HPPD: A newly recognized autosomal dominant disorder involving hypertelorism, preauricular sinus, punctal pits, and deafness mapping to chromosome 14q31

Authors

  • Srirangan Sampath,

    1. Department of Genetics, LSU Health Sciences Center, New Orleans, Louisiana
    Current affiliation:
    1. McKusick-Nathans Institute of Genetic Medicine, Johns Hopkins University, Baltimore, MD.
    Search for more papers by this author
    • Postdoctoral Fellow.

  • Bronya J.B. Keats,

    1. Department of Genetics, LSU Health Sciences Center, New Orleans, Louisiana
    Current affiliation:
    1. Research School of Biology, Australian National University, Canberra, Australia.
    Search for more papers by this author
  • Yves Lacassie M.D., FACMG

    Corresponding author
    1. Department of Pediatrics, LSU Health Sciences Center, New Orleans and Children's Hospital, New Orleans, Louisiana
    • Division of Clinical Genetics, Department of Pediatrics LSUHSC and Children's Hospital New Orleans, 200 Henry Clay Avenue, New Orleans, LA 70118.
    Search for more papers by this author

  • How to Cite this Article: Sampath S, Keats BJB, Lacassie Y. 2011. HPPD: A newly recognized autosomal dominant disorder involving hypertelorism, preauricular sinus, punctal pits, and deafness mapping to chromosome 14q31. Am J Med Genet Part A 155:976–985.

Abstract

We report on a novel autosomal dominant disorder with variable phenotypic expression in a three-generation family; the major features include hypertelorism, preauricular sinus, deafness, and punctal pits with lacrimal-duct obstruction. We ruled out the involvement of EYA1, SIX1, and SIX5 as candidate genes by direct sequencing of their exons and by SNP-based linkage analysis. Subsequent SNP-based whole-genome genotyping and parametric multipoint linkage analysis gave lod scores >1 at 14q31 (LOD = 3.14), 11q25 (LOD = 1.87), and 8p23 (LOD = 1.18). By genotyping additional microsatellite markers at two of these three loci and using an expanded phenotype definition, the LOD at 14q31 increased to 3.34. Direct sequencing of the gene exons within the 14q31 critical interval and a custom aCGH experiment did not show any pathogenic mutation or copy-number changes. Further sequencing of 21 kb of promoter regions showed a novel polymorphism 1,249 bp upstream from the SELIL start codon that segregated with the disease haplotype. Cloning the novel polymorphism into luciferase reporter constructs resulted in a 20% reduction in the expression levels. The identification of this family with a distinctive clinical phenotype and linkage to a novel locus at 14q31 supports the existence of a new syndrome of the branchial cleft. © 2011 Wiley-Liss, Inc.

INTRODUCTION

Congenital branchial anomalies presenting as sinuses, fistulae, and cysts account for 20–30% of the head and neck malformations seen in the pediatric population [Schroeder et al., 2007]. These anomalies, classified by the cleft and pouch from which they originate, are thought to arise due to incomplete obliteration and persistence of embryonic branchial clefts and pouch structures [Benson et al., 1992]. Preauricular sinuses or ear pits, visible as small depressions on or in front of the anterior margin of the ascending helix, are a common congenital anomaly with an incidence of 0.1–0.9% in the United States [Tan et al., 2005].

Inherited and sporadic forms of preauricular sinuses exist in a number of syndromes, often presenting with variable expression and reduced penetrance [Tan et al., 2005]. Some syndromic forms of preauricular sinuses are frequently associated with renal malformations and hearing defects. Preauricular sinus and hearing loss are the most common features observed in patients with branchio-oto-renal (BOR) syndrome (OMIM 113650 and 610896) and the allelic branchio-otic (BO) syndrome (OMIM 602588 and 608389) [Smith and Schwartz, 1998; Kochhar et al., 2007]. Besides BOR/BO, other syndromes with overlapping phenotypes such as oto-facial-cervical syndrome (OMIM 166780), Townes–Brocks syndrome (OMIM 107480), and branchio-oculo-facial syndrome (OMIM 113620) also manifest varying degrees of preauricular sinus and hearing loss. BOR/BO syndrome is an autosomal dominant disorder with high penetrance and variable expressivity, often within the same family [Heimler and Lieber, 1986; Konig et al., 1994]. Patients with BOR/BO syndrome may have conductive, sensorineural, or mixed type hearing loss which may be stable or progressive with severity ranging from mild to profound [Fraser et al., 1980]. To date, mutations in three genes, EYA1 [Abdelhak et al., 1997], SIX1 [Ruf et al., 2004; Ito et al., 2006], and SIX5 [Hoskins et al., 2007], have been found in patients with BOR/BO syndrome.

