Genes Involved in Dentin Formation

COL1A1 and COL1A2

Type I collagen, which comprises the bulk of the organic phase of dentin, is a triple helix containing two proa1(I) and one proa2(I) chain. The two chains are encoded by the COL1A1 and COL1A2 loci, at 17q21.3-q22 and 7q22.1, respectively. Assembly of the collagen triple helix begins at the C-terminus and proceeds towards the N-terminus. Thus mutations at the C-terminus are typically associated with a more severe phenotype than those nearer the N-terminus [Steiner et al., 2005]. OI type I usually results from mutations in COL1A1 that result in haplosufficiency of the proa1(I) chain. Haploinsufficien-cy leads to half of the amount of type I collagen. No abnormal collagen is produced. These individuals do not have DI. Individuals with OI type I with DI (OI type IB in the classification of Sillence et al. [1979]) have missense mutations in COL1A1. A missense mutation is a nucleo-tide substitution in the coding region of a gene that changes a nucleotide codon so that it codes for a different amino acid in the protein, resulting in an abnormal collagen fibril. Incorporation of an abnormal collagen fibril interferes with normal triple helical formation and produces OI through a dominant negative effect. In this case, three fourths of the collagen produced are abnormal. Missense mutations also underlie OI types II, III, and IV. The ultimate phenotype depends upon the particular substitution, with substitutions of glycine in the proa1(I) chain by arginine, valine, glutamic acid, aspartic acid, and tryptophan usually resulting in OI type II if they occur in the carboxylterminal 70% of the triple helix. DI may or may not be present in these cases of OI due to dominant negative effects. Recently, a patient with type I EDS, a disorder characterized by joint hypermobility, skin hyperextensibility and tissue fragility, in whom a C to T transition (c.934C>T; p.R134C) in COL1A1 was identified, was shown to have structural dentin defects despite a clinically unaffected dentition [De Coster et al., 2007]. This same mutation has been reported in 2 additional type I EDS patients, but no histological and ultra-

structural studies have been performed on teeth from those patients [Nuytinck et al., 2000].


The SIBLING (small integrin-binding ligand N-linked glycoprotein) family of proteins includes dentin sialo-phosphoprotein (DSPP), osteopontin [also known as SPP1 (secreted phosphoprotein)], IBSP (integrin binding sialoprotein), DMP1 (dentin matrix protein 1), and MEPE (matrix extracellular phosphoglycoprotein) [Fisher et al., 2001]. Although all members were originally isolated from tooth or bone, it is now apparent that at least some SIBLINGs are expressed in non-mineralized tissues. The SIBLING genes are syntenic (on the same chromosome), clustered in a 0.37-Mb region on chromosome 4q. The encoded proteins contain an integrin binding tripeptide (Arg-Gly-Asp) and several conserved N-glycosylation and phosphorylation sites. The expression levels of the different SIBLINGs vary greatly between tissues. For example, OPN is expressed in bone at a level 70 times higher than that in dentin [Butler et al., 2003]. Although all SIBLINGS are thought to function in the mineralization process, their expression outside of mineralized tissues suggests additional roles. The recent finding of DSPP expression in embryonic kidney and lung has led to the hypothesis that DSPP may regulate branching morphogenesis [Alvares et al., 2006]. The only two SIBLINGS known to be associated with human genetic disorders are DMP1 and DSPP. Mutations in DMP1 were recently shown to underlie autosomal recessive hypophosphatemic rickets [Feng et al., 2006; Lorenz-Depiereux et al., 2006].


DSPP is located at 4q22.1 and consists of five exons spanning 8,343 bp. DSPP is expressed in dentin at levels that are hundreds of times higher than that of other tissues. Three distinct protein products result from proteolytic cleavage of the initial 1253-amino acid protein product, DSPP: DSP (amino acids 16-374), dentin glycoprotein (DGP; amino acids 375-462), and DPP (amino acids 463-1253; fig. 2). The initial cleavage by a yet unidentified protease releases DPP, a process that occurs almost immediately upon formation of DSPP [Yamakoshi et al., 2006]. Matrix metalloproteinase (MMP) 20 then cleaves DSP-DGP to generate DSP and DGP, while MMP2 catalyzes other cleavages on DSP. DPP is a highly repetitive protein with a great degree of phosphorylation and is thought to be involved in the nucleation and control of mineralization in dentin [George et al., 1996]. DPP contains repeats of aspartic acid (Asp) and phosphoserine

