Enamelin Is Critical for Ameloblast Integrity and Enamel Ultrastructure Formation

Mutations in the human enamelin gene cause autosomal dominant hypoplastic amelogenesis imperfecta in which the affected enamel is thin or absent. Study of enamelin knockout NLS-lacZ knockin mice revealed that mineralization along the distal membrane of ameloblast is deficient, resulting in no true enamel formation. To determine the function of enamelin during enamel formation, we characterized the developing teeth of the Enam−/− mice, generated amelogenin-driven enamelin transgenic mouse models, and then introduced enamelin transgenes into the Enam−/− mice to rescue enamel defects. Mice at specific stages of development were subjected to morphologic and structural analysis using β-galactosidase staining, immunohistochemistry, and transmission and scanning electron microscopy. Enamelin expression was ameloblast-specific. In the absence of enamelin, ameloblasts pathology became evident at the onset of the secretory stage. Although the aggregated ameloblasts generated matrix-containing amelogenin, they were not able to create a well-defined enamel space or produce normal enamel crystals. When enamelin is present at half of the normal quantity, enamel was thinner with enamel rods not as tightly arranged as in wild type suggesting that a specific quantity of enamelin is critical for normal enamel formation. Enamelin dosage effect was further demonstrated in transgenic mouse lines over expressing enamelin. Introducing enamelin transgene at various expression levels into the Enam−/− background did not fully recover enamel formation while a medium expresser in the Enam+/− background did. Too much or too little enamelin abolishes the production of enamel crystals and prism structure. Enamelin is essential for ameloblast integrity and enamel formation.


Introduction
Dental enamel, which has the hardness of mild steel [1], is formed by ameloblasts. Ameloblasts are a tooth-specific cell type, that secrete three major enamel-specific matrix proteins: amelogenin, ameloblastin, and enamelin. These matrix proteins are expressed by secretory stage ameloblasts and are critical for normal enamel biomineralization. After expanding the enamel matrix to its final dimensions, secretory stage ameloblasts undergo a transition to maturation stage that alters their expression of matrix proteins. During the maturation stage, degradation and reabsorption of matrix proteins becomes a priority of ameloblasts, because such events allow space to be freed up for enamel crystals to increase in width and thickness, leading to the hardening of dental enamel. Mature dental enamel contains less than 1% of organic matter [2]. Based on studies of gene-targeted mouse models, absence of amelogenin [3], ameloblastin [4] or enamelin [5], results in significantly decreased enamel thickness and altered prism structure.
In humans, enamelin gene mutations result in autosomal dominant hypoplastic amelogenesis imperfecta (OMIM 104500 AIH2 and 204650 AI1C). Individuals with both ENAM alleles defective typically present with severe enamel hypoplasia, while individuals with one allele mutated have localized, pitted defects and/or horizontal grooves on their teeth [6,7]. Enamelin was first isolated from developing porcine enamel matrix [8]. Full length enamelin is a large glycoprotein having apparent molecular weight ,186 kDa [9]. In the developing enamel, rapid cleavage of enamelin mediated by matrix metalloproteinase 20 (MMP20) takes place immediately upon secretion. The C-terminus of enamelin concentrates near the mineralization front where crystal initiation and elongation takes place, while the major cleavage product, 32 kDa enamelin, distributes throughout the developing enamel layer and accumulates to make up about 1% of the total enamel matrix [10]. Compared to amelogenin, the major extracellular constituent comprising ,90% of the developing enamel matrix, enamelin and its cleavage products are present at a lesser quantity, only about 5% of the total enamel matrix. The importance of enamelin in dental enamel formation has been demonstrated unequivocally by a chemically induced mouse model [11], gene-targeted mouse model [5] and human mutations giving rise to hypoplastic amelogenesis imperfecta [12]. The consistent observation of abnormally thin and disorganized enamel in Enam null mice and in human AI suggests that enamelin function is required to direct crystal formation and to achieve prism structural organization, and optimal enamel thickness. Loss of tooth or enamel formation during evolution is associated with molecular decay of the enamelin gene [13].
The abnormal enamel development of Enam +/2 and Enam 2/2 mouse teeth was illustrated by morphological analyses of the enamelin (Enam ID:13801) knockout lacZ knockin mouse model [5]. Based on the observation of more than 10 generations of the Enam 2/2 breeding colony, enamelin absence affected litter size, weight gain and survival of the young pups if the Enam 2/2 mice were not maintained on soft diet [14]. Furthermore, because the abnormally thin enamel altered inter-dental contacts, secondary effects of molar drifting were observed in the Enam 2/2 mice, such as decreased bone volume between the first and second molars [15]. This observation indicated that when the space distal to the first molar was open due to the absence of enamel, the teeth responded by shifting to close the space. However, bone density was not altered among the Enam 2/2 mice, which suggested that the observed changes in periodontal bone level were secondary to molar drifting and super-eruption caused by the lack of occlusal and interdental contacts due to enamel hypoplasia and attrition. At the histological level, the most striking effect of enamelin absence was the lack of mineral formation at the mineralization front that ultimately resulted in no mineralization of the enamel layer as demonstrated by von Kossa staining of non-demineralized PN7 first molar sections [5]. In the Enam 2/2 mice, premature and excessive apoptosis of ameloblasts contributed to overall cellular disorganization and enamel hypoplasia [16].
In this study, our objective was to define the enamel ultrastructural defects of the enamelin knockout lacZ knockin mouse and to determine the effect of enamelin transgene expression in the knockout background. Experiments were conducted to characterize the cellular and structural anomalies of enamel occurring in the Enam +/2 and Enam 2/2 mouse teeth. Transgenic mice over expressing enamelin at minimal, medium and high levels were established, characterized and bred with the Enam 2/2 mice separately. The effect of transgenic enamelin on enamel formation in the Enam 2/2 background was assessed morphologically. We hypothesized that if the amount of enamelin comparable to that of the wild type could be introduced into the Enam 2/2 background, then the defective enamel structure and thickness should be restored.

