Confirmation of a Phenotypic Entity for TSPEAR Variants in Egyptian Ectodermal Dysplasia Patients and Role of Ethnicity

Ectodermal dysplasia (ED) are hereditary disorders characterized by the disturbance of the ectodermal development of at least two of four ectodermal tissues: teeth, hair, nails and sweat glands. Clinical classification of ED is challenged by overlapping features, variable expressivity, and low number of patients, hindering full phenotypic spectrum identification. Disease-causing variants in elements of major developmental pathways, e.g., Ectodysplasin/NFκB, Wnt, and Tp63 pathways, have been identified in fewer than half of ED phenotypes. Whole-exome sequencing (WES) was performed for ten Egyptian ED patients presenting with tooth agenesis, normal sweating, scalp hypotrichosis, and sharing characteristic facial features. WES was followed by in silico analysis of the effects of novel detected genetic variants on mRNA and protein structure. The study identified four novel rare pathogenic and likely pathogenic TSPEAR variants, a gene which was recently found to be involved in ectodermal organogenesis. A novel in-frame deletion recurred in eight patients from six unrelated families. Comparing our cohort to previously reported TSPEAR cohorts highlighted the influence of ethnicity on TSPEAR phenotypic affection. Our study expands the clinical and mutational spectrum of the growing TSPEAR associated phenotypes, and pinpoints the influence of WES and in silico tools on identification of rare disease-causing variants.


Introduction
Ectodermal dysplasia (ED) is a nosologic group of hereditary disorders of the development and/or homeostasis of two or more tissue derivatives of the human ectoderm [1]. In 1970s, Freire-Maia classified 186 ED phenotypes into two groups: group A, in which at least two of four classical ectodermal derivatives are affected, these being the hair, teeth, nails, and sweat glands, and group B in, which only one of the classical ectodermal derivatives is affected, together with at least one other tissue of ectodermal origin, e.g., central nervous system, melanocytes, adrenal medulla, or lacrimal, Meibomian, mammary and thyroid glands. Eleven subgroups stem from group A according to the affected structures, e.g., hair/teeth/nails/sweat glands, hair/teeth/nails, hair/nails, teeth/sweat glands, etc. [2][3][4][5][6][7]. Concomitant disturbance of tissues originating from other embryonic layers has been reported, e.g., cardiofaciocutaneous syndrome (OMIM #115150) and cleft lip/palate-ED syndrome (OMIM#225060) [1]. A consensus on ED classification remains challenging patients and one NSTA patient all presented with normal hearing [27][28][29]. Among a total of 22 previously reported unrelated families with probands harboring biallelic TSPEAR disease-causing variants, the majority of TSPEAR associated phenotypes fall within ED or TA spectra [24,29].
We identified four rare pathogenic and likely pathogenic TSPEAR variants in a North African cohort of ten Egyptian ED patients descending from eight unrelated families. Our patients were initially found to be negative for EDA, EDAR, EDARADD, and WNT10A disease-causing variants, and thus were opted for whole exome sequencing (WES). Patients presented with TA, scalp hypotrichosis, normal sweating, and shared common characteristic facial features. Our study expands the clinical and mutational spectrum of TSPEAR associated phenotypes, and highlights the power of WES in identification of rare diseasecausing variants.

Ethical Aspects
All patients, parents, legal guardians, and available family members signed written informed consents to be included in our study after thorough explanation and discussion. Study design and ethical procedures were approved by the institutional review boards of the National Research Centre (NRC) of Egypt and the of the American University in Cairo (AUC) according to the Declaration of Helsinki. Patients were recruited from the Genodermatoses and Oro-Dental Genetics clinics of the Medical Research Excellence Centre, NRC, Egypt.

