Identification and characterization of amelogenin genes in monotremes, reptiles, and amphibians

Abstract

Two features make the tooth an excellent model in the study of evolutionary innovations: the relative simplicity of its structure and the fact that the major tooth-forming genes have been identified in eutherian mammals. To understand the nature of the innovation at the molecular level, it is necessary to identify the homologs of tooth-forming genes in other vertebrates. As a first step toward this goal, homologs of the eutherian amelogenin gene have been cloned and characterized in selected species of monotremes (platypus and echidna), reptiles (caiman), and amphibians (African clawed toad). Comparisons of the homologs reveal that the amelogenin gene evolves quickly in the repeat region, in which numerous insertions and deletions have obliterated any similarity among the genes, and slowly in other regions. The gene organization, the distribution of hydrophobic and hydrophilic segments in the encoded protein, and several other features have been conserved throughout the evolution of the tetrapod amelogenin gene. Clones corresponding to one locus only were found in caiman, whereas the clawed toad possesses at least two amelogenin-encoding loci.

One of the major innovations accompanying the emergence of jawed vertebrates was the development of teeth, presumably from dermal scales (1). The concurrent development of jaws and teeth generated a new type of feeding device—biting structures that enabled gnathostomes to colonize new environmental niches and thus allow their adaptive radiation. To understand how this innovation occurred, it is necessary to delineate the evolutionary history of the genes involved in tooth development, in particular those responsible for the formation of the two characteristic tooth components, the enamel and the dentin. Tooth enamel is one of the most highly mineralized tissues known (2). It is formed in the early stages of odontogenesis by ameloblasts, which synthesize and secrete several matrix proteins. Later, during the stage of enamel maturation, the production of matrix proteins wanes, and the proteins already produced are gradually replaced by hydroxyapatite crystals. The main matrix protein, amelogenin, is thought to be involved in the regulation of enamel crystallite formation, presumably by providing the hydrophobic environment necessary for the initiation and growth of calcium hydroxyapatite crystals (2).

Amelogenin-encoding cDNA clones have been isolated from several representatives of mammals, including humans (3), cattle (4), pig (5), rat (6), mouse (7), and opossum (8). The presence of amelogenin-encoding genes in wallaby (9) and platypus (9) has been indicated by Southern blot hybridization, but no sequence has been presented to date. In the bovids and anthropoid primates, two amelogenin genes have been identified, one on the X and the other on the Y chromosome (10–15); in rodents, only the X linked gene has been found (14, 15). Immunohistochemical analysis has indicated the possible existence of amelogenin-like compounds in reptiles and amphibians (16, 17). The nature of an enamel-like material, sometimes referred to as enameloid, remains unresolved and controversial. As part of a systematic effort to elucidate the emergence of teeth in evolution, we have initiated a study aimed at cloning the major tooth-forming genes of species representing gnathostome classes. Here we describe amelogenin-encoding cDNA clones isolated from reptiles and amphibians as well as single-exon PCR products from monotremes.

MATERIALS AND METHODS

Source and Isolation of DNA.

DNA samples from platypus (Ornithorhynchus anatinus) and the short-nosed echidna (Tachyglossus aculeatus) were provided by Robert W. Slade (Queensland Institute for Medical Research, Royal Brisbane Hospital, Australia). Tissue samples from 3-day-old smooth-fronted caimans (Paleosuchus palpebrosus) and adult African clawed toads (Xenopus laevis) were kept frozen at −70°C until their use. Genomic DNA was isolated from the tissues by phenol-chloroform extraction (18).

cDNA Library Construction and Screening.

The caimans and the African clawed toads were killed under anesthesia, and their jaws were removed and immediately frozen in liquid nitrogen. The frozen tissues were homogenized to a fine powder, and total RNA was extracted (19). Poly(A)⁺ RNA isolation and cDNA synthesis were performed with the mRNA purification kit (Pharmacia) and the TimeSaver cDNA synthesis kit (Pharmacia), respectively. The cDNA was inserted into the EcoRI-digested λ gt10 vector (Stratagene), and the cDNA library was in vitro-packaged with the help of the Gigapack cloning kit (Stratagene) and used to transform competent Escherichia coli NM514 bacteria. The initial titers of the libraries were 1.2 × 10⁶ plaque-forming units (pfu) for the caiman, and 6 × 10⁵ pfu for the African clawed toad. The caiman and African clawed toad libraries were amplified once to a titer of 1.5 × 10¹¹ pfu/ml and 1.8 × 10¹¹ pfu/ml, respectively.