Protocol approval
Animal protocols were reviewed and approved by the University of Michigan Institutional Animal Care and Use Committee.

Beta-galactosidase staining
Beta-galactosidase whole-mount staining of embryonic day 13.5, newborn, and postnatal day 7 (PN7) and day 14 (PN14) mice from wild type, enamelin knockin heterozygous (Enam +/2 ) and homozygous (Enam 2/2 ) mice were processed with the removal of intestines. Skin was removed from samples for genotyping. Experiments were done in triplicate. Separately, internal organs from PN7 Enam 2/2 mice were processed for cryosectioning and bgal staining. Day 14 mice were perfused via the heart with 4% paraformaldehyde (PFA) and 0.1% glutaraldehyde (GA) and all other samples were collected following conventional protocol. Half maxillae were collected and fixed overnight, washed with phosphate-buffered saline (PBS, 3615 min), then decalcified in 10% ethylenediamine tetraacetic acid (EDTA) up to 7 days according to the age of the mice. The PN7 decalcified tissues were washed in PBS (3615 min) and embedded in Tissue-Tek O.C.T. Compound (ProSciTech, Queensland, Australia). Whole mouse samples and PN7 tissue blocks sectioned at 10 mm thickness were post-fixed for 5 min in 0.5% GA, washed in 100 mM, pH 8.0 Hepes buffer (365 min), and incubated at 45uC for 2 h in freshly prepared X-gal staining buffer, pH 8.0, containing 1 mg/ml Xgal, 100 mM Hepes, 5 mM potassium ferricyanide, 1 mM MgCl 2 , 2% Triton X-100. Tissue sections were rinsed and stored in 100 mM, pH 8.0 Hepes buffer for counterstaining with hematoxylin then observed under a dissection microscope (Nikon SMZ1000) or light microscope (Nikon Eclipse E600). All images were captured using a digital camera (Nikon Digital Sight) and NIS-Elements basic research imaging software V4.10 (Mager Scientific, Dexter, MI, USA).

Immunohistochemistry
The techniques used for immunohistochemistry were described previously (Simmer et al., 2009). PN7 mouse heads were dissected and fixed by immersion in ice cold 4% PFA and 1% GA in a 67 mM phosphate buffer, pH 7.4 overnight. PN14 mice were perfused with ice cold 4% PFA and 1% GA in a 67 mM phosphate buffer, pH 7.4. Mouse maxillae and mandibles were demineralized using 10% EDTA for 5 to 7 days. The fixed maxillary and mandibular processes were dissected, embedded and sectioned at 5 mm. The slides of the maxillary molars were deparaffinized 36 in xylene, rehydrated and treated with peroxidase blocking agent. The sections were incubated with serum blocking solution and then incubated with amelogenin antibodies (rm179 at 1:2,000) at room temperature for 30 min and enamelin antibodies (mENAM 223-236 at 1:100) at 4uC for overnight. Negative control slides were incubated with preimmune serum at the same concentrations as the primary antibodies. Following incubation with primary antibodies (or pre-immune serum in the negative control), the slides were washed 36 with PBS, incubated with biotinylated secondary antibody, washed 36, incubated with streptavidin-enzyme conjugate, washed 36, incubated with substrate-chromogen mixture, washed 36, counterstained with hematoxylin, and mounted with Aqueous Mounting Solution (Invitrogen, Carlsbad, CA) for microscopy and photography.

