affinitypurified polyclonal antibody against enamelin (30) (used originally to screen the enamelin cDNA clones from the expression library) with nonconjugated synthetic peptides (73, 74, 75) produced according to the tuftelin-deduced sequences. Reaction with preimmune sera served as control. The results indicate that all three synthetic peptides made to the deduced protein reacted with the polyclonal antibody against enamelin and that the highest activity occurred with peptide 75. Fig. 6 shows the immunological relationship of the deduced tuftelin protein to enamel protein extracts using Western analysis. In this experiment an enamelin-enriched fraction (obtained by the method of Termine et al. (6)) (see also Refs. 5 and 49) was reacted with synthetic peptide antisera LF-73 (antibody produced to the peptide sequence near the COOH-terminal end of the deduced protein). The results reveal that the antisera reacted with 66-58-, 48-, and 28-kDa protein bands. Indeed, enamelins have been reported to possess molecular masses of 66,58,48,and 28 kDa (17,19,44,45).
Expression and Ultrastructural Localization of Tuftelin in Enamel-In order to determine the expression and ultrastructural localization of tuftelin, indirect immunohistochemistry, indirect immunofluorescence, and high resolution protein Agold immunochemistry were performed employing the synthetic peptide antisera (see Figs. 7 and 8). Fig. 7A shows the indirect immunohistochemistry (using horseradish peroxidase as second antibody) and immunofluorescence (using FITC antibody as second antibody) on paraffin-embedded and cryostat sections (5 pm) of forming bovine enamel, respectively, employing synthetic peptide antisera . Reaction with preimmune sera served as controls. The antibodies reacted with the ameloblast cells, in particular cDNA-derived Enamelin Protein Sequence and Localization with the secreting region (Tomes's process) but not with the odontoblasts (dentin-secreting cells) Fig. 7A( I). In addition, the antisera reacted with developing enamel extracellular matrix but not with dentin ( Fig. 7 A ( 2 ) ) . Fig. 7B shows the indirect immunofluorescence (using FITC antibody as second antibody) on 30-40-pm longitudinal ground sections of mature, erupted, unetched, and HC1-etched bovine and human enamel, employing the synthetic peptide antisera as first antibodies. Reaction with preimmune sera served as control.
The results show reaction of antibodies with bovine ( Fig.  7 B ( I ) ) and human ( Fig. 7 B ( 2 ) ) tuft proteins radiating from the dentin-enamel junction toward the enamel surface. The reaction on a light-microscopic level was localized mainly interprismatically, between the enamel prisms. In the human mature enamel, the typical horseshoe pattern was seen ( Fig.  7 B ( 2 ) ) . No definite reaction occurred with the underlying dentin. In order to determine the ultrastructural localization of tuftelin protein in developing bovine enamel, the protein-A gold method (48, 49) was used with synthetic peptide antisera as first antibodies and conjugated gold particles of 10 nm in size. The grids were inspected under a Phillips 300 electron microscope. Fig. 8A shows reaction of tuftelin antisera with developing bovine enamel and Fig. 8B the control, using preimmune sera. These results reveal that tuftelin, present in the extracellular enamel, is mainly associated with the crystal component.

DISCUSSION
We have determined for the first time the DNA sequence for an enamel-specific gene coding for one acidic enamelin protein, tuftelin. The results reveal that the protein deduced from the sequenced cDNA is a novel protein with many hydrophilic residues and an isoelectric point of an acidic Sections treated with preimmune sera were used as control (2). Indirect immunohistochemistry of paraffin-embedded thin sections (5 pm thick) of a developing bovine tooth depicting ameloblast ( 3 ) and odontoblast ( 4 ) cells (photographed from the same antibody-treated section) utilizing LF-74 synthetic peptide antisera as first antibody and conjugated alkaline phosphatase as second antibody. A, ameloblast; D, dentine; 0, odontoblast. In B, reaction of synthetic peptide antisera LF-75 with a 30-40-pm thick longitudinal section of mature erupted bovine enamel. I , unetched tissue; 2, enamel which was etched with HCl leaving behind tuft protein residues (7") radiating from the enamel dentin junction toward the enamel surface; 3, reaction of synthetic peptide antiserum LF-75 with a nonetched 30-40-pm thick longitudinal section of mature erupted human enamel showing tuft proteins which fluoresce; 4, a cross-section of enamel prisms. The surrounding of each prism (horseshoe pattern) clearly fluoresces. The FITC filter set consists of Exciter filter BP 450-490 nm, dichromatic beam splitter FT 510 nm, and a barrier filter LP 520.
protein. The cDNA sequence contains a potential start codon (AUG) at position 182 with a homologous Kozak sequence (the consensus sequence for a eukaryotic initiation site) and a stop codon (TAG) followed by a potential poly(A) initiation site and a poly(A) tail. This results in a predicted 389-amino acid protein of 43,814 daltons.
However, glycosylation of the nascent protein would result in a significantly higher apparent molecular size on SDS gels.