PCR Amplification.

The amelogenin genes of the monotremes (platypus, echidna) and the caiman were amplified by using primers based on a comparison of human, cattle, pig, rat, mouse, and marsupial sequences. In the case of the monotremes, primers AM1 (sense; 5′-TATGGTTACGAA-CCCATGGGTGGATGG-3′) and AM2 (antisense; 5′-ATCCACTTCTTCCCGCTTGGTCTTGTC-3′) were used to amplify 420-bp fragments of the amelogenin exon 6 sequence. In the caiman, primers AM6 (sense; 5′-GAA-CCCATGGGTGGATGGCTGCACCA-3′) and AM2 were used for the original amplification, and primers AM6 and Tu1360 (antisense; 5′-GGCAGCAGTGGGGGCAGAGGCTG-3′) were then used for half-nested PCR to amplify a 370-bp fragment of amelogenin exon 6 sequence from genomic DNA. The primers AM8 (sense; 5′-CAGACTCTCACACCTCACCACCA-3′) and AM7 (antisense; 5′-TTGTTGCTGTGGTATAGGCATCAT-3′), designed within the 370-bp product amplified by the primers above, were used to recover inserts by anchored PCR. In the case of the African clawed toad, primer AM10 (sense; 5′-CCTGGTTATGTCAACTTCAGTTATGA-3′), based on the caiman amelogenin sequence, was used to amplify a fragment of about 600 bp by anchored PCR. The primer AM22 (antisense; 5′-CATCATAGATTGGTA-CCATTT-3′), designed within the fragment of the toad amelogenin product amplified by the above primers, was used to recover the 5′ region of the inserts by anchored PCR. To determine the organization of the caiman and toad amelogenin genes, primers were designed to amplify regions of the individual exon–intron boundaries (Table 1). The primer AM57 (antisense; 5′-ATTCTGGCTCTCGTGGTCAGGTTT-3′) was used to confirm the identities of the second toad amelogenin clone. Genomic DNA (100 ng/μl) or lysate of the cDNA libraries (1 μl) were amplified by PCR in 50 μl of PCR buffer (1.5 mM MgCl₂/200 μM dNTP/10 mM Tris, pH 8.5) in the presence of the sense and antisense primers and 2.5 units of Taq polymerase (Pharmacia). Amplifications were performed in the RoboCycler Gradient 96 (Stratagene) in 35 cycles, each cycle consisting of 1 min of denaturation at 94°C, 1 min of annealing at the annealing temperature, and 3 min of extension at 72°C. The final extension was for 10 min at 72°C.

Table 1.

Oligonucleotide primers used in the determination of exon-intron boundaries

Primer designation	Sequence	Specificity exon, codon	Orientation
Caiman
AM-19	TAGAGAATTTAGCTGGAGTACTTC	E1, 5′UTR	S
AM-20	AGTGATCAACATCCAGCCCTCCAT	E2, 1–8	A
AM-14	ATCACTTGCCTACTAGGTGCA	E2, 7–13	S
AM-16	TCATAACTGAAGTTGACATAACC	E3, 27–34	A
AM-9	CATCATCCTGGTTATGTCAACTT	E3, 24–31	S
AM-15	TTGTCTCATCAGGCTCTGGTACCA	E5, 41–48	A
AM-12	TTAACACCTTTGAAATGGTACCA	E5, 36–43	S
AM-13	TAACATTGGCTGGTGTAGCCATCC	E6, 60–67	A
AM-17	TGGCGGCCAATGGACAAGACCAA	E6, 209–216	S
AM-18	ACTCCAGTGGAATGATGGATTCTTGA	E7, 3′UTR	A
African clawed toad (toad 1)
AM-24	CTAATGCTAACAGCTCTCATT	E2, 5–11	S
AM-25	CTCATAACTGAAGTTCACATACCC	E3, 27–34	A
AM-26	CAGCATCCTGGGTATGTGAACTT	E3, 24–31	S
AM-27	CTGATGTGTCATCATAGATTGGTA	E4, 42–49	A
AM-28	TTATCACCTTTGAAATGGTACCAA	E5, 36–43	S
AM-29	AATTGGGTTCTGAAGCCAGCCAGA	E6, 59–66	A
AM-40	TATCCAAATTACGGCTATGAACCT	E6, 50–57	S
AM-39	AAAGTACAGTAAACAATATTCTG	E7, 3′UTR	A

A, antisense; S, sense; E, exon; UTR, untranslated region. 