SEM analyses
Ethanol dehydrated, air-dried, fractured mandibular incisors from 7-week-old wild type, Enam +/2 , Enam 2/2 and transgenic mice were mounted with conductive carbon cement onto metallic stubs, de-gassed in a vacuum desiccator overnight, sputter coated with Au-Pd thin film to increase conductivity, and imaged at Amray EF 1910 Scanning Electron Microscope operating at an accelerating voltage of 3-5 kV. For backscattered SEM, samples were polished with silicone carbide paper (South Bay Technology, Inc. San Clemente, CA) followed by 4 h of polishing with 1.0 micro alumina abrasive with Multitex Polishing Cloth using a Buehler Supermet 2 Position Polisher (Lake Bluff, IL), then sonicated and rinsed with water. The finely polished tooth surface was coated with carbon and imaged using the Cameca SX-100 Electron Microprobe Analyzer (Cameca, Gennevilliers Cedex, FR) at the University of Michigan Electron Microbeam Analysis Laboratory (EMAL) using the backscatter mode at a beam current of 15 kV and 10 nA.

TEM evaluation
Transmission electron microscopy (TEM) was used to determine ameloblast and enamel morphology and to investigate the secretion and localization of amelogenin. Briefly, one-week-old mandibles of wild type, Enam +/2 and Enam 2/2 mice were fixed in 4% PFA plus 1.0% GA in 0.1 M sodium cacodylate buffer, pH 7.3. Mandibles were left undecalcified for embedding in Epon epoxy resin (Cedarlane, Burlington, ON), or were decalcified for immunogold labeling in 8% EDTA over 2 weeks followed by embedding in LR White acrylic plastic (London Resin Company, Berkshire, UK). Samples destined for embedding in Epon for morphological analysis were additionally osmicated for 1 h in potassium ferrocyanide-reduced 1% osmium tetroxide. Prior to embedding, all samples were dehydrated in a graded ethanol series, infiltrated with the embedding media, placed into mounting molds and the blocks were polymerized at 55uC for 2 days. Thin sections (80 nm) were cut using a Leica Ultracut E ultramicrotome (Leica, Wetzlar, Germany) followed by staining with uranyl acetate and lead citrate after which the sections were viewed in a FEI Technai 12 transmission electron microscope (Hillsboro, OR) operating at 120 kV and equipped with a 792 Bioscan 1 k61 k wide-angle multiscan CCD camera (Pleasanton, CA). For immunogold labeling of amelogenin prior to TEM evaluation, LR White sections were incubated with polyclonal rabbit anti-porcine 25 kDa amelogenin (Dr. T. Uchida, Hiroshima University, Japan) followed by protein A-colloidal gold (14 nm) conjugate (Dr. G. Posthuma, University of Utrecht, Utrecht, The Netherlands) [17].
Generation of enamelin transgenic founders and breeding with Enam 2/2 mice Six PCR primers were designed to amplify target sequences and to introduce rare (8 base cutter) restriction sites (Fig. S1). The AmelX promoter (59AmelX, 4655 bp), the Enam cDNA (Enam, 3845 bp) and AmelX downstream (39AmelX, 1143 bp) sequence were amplified, subcloned into pCR2.1-TOPO (3931 bp) and sequence confirmed. Plasmids having the 59 ends of the PCR products on the NotI side of the vector were used to construct the Enam transgene. The plasmid containing AmelX promoter sequence (4.6 kb) and Enam cDNA sequence (3.8 kb) were restricted with AscI and NotI, ligated and sequenced. The AmelX promoter-Enam cDNA fusion to AmelX 39 sequence involved SrfI and SgfI restriction (Fig. S2). Following DNA sequencing confirmation (Fig. S3), the correct construct of 13.5 kb was restricted with NotI-SrfI to yield the 9.6 kb Enam transgene with Amelx regulatory sequences, which was microinjected into fertilized C57BL/6 X SJL F2 oocytes and surgically transferred to recipients at the Transgenic Animal Model Core at the University of Michigan. A total of 13 independent lines were generated and bred with C57BL/6 mice. Germline transmission was determined by PCR analyses of genomic DNA obtained from tail biopsies of the offspring. The transgenic lines designated as line 2, 3, 7, 11, and 12 were propagated and their offspring characterized. Offspring carrying the Enam transgene (Enam +/+,tg ) were mated with Enam 2/2 mice. Two breeding strategies were used. Enam +/+,tg mice were used to breed with C57/BL6 wild type mice to generate transgenic mice (Enam +/+,tg ) or wild type mice (Enam +/+ ); Enam +/ +,tg mice were used to breed with Enam 2/2 mice to generate F1 offspring. Enam transgene positive F1 (Enam +/2,tg ) were used to breed with Enam null mice again to generate F2 offspring. Four genotypes were generated: Enam 2/2,tg , Enam +/2,tg , Enam 2/2 and Enam +/2 . Genotyping was carried out using the three primer sets, Enam tg, Enam 4&5, and lacZ. (Fig. S4).