(Pse) mainly as (Asp-Pse-Pse) and (Asp-Pse). There is extensive variability in the number of repeats in normal individuals [Fisher, pers. commun.]. After synthesis and cleavage, DPP moves quickly to the mineralization front and binds to collagen. DSP is a heavily glycosylated protein that forms covalent dimers by intermolecular disul-fide bridges [Yamakoshi et al., 2005], but whose function is not entirely known. DGP was recently isolated from porcine odontoblasts and found to contain four phos-phorylated serines and one N-glycosylated asparagine. Given the posttranslational modifications and amino acid conservation across species, it has been proposed that DGP plays a role in dentin biomineralization [Yamakoshi et al., 2005].

DSPP knockout mice develop abnormalities similar to those seen in DI type III [Sreenath et al., 2003]. Histo-logically, large areas of unmineralized dentin were seen as well as an irregular border between the unmineralized and mineralized dentin. In addition, there were areas of partially mineralized dentin between areas that were fully mineralized.

To date only mutations in DSPP have been found to underlie DI types II and III and DD type II. Although 11 mutations have been reported in the literature, two of the published disease-causing mutations are actually normal sequence variants: p.R68W and Del1160-1171; ins1198-1199 (table 2). Holappa et al. [2006] found the p.R68W change in 15% of Finnish controls. Our laboratory found p.R68W with allele frequencies of 6 and 16% in Caucasian and African-American controls, respectively. We have also studied a branch of the Brandywine family reported by Dong et al. [2005]. Although we identify the same alteration in the hypervariable region, there is considerable variation in this region in normal controls, with similar deletions and insertions found. This normal variation is apparent when trying to align the genomic sequence with the DSPP cDNA reference sequence [Beattie et al., 2006]. We identified a second nucleotide change in the genomic sequence (g.49C>T) of the DSPP gene in this family, predicted to result in a p.P17S missense alteration, that segregates with the phenotype and causes endoplasmic re-ticulum retention of the protein in a functional assay. Thus, we propose that the p.P17S substitution is the true causative mutation in the Brandywine family. Of the nine confirmed DSPP mutations, all occur within the DSP region of DSPP. It should be pointed out that the effect of the mutation on the protein is simply a prediction, as more than one consequence may be possible. For example, the c.52G>T alteration might result in the missense alteration, p.V18F. On the other hand, this alteration occurs as

Fig. 2. Diagram of the DSPP gene and corresponding protein structure. Mutations are shown above the genomic structure. Gray areas correspond to untranslated regions. The white box corresponds to the signal peptide [amino acids (aa) 1-15]. DSP is shown as black regions (aa 16-382). DGP, represented by a stippled box, corresponds to aa 383-462. The highly repetitive protein, DPP, is shown as a striped box (aa 463-1253).

Fig. 2. Diagram of the DSPP gene and corresponding protein structure. Mutations are shown above the genomic structure. Gray areas correspond to untranslated regions. The white box corresponds to the signal peptide [amino acids (aa) 1-15]. DSP is shown as black regions (aa 16-382). DGP, represented by a stippled box, corresponds to aa 383-462. The highly repetitive protein, DPP, is shown as a striped box (aa 463-1253).

Table 2. Mutations in DSPP resulting in dentin defects






Exon 2

p.Y6D p.A15V p.P17T p.P17S

c.16T>G c.44C>T c.49C>A c.49C>T

g.16T>G g.44C>T g.49C>A g.49C>T


Rajpar et al. [2002] Malmgren et al. [2004] Xiao et al. [2001] this report [2007]

Intron 2


c.52-3C>G c.52-3C>A

g.1194C>G g.1194C>A


Kim et al. [2004] Holappa et al. [2006]

Exon 3

p.V18F or p.V18_Q45del




Xiao et al. [2001] Kim et al. [2005] Song et al. [2006] Holappa et al. [2006]