Cloning, Sequencing, and Blotting.

One microgram of a PCR product was isolated from an agarose gel (Gibco/BRL) by using the Qiagen (Hilden, Germany) extraction kit. The isolated DNA was ligated to SmaI-digested pUC18 plasmid vector with the SureClone ligation kit (Pharmacia) and used to transform competent E. coli XL-1 blue bacteria. Transformants were grown overnight in Luria–Bertani (LB)-ampicillin broth, and minipreps were prepared according to the standard protocol (20). Two to five micrograms of DNA were used in the dideoxy sequencing reactions with the AutoRead sequencing kit (Pharmacia). The reactions were then processed by the Automated Laser Fluorescent (ALF) sequencer (Pharmacia). To determine the exon-intron organization of the caiman and toad amelogenin genes, the genomic sequence data were compared with the cDNA sequences and exon–intron boundaries were identified by comparison with the published consensus sequences (21).

For Southern blots, 10 μg of genomic DNA was digested with the appropriate restriction enzyme, and the resulting fragments were separated in 0.8% agarose gels (Gibco/BRL) and transferred to a hybridization membrane (Hybond-N⁺, Amersham) by using the VacuGene blotting system (Pharmacia). The filters were incubated for 5 h in a prehybridization solution containing 1× Denhardt’s solution (0.02% polyvinylpyrrolidone/0.02% Ficoll/0.02% BSA)/5× standard saline citrate (SSC, 1× SSC = 0.15 M sodium chloride/0.015 M sodium citrate, pH 7)/0.2% SDS/5 × 10⁶ cpm of the probe, which was labeled by using the random-priming method with ³²P to a specific activity of 1 × 10⁹ cpm/μg with the Ready-To-Go DNA labeling kit (Pharmacia). Filters were washed in 2× SSC/0.1% SDS for 20 min at room temperature and then used to expose XAR5 film (Kodak) with intensifying screens for 3–5 days.

Data Analysis.

The nucleotide sequences and inferred protein sequences were aligned with the aid of the gcg package (Genetic Computer Group, Madison, WI) and the seqpup (ref. 22; available at http://iubio.bio.indiana.edu/soft/molbio) computer program. The evolutionary relationships were then evaluated by the neighbor-joining algorithm (23).

RESULTS AND DISCUSSION

PCR amplification of platypus and echidna genomic DNA using the primer pair AM1/AM2 yielded in both instances a 420-bp fragment that cloning and sequencing revealed to be homologous to part of exon 6 of the amelogenin gene (Figs. 1 and 2). Half-nested PCR with the primer pairs AM6/AM2 and AM6/Tu1360 and caiman genomic DNA yielded a 370-bp amplification product; the product’s sequence, homologous to amelogenin exon 6 sequence, was then used to design primers AM7 and AM8 for the amplification of a cDNA clone by anchored PCR. Four clones could be amplified from the cDNA library. One, amplified with the AM8 primer, was 600 bp long and contained the 3′ untranslated region (UTR). Another, amplified with the AM7 primer, was 420 bp long and encompassed a portion of the 5′UTR 48 bp upstream from the initiation codon to the coding sequence to codon 110. The third clone, amplified by the same primer as the second clone, was 390 bp long and encompassed roughly the same cDNA segment as the second clone (it started 63 bp upstream from the initiation codon); the shorter length of the third clone resulted from the deletion of exon 3, presumably by alternative splicing. Several isoforms generated by alternative splicing have been described for amelogenin transcripts in human, cattle, pig, rat, and mouse (5, 7, 11, 12, 24). Apart from the difference in exon 3, the two clones were identical in their sequence and were probably derived from the same gene. The fourth clone, amplified by the use of primer AM18 designed on the basis of the 3′UTR sequence of the first clone, was 790 bp long and extended from the 5′UTR to the 3′UTR. The coding sequence of this clone consisted of 597 bp encoding a polypeptide chain of 199 amino acid residues (Figs. 1 and 2). Exon–intron boundaries of the caiman amelogenin gene were determined by sequencing the relevant regions of genomic DNA. They corresponded to the borders reported for the mammalian amelogenin gene (10, 12) except for the boundary between exon 1 and intron 1 (part of the 5′UTR), which in the caiman gene is 7 bp upstream of the mammalian splice site. The putative AATAAA polyadenylation signal of the caiman gene is located 13 bp upstream from the poly(A) tail.