Patient Phenotyping and Inclusion Criteria
Patients were included based on: (1) clinical diagnosis of ED in which at least two ectodermal tissue derivatives (hair, nails, teeth and skin) were impaired, or diagnosis of NSTA; and (2) previous molecular exclusion of EDA, EDAR, EDARADD, and WNT10A disease-causing variants using targeted next generation sequencing (NGS) panel. Detailed medical and family histories were recorded, and three generation pedigrees were constructed. Thorough medical examination of the patients was performed with special emphasis on skin, hair, nails and other ectodermal elements. Hypohidrosis was assessed by patients' reports of unexplained fever attacks that are unresponsive to antipyretics, flushing with warm temperature, and intolerance to hot temperatures (35-40 • C), as well as clinical examination of perspiration patterns with emphasis on armpits, palms, soles, and other focal areas (craniofacial, buttocks, back, groin, and breasts).
Detailed oral and dental examination of the patients and parents was performed. Panoramic dental radiographs were used to assess dental phenotype. Oligodontia describes the agenesis of six teeth or more while hypodontia describes the agenesis of fewer than six teeth, excluding third molars [30]. Clinical evaluation ensured that patients met the inclusion criteria for ED and differentiated between syndromic and NSTA. Other associated dental anomalies, such as conical teeth, were also recorded. Conical teeth denote teeth exhibiting a sharply pointed crown or incisal edge [31]. All patients were referred for hearing loss testing.

DNA Extraction and Whole Exome Sequencing (WES)
Genomic DNA was extracted from peripheral blood samples of all participants using QIAamp DNA Mini Kit (Qiagen, Hilden, Germany). Quality and quantity of DNA samples of patients were assessed using fluorometric Denovix Qubit™ dsDNA BR Assay Kit (Ther-moFisher, Waltham, MA, USA). DNA samples from 10 patients were sequenced using the Twist Human Core Exome Plus kit (Twist Bioscience, San Francisco, CA, USA) and NovaSeq 6000 system (Illumina, San Diego, CA, USA) according to the manufacturer's protocol. Libraries were prepared in paired-end mode (2 × 100 bp) for an output of 6 GB per sample, and an average coverage of 50X. Sequencing reads were demultiplexed using Illumina bcl2fastq (2.20) and adapter sequences were trimmed using Skewer (version 0.2.2) [32]. The quality of the generated FASTQ files was analyzed with FastQC software (version 0.11.5; Illumina, San Diego, CA, USA). BAM files were generated using Burrows Wheeler Aligner (BWA) by aligning reads against the GRCh37/hg19 reference genome. Variant calling files were generated using previously described pipeline [33].

Variant Segregation
Sanger sequencing was used to confirm that prioritized variants segregated consistently among parents and available family members according to the predicted mode of inheritance. We designed primers targeting TSPEAR exons which harbor the filtered variants of interest using Primer3 tool [47], Table 1. PCR was carried out as previously described [48]. Reactions were sequenced according to manufacturer's recommendation using the Big Dye Termination kit (Applied Biosystems, Waltham, MA, USA), and ABI Prism 3500 Genetic Analyzer (Applied Biosystems, Waltham, MA, USA). Variants were named based on Human Genome Variation Society nomenclature recommendations [49]. The standards of the American College of Medical Genetics and Genomics (ACMG) were used to classify the level of variant pathogenicity, i.e., pathogenic, likely pathogenic, variant of unknown significance (VUS), benign, or likely benign [50]. In silico tools were used to predict the potential impact of the missense variants on Tspear structure. Since Tspear had no PDB (Protein Data Bank) crystal structure, we retrieved a predicted protein structure model covering the entire Tspear amino acid chain generated by AlphaFold [51] from the Uniprot database (Uniprot ID: Q8WU66) [52]. Two different in silico tools were used: PREMPS and Missense3D [53,54]. PremPS predicts the effect of missense variants on protein stability via calculation of the change in the unfolding free energy (∆∆G) for each mutated protein, i.e., the changes in Gibbs free energy between folded and unfolded states of the protein. PREMPS also shows the location of the mutated residue, i.e., on the surface or core of the protein, and provide the predicted 3D structure of the mutated protein with emphasis on the changes in the types of bonding at the mutated residue [53]. Missense3D predicts the mutated protein 3D structure to identify changes in conformational features including changes in cavity volume, buried versus exposed states of target residue, changes in charge and hydrophobicity, and changes in relative solvent accessibility (RSA) [54].

