Novel DLX3 variants in amelogenesis imperfecta with attenuated tricho‐dento‐osseous syndrome

Abstract Objectives Variants in DLX3 cause tricho‐dento‐osseous syndrome (TDO, MIM #190320), a systemic condition with hair, nail and bony changes, taurodontism and amelogenesis imperfecta (AI), inherited in an autosomal dominant fashion. Different variants found within this gene are associated with different phenotypic presentations. To date, six different DLX3 variants have been reported in TDO. The aim of this paper was to explore and discuss three recently uncovered new variants in DLX3. Subjects and Methods Whole‐exome sequencing identified a new DLX3 variant in one family, recruited as part of an ongoing study of genetic variants associated with AI. Targeted clinical exome sequencing of two further families revealed another new variant of DLX3 and complete heterozygous deletion of DLX3. For all three families, the phenotypes were shown to consist of AI and taurodontism, together with other attenuated features of TDO. Results c.574delG p.(E192Rfs*66), c.476G>T (p.R159L) and a heterozygous deletion of the entire DLX3 coding region were identified in our families. Conclusion These previously unreported variants add to the growing literature surrounding AI, allowing for more accurate genetic testing and better understanding of the associated clinical consequences.

. All of the individuals reported to date presented with taurodontism and enamel hypoplasia. Robinson and Miller (1966) reported that the hair of affected individuals is classically coarse, kinky/curly, brittle and rough, with profuse shedding. This usually normalizes with age. Skeletal abnormalities within the head and neck have also been associated with TDO, including a lack of mastoid pneumatization, increased thickness of the cranial bones, increased bone density and a shortened mandibular ramus (Crawford & Aldred, 1990). The thickening of the bone in TDO may be of concern with regard to the increased risk of fracture or macrocephaly (Al-Batayneh, 2012;Shapiro, Quattromani, Jorgenson, Young, & Opitz, 1983). However, Hart et al. (1997) stated that the bone changes are not associated with any pathology. Nails have also been reported to show splitting and to break easily Wright, Kula, Hall, Simmons, & Hart, 1997). Atopic dermatitis was also noted in one individual with TDO (Mayer, Baal, Litschauer-Poursadrollah, Hemmer, & Jarisch, 2010).
The clinical abnormalities associated with TDO are usually identified within the first year of life. Differentiating between TDO and amelogenesis imperfecta of the hypomaturation-hypoplasia type with taurodontism (MIM #104510) can be difficult and is typically based upon other syndromic findings, such as kinky hair and bony changes. Both conditions have been associated with DLX3 mutations suggesting the two conditions may be part of the same syndromic spectrum (Dong et al., 2005). However, Price et al. (1999) and Crawford and Aldred (1990) considered the two syndromes different entities. To address the controversy diagnosing TDO, Seow (1993) created criteria comprising generalized enamel defects; severe taurodontism affecting the mandibular first permanent molars; autosomal dominant inheritance; and at least one other feature (i.e., nail changes; bone sclerosis; or curly, kinky or wavy hair at a young age). However, because of the known clinical variations in the phenotypes of TDO, these criteria are not widely used.
This report presents two novel DLX3 variants and heterozygous deletion of DLX3 associated with AI. All families presented with AI prior to genetic testing. Subsequent to the identification of a DLX3 variant in each family, affected individuals were reassessed clinically and were considered to have attenuated versions of TDO. We also report analysis of the hair phenotypes in family 1 and discuss the likely consequences of these newly identified mutations on the function of the DLX3 protein.

| Patients
Eleven individuals from family 1, three individuals from family 2 and one individual from family 3 ( Figure 1) were recruited following informed consent in accordance with the principles outlined in F I G U R E 1 Pedigrees for families 1, 2 and 3. Probands wholeexome sequenced/targeted clinical-exome sequenced marked with arrows. * indicates that the DNA was of insufficient quality and quantity to sequence. •: Affected female, □: Unaffected male

| Whole-exome and targeted clinical exome sequencing and analysis
Genomic DNA from individual IV:1 from family 1 (marked with an arrow on the pedigree, Figure 1) was subjected to whole-exome se- the human reference genome (GRCh37) using the Burrows Wheeler aligner (BWA) (Li & Durbin, 2009 Genes known to be associated with autosomal dominant AI were then identified from the variants list and segregated with Sanger sequencing for all family members for which DNA was available. Sanger sequencing was performed using the BigDye Terminator Biosoftware, Rouen, France) and further filtered to identify variants within genes known to be associated with amelogenesis imperfecta.
For family 2, Sanger sequencing to confirm segregation was carried out as above ( Figure 2).
Primer sequences for Sanger sequencing of both point mutations can be found in Supporting Information Table S1. Gene lists used for filtering for both families are available in Supporting Information Table S2.
Different methods were employed for each family because families 2 and 3 were part of a pilot study for NHS screening of known genes for AI and screened patients from the newly implemented NHS screening service, respectively (Holland, 2017). Family 1 were recruited as part of ongoing research into the genetic basis of AI, based at the Faculty of Medicine and Health, University of Leeds. and operated at an accelerating voltage of 5 kV using back-scatter electron detection. One strand of hair from each of the three available hair samples was analysed ( Figure 3).