Assessment of Enam transgene expression levels by ELISA
Day 5 molars from the right mandibles and maxillae of mice with specific genotypes were extracted and incubated on a rotator with 1 mL of HF solution (0.17 N HCl + 0.95 N formic acid) that contained both the protease inhibitors (Cocktail Set-III-EDTA free, EMD Millipore Corporation, Billerica, MA) and phenanthroline at 1 mM concentrations for 3 h at 4uC. Samples were centrifuged (10000 rpm, 3 min) to remove residual insoluble material, and then the supernatant was neutralized with 6 N NaOH with a final pH ,5. Desalting and buffer exchange was carried out by using 0.01% formic acid and Amicon 4 mL ultracentrifuge concentrators (molecular weight cutoff = 3 kDa; EMD) according to manufacturer instructions. After desalting, sample volume was raised to 1 mL using 0.01% formic acid and frozen at 280uC for more than 1 h, then lyophilized overnight. Lyophilized samples were re-dissolved in 250 mL 0.1 M bicarbonate buffer (pH 9.6) containing protease inhibitors.
To determine the quantity of enamelin, samples were diluted to the desired concentration, 1/50 th of the total protein, in 50 mL of coating buffer (0.1 M sodium bicarbonate, pH 9.6) and coated onto the wells of 96-well microplates (Immulon 2 HB, Fisher Scientific, Pittsburgh, PA), incubated overnight at 4uC, washed three times with PBS containing 0.01% Tween 20 (PBS-T) and then blocked for 1 h at room temperature with 4% skimmed milk in PBS-T. The plates were then washed three times with PBS-T and 50 mL of mENAM 223-236 antibody (1:1,000 dilution in 4% skimmed milk in PBS-T) was added and incubated for 2 h at room temperature. The plates were washed three times with PBS-T, incubated with 50 mL (5000-fold diluted in 4% skimmed milk in PBS-T) of HRP-conjugated anti-rabbit IgG antibody from donkey (GE Healthcare, Pittsburgh, PA) for 1 h at room temperature. The wells were washed three times with PBS-T before 50 mL of 0.022% of 29, 29-azino-bis-3-ethylbenzthiazoline-6-sulfonic acid in citric acid with 0.05% hydrogen peroxide was added to each well for 15 min, and then the absorbance was measured at 405 nm on a Microplate Reader 680 (BioRad, Hercules, CA). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as internal control for normalization. Equal quantity of protein, 1/50 th of the total protein of two Day 5 teeth, in 50 mL coating buffer (0.1 M sodium bicarbonate, pH 9.6) was used in the ELISA assay. Rabbit polyclonal anti-GAPDH antibody with a 1:500 dilution in 4% skimmed milk was used as a primary antibody and HRPconjugated anti-rabbit IgG antibody at 1:5000 from donkey (GE Healthcare) was used as secondary antibody to detect the GAPDH.

Von Kossa staining
Day 4 mandibles of offspring from wild type, line 2 and line 3 Tg mice were fixed by immersion in 4% PFA/0.1% GA, dehydrated in increasing concentrations of ethanol, embedded in paraffin and sectioned at 5 mm. Sections were selected, deparaffinized and rehydrated. Von Kossa staining for mineral was performed by applying 1% silver nitrate in distilled water to the sections and exposing them to ultraviolet light wavelength 302 nm for 20 min. Unreacted silver was removed with 5% sodium thiosulfate for 5 min [18]. Sections were washed 3 times in distilled water, counterstained with methyl green for 3 min, rinsed in distilled water, dehydrated through graded ethanol, cleared in xylene and coverslips were mounted using Permount (Fisher Scientific, Fair Lawn, NJ) for microscopy and photography.