Comparison of Phenotypic Variabilities among TSPEAR Cohorts in Relation to Ethnicities
We compared clinical phenotypes and dysmorphic facial features of our studied cohort of patients to previously published cases that were similarly identified to harbor biallelic TSPEAR disease causing variants [24,25,[27][28][29]55,56]. Patients were grouped according to their reported ethnicities into: North African, Middle Eastern, European and others. The involvement of different ectodermal elements in ED phenotype of these cases was also compared. Data are summarized as frequencies and percentages for categorical variables. Chi-squared test for independence (χ2) was used to compare each clinical feature across different ethnicities. Statistical analyses were performed using IBM SPSS Statistics for Windows, version 19.0. (IBM Corporation, Armonk, NY, USA). p-values < 0.001 were considered statistically significant.

Clinical Features
Ten patients descending from eight unrelated families of Egyptian origin were recruited. All families showed parental consanguinity except family 3, see Figure 1. Clinical data are summarized in Table 2. Patients showed normal sweating except P5, who had hyperhidrosis of palms and soles. The involvement of teeth was consistent among the entire cohort in the form of oligodontia except P7, who showed total agenesis of all teeth, see Figure 2. Conical-shaped anterior teeth were observed in all patients with the exception of P7, as she was completely edentulous, and P8, as she lacks any anterior teeth. Retained deciduous teeth and delayed eruption of teeth were found in two patients (P1 and P10), see Figure 3. None of the parents and family members showed ED-related features except the mother of P1 and the parents of P2, who had hypodontia milder than their respective probands or any other probands, see Figure 2.
Patients showed common characteristic facial features: broad forehead, short philtrum, prominent and broad nasal root, broad nose, low set ears, and thick lips, see Figure 4A. Thick and everted lips, particularly the lower lip, were common oral findings: thick lower lips were found in eight patients and everted lower lips were also observed in eight patients, see Table 2. Some patients had malar hypoplasia (P2, P4, P6 and P7), see Figure 4A. Common hair-related features were scalp hypotrichosis, more prominent on the anterior part of the scalp, and high anterior hairline, except for P2, who had a normal hairline, and density, see Figure 4A,B. Five patients had hypotrichosis or atrichia of eyebrows (P5-P9). Dysplastic nails were observed in four patients (P1, P2, P8 and P9), see Figure 4C. Hyperkeratotic skin was observed in two patients (P1 and P5), and P1 had severe palmoplantar keratoderma, see Figure 4D. Uncommon findings were skeletal abnormalities in P1, delayed motor and mental milestones in P6, and decreased salivary flow in P7.

TSPEAR Variants' Identification, Segregation and MAF
Exome analyses of the ten studied ED patients identified four different novel TSPEAR variants (NM_144991.3, NP_659428.2), see Figure 5. P1 had a homozygous frameshift deletion (c.44delC; p.(Gly17Alafs*34)) in exon 1 that substitutes glycine residue with alanine and is predicted to create a frameshift of 33 amino acids. The resulting mRNA is predicted to undergo NMD by NMDEscPredictor tool [46], see Figure 6A. Variant segregation identified the heterozygous form of c.44delC in mother of P1, see Figure 5. The variant was not found in any of the large population databases (dbSNP, 1000G and gnomAD), nor in our in-house database of 55 Egyptian exomes. According to ACMG guidelines, c.44delC is classified as a "pathogenic" variant, see Table 3.   Note: Polyphen-2 score is the probability that a substitution is damaging. MutationTaster2 provides the probability of the prediction, and a score in case of amino acid substitutions according to an amino acid substitution matrix. PROVEAN score ≤ −2.5 is predicted to be damaging. SIFT scores < 0.05 are considered deleterious or not tolerated. Mutation Assessor ranks the functional impact of missense variants as neutral, low, medium, and high, with scores from 0 to 1, high impact, i.e., deleterious variants are close to 1. CADD score ≥ 20 predicts the missense variant is among the top 1% of the most deleterious substitutions of the human genome. The GERP score values are positive for conserved positions/constrained loci; in the case of multiple base deletions, the highest score among the deleted bases is displayed. ACMG classification is assigned according to levels of evidence (reference in text). N/A: not available. Abbreviations of the in silico tools' names are in text.
P2 had a heterozygous missense variant (c.668C>T; p.(Ser223Leu)) in exon 5 substituting serine for leucine. The variant was inherited from the father, who carries the same heterozygous missense variant, unlike the wildtype mother, see Figure 5. Exome analysis of P2 neither identified other variants in TSPEAR nor in other ED-related genes. MAF of c.668C>T are less than 0.1% in dbSNP, 1000G, and gnomAD, and the minor allele was not found in our in-house database. The variant is classified as "likely pathogenic" according to ACMG guidelines, see Table 3.
P3 and P4 are affected siblings, both of whom had compound heterozygous TSPEAR variants: a missense (c.1423G>C; p.(Gly475Arg)) in exon 9 and an in-frame deletion (c.1788-1790delAGA; p.(Glu596del)) in exon 11. Segregation analysis showed that the missense allele was maternally inherited while the in-frame deletion allele was paternally inherited, see Figure 5. The same in-frame deletion c.1788-1790delAGA (p.Glu596del) was identified in six patients (P5 to P10) in homozygous form and variant segregation confirmed germline inheritance in their families, see Figure 5. MAF for c.1788-1790delAGA are less than 0.026% in dbSNP, and gnomAD, and the minor allele was found in neither the 1000G database nor in our in-house database. The missense c.1423G>C variant has MAF of less than 0.0007% in gnomAD, and the minor allele was not reported in dbSNP, 1000G, or in our in-house database. Both c.1788-1790delAGA and c.1423G>C are classified according to ACMG guidelines as "pathogenic" and "likely pathogenic" variants, respectively, see Table 3.
The four identified TSPEAR variants are located in highly conserved regions in functional protein domains, see Figure 6B. This was also evident in their positive GERP scores, which refers to constrained genomic regions across different species, see Table 3. Moreover, the average CADD score for the two missense variants (c.668C>T and c.1423G>C) was approximately equal to 25, which predicts these variants to be among the top 1% of the most deleterious substitutions of the human genome, see Table 3. Both c.668C>T and c.1423G>C variants are reported in ClinVar; however, no clinical criteria/phenotypes are provided for their entries.