| iTasser
Predicted protein outcome for the missense variant found in family 2 was assessed using the online protein structure and predictions resource iTasser (Yang & Zhang, 2015). Wild-type and mutant residues F I G U R E 3 Samples of hair from family 1, imaged longitudinally using Hitachi S-3400N scanning electron microscope to investigate differences in hair structure between affected and unaffected individuals. a = V:2 (affected), b = IV:4 (affected), c = V:1 (unaffected) normal area, d = V:1 (unaffected) irregular area. Scale bar = 50 µm of DLX3 were input and the predicted outcomes for active sites and the secondary and tertiary structures of the proteins compared.

| RE SULTS
One mixed race (family 1-Afro-Caribbean and Caucasian) British family and one Caucasian British family (family 2), both with autosomal dominantly inherited hypoplastic AI, and one British Caucasian individual affected with hypoplastic AI but with no family history of AI (family 3) (Figure 1), were recruited to this study as part of a larger cohort of AI families. Families 1 and 2 did not display any clinically obvious co-segregating health problems, but a kinky, curly hair phenotype was reported in family 1. The affected member of fam- (Supporting Information Figure S1). Clinically evident bony changes (e.g., mandibular prognathism, Robinson & Miller, 1966) were not reported.
Bitewing radiographs (Supporting Information Figure S1 reported. Delayed eruption of the upper lateral permanent incisor teeth was a feature, along with talon cusps evident on radiography on these teeth, but no bony changes were identified in the gnathic bones on dental radiographs, nor were they reported by family members. One tooth in the permanent dentition had internal resorption and subsequent fracture. A curly/kinky hair phenotype was identified in II:1 in early childhood, but this normalized with age and the individual's parents were not aware of a kinky/curly hair phenotype in their childhoods.
For all of our families, there was no clinical justification for formal bone assessment involving ionizing radiation to confirm or exclude a phenotype consistent with TDO.
DNA from individual IV:1 (family 1) was screened by WES. p.R159L (NM_005220.2) affecting a base in the second exon of the DLX3 transcript. Sanger sequencing confirmed segregation of these variants with disease in all available family members (Figure 2).
Supporting Information Figure S3 shows the Integrated Genomics Viewer reads for family 1 to confirm the presence of a 1-bp deletion in clonal sequencing.
The c.574delG variant identified in family 1 is predicted to result in a frameshift p.(E192Rfs*66) (NP_005211.1). However, the transcript produced is likely to escape nonsense-mediated decay because it is in the final exon (Isken & Maquat, 2008). The variant transcript is predicted to encode a protein of 256 amino acids (compared to the 287 amino acid wild-type protein) with 65 incorrect amino acids incorporated at the C-terminus.
The variant from family 2 incorporates a hydrophobic leucine residue instead of a positively charged arginine residue. When the p.R159L-mutated DLX3 protein identified in family 2 was input into iTasser, the resultant protein structure was significantly different from that predicted for wild-type DLX3. Protein p.159L is predicted to have a more tightly packed arrangement, with additional helical regions (Supporting Information Figure S4) and loss of predicted methionine and propanoic acid ligand-binding sites.