Enamelin expression determined by lacZ histostaining
The enamelin KO/lacZ KI construct replaced the Enam gene segment from the start codon in exon 3 to the 59 end of intron 7, keeping the entire 59 regulatory region inclusive of exon 1/intron 1, exon2/intron 2, and the 59 end of exon 3 intact [5]. Enamelin temporal and spatial expression pattern was characterized using bgal staining of embryonic day 13.5, newborn and postnatal day 7 (PN7) Enam 2/2 mice. LacZ positive staining was observed in developing incisors and molars from day 7 and on. Although greenish blue stain was initially observed in the intestine, when the intestinal contents were washed clean, no further positive staining occurred. Histologically, no positive staining for enamelin expression was detected in major organs of PN7 mice, which included lung, liver, kidney, stomach, small intestines, large intestine and cartilage of the long bone (Fig. 1). LacZ-positive staining in the histological sections of the developing teeth from PN 5 (Fig. 2), 7 (Fig. 3), and 14 ( Fig. 4) Enam +/2 and Enam 2/2 mice was specifically associated with ameloblasts while there was no positive staining among the comparable wild type negative control samples. During the secretory stage, normal enamel matrix and ameloblasts were seen in Enam +/2 molars (Fig. 2G-J), while abnormal accumulation of enamel matrix proteins and detachment of ameloblasts were apparent in all null mouse teeth ( Fig. 2K-N). Although some irregularity of the ameloblast cell layers, specifically homogeneity and continuity, existed among Enam +/2 molars at all time points, there was no apparent variation of enamel thickness of these teeth. The pathologic changes in the developing Enam 2/2 molars increased at PN7 during which time ameloblasts completely lost their polarity and aggregated abnormally ( Fig. 3G-I). The enamel space was irregularly thin except for a significant bulge was often observed on the mesial cuspal slope. At PN14, the Enam 2/2 molars failed to achieve the normal enamel thickness (Fig. 4G). The ameloblasts lacked a Tomes' process and tall columnar shape, but retained lacZ expression and deposited matrix and amorphous calcifications along the molar mesial and distal cusp slopes (Fig. 4H).

Enamelin expression determined by IHC
Despite the cellular abnormalities, amelogenin was secreted but enamelin was absent in all PN5 and PN14 Enam 2/2 samples (Fig. 5). In normal development, amelogenin and enamelin are deposited in the enamel space. Because of demineralization during sample processing, the enamelin matrix is not preserved. Therefore in this experiment, amelogenin and enamelin were detected only at the DEJ and ameloblast distal membrane. Enamelin expression was detected in both the wild type and the Enam +/2 molars (Fig. 5C, G, O, S). Abnormal ameloblast aggregations and cyst-like bulges containing amelogenin were observed in Enam 2/2 molars (Fig. 5I, J, U, V). Ameloblasts demonstrated premature and excessive apoptotic activities as early as PN5 in Enam 2/2 molars (Fig. 2K-N). In a previous report, we observed apoptosis in some ameloblasts in postsecretory transition along the mesial and distal cuspal slopes of Enam +/2 molars, but the levels were not different from those observed in the wild type molars [16].

Ultrastructural analyses
A dosage effect of enamelin absence was observed in 7-week-old incisors under SEM (Fig. 6). Organized rods were present in the Enam +/2 samples but these rods fractured differently and were spaced farther apart compared to wild type samples (Fig. 6B, C, F-G). There were no organized crystals in the enamel space of Enam 2/2 incisors and the tooth surfaces were rough and highly irregular (Fig. 6J, K). There were no appreciable differences in dentin structure among all samples (Fig. 6D, H, L). Using TEM, organized crystals of hexagonal shape were present in some areas of the Enam +/2 samples (Fig. 7A-B) while in different areas heterogeneous mineral phases were observed in the outermost part of the developing enamel layer (Fig. 7C-D). Amelogenin was produced by Enam 2/2 ameloblasts but it pooled in the enamel space as well as inside and between ameloblasts. Abnormal mineral mixed with amelogenins were present extracellularly ( Fig. 7E-H). Enamel and dentin hardness in Enam +/2 mandibular incisors was no different from wild type (data not shown). There was no true enamel in the Enam 2/2 samples, making hardness impossible to measure.

Enam transgenic models
The mouse amelogenin (Amelx) 59 and 39 regulatory regions and the Enam coding sequences were separately amplified from genomic DNA template using oligonucleotide primers that introduced rare 8 base restriction sites (Fig. S1). The amplification products were ligated into the pCR2.1TOPO cloning vector and subsequently excised and combined to produce the Enam transgenic construct that included 4655 bp of Amelx 59 promoter sequence extending into exon 2 but short of amelogenin translation imitation site, the complete Enam cDNA sequence including translation initiation and termination sites, and 1143 bp of 39 Amelx sequence that included three transcription termination        (Fig. S2). The construct was characterized by DNA sequencing, which verified that the sequence was unaltered by the cloning process (Fig. S3). The transgene construct was excised from the vector by NotI and SrfI digestion and used for blastocyst microinjection, which yielded 13 positive founder lines. The strategy used for breeding and genotyping the transgenic mice is demonstrated in Fig. S4. Three out of the 13 transgenic lines never produced offspring, 3 female founders were poor breeders and unable to produce sufficient transgene-positive offspring to keep the line viable, one founder line never produced positive pups and was terminated after 6 litters, and one founder line generated pups with almost undetectable transgene levels was also excluded. Multiple viable transgenic lines were evaluated for their levels of enamelin transgene expression in developing teeth using ELISA protein quantification methods. Lines 7 and line 11 expressed the transgenes at low levels, and were designated 7 (L) and 11 (L) . Lines 2 and 12 expressed their transgene at medium levels, and were designated 2 (M) and 12 (M) . These transgenes were still expressed at less than half of normal Enam expression levels. Line 3 expressed its transgene at a higher level than normal Enam expression and was designated 3 (H) (Fig. 8A). GAPDH levels were determined to be similar in all genotypes (Fig. 8B). Expression of Enam transgenes at medium or high levels in wild type mice altered both the appearance and surface texture of the enamel (Fig. S5). Transgenic overexpression of Enam produced whitish calcospherites on the mandibular incisor surface, which contributed to the chalky white appearance of those teeth (Fig. S5 J, O, T).