In Silico Predicted Protein Structural Alterations of Missense TSPEAR Variants
The PremPS tool predicted that the two missense (c.1423G>C; p.(Gly475Arg)) and (c.668C>T; p.(Ser223Leu)) variants are destabilizing, with ∆∆G values of 0.53 and 0.9 kcal/mol, respectively. For the (c.1423G>C; p.(Gly475Arg)) variant, the wildtype uncharged Gly475 residue has no side chain, is buried in the protein core, and interacts with His443, His449, Tyr477, Thr495, and Phe496 via polar interactions, see Figure 7A. The mutant Arg475 residue introduces destabilizing conformational changes via its long positively charged side chain; the long aliphatic side chain forms multiple hydrophobic interactions, particularly with His449 and Phe496, and its positively charged amino groups result in increased polar interactions, see Figure 7B,C. Missense 3D software predicted Gly475Arg mutation to increase relative solvent accessibility (RSA) from 0 to 22%, i.e., the buried glycine switches to an exposed arginine, with consequential damaging conformational changes. For the (c.668C>T; p.(Ser223Leu)) variant, the change from wildtype polar Ser223 residue, which is located on the protein surface, to the mutant non-polar Leu223 is predicted to cause loss of polar interaction with Arg228 and the gain of a new hydrophobic interaction with Pro227, thus destabilizing the protein, see Figure 7D-F. Missense 3D predicted a change in RSA from 27.6% in the case of Ser223 to 53.6% for Leu223, and the expansion of the surface pocket by 41.7 Å 3 ; however, predictions did not pass the software's criteria for structural damage (expansion/contraction of the surface pocket volume of ≥ 70 Å 3 , and change in exposed/buried amino acid state where RSA is less than 9% for buried residue and difference in RSA is at least 5%).