Figure 1.

Structural organization of the human, caiman, and African clawed toad amelogenin genes. Exons, indicated by boxes, are numbered 1–7, and the numbers below exons indicate the length in base pairs. The variable length of human X and Y exons is indicated by two numbers separated by a slash. Exons known to be alternatively spliced are indicated by hatched boxes. The first exon–intron boundary in the toad amelogenin genes has not been identified.

Figure 2.

Amino acid alignment of amelogenin sequences. A simple-majority consensus sequence is shown at the top. Identity with the consensus is indicated by a dash (-); an asterisk (∗) indicates an alignment gap, and a dot (⋅) indicates an unknown sequence. The common names of the species are shown together with accession codes for sequences obtained from databases.

The caiman-based primer AM10 was also used in anchored PCR to amplify two clones from a toad cDNA library. The clones were of similar length, 650 bp and 680 bp, respectively, and encompassed the same region, from codon 35 to the 3′UTR. The 5′ ends of the two transcripts were obtained by anchored PCR with the primer AM22 positioned to give a 16-bp overlap with each of the two clones. The two clones differed by numerous substitutions, as well as by some insertions and deletions (Fig. 2). To establish their identities, primer AM39, based on the sequence of the 3′UTR, was used in anchored PCR to amplify a full-length coding sequence of the toad-1 clone, and the primer AM57 was used to amplify almost the entire coding sequence of the toad-2 clone. The differences between the two clones could thus be verified, and hence it could be established that the clones were derived from different genes. The full-length sequence of the toad-1 contained 519 bp of coding sequence specifying a polypeptide 173 residues long, whereas the full-length sequence of the toad-2 comprised 552 bp of coding sequence translatable into a polypeptide chain 184 residues long (Fig. 2). Exon–intron boundaries of the toad-1 amelogenin gene were determined by the same method as was the caiman gene. The boundaries were the same as in the mammalian amelogenin gene, with the intron length ranging from 1.2–1.5 kb (Table 2), and the AATAAA polyadenylation signals of toad-1 and -2 genes were located 19 bp and 20 bp upstream of the poly(A) tail, respectively.

Table 2.

Exon-intron organization of caiman and African clawed toad amelogenin genes

Gene	Size		Sequence of exon-intron junctions
	Exon, bases	Intron, kb	5′ Boundary	Intron	3′ Boundary
Caiman
1		1.3	TTC TAC AG	gtaaacct… . .tttttcag	G TAC TAT
2	73	0.6	GCT ATA CCA	gtgagtat… . .tttaacag	TTG CCT CCC
3	48	1.5	AGT TAT GAG	gtaaaaca… . .ccctgaag	GTG TTA ACA
5	45	0.8	AGA CAA CCG	gtaaacat… . .ccctgtag	TAT TCA TCC
6	447	1.3	GAG GAA ATA	gtaagaag… . .tctttcag	GAT TAA AGA
7	>200
African clawed toad (toad 1)
2		1.2	TCT GTT CCT	gtaagtat… . .atttgcag	CTT CCG CCT
3	48	1.5	AGT TAT GAG	gtatgtca… . .tatccaag	ATA TTA TCA
5	45	1.5	ACA CAT CAG	gtaagaat… . .tttcccag	TAT CCA AAT
6	369	1.3	GAG GAA CTG	gtaaatat… . .ttttatag	GAT TAG AAG
7	>160

To test whether the two toad amelogenin genes are alleles or whether they are derived from two loci, we performed Southern blot analyses. We digested genomic toad DNA with the HindIII, EcoRI, and TaqI restriction endonucleases, blotted the digests, and hybridized the blots with a ³²P-labeled, nearly full-length cDNA probe. In all three blots, two distinct hybridizing bands were clearly recognizable, their sizes being 4.5 and 1.5 kb, 2.4 and 1.8 kb, and 3.2 and 2.0 kb for the HindIII, EcoRI, and TaqI digests, respectively (data not shown). This result is consistent with the presence of two amelogenin loci in the toad genome.