Enam transgene rescue
Five founder lines selected by their demonstration of germline transmission and Enam +/+,tg expression levels were crossed with Enam 2/2 mice to generate offspring with 4 different genotypes, Enam +/2 , Enam 2/2 , Enam +/2,tg , Enam 2/2,tg (Fig. S4). Developing teeth from offspring with the correct genotype were then compared. Blunting of the incisal edge and cuspal tips of molars suggested altered enamel hardness in Enam +/2 , Enam 2/2 , and Enam 2/2,tg mice (Fig. 9A). Comparing offspring from line 12 (M) , line 7 (L) and the wild type, the enamel phenotype was partially recovered when enamelin transgene was expressed in the Enam 2/ 2 background [Fig. 9B, lines 12 (M) and 7 (L) ]. When higher than normal levels of enamelin were expressed in the Enam 2/2 background, such as in the case of line 3 (H) , the enamel covering the molars and incisors appeared to be abnormal in thickness and lacked prism structure as revealed by backscatter and conventional SEM analyses (Fig. 9B, line 3). Close to normal enamel density, structure and thickness were observed only in Enam +/2,tg of line 12 (M) while in the same genetic background Enam +/2,tg of line 7 (L) the enamel prism and thickness was only partially restored Figure 8. Enamelin transgene expression as determined by ELISA. Graph A presents the enamelin expression as detected by mENAM 223-236 anti-peptide antibodies in wild type, Enam +/2 , Enam 2/2 , Enam +/2,tg , Enam 2/2,tg mouse molars from 5 different transgenic lines. Lines 7 and 11 are low expressers, lines 2 and 12 are medium expressers, and line 3 is high expresser. Graph B demonstrates GAPDH expression levels of test samples, which were assessed to be comparable among all test samples. doi:10.1371/journal.pone.0089303.g008 (Fig. 9B). Also in line 7 (L) , recovery was best near the DEJ where enamel crystals seemed to form normally followed by the usual decussated rods with surrounding interrod material although outer enamel was never rescued. Histologically, the most apparent finding was ameloblast pathology and cyst formation in the absence of enamelin, which was evident in both the maxillary and mandibular incisors of the Enam 2/2 and Enam 2/2,tg mice but only the mandibular incisors of the Enam +/2 mice (Fig. 10). Introducing enamelin at a low expression level did not alleviate the pathologic changes of ameloblasts (Fig. 10G-H).
Positive von Kossa reaction of the enamel and dentin of the wild type molars were evident (Fig. 11A, G), while only the dentin layer demonstrated positive reaction in the Enam 2/2 molars (Fig. 11B,  I). Molars from PN4 line 2 (M) and line 3 (H) Enam +/+,tg and Enam 2/ 2,tg were also tested. Although enamelin transgene expression in line 2 (M) and line 3 (H) did not rescue the enamel thickness or structures, von Kossa staining detected the presence of mineral in the presumed enamel space of the developing molars (Fig. 11C, E). Positive reaction in the enamel layer of the Enam +/+,tg molars from line 2 (M) (Fig. 11C) and line 3 (H) (Fig. 11E) revealed appreciable differences in staining intensity. Enamel mineralization as demonstrated by von Kossa positive reaction was present in Enam 2/ 2,tg samples from line 2 (M) (Fig. 11D, H) and line 3 (H) (Fig. 11F, H).