One potential N-glycosylation site was determined in the present 44-kDa deduced protein. Enamelins have, indeed, been reported to be glycosylated (6,43). The prediction of the number and location of 0-linked glycosylation sites that might very well also exist in the deduced protein is more difficult. In this respect, the molecular weight of the deduced protein

A B
FIG. 8. Ultrastructural localization of tuftelin protein in developing bovine enamel using protein-A gold method (48, 49) and synthetic peptide antisera as first antibodies. The grids were inspected under a Philips 300 electron microscope. A, reaction of tuftelin antisera with developing bovine enamel ( X 65,250); B, shows reaction of preimmune sera (control) with developing bovine enamel ( X 65,250); E, enamel. fits well with the range of molecular weights reported for the enamelin proteins (3,6,18,19,30,44,45).
Western blotting of enamelin protein bands (found in enamelin-enriched fraction) (46) revealed that synthetic peptide antiserum LF-73 (antibody produced to the peptide sequence near the COOH-terminal end of the deduced protein) recognizes 66-58-, 48-, and 28-kDa enamelins that are typically recognized by affinity-purified polyclonal antibodies (30) and have been shown to be synthetized by the ameloblast cells.
The amino acid composition of the deduced enamelin protein is similar to tuft proteins (proteins which remain in the mature enamel (41). This suggests that at least some of the enamelins which are secreted during the early stages of enamel formation remain throughout the development of the enamel and are found (perhaps partially degraded) in the mature tissue. Further support for this idea comes from recent independent studies. Robinson et al. (42) produced antibodies to tuft proteins which reacted with 50-70-kDa enamelin components in the developing enamel matrix and with secretory organelles of ameloblasts, indicating that some components of tuft are early secretory products (nonamelogenin enamelinlike) of ameloblasts.
Their antisera also reacted with some minor components (25 kDa), which the authors suggest were not amelogenin. Furthermore, Amizuk and Ozawa (47) showed that specific polyclonal antibodies against enamelin proteins cross-reacted with components of tuft proteins. Our present indirect immunohistochemical studies (Fig. 7) show that the synthetic peptide antisera, which reacted with purified enamelin protein (Figs. 5 and 6), also reacted well with the tuft proteins radiating from the enamel dentin junction towards the ename surface both in bovine and human mature enamel. For this reason, we propose the name tuftelin for this specific enamel protein.
On a light-microscope level, the reaction of the synthetic peptide antisera with the mature enamel occurred mainly interprismatically, surrounding the enamel prisms. In the human tooth (Fig. 7), this reaction occurs in a typical horseshoe pattern. This agrees well with recent high resolution protein-A gold studies of Amizuk and Ozawa (47), who reacted polyclonal antibodies against enamelin with mature enamel proteins.
Recently, our studies using high resolution protein A-gold immunohistochemistry employing these synthetic peptide antisera confirmed earlier findings (3, 6-10) that enamelins in the developing extracellular enamel matrix are localized mainly at the crystal surfaces (Fig. 8).
Our results reveal cross-reactivity between enamelins of different species, such as bovine and human (see Fig. 7) and also shark enameloid (50). These immunological data suggest that not only has the enamelin structure been highly conserved during the 450 million years of vertebrate evolution (27), but that this extracellular matrix protein has an important role in enamel mineralization. Southern hybridization with bovine genomic DNA indicates the existence of one tuftelin gene containing one or more introns. EcoRI has one internal site in the cDNA, predicting two or more radioactive DNA bands (if only one gene exists), but Hind111 has no internal restriction sites and thus the appearance of two or more radioactive bands indicates one or more introns. The existence of one or more introns suggests the possibility of RNA processing as the basis for the existence of two to four nascent enamelin proteins (19, 45), but this must await further analysis. It is also possible, however, that enamel contains a number of unrelated enamelins coded by different genes.
The secretion of relatively high concentrations of enamelin at the dentin enamel junction area, already in very early stages of development, as reported by Robinson et al. (13), Deutsch et al. (17), and more recently by Slavkin et al. (18), has suggested to us (5, 17) that these enamelin proteins are associated with the mineralization of the hypermineralized enamel region adjacent to the dentin enamel junction area present in the very early forming enamel of all species, thus creating a mineralization front in enamel. The exact role, however, is still unclear. Some have suggested enamelin is a template or nucleator for mineralization (3, 18, 20). More recently, it has been suggested (51) that a component of enamel matrix protein (enamel sheath, which represents in part the presence of enamelins) could bind to collagen fibers of the underlying dentin. Such chemical interactions between the two matrices may serve to promote enamel crystal nucleation. Preliminary results' show that some reaction of our enamelin synthetic peptide antisera on the ultrastructural level (protein-A gold method) could be seen at the dentin enamel junction area in the vicinity of the collagen fiber tip neighboring the enamel crystals.
Finally, an NH2-terminal decapeptide sequence of a 22-kDa unidentified protein band isolated from enamelin extract has just been published (53). It contains a -Pro-Ser-Ser-X-X-Ala-Gln-sequence. A homologous sequence can be found in our tuftelin sequence: -Pro-Ser-Pro-Pro-Ala-Gln-(previous sequence has an extra serine). This may indicate the published peptide sequence to be a tuftelin derivative or fragment.