Novel KLK4 and MMP20 Mutations Discovered by Whole-exome Sequencing

Abstract

Non-syndromic amelogenesis imperfecta (AI) is a collection of isolated inherited enamel malformations that follow X-linked, autosomal-dominant, or autosomal-recessive patterns of inheritance. The AI phenotype is also found in syndromes. We hypothesized that whole-exome sequencing of AI probands showing simplex or recessive patterns of inheritance would identify causative mutations among the known candidate genes for AI. DNA samples obtained from 12 unrelated probands with AI were analyzed. Disease-causing mutations were identified in three of the probands: a novel single-nucleotide deletion in both KLK4 alleles (g.6930delG; c.245delG; p.Gly82Alafs*87) that shifted the reading frame, a novel missense transition mutation in both MMP20 alleles (g.15390A>G; c.611A>G; p.His204Arg) that substituted arginine for an invariant histidine known to coordinate a structural zinc ion, and a previously described nonsense transition mutation in a single allele of FAM83H (c.1379G>A; g.5663G>A; p.W460*). Erupted molars and cross-sections from unerupted parts of the mandibular incisors of Mmp20 null mice were characterized by scanning electron microscopy. Their enamel malformations closely correlated with the enamel defects displayed by the proband with the MMP20 mutation. We conclude that whole-exome sequencing is an effective means of identifying disease-causing mutations in kindreds with AI, and this technique should prove clinically useful for this purpose.

Introduction

Non-syndromic amelogenesis imperfecta (AI) is a collection of inherited dental enamel malformations that can be caused by defects in many genes and can be manifested by diverse enamel phenotypes. X-linked AI is caused by defects in AMELX. Autosomal-dominant forms of AI can be caused by ENAM (Rajpar et al., 2001) and FAM83H (Kim et al., 2008) mutations, while autosomal-recessive AI can be caused by defects in MMP20 (Kim et al., 2005), KLK4 (Hart et al., 2004), WDR72 (El-Sayed et al., 2009), and C4orf26 (Parry et al., 2012). These genes account for about half of all non-syndromic AI cases (Chan et al., 2011). The other causative genes are currently unknown.

The term ‘amelogenesis imperfecta’ is also used to describe the presence of an enamel phenotype in syndromes. Syndromic forms of AI and their associated causative genes include junctional epidermolysis bullosa (LAMA3, LAMB3, LAMC2, COL17A1) (Buchroithner et al., 2004; Almaani et al., 2009), cone-rod dystrophy and AI (CNNM4) (Parry et al., 2009), AI with gingival fibromatosis syndrome (FAM20A) (O’Sullivan et al., 2011), enamel-renal syndrome (FAM20A) (Wang et al., in press), tricho-dento-osseous syndrome (DLX3) (Price et al., 1998), autoimmune polyendocrinopathy candidiasis ectodermal dystrophy (AIRE) (Pavlic and Waltimo-Sirén, 2009), oculodentodigital dysplasia (GJA1) (van Es et al., 2007), and Kohlschütter-Tönz syndrome (ROGDI) (Schossig et al., 2012), among others. Many of these conditions can be confused with non-syndromic forms of AI, since the enamel malformations may be the only apparent, or earliest-onset, phenotype in some cases.

As the number of candidate genes proven to cause inherited enamel defects has risen, genotype-phenotype correlations have been used to prioritize candidate genes for mutational analyses to identify the genetic defect. However, enamel phenotypes associated with many AI candidate genes are not distinctive, limiting the usefulness of genotype-phenotype correlations in predicting the causative gene, especially in recessive cases. Whole-exome sequencing identifies sequence variations in coding regions and intron junctions throughout the genome. Variations within candidate genes can then be analyzed to determine if they are disease-causing, and the results should prove useful in the diagnosis by dentists of patients with inherited disorders. In this study, we used whole-exome sequencing to identify disease-causing mutations in KLK4, MMP20, and FAM83H in our patients with amelogenesis imperfecta.

Materials & Methods

The human study protocol and participant consents were reviewed and approved by the Institutional Review Board at the University of Michigan. Study participants signed appropriate written consents after an explanation of their contents and after their questions about the study were answered. All animal procedures were approved by the University of Michigan Committee of Use and Care of Animals.