In eutherian mammals, the amelogenin molecule has been described as consisting of three regions (26): the N-terminal tyrosine-rich amelogenin protein (TRAP) sequence of some 44–45 residues; the hydrophobic core sequence of some 100–130 residues; and the acidic hydrophilic C-terminal sequence of some 15 residues. The amino acid alignments of the monotreme, reptilian, and amphibian amelogenin sequences with the sequences of the eutherian mammals (Fig. 2), as well as the hydrophilicity plots of all of these molecules (Fig. 3) indicate that this division is a general feature of the amelogenin molecule. The six tyrosine residues of the TRAP region are conserved in all of the amelogenins for which sequence information is available. This region also contains two other conserved features, the N-linked glycosylation site at residue 30 and the serine phosphorylation site at position 32 (Fig. 2). Whether amelogenins are glycosylated remains controversial (27). The conservation of the one glycosylation site suggests, however, that it may be functional. Like the eutherian mammals, the reptiles and amphibians possess a hydrophobic core region rich in Pro and Gln residues in their amelogenin molecules (caiman 27% Pro, 14% Gln; toad-1 product 22% Pro, 11% Gln; and toad-2 product 20% Pro, 13% Gln, as compared with 23–28% Pro and 13–17% Gln in the amelogenin of eutherian mammals). The application of the 3D-1D compatibility algorithm (28) to this region has revealed the presence of a conserved α-helix extending between positions 39 and 48 (Fig. 2). The C-terminal portion of the core region contains multiple tripeptide repeat sequences (Pro-X-Gln)_n that are apparently responsible for the variability observed among the different proteins. The hydrophilicity of the C-terminal region of the amelogenin molecule has been conserved during the entire period of tetrapod evolution (Fig. 3).

Figure 3.

Hydrophilicity plot of amelogenins from selected species prepared by using the method of Kyte and Doolittle (25). The plots share the following characteristics: the hydrophobic leader peptide of about 20 amino acid residues is followed by a short hydrophilic segment (10 residues), another short hydrophobic segment (10 residues), and a larger hydrophilic segment (about 20 residues). The conserved α-helical segment lies on the border of the hydrophobic and hydrophilic domains (approximately corresponding to residues 35–45). This shared hydrophobicity signature is followed by a more variable internal segment (residues 60–110) of irregularly alternating hydrophilicity and hydrophobicity. The repeat region (residues 110–190) and the C-terminal region are hydrophilic. Although the primary sequence of amelogenin varies greatly between species, the hydropathy pattern is well conserved.

In some eutherians (human, cattle) the amelogenin genes exist as copies on both the X and Y chromosomes. The X and Y linked loci do not evolve in concert, and several differences, including substitutions and indels, have arisen between the copies. It has been suggested that the human X and Y linked genes have been independently evolving for 45 million years (Myr) (14, 29). It is also probable that the Y linked gene, although containing no inactivating mutation, is not under strong functional constraint because it is normally masked by the X linked gene and cannot provide protection when the X linked gene is inactive, as in X linked amelogenesis imperfecta (30). The eutherian Y linked copy may therefore be tending toward pseudogene status. Even if the amelogenin gene is autosomal, as in marsupials and monotremes, there is some evidence for the presence of multiple amelogenin loci (9). The present work indicates that at least two amelogenin loci exist in the toad, a known polyploid (31). On the other hand, there is no evidence for multiple expressed loci in the caiman.

In eutherian mammals, the average nonsynonymous evolutionary rate of the amelogenin genes (0.5 × 10⁻⁹ per nonsynonymous site per year) is less than the average found for 40 genes (0.9 × 10⁻⁹ per site per year) by Li et al. (32). Slow evolution of the gene in eutherians is also indicated by the identity of inferred mouse and rat amelogenin amino acid sequences; the species might have diverged as long as 40 Myr ago (33). The synonymous substitution rate of the amelogenin gene is also low. In human-rodent comparisons the number of synonymous substitutions per site averages 23.1%, which is lower than similar comparisons for 28 human-rodent gene pairs made by O’hUigin and Li (34).