Here, we describe a previously undescribed autosomal dominant disorder in a three generation family with a constellation of abnormalities including preauricular sinus, punctal pits, hearing loss, hypertelorism (distance between orbits >97th centile), and abnormal palmar flexion creases that do not fit any of the clinically well-documented syndromes involving the first branchial arch. After eliminating linkage to three known candidate genes, we mapped the disease gene to a novel locus at chromosome 14q31.1–31.3. Mutation screening of all the critical interval gene exons and copy-number analysis did not show any pathogenic mutation. Sequencing 21 kb of promoter regions within this interval identified a novel polymorphism upstream of SEL1L that showed modest effect in luciferase reporter constructs. Given the marginal effect on expression and the fact that this promoter variant was found in control populations we do not consider the SEL1L promoter variant to be causative. We believe that the collection of manifestations seen in this family is unique, leading us to name this novel autosomal dominant syndrome HPPD (Hypertelorism, Preauricular sinus, Punctal pits, and Deafness).

PATIENTS AND METHODS

Clinical Description

A 29-day-old African-American male was admitted at Children's Hospital New Orleans for the evaluation of suspected seizures, an abnormal newborn Brainstem Auditory Evoked Response (BAER) test, and a paternal family history of hearing loss and preauricular pits. There was no consanguinity; the 32-year-old father was profoundly deaf and had one deaf child and one with normal hearing from previous unions, while the 31-year-old mother was healthy as was her older son, a half-sib of the propositus (not included in this study). The pregnancy was uneventful. The propositus was born full term, via normal spontaneous vaginal delivery with a birth weight of 2,970 g and a birth length of 49 cm. On physical examination, bilateral preauricular sinus, bilateral punctal pits, lacrimal duct obstruction, hypertelorism (interpupillary (IP) distance >97th centile), abnormal palmar flexion creases, and penoscrotal inversion (shawl scrotum) were observed (Fig. 1A). In the left palm the transverse distal crease runs across the palm (special form). In the right palm, a bridge between the vertical and transverse proximal palmar flexion creases was observed. Both palms had vertical creases in the fourth interdigital areas (Fig. 1A-d). A cranial CAT scan, renal ultrasound, and KUB were all normal.

Figure 1.

A: The propositus and (B) affected father showing major findings: preauricular sinus (a), punctal pits (b), hypertelorism (c), and crease in the 4th interdigital area (d). A special form of the transverse distal (A-d; Lt), and a bridge between vertical and transverse proximal flexion crease (A-d; Rt), and penoscrotal inversion (shawl scrotum) (A-e) were seen only in the propositus.

Examination of the father showed similar findings to those of the propositus (Fig. 1B) except for the absence of the penoscrotal inversion. Both father and son also had bilateral distal axial triradii. The presence of similar clinical findings in the father and son, and the preliminary family history obtained from the mother, led to the study of this family.

Members of the family were entered into the study after evaluation either in the Ambulatory Care Center at Children's Hospital or during visits to their homes in New Orleans. Several different measurements (OFC, inner canthal, IP and outer canthal distance, ears, etc.) were taken, and any unusual findings were documented by photograph. Ocular hypertelorism was defined as an increased IP distance (>97th centile) for age and OFC. A diagnosis of hearing loss in family members other than the propositus was based on use of hearing aids. About 5–10 ml of peripheral venous blood was collected from 24 family members in EDTA tubes and frozen at −80°C until use. The study protocol was approved by the LSUHSC and Children's Hospital Institutional Review Boards.

DNA Extraction and PCR Analysis

Genomic DNA was extracted from frozen whole blood using the Gentra Puregene Blood Kit (Qiagen, Valencia, CA) and used directly for polymerase chain reaction (PCR) and genotyping applications.

Amplification by PCR was carried out using custom primers designed for all exons of EYA1, SIX1, and SIX5 genes, and for the genes within the 14q31 critical interval (see Supplementary Material and Supplementary Tables I and II online). Primers were placed in the introns and 5′/3′ UTR regions to guarantee sequencing of the whole exons together with their splice junctions. For large exons, overlapping primer pairs amplified exons along with at least 50–100 bp of intronic sequences. To sequence the promoter regions, 3,000 bp upstream from the start codon of the six RefSeq genes and one transcript were PCR amplified using four overlapping primer pairs (see Supplementary Table III online). Sequencing was performed bidirectionally, unless one of the primers failed (owing to the presence of long stretches of homopolymer nucleotides in the UTR and introns). If this happened, sequencing was repeated with the successful primer. None of the exons or promoters failed sequencing completely.