DNA Isolation and Characterization

Peripheral whole blood (5 cc) was obtained from recruited members of each family. Genomic DNA was isolated with the QIAamp DNA Blood Maxi Kit and protocol (Qiagen Inc., Valencia, CA, USA), and its quality and quantity were determined by spectrophotometry at OD₂₆₀ and OD₂₈₀. DNA samples, one from each proband of the 12 AI kindreds showing a simplex or autosomal-recessive pattern of inheritance, were submitted for whole-exome sequencing (Edge BioSystems, Gaithersburg, MD, USA). Genomic DNA (3 µg) from each proband was nebulized into ~300-bp fragments, which were tailed and ligated to adapters, amplified, enriched, and analyzed by means of an Agilent 2100 Bioanalyzer prior to being sequenced in paired-end reads, 75 to 100 bases per read (Life Technologies SOLiD™ System, Carlsbad, CA, USA). For the 12 samples analyzed, the average depth of coverage of the targeted exome was 73x. The percentage of target bases not covered was 6.2%.

The annotated results were inspected to search for potential disease-causing sequence variations in the following candidate genes: AMELX, ENAM, FAM83H, KLK4, MMP20, WDR72, FAM20A, C4orf26, CNNM4, AMBN, AMTN, ODAM, SP6, COL17A1, ROGDI, LAMA3, LAMB3, LAMC2, DLX3, GJA1, AIRE, and ABCC6. In the three families with identified mutations, the exon carrying the suspected defect (KLK4 coding exon 3, MMP20 exon 4, and FAM83H exon 5) inclusive of adjoining intron sequences was amplified from all recruited family members and characterized by Sanger DNA sequencing.

Scanning Electron Microscopy (SEM)

Mandibular molars and incisors (sectioned at level 8 in the maturation stage and polished) from 9-week-old wild-type and Mmp20 null mice were imaged with a Field Emission Gun Scanning Electron Microscope (FEG-SEM; Amray 1910 Field Emission Scanning Electron Microscope) at the Microscopy and Image Analysis Laboratory at the University of Michigan as described previously (Hu et al., 2011). Mandibular incisors (sectioned at level 3 at the end of the secretory stage) were examined by means of a Hitachi S-3000N variable pressure scanning electron microscope in backscatter mode as described previously (Smith et al., 2011).

Results

Family 1: KLK4

The proband (IV:1) of family 1 was a 9-year-old girl from Turkey with a hypomaturation form of amelogenesis imperfecta (Fig. 1) and no history of systemic disturbances. Her father (III:5) and mother (III:6) were first cousins (Fig. 1A), and their enamel appeared to be normal (Appendix Fig. 1). The enamel covering the proband’s teeth appeared normal in size and shape, but was discolored brown, chipped on multiple teeth, and secondarily affected with dental caries (Fig. 1B). The proband had a class III molar relationship and may have lost vertical dimension. Although eruption of the maxillary incisors was delayed, there was no anterior open bite. Radiographically, the enamel was of normal thickness, but only slightly more radio-opaque than dentin (Fig. 1C). Whole-exome sequencing of the proband’s genome identified a single-nucleotide deletion (g.6930delG; c.245delG; p.Gly82Alafs*87) in both KLK4 alleles that shifted the reading frame at codon 82 of 255. The frameshift is in the third of 5 coding exons (Appendix Fig. 2), so the mutant KLK4 mRNA transcripts might have been degraded by nonsense-mediated decay. If translated, the mutant protein would not be catalytically active, although it could potentially cause secondary pathology through dominant- negative or gain-of-function effects. The homozygous mutation was confirmed by Sanger sequencing of PCR amplification products. Both parents were heterozygous for the deletion (Fig. 1D). The simplex pattern of inheritance of the pedigree was thus shown to be autosomal-recessive.

Figure 1.

Family 1. (A) Pedigree of consanguineous family from Turkey with enamel malformations in the proband. (B) Frontal, lateral, and occlusal photos of the proband at age 9 yrs. The enamel is chipped and shaded brown. (C) Panorex radiograph of the mixed dentition showing enamel of normal thickness that contrasts only slightly with dentin. (D) DNA sequencing chromatograms showing that the single-nucleotide deletion (c.245delG; position marked by arrowhead) is homozygous in the proband and heterozygous in his parents.

Family 2: MMP20

The proband (III:2) of family 2 was a 14-year-old boy who was the only family member with a dental phenotype (Fig. 2A). His chief complaint concerned his dentition, which exhibited generalized hypoplastic (thin) enamel that abraded easily and was sensitive to thermal changes, particularly cold (Fig. 2B). Bitewing radiographs showed a thin enamel layer only slightly more radio-opaque than the underlying dentin. Whole-exome sequencing identified an apparent disease-causing mutation (g.15390A>G; c.611A>G; p.His204Arg) in both alleles of MMP20 that was confirmed by Sanger sequencing. This sequence variation is absent from the dbSNP and 1000 Genomes Project Pilot Data databases. His²⁰⁴ coordinates a structural zinc ion (Maskos, 2005) and is conserved in all known MMP20 sequences. The distribution of this sequence variation among the five members of the nuclear family was determined (Fig. 2C). The parents and the proband’s younger brother (III:3) were heterozygous for the c.611A>G transition mutation, while the older brother had the wild-type sequence in both alleles. These analyses strongly support the conclusion that the homozygous MMP20 p.His204Arg missense mutation caused the recessive enamel malformations in the proband.