| D ISCUSS I ON
Here, we report the first families segregating the DLX3 c. ing the breakpoints of this deletion, it is difficult to know whether this phenotype is wholly attributable to the DLX3 deletion, or whether other genes flanking DLX3 may have also been deleted.
The variants discussed in this paper were detected using two different methods; for family 1, whole-exome sequencing was used, whilst for families 2 and 3 targeted clinical exome sequencing was used. Clinical exome sequencing has the advantages over wholeexome sequencing of providing higher coverage for targeted areas, a lower cost per sample analysed and easier interpretation. However, whole-exome sequencing allows the identification of variants in new genes for AI.
DLX3 is comprised of 3 exons and is located on the long arm of chromosome 17, along with other members of the distal-less family of genes (DLX1-6). It is expressed in the placenta and is considered to have a pivotal role in hard-tissue formation, possibly explaining why the DLX3 null genotype is fatal in embryo mice (Hwang, Mehrani, Millar, & Morasso, 2008;Morasso, Grinberg, Robinson, Sargent, & Mahon, 1999). DLX3 is crucial for hair cycling (Hwang et al., 2008), patterning of the embryonic ectoderm (Park & Morasso, 2002), and is implicated in bone formation (Hassan et al., 2004). It is a transcriptional activator, expressed in the differentiated epidermal granular cell layer and in the matrix cells of hair follicles. It has also recently been shown to affect ion transport in amelogenesis through regulation of genes in the solute carrier (Slc) family (Duverger, Ohara, Bible, Zah, & Morasso, 2017).
The phenotype of the affected individuals in all three of our families is similar to that seen in TDO, but without clinically obvious bony changes, nail features and dermatitis. Conducting bone scans and radiographs using ionizing radiation to characterize any bone changes could not be justified in the context of the clinical presentations. The bony changes in TDO are considered to be variable and can be difficult to characterize, despite reportedly being present in 65%-80% of patients (Price et al., 1998). It has also been suggested that bony changes progress with age, meaning that while not currently clinically apparent, the bony features may be more noticeable as the younger affected individuals age Price et al., 1998). The increase in bone density seen in TDO patients has been hypothesized to be due to decreased osteoclastic bone due to the increased IFN-γ expression by immune cells, although the pathway is not fully elucidated (Choi et al., 2009). It is currently unclear whether bones outside of the craniofacial area are also affected in TDO; however, Haldeman et al. (2004) demonstrated that bone density was increased at the radius and ulna sites in individuals with TDO at ages less than 30 years but showed no obvious association between 30 and 74 years of age. Bone density of the spine and hip was also elevated in individuals with TDO, even at a young age. One of the more common, and easily identifiable, bony features of TDO is mandibular prognathism. In a study by Nguyen, Phillips, Frazier-Bower, and Wright (2013), TDO patients were compared to wild-type controls and no mandibular prognathism was identified. However, a retrusive maxilla, increased mandibular length and increased ramus height were common to most TDO cases, possibly giving the appearance of a prognathic mandible in affected individuals.
DLX3 has three main domains: the N-and C-terminus transactivation domains, and a central homeodomain, encoded by exons 2 and 3. The homeodomain can interact directly with DNA in a sequence-specific way and regulates the expression of target genes throughout numerous developmental processes (Feledy, Morasso, Jang, & Sargent, 1999). All of the previously reported mutations in DLX3 (each with an attenuated TDO-like phenotype) have affected residues within, or adjacent to, the homeodomain (Figure 4), altering the structure of this region and highlighting its importance in the pathogenesis of TDO. The addition of these variants to the existing literature highlights the different presentations associated with DLX3 variants and demonstrates, for all of our families, a more restricted phenotype than classical TDO.
The literature around TDO states that the hair of affected individuals straightens with age in 54% of affected individuals , and members of family 1 reported that their kinky/curly hair phenotype noted in childhood straightened with age. However, this changing phenotype does not seem to have been structurally demonstrated with SEM imaging of affected individuals from family 1, as the hair from the affected adult (IV:4) was morphologically similar to that from the affected child (V:2), with a woody, brittle appearance and almost complete lack of cuticle. In the unaffected individual, V:1, occasional, patchy, irregular cuticle formation was identified in a largely normal shaft. We have no information as to whether individual V:1 had undergone chemical treatments (e.g., straightening or colouring), which may have led to damage (Figure 3). The lack of a suitable control and the absence of a "hair history" for members of family 1 makes associating the DLX3 genotype with the curly/kinky hair phenotype, and the structural changes seen with SEM difficult in this case. The affected individual from family 3 was reported to have a kinky/curly hair in early childhood, but this straightened over time.
It is difficult to compare our findings to the paper by Wright et al (1997) as SEM was not conducted on the hair samples in that cohort.
It would be advantageous to have hair from family 3 to ascertain any structural changes to the hair shaft which may corroborate an abnormal hair phenotype, rather than relying solely on patient reports.
It is unknown whether delayed eruption of the permanent upper lateral incisor teeth in family 3 reflects a further phenotypic variation in TDO. In a TDO case described by Jain, Kaul, Saha, and Sarkar (2017), there was early eruption of the permanent first molar teeth.
Otherwise, the case had classical features of TDO without a family history. The authors suggest the mutation arose sporadically. This case was associated with a 4-bp deletion, but the exact location is not discussed in the report. Jain et al. (2017) hypothesized that the precocious eruption may be due to increased osteoblastic activity around the erupting tooth. As no bony abnormalities were identified clinically or on dental radiographs in family 3, it is difficult to determine the cause of the delayed eruption for our case. DNA from family 3, to confirm the deletion and to determine the exact break- The mechanism by which DLX3 variants cause disease has not been fully elucidated, with numerous mechanisms hypothesized (Choi et al., 2008;Dong et al., 2005;Duverger et al., 2008Duverger et al., , 2017Feledy et al., 1999). There are conflicting reports suggesting both haploinsufficiency and a dominant gain-of-function effect (Duverger et al., 2008;Li & Roberson, 2017;Nieminen et al., 2011).
Of the DLX3 variants previously reported to date, the c.571_574del-

CO N FLI C T O F I NTE R E S T
No conflict of interests declared.