Comparison of Clinical Phenotypes and Dysmorphic Facial Features of TSPEAR Cohorts of Different Ethnicities
The clinical features of our TSPEAR-ED cohort (ten patients from eight unrelated families) of Egyptian, i.e., North African origin are detailed in comparison to 28 previously published TSPEAR cases from 22 unrelated families of different ethnicities, see Table 4. Phenotypes associated with biallelic TSPEAR variants can be categorized into: (1) ED with or without TA (57.9%); (2) TA without other ectodermal features (e.g., isolated TA or TA with other non-ED-related features) (21.1%); (3) SNHL without ED features (18.4%); and (4) SNHL with ED, which was reported in only one patient (2.6%). Only four previously reported patients had ED without TA; however, they had conical-shaped teeth, which does not fulfill the criteria of TA; still, it suggests the involvement of teeth [24]. Clinical presentation of dysmorphic facial features showed statistically high significant dependence (p = 0.0001) on ethnic origin. North African as well as Middle Eastern patients were reported to have characteristic dysmorphic facial features in contrast to patients of European origin. The details of these dysmorphic features are summarized in Table 4. The involvement of different ectodermal elements in a total of 24 TSPEAR-ED cases showed that teeth were the most affected ectodermal derivative (82.6%), followed by the hair (78.3%), nails (43.5%) and sweat glands (39.1%), see Table 4.

References
Current study [24,25,[27][28][29] [24] [24,27,55,56] Abbreviations: N/A: not available. * Others includes reports of African American, Ashkenazi Jewish, Asian, Korean, Caucasian or unspecified ethnicities. ** TSPEAR-associated phenotypes refers to the classification of patients as ED (two or more of the four classical ectodermal elements are affected including teeth) or TA (missing teeth with no other ectodermal elements involved) or SNHL (sensorineural hearing loss with no ED features) or SNHL & ED (sensorineural hearing loss accompanied by ED features). *** Ectodermal elements involvement in ED phenotype refers to the ectodermal element/s affected in the subset of patients classified as ED per each ethnicity.