Discussion
Amelogenesis imperfecta is a diverse group of disorders affecting dental enamel formation; when clinical phenotypes and mode of inheritance are considered, 15 subtypes can be distinguished [19]. Single allele mutations of ENAM result in hypoplastic enamel where the affected enamel is thin and rough and the teeth are yellow and sensitive to thermal changes. Mutations in both alleles often lead to complete absence of enamel. Patients with such a condition require extensive and frequent dental care, which causes a significant psychological and financial burden. To improve diagnosis, treatment and long-term prognosis of patients with AI, understanding the genetic control of amelogenesis is paramount. The enamelin knockout and transgenic mouse models described here make possible the study of human hypoplastic AI arising from enamelin gene defects.
In this study, we conclude that enamelin is essential for the formation of dental enamel with its distinctive crystal and the prism structures. Characterization of the enamelin gene-targeted mouse models reveals the essential function of enamelin in maintaining ameloblast integrity and directing the assembly of enamel ultrastructure. Based upon the analysis of the knockout mouse model, we determined the temporal and spatial expression of enamelin, characterized the structural and cellular defect of amelogenesis, which provided the pathologic basis of enamel hypoplasia. Furthermore, the study of selected enamelin transgenic mouse lines, determination of transgene expression by protein quantification methods, and introduction of enamelin transgenes into the knockout background to recover enamel defects demonstrated a dosage effect of enamelin.
Considering that regulatory elements may reside in intronic regions, introns 1 and 2 were preserved in the design of Enam KO/ Figure 9. Comparison of incisor enamel in wild type mice with incisor enamel from littermates of a cross between Enam +/2,tg and Enam 2/2 mice. (A) The Enam +/2 mice without the Enam transgene have chalky enamel that readily chips from dentin, although it has slightly decreased thickness and seemingly normal rod organization except for the outer enamel layer. The Enam 2/2 mouse has a mineral crust covering dentin that is very thin, and shows none of the characteristic of enamel. (B, BEI and fractured surface SEM of line 12). The Enam +/2,tg from line12 (M) had normal looking enamel that was similar to the wild type in thickness, prism pattern, and resistance to wear but in line 12 ( lacZ KI construct. Enamelin expression as demonstrated by bgalactosidase staining is specific to the developing tooth organ, restricted to the secretory and early maturation stage ameloblasts. Expression of enamelin in odontoblasts or osteoblasts was not observed. In the developing molars from the Enam 2/2 mouse, the ameloblast layer becomes increasingly crowded as tooth formation progresses. By PN5 small groups of ameloblasts lose contact with the matrix covering dentin and form nodules within the stratum intermedium. Ameloblasts continue to synthesize and secrete amelogenin, but show atypical features including expanded endoplasmic reticulum. An extracellular layer of organic material abundant in amelogenins is deposited on the surface of normal appearing dentin although no mineralization takes place. Mineral formation at the ameloblast-matrix interface (mineralization front) is absent and no enamel crystal ribbons can be observed at the early secretory stage (PN5). At the secretory stage, the dentinoenamel junction (DEJ) appears to be highly irregular as shown in Fig. 7F. Amorphous, granular deposits of pathological calcification are observed atop the DEJ and between ameloblasts. In the Enam +/2 mice, the enamel crystals have the classic hexagonal appearance but are less densely packed than in wild type. Ameloblasts in the Enam 2/2 mice have poorly formed Tomes' processes and undergo pathological alterations that lead to ectopic calcifications and the formation of a thin, highly irregular mineral crust instead of true enamel.
There exist appreciable differences among the defective enamel found in amelogenin, ameloblastin and enamelin knockout mice. Although the absence of amelogenin [20] and ameloblastin [4] also lead to hypoplastic enamel defects, the degree of hypoplasia is most severe in enamelin knockout mice, which produced no true enamel. In case of amelogenin absence, ameloblasts lacking Tomes' processes are able to make a thin layer of enamel that is 10-20% the thickness typical of wild type enamel. When mutant ameloblastin Ambn D5-6 is present, the ameloblasts detach from the underlying matrix at the start of the secretory stage and generate a thin enamel with irregular prism pattern [21,22]. When enamelin is absent, ameloblasts soon become pathological, unable to adhere to underlining surface, and they prematurely undergo apoptosis [16]. Cystic changes often occur at what is normally the start of the secretory stage, which was most apparent on the mesial cuspal slope of the first molar and the enamel organ of the mandibular incisor of the PN7 Enam 2/2 mice. In the Enam 2/2 mouse incisors, the matrix-like materials pooled irregularly and primarily within the enamel organ resulting in a very thin enamel space. In the Enam +/2 mouse mandibular incisors, the cysts appear late in the secretory stage and carry forward into the maturation stage, which causes additional problems to getting full maturation of the enamel. It ends up chalky white and soft on the erupted portion of the tooth. These observations suggest that cell-matrix interactions during enamel formation is a plausible regulatory mechanism of ameloblast activities and the major enamel matrix proteins, amelogenin, ameloblastin and enamelin are likely conducting different functions in a synchronized fashion to produce well organized and fully mineralized dental enamel.