Zhang et al. [2001] Song et al. [2006]

Intron 3





Xiao et al. [2001]

Exon 4





Malmgren et al. [2004]

Exon 5



Dong et al. [2005]



All numbering assumes the A of the ATG start codon as nucleotide 1. The reference is NM_014208. a This alteration has been shown to be a normal variant and is not disease causing.

the first nucleotide of exon 3 and may thus disrupt splicing, causing skipping of exon 3 (p.V18_Q45del). Site-directed mutagenesis was used to introduce each of the nine mutations and the p.R68W polymorphism into the human DSP mRNA which were then transfected into odon-toblast cells. The ability of the cells to secrete DSP was evaluated by Western blot analysis. This functional assay demonstrated that with the exception of the nonsense mutation, p.Q45X, all other mutations are associated with endoplasmic reticulum retention, consistent with a dominant negative mutation [Choi et al., submitted].

The finding that the same mutation produces different phenotypes suggests that factors other than the specific DSPP mutation contribute to the ultimate phenotype. There are some families with dentin defects linked to 4q22.1 in which no DSPP mutation can be identified [Beattie et al., 2006]. Family A in the paper by Malmgren et al. [2004] is also an example of linkage to 4q22.1 with out the underlying defect identified since the predicted causative mutation (p.R68W) has since been shown to be simply a non-synonymous single nucleotide polymorphism that segregated with the phenotype. Whether these families have undetected mutations in DSPP, e.g. intronic changes, promoter alterations, deletions of exons, or mutations in the hypervariable region which is typically not analyzed, or have involvement of another gene in the 4q22.1 region remains to be determined.


As noted above, MMP2 and MMP20 were recently shown to be involved in the processing of DSPP [Yamako-shi et al., 2006], specifically to generate DSP and DGP. Mutations in MMP20 result in amelogenesis imperfecta without an obvious dentin abnormality [Kim et al., 2005; Oz-demir et al., 2005]. It has been suggested that molecular redundancy may explain this lack of phenotype [Yamako-shi et al., 2006]. MMP8 was recently identified as the major collagenase in human dentin [Sulkala et al., 2007]. Other metalloproteases synthesized by odontoblasts include MMP9, and membrane-bound MMP14 (MT1-MMP). KLK4 has also been detected in dentin extracts but its role in processing dentinal proteins is not known [Yamakoshi et al., 2006], but its expression by odontoblasts might be associated with the hypermineralization of enamel above the dentinoenamel junction [Fukae et al., 2002].

Other Genes Expressed in Dentin

A variety of other genes are expressed in dentin. Many of these genes, if mutated, would not be expected to give rise to isolated enamel defects. Patients with EDS type VIIC, who have a defect in the ADAMTS2 gene which encodes the enzyme responsible for removing the N-termi-nal propeptide in procollagen types I-III, exhibit dentin defects [De Coster et al., 2006]. Recently, HMGB1 (high mobility group box 1) was shown to be highly expressed in dentin at regions undergoing mineralization [Sugars et al., 2007]. HMGB1-/- mice die within 24 h of birth due to hypoglycemia [Calogero et al., 1999]. PHEX, mutated in X-linked hypophosphatemic rickets (XLH), encodes an en-dopeptidase postulated to play a role in the processing of DMP1 and DSPP [Qin et al., 2004]. Loss of function mutations in the gene encoding fibroblast growth factor 23 (FGF23), a phosphaturic protein, result in autosomal recessive hyperphosphatemic familial tumoral calcinosis (OMIM 211900) [Benet-Pag├Ęs et al., 2005] while gain of function mutations produce autosomal dominant hypophosphatemic rickets (OMIM 193100) [ADHR Consortium, 2000]. Hyperphosphatemic familial tumoral calcinosis can also result from mutations in GALNT3 which encodes a glycosyltransferase involved in initiating O-gly-cosylation [Ichikawa et al., 2005; Specktor et al., 2006]. The demonstration that both hyper- and hypophosphatemic disorders have dentin defects underlies the importance of phosphate homeostasis in normal dentin development.

Was this article helpful?

0 0

Post a comment