Discussion
Our study expanded the limited reports of TSPEAR variants by adding ten patients descending from eight unrelated Egyptian families to the previously published 28 probands from 22 unrelated families of different ethnicities [24,25,[27][28][29]55,56]. All of our ten reported TSPEAR-ED patients harbored novel TSPEAR variants, and they all presented with TA, scalp hypotrichosis, normal sweating, and common characteristic facial features. Similar to the majority of reported cases with biallelic TSPEAR variants, all of our patients had normal hearing, which did not support the disputed association between TSPEAR and SNHL [57]. Four of seven previously reported TSPEAR-SNHL cases were identified to have mutations in other hearing-loss-associated genes, e.g., GJB2, GJB6, TMPRSS3, and SLC26A4 [24,58]. The other three TSPEAR-SNHL cases were three siblings from one family in which a homozygous frameshift TSPEAR variant segregated with SNHL [25]. The hearing loss expert panel of ClinGen database cited this family as the source of conflicting evidence regarding TSPEAR and SNHL association, since the same TSPEAR variant was identified in two ED families and one NSTA family with normal hearing [24,25,59]. It was suggested that the SNHL phenotype might have reduced penetrance or that TSPEAR variants might not be the monogenic cause of SNHL owing to the locus heterogeneity of SNHL [24].
Our ten reported TSPEAR-ED patients shared characteristic dysmorphic facial features: broad forehead, short philtrum, prominent and broad nasal root, broad nose, low set ears, and thick and everted lips. We showed that TSPEAR-associated dysmorphic facial features vary according to ethnic origin (p = 0.0001), with an emphasis on patients from North African or Middle Eastern origin. The influence of ethnicity on disease phenotype has been described before in rare and common disorders [60,61]. Our reported facies were distinguishable from previously reported patients from other ethnicities, see Table 4. Some features were shared with patients from Middle Eastern origin, e.g., scalp hypotrichosis, which was more prominent on the anterior of the scalp in six of our ten reported TSPEAR-ED patients, has similarly been reported in three Palestinian patients [27]. Everted lips were common oral finding in our patients. Bondarets and McDonald [62] reported that everted lips are characteristic feature of HED and that tooth agenesis results in poor development of the alveolar ridge with a consequent reduction in lower facial height in both HED and NSTA. The marked reduction in facial height was severe enough in HED to cause everted lips, but not in isolated tooth agenesis. Retained primary teeth and delayed dental development seen in patients P1 and P10 were among dental anomalies that have previously been reported to be found in association with tooth agenesis [63,64].
The involvement of different ectodermal elements in the TSPEAR-ED phenotype showed that teeth and hair are affected at higher frequency than nails and sweat glands, see Table 4. The involvement of TSPEAR in both hair and teeth ontogenesis was found to be mediated functionally via Notch signaling, particularly Notch1 [27]. Notch1 is one of the four transmembrane receptors of the highly conserved Notch signaling pathway that is essential for deciding cellular fate via cell-to-cell communication during embryogenesis, and also in maintenance of homeostasis in adulthood [65,66]. The extracellular compartment of Notch receptors transduces cell-to-cell signals by interacting with transmembrane ligands of neighboring cells, e.g., Delta-like and Jagged ligands. Ligand-receptor binding results in proteolytic cleavage of the Notch receptor's intracellular domain (NICD), which translocates to the nucleus, forming a ternary protein complex with the DNA-binding factor (RBP-J) and Mastermindlike-1 (MAML) necessary for downstream transcriptional regulation of target genes [67]. Microarray expression data of primary human keratinocytes following TSPEAR knock down showed down-expression of NOTCH1, and altered expression of genes encoding proteins within NOTCH1 interactome, e.g., CDCA7, DLL1, IGFBP3, and TP63, and/or genes encoding proteins involved in hair and tooth development, e.g., KRT1, KRT10 and WNT10A [27]. Further evidence of TSPEAR-NOTCH1-mediated function was down-expression of NOTCH1 in the epidermis of a skin biopsy obtained from an ED patient bearing the homozygous TSPEAR loss of function (c.1726_ 1728delGTCinsTT, p.(Val576Leufs*38)) variant [27]. It is unknown how Tspear and Notch1 proteins interact or if their interaction is mediated by proteins involved in previously identified ED pathways; thus, the TSPEAR-ED phenotype might prospectively fit into one of the classified ED pathway clusters or propose an additional cluster.
In mice, Tspear is expressed in the enamel, as well as different compartments of the murine hair follicle, e.g., hair bulb, shaft, outer and inner sheaths. Knock down of murine TSPEAR in vitro hair follicle culture decreased hair bulb size, arrested hair growth, increased apoptosis, and reduced Notch1 expression [27]. Murine models support the critical role of Notch signaling in dental stem cell differentiation and enamel formation throughout early embryogenesis [27,68]. On the other hand, Notch signaling was found to be dispensable for initial placodal formation and hair follicle differentiation but indispensable for subsequent cellular differentiation postnatally [69][70][71][72]. The critical role of Notch signaling in tooth and hair development might explain the higher occurrence of teeth and hair manifestations in the TSPEAR-ED phenotype.
While TSPEAR expression in nails, sweat and salivary glands has not been studied, Notch signaling is involved in their organogeneses in mice. Downregulation of Notch1 directs sweat gland differentiation, while its upregulation is required for salivary glands' differentiation and growth and promoting mitosis in the nail matrix [73][74][75]. The ten reported patients herein had ED without hypohidrosis, but only one patient showed decreased salivary flow, which has not been reported in TSPEAR-ED cases before. Relatively lower occurrence of sweat gland involvement in TSPEAR-ED cases might be attributed to the differences in the roles of Notch signaling in these tissues or the differences in the expression or functional role of TSPEAR in their ontogeneses.