Although the overall evolutionary rate for amelogenin is slow, some regions of the polypeptide appear to evolve rapidly and undergo insertions or deletions of codons readily. This is apparent in the proline/glutamine repeat region (residues 120–200) of exon 6. The cattle X linked, as well as the opossum and caiman sequences, are up to 28 aa longer in this region than other sequences. Because the repeat region lies within an exon, these indels are not a result of alternative splicing. The lack of constraint in the repeat region is seen in the absence of similarity between mammalian, caiman, and toad sequences at positions 70–190 of the alignment (Fig. 2). The variability of the repeat region even between closely related genes suggests that there is a strong propensity for indels to occur frequently, perhaps by replication slippage between the repeat motifs.

To determine whether functional constraints limit the evolutionary rates in any particular part of the gene, the dependence of nonsynonymous substitution rates on their position within the gene was examined. A sliding window of 30 codons was used within which nonsynonymous substitution rates were measured for several gene pairs. In this way, tendencies apparent in particular pairwise comparisons can be verified by checking for the same tendency in other comparisons. Because of their uncertain functional status, the Y linked genes were omitted from the study. The analysis shows that the region from residues 30–40 (Fig. 4) and involving the C-terminal part (residues 200–220) tends toward conservation in pairwise comparisons involving caiman, toad, human, rat, and cattle. Analysis of peptide secondary structure predicts that α-helices form in the first conserved region (residues 35–45) in all eutherian as well as in reptilian and amphibian amelogenins (not shown), and that this structure is subject to strong functional constraint. The repeat region and its flanks (residues 60–200) is most divergent in all comparisons. By contrast, comparisons of human or cattle X and Y linked genes show that not all of these regions are well conserved, presumably because of reduced functional constraints in the Y linked genes.

Figure 4.

Plot of nonsynonymous substitution percent (K_a%) against amino acid residue position (numbering follows that of Fig. 2) in selected pairwise comparisons of human (H), cattle (Ct), rat (R), caiman (C), and toad-2 (T) amelogenin sequences. An initial sliding window of 30 codons was used to estimate K_a by the method of Li et al. (32), and a further sliding window of 30 codons was used to smooth the values obtained. Small alignment gaps (<8 residues) were ignored in the plot; larger gaps are indicated by vertical lines.

A phylogenetic tree based on protein sequences of all available amelogenin sequences (Fig. 5) shows that the sequences group as expected, giving distinct clades for eutherian and noneutherian mammals, as well as a branch containing the caiman sequence only and a clade of the two toad sequences. We interpret the tree as indicating that the multiple copies of amelogenin arose in different ways in several species. The sequences may represent polymorphisms in the rat, sex-linked divergences in humans and cattle, and polyploidization in the toad. In species possessing the Y linked sequence (human and cattle), the branch length joining it to the X linked sequence is long, indicating that more replacements have occurred in the Y linked lineage than in the X linked lineage. This observation supports the conjecture that the Y linked genes are less functionally constrained. Similarly, the toad-2 sequence has a longer terminal-branch length than the toad-1 sequence. Again, this may indicate a difference in functional constraint, although the long divergences from other sequences and the presence of many indels in the variable region may obscure the relationships between the toad sequences.

Figure 5.

Neighbor-joining tree of sequences shown in Fig. 2. Distances were estimated from the proportion of differences in pairwise comparisons following exclusion of gaps. The common names of the species are indicated together with accession codes for sequences obtained from the databases. Numbers on the nodes indicate the percent recovery of that node in 500 bootstrap replications.

The availability of amelogenin sequences in representatives of reptiles and amphibians should facilitate the search for homologous genes in fishes on the one hand, and birds on the other. The former were apparently the first vertebrates to develop teeth (35), and the latter lost the ability to form teeth more than 120 Myr ago. [Although the family of neotropical wood-quails is referred to as Odontophorinae, i.e., “tooth-bearing,” these birds, too, are in fact toothless. The designation refers to the serrated or “toothed” appearance of the lower jaw’s cutting edge, which, like the rest of their stout bill, has ontogenetically nothing in common with true teeth (36). Similarly, the “egg tooth,” a small sharp projection on the upper jaw of embryos in many bird species used in opening the egg shell during hatching, is histologically and embryologically unrelated to true teeth (37)]. Turtles, alone among the reptiles, have also been toothless for more than 100 Myr. It will be interesting to find out whether amelogenin and other genes involved in tooth formation have been retained by birds and turtles and used in functions other than tooth formation.

ABBREVIATIONS

Myr

million years

pfu

plaque-forming units

UTR

untranslated region