Introducing appropriate transgenes capable of expressing proteins at the quantity comparable to endogenous amounts in a null background to rescue dental defects has been demonstrated by Kulkarni et al., Chun et al., and Gibson et al. In the case of DSPP absence, re-introducing DSP only improves the dentin volume but not the mineral density [23]. In the case of Ambn D5-6 partial deletion mutant, expression of its transgene at somewhat higher than the normal levels almost fully rescues the enamel defects in the mutant background [24]. Most interestingly, mouse amelogenin180-87 cleavage products rescues enamel mineral density and increases thickness while introducing both amelo-genin180-87 and leucine-rich amelogenin peptide, which contains residue 1-33 plus 13 residues from the most C-terminus of the amelogenin180, restores enamel prism structure and further improves enamel thickness in the Amelx knockout mice. This study demonstrated functional importance of different amelogenin cleavage products in enamel formation [20]. In our study, complete rescue of the enamel defect in the enamelin null background was not achieved by introducing three different levels of transgene expression from the amelogenin promoter whereas typical enamel thickness and prism structures were observed when near normal levels of the transgene were introduced into the heterozygous Enam +/2 background where partial defects were present. The collective results suggest there exists a very tight control of enamelin dosage in order to achieve proper enamel formation. This is further demonstrated by the mineralization defects seen in wild type mice starting with the lowest expression levels of ''extra'' enamelin (Fig. S5). Matrix protein stained blue in the enamel space during secretory stage was absorbed in the maturation stage allowing crystal maturation to take place (arrowheads). (C) Maxillary incisors in Enam +/2 mice appear similar to wild type. (D) However, the mandibular incisors in these mice consistently demonstrated disturbed ameloblasts resulting in cyst formation within the enamel organ beginning in late secretory stage (arrow). (E) In Enam 2/2 maxillary incisors, there was no apparent enamel matrix deposition; (F) in the mandibular incisors, soon after the onset of the secretory stage, ameloblasts showed pooling of secreted material at their apices and gradual loss of polarity and organization within the enamel organ (arrowheads). (G, H) In line 2 (M) , Enam 2/2,tg incisors, ameloblasts were polarized and deposited matrix during early secretory stage. Soon after that ameloblast layer became detached and cystic formation was evident on both maxillary and mandibular incisors (arrows). These 0.5 mm thick toluidine blue stained sections of EDTAdecalcified incisors were obtained from mice perfused with 2.5% GA and embedded in Epon. Bar = 500 mm for all panels. doi:10.1371/journal.pone.0089303.g010 The most intriguing finding of this study is that in the Enam 2/2 mice, there is absence of proper enamel matrix accumulation and organized enamel crystals growth resulting in no true enamel formation. The mineralized crust later formed on the dentin surface of the Enam 2/2 mice is likely a product of pathologic calcification. Re-introduction of enamelin allowed initial mineralization to take place at the DEJ during early secretory stage, although the specific dosage of the introduced enamelin dictates how well the enamel ultrastructure is recovered. Such evidence and the observation of restricted localization pattern of C-terminus enamelin have led us to hypothesize that enamelin is a critical matrix molecule in promoting crystal initiation and elongation during a specific stage of amelogenesis. Furthermore, ameloblast integrity is jeopardized in the absence of enamelin, which suggests the potential involvement of enamelin in maintaining a functional cell-to-matrix interface. Thus, in the absence of enamelin, ameloblast Tomes' process, mineralization front and true enamel formation cannot be attained. Dental enamel forms by the deposition of characteristic thin, non-crystalline, mineral ribbons along a mineralization front that are physically closely associated with the secretory surfaces of the ameloblast plasma membrane. The mineralization front apparatus is the key to enamel formation [25]. Following the formation of initial enamel, enamel matrix components such as amelogenin, enamelin and ameloblastin may self-assemble into nanostructures allowing crystal elongation and organization to take place repetitively with precision to form rods and interrod enamel which subsequently arranged into an intricate decussating pattern [26]. To advance our understanding of the basic mechanism of biomineralization, it will be important to visualize the mineralization front, to discover the pattern of the matrix protein assembly, and to understand their functional relationship with the calcium phosphate phases deposited within the extracellular enamel space.  initiates in the 59AmelX region at the start of the exon 1, which is non-coding. Intron 1 (1277 bp) of AmelX is removed by RNA splicing. The AscI site connects 59AmelX, including 10 nucleotides in exon 2, to the Enam cDNA sequence (3845 bp). The splice junction at the start of exon 2 is indicated by hash marks in the expanded sequence surrounding the AscI site. The Enam cDNA sequence is immediately downstream of the AscI site and is in lower case and boxed. The AmelX and Enam translation initiation codons are underlined.