Tspear is predicted to function via its two evolutionarily conserved domains, the thrombospondin-like laminin G domain and the seven EARs domain. Based on sequence similarity, Tspear is considered to be apart of a protein superfamily comprising EARs, which form a seven-bladed β propeller domain that presumably acts as a ligand binding domain, see Figure 6C [25,26,76]. Approximately 53% of the 30 TSPEAR disease-causing variants, including our four identified TSPEAR variants, are missense variants, which are predicted to affect one of these Tspear domains [77]. Other types of TSPEAR disease-causing variants include small frameshift deletions predicted to produce mRNA that is NMD susceptible, including the novel (c.44delC; p.(Gly17Alafs*34)) variant identified in ED patient (P1). Two previously reported small frameshift deletions featuring the ED phenotype, (c.38delT; p.(Leu13Argfs*38)) and (c.454_457delCTGG; p.(Leu152Trpfs*29)), were NMD susceptible, while the (c.1505delA; p.(Lys502Argfs*67)) variant, featuring TA, was predicted to escape NMD [24,27].
Interestingly, the mother of P1, a heterozygous carrier of the (c.44delC; p.(Gly17Alafs*34)) variant, showed mild hypodontia in the form of a missing upper later incisor, see Figure 2. Mild phenotypic features of heterozygous carriers of recessive alleles have previously been reported in carrier parents of ED patients harboring EDAR disease-causing variants [48,78], and similarly in a carrier father of a TA patient who carried a TSPEAR missense (c.1877T>C; p.(Phe626Ser)) variant [28]. Still, there is still the possibility that these parents carry a deep intronic TSPEAR variant, or alternatively, a variant of another TA-causing gene. In another proband, P2, a heterozygous likely pathogenic TSPEAR missense variant (c.668C>T; p.(Ser223Leu)) was identified as being paternally inherited. The heterozygous carrier father had hypodontia milder than that of P2, but with no other ED-related features, see Figure 2. The second allele responsible for the ED phenotype could not be identified in P2, and we suspect that the second allele might be outside the WES capture region, e.g., noncoding or deep intronic variant, which necessitates the whole-genome or RNA sequencing methods [79][80][81]. The second allele may be maternally inherited, since the mother of P2 also had mild hypodontia, see Figure 2.
We identified a novel in-frame deletion (c.1788-1790delAGA; p.(Glu596del)), affecting the sixth EAR domain of Tspear through deletion of its 596th glutamic acid residue. This is the first in-frame deletion to be reported for TSPEAR; it was identified in homozygous form in six patients, and in heterozygous form in two other patients. The heterozygous c.1788-1790delAGA allele was found in trans with another TSPEAR novel missense variant (c.1423G>C; p.(Gly475Arg)) affecting the fourth EAR of Tspear, see Figure 6B. Having a relatively large number of phenotypically characterized ED patients from the same ethnic origin harboring the same disease-causing variant might propose a founder effect for c.1788-1790delAGA variant. We have previously shown that the HED mutation spectrum in Egyptians was different from that in other studied cohorts, whether this was due to the genes responsible for the phenotype or the incidence of founder mutations owing to high percentage of consanguineous marriages (~33-35%) among Egyptians [48,82,83]. Moreover, the recurrence of the c.1788-1790delAGA variant in eight ED patients shows the phenotypic heterogeneity of that variant. For example, variable patterns of TA were evident, as P7 had total anodontia (28 missing teeth) in contrast to P10, who had ten missing teeth.
Our four reported TSPEAR variants were found to be rare and located at conserved genomic locations by mining large population and in-house databases, and reviewing GERP and CADD scores, respectively [41,42]. For missense variants, recent advances in protein modelling and machine learning have allowed in silico variant visualization and analysis of the two missense TSPEAR variants, (c.1423G>C; p.(Gly475Arg)) and (c.668C>T; p.(Ser223Leu)). The predicted changes in the protein folding energy (∆∆G values>0.5), as well as the predicted changes in the types of non-covalent bonding between the amino acid residue of interest and the surrounding residues, were evident of destabilization of the mutant proteins. The damaging effect of (c.1423G>C; p.(Gly475Arg)) variant on the Tspear protein conformation is predicted to be more severe than (c.668C>T; p.(Ser223Leu)) variant which can be deduced given the greater difference in size, charge and polarity between glycine and arginine compared to the difference in size and polarity between serine and leucine.

Conclusions
We expanded the TSPEAR mutational spectrum by identifying four novel variants in TSPEAR in ten Egyptian patients featuring TA, scalp hypotrichosis, normal sweating, and characteristic facial features, confirming and expanding the clinical distinction of TSPEAR-ED as an entity, as well as concluding the role ethnicity plays in TSPEAR phenotypic spectrum. Our results do not support the association between TSPEAR and SNHL, and show evidence of complete penetrance of TA in our cohort in contrast to other ectodermal elements. We underscore the power of WES for broadening the molecular spectrum of unidentified ED cases. Nonetheless, WES is limited by its capture region, and in some cases whole-genome sequencing or RNA sequencing might be useful in the identification of deep intronic and non-coding variants. Furthermore, functional characterization of Tspear protein-protein interactions could help in the delineation of the mutational and clinical TSPEAR profiles.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. Whole exome sequencing data are not publicly available due to ethical considerations of patients' confidentiality.