Figure 2.

Family 2. (A) Pedigree of family with a simplex pattern of inheritance. DNA was obtained from the five persons in the nuclear family (black dots). (B) Oral photographs and bitewing radiographs of the proband at age 14 yrs. Most of the occlusal enamel had abraded from the mandibular first molars. The radiographs show a thin layer of enamel that is only slightly more radio-opaque than dentin. (C) DNA-sequencing chromatograms show that the parents were both heterozygous for the MMP20 c.611A>G transition mutation, while the proband was homozygous for the mutation (p.His204Arg) and his older brother was homozygous-normal. The sequence analysis determined that the mutation was inherited in a recessive pattern.

Scanning electron microscopy of erupted first mandibular molars from 9-week-old Mmp20 null mice showed severe abrasion of the occlusal surfaces (Fig. 3A). A polished, late-maturation-stage cross-section of an Mmp20 null mouse mandibular incisor, where the enamel had reached full thickness but had not yet erupted and suffered abrasion, showed enamel thinner than that in the wild-type mouse (27 µm vs. 114 µm at the height of curvature) (Fig. 3B), and a smaller cross-sectional area (26,577 µm² vs. 70,389 µm²). Backscatter SEMs of a polished cross-section of a mandibular incisor at the end of the secretory stage was already much thinner than that of the wild-type mouse and less mineralized in the deeper enamel near the dentin surface (Fig. 3C). The Mmp20 null mouse enamel defects closely resemble those of the proband.

Figure 3.

Mmp20 null mouse enamel. (A) SEMs of occlusal surfaces of mouse first molars at 9 wks. The Mmp20 null molar shows extensive abrasion, with enamel having chipped away from the biting surfaces (arrowheads). (B) SEM of mandibular incisors cross-sectioned at an unerupted location in the late-maturation stage. The Mmp20 null enamel is only 1/3 as thick as that of the wild-type mouse. (C) SEM in backscatter mode of mandibular incisors cross-sectioned at a position near the transition from secretory to maturation stages. The Mmp20 null enamel is thin, hypermineralized at the surface, but hypomineralized near the dentin. d, dentin; e, enamel. Bars = 100 µm.

Family 3: FAM83H

The proband of family 3 was a 9-year-old Caucasian girl. The condition was reported to be inherited from the paternal side of her family. The proband’s enamel was brown, soft with the consistency of hard cheese, did not contrast with dentin on radiographs, and abraded easily (Appendix Fig. 3). Oral hygiene was poor. Based upon her enamel phenotype, we screened WDR72, but no mutations were found. Whole-exome sequencing identified a heterozygous mutation in FAM83H (c.1379G>A; g.5663G>A; p.W460*) that was confirmed by Sanger sequencing. This mutation was previously shown to cause autosomal-dominant AI (Wright et al., 2009), which is the reported pattern of inheritance in family 3, where a defect in a single FAM83H allele caused the enamel phenotype.

Discussion

KLK4 is a serine proteinase that cleaves extracellular matrix proteins during the transition and maturation stages of amelogenesis (Lu et al., 2008). Klk4 null mice, with lacZ inserted in place of the Klk4 coding region, have soft enamel that undergoes rapid attrition (Simmer et al., 2009). No phenotype outside the dentition has been observed in these mice. Xgal staining of various tissues in the lacZ knock-in mice showed that Klk4 is expressed at relatively high levels by maturation-stage ameloblasts and at trace levels in salivary glands and the prostate (Simmer et al., 2011). Previously, a homozygous KLK4 mutation introducing a premature stop codon was shown to cause hypomaturation AI in identical twins, with no signs or symptoms elsewhere in the body (Hart et al., 2004). Disease-causing mutations in KLK4 are extremely rare. This is only the second novel disease-causing KLK4 mutation ever reported, although it has been a candidate gene for AI since it was originally cloned from developing teeth in the 1990s.

