Novel MMP20 and KLK4 Mutations Causing Amelogenesis Imperfecta

Amelogenesis imperfecta (AI) is a heterogeneous collection of inherited enamel malformations. Genes known to participate in the etiology of non-syndromic AI are AMELX, ENAM, FAM83H, MMP20, KLK4, WDR72, and C4orf26, although defects in these genes account for less than 50% of all cases. Objective: To use exome sequencing combined with bioinformatics to identify the disease-causing mutations in 12 AI families. Methods: AI families with recessive or simplex patterns of inheritance were recruited. Pedigree construction and phenotypic characterization were based upon family history, clinical examinations, intraoral photos and dental radiographs. A sample of each proband’s DNA was submitted for exome capture/next generation sequencing (Edge BioSystems, Gaithersburg, MD USA). Sequence variations in known and suspected AI candidate genes were analyzed to identify probable disease-causing mutations. Then the distribution of these variations throughout their respective pedigrees were determined by standard mutational analyses. Results: We identified three AI-causing mutations: a heterozygous FAM83H (p.W460*) mutation that had previously been characterized, and novel homozygous MMP20 (p.His204Arg), and KLK4 (p.Gly82Alafs*87) mutations. The MMP20 defect substitutes arginine for an invariant histidine that coordinate a structural zinc ion critical for MMP20 structure. The proband’s enamel was thin and less mineralized than normal, and more closely resembled the enamel defects in MMP20 null mice than the enamel in the few previously reported MMP20 mutations. The KLK4 mutations was a frameshift that is predicted to have caused nonsense mediated decay of KLK4 mRNA. The proband’s phenotype was severe enamel hypomaturation. We continue to analyze the data to identify potential AI-causing mutations in genes not yet associated with AI in the remaining families. Conclusion: Combined exome sequencing and bioinformatics can screen all suspected AI candidate genes in a single step and is more efficient than using mutational analyses prioritized using genotype-phenotype correlations. This study was supported by NIDCR grant DE015846.