MMP20 is secreted by secretory- and early-maturation-stage ameloblasts (Hu et al., 2002). The enzyme cleaves enamel extracellular matrix molecules, like amelogenin (Ryu et al., 1999) and ameloblastin (Chun et al., 2010). In the absence of MMP20, uncleaved enamel proteins accumulate in the secretory-stage enamel matrix (Yamakoshi et al., 2011). Although a major function of MMP20 is to facilitate removal of enamel matrix proteins to enhance mineralization, the complex enamel phenotype that develops in its absence seems to involve the disruption of multiple processes.

Disease-causing mutations in MMP20 are extremely rare. This is only the fifth novel disease-causing MMP20 mutation reported to date (Appendix Fig. 4). In all cases, there was no evidence of systemic abnormalities. Although we observed close similarity between the enamel malformations in our proband and those in Mmp20 null mice, the enamel phenotypes observed in persons with different MMP20 mutations vary (Appendix Fig. 4). Part of the variation in the enamel phenotypes may relate to the susceptibility of the malformed crowns to staining and abrasion following eruption. The enamel phenotype typically increases in severity with age, due to progressive deterioration.

FAM83H first came to the attention of scientists with the discovery that mutations in FAM83H cause autosomal-dominant hypocalcified amelogenesis imperfecta (ADHCAI) (Kim et al., 2008). FAM83H encodes an intracellular protein of unknown function that is associated with the Golgi apparatus or trans-Golgi network (Ding et al., 2009). To date, 20 novel FAM83H disease-causing mutations have been reported, and all are nonsense or frameshift mutations in the last coding exon that truncate the protein between Ser²⁸⁷ and Glu⁶⁹⁴ (Appendix Fig. 5). The N-terminal region of the mutant protein is all that is translated, sometimes along with a large frame-shifted segment. The N-terminal domain shares homology with the phospholipase D superfamily and is assumed to form dimers and to interact with other protein(s). Quaternary structures containing the truncated protein are thought to mediate dominant-negative effects in ameloblasts or their progenitors, and the resulting pathology causes enamel malformations.

Recently a single-nucleotide deletion in exon 5 of dog Fam83h showed no discernible phenotype in heterozygotes (Forman et al., 2012). Even the homozygous mutants showed no obvious enamel abnormalities, but suffered from congenital keratoconjunctivitis sicca and ichthyosiform dermatosis (CKCSID), a disease not closely analogous to any human condition. The frameshift in dogs (c.977delC; p.Pro326Hisfs*258) is analogous to the types of mutations that cause ADHCAI in humans (Appendix Fig. 6). The dog results are mystifying. Perhaps the mutant FAM83H protein in dogs is degraded, or its mRNA transcript undergoes nonsense-mediated decay, so that the dominant-negative or gain-of-function effect is not observed.

MMP20 and KLK4 are the major extracellular proteases in developing enamel. When these enzymes were discovered, it was anticipated that defects in the genes encoding them might account for a high proportion of recessive AI cases, especially in families with hypomaturation AI, where the enamel shows normal thickness but is softer than normal and appears to contain residual enamel proteins. It now appears that defects in many genes cause autosomal-recessive AI. In this study, we were able to pinpoint the disease-causing mutation in three of 12 families by searching for mutations among a list of candidate genes. Since no probable disease-causing defects were observed among the many AI candidate genes in nine of the families, it is clear that more genes contribute to the etiology of autosomal-recessive AI than are currently recognized. This is consistent with recent studies reporting that mutational analyses of all the known AI candidate genes cannot identify the disease-causing mutation in most recessive AI cases (Chan et al., 2011). Mutation analyses of known AI candidate genes identified the disease-causing mutation in only nine of 35 families with recessive AI (Wright et al., 2011). Whole-exome sequencing is a rapidly evolving, high-throughput technology. It is possible that a disease-causing sequence variation was missed because of limitations of the sequencing platform and analysis pipeline; however, many common single-nucleotide polymorphisms (SNPs) were identified among the AI candidate genes in our study. Therefore, it is not because of the limitations of whole-exome sequencing that we identified the disease-causing mutation in only three of 12 probands, but because of our lack of knowledge of the full spectrum of genes that causes recessive forms of AI. Further analyses of the whole-exome sequences in these families may help us identify new AI candidate genes.

Recently, whole-exome sequencing added FAM20A (O’Sullivan et al., 2011) and C4orf26 (Parry et al., 2012) to the list of candidate genes that causes syndromic and non-syndromic AI, respectively. Whole-exome sequencing is an efficient means of identifying the disease-causing mutation in kindreds with AI, as demonstrated in this report. Whole-exome sequencing makes it practical for dentists to consult clinical geneticists to order tests that can identify the disease-causing mutations in their patients with inherited enamel malformations to make a gene-based diagnosis of their type of AI and to determine whether or not systemic abnormalities should be anticipated.