Genotyping

We used a SNP-based genotyping approach for linkage analysis in the three candidate loci: EYA1, SIX1, and SIX5. Genotyping of SNPs was carried out by PCR amplification of 200–300 bp flanking each SNP followed by direct sequencing of the PCR products (primers available on request). The Illumina HumanLinkage-12 panel was used for whole-genome genotyping following the manufacturer's protocol (Illumina, San Diego, CA) (see Supplementary Material online), and microsatellite markers were genotyped in two chromosomal regions (14q31 and 11q25) with lod scores >1.5 (see Supplementary Material and Supplementary Table IV online).

Linkage Analysis

Multipoint linkage analysis was performed using GENEHUNTER v1.2 program built into the easyLINKAGE v5.08 graphical user interface [Lindner and Hoffmann, 2005]. For the three candidate loci (EYA1, SIX1, and SIX5) and whole-genome SNP genotypes, parametric lod scores were calculated using an autosomal dominant mode of inheritance, 90% penetrance, and a disease allele frequency of 0.1%. Preliminary analysis was carried out with a strict definition of affected status. Individuals were considered “affected” if they manifested two or more of the four main phenotypic features (hypertelorism, abnormal palmar variation, preauricular sinus, and deafness); individuals with one phenotypic feature were considered “unknown,” and individuals with no phenotypic features were considered “unaffected.”

A map file containing the SNP markers, their chromosome number, base pair position (NCBI build 36.3), and an interpolated deCODE genetic map position was used. For whole-genome linkage analysis SNPs were analyzed in sets of 50 and 100 markers. Loci showing positive lod scores were further analyzed with a less stringent phenotype definition and by incorporating microsatellite marker information. In this case, individuals were called “affected” if they manifested one or more phenotypic features, and individuals with no phenotypic features were considered “unaffected.” Haplotypes were generated using GENEHUNTER and visualized using HaploPainter v029.5 [Thiele and Nurnberg, 2005].

Copy-Number Analysis

Copy-number analysis was performed on a subset of four individuals (Fig. 2: III-5, III-6, IV-7, and IV-8) using array-comparative genomic hybridization (aCGH). A custom Agilent 4 × 44K oligonucleotide microarray was designed (Agilent Technologies, Santa Clara, CA) with 30,000 oligo probes in 20 Mb (hg18, chr14:73,165,000–93,165,000) spanning the 14q31 critical interval. The probes spanned both coding and noncoding sequences. The median probe spacing was 422 bp, with a minimum and a maximum probe spacing of 192 and 26,280 bp, respectively. The aCGH experiment was performed according to the manufacturer's protocol (see Supplementary Material online) described in Agilent Oligonucleotide Array-Based CGH for Genomic DNA Analysis v5.0 (Agilent Technologies).

Figure 2.

Pedigree with haplotypes of microsatellite and SNP markers at 14q31. The four main phenotypic features are included. Horizontal lines above the symbol indicate individuals who were phenotypically evaluated and the arrow indicates the propositus (IV-8). Individuals whose affection status is unknown are shown with a ? symbol. Haplotype analysis show that the proximal and distal recombination events are at D14S1008 and D14S1033 (compare with Fig. 4). Typing additional promoter markers narrowed the proximal boundary to rs61980901. With the exception of IV-13, individuals with at least one of the four major phenotypic features carried the disease haplotype. The markers and their corresponding genetic positions are shown on the left.

Transient Transfection and Reporter Assay

Transient transfection was done in HEK293 and HeLa cell lines (ATCC) using pGL3 luciferase reporter constructs (see Supplementary Material online). The cell lines were cultured at 37°C (5% CO2) in either 6- or 12-well culture dishes with Dulbecco's modification of Eagle's medium (Mediatech, Inc., Manassas, VA) containing 4.5 g/L glucose, L-glutamine, sodium pyruvate, and supplemented with 10% fetal bovine serum (Mediatech, Inc.) and 100 IU/ml of penicillin/streptomycin (Invitrogen, Carlsbad, CA). The cells were transfected with FuGENE HD Transfection reagent (Roche Applied Science, Indianapolis, IN). Briefly, the cells were plated in triplicate the night before transfection in 6-well (200,000 cells/well) or 12-well (100,000 cells/well) dishes. The following day, 50 or 100 µl (6- and 12-well, respectively) of the transfection complex containing 0.5–1.0 µg of one of the experimental firefly luciferase reporter vectors (pGL3-Basic, pGL3-Basic-A, and pGL3-BASIC-C) and 50–100 ng of internal control vector pRL-TK (Promega, Madison, WI) were combined with FuGENE transfection reagent (3:2) according to manufacturer's protocol, incubated at room temperature for 30 min and added to the cells. pRL-TK containing a Renilla luciferase cDNA cloned downstream of the herpes simplex virus thymidine kinase promoter served as a control for transfection efficiency. The cells were lysed with Passive Lysis Buffer (Promega) 24–36 hr after transfection and assayed for firefly and Renilla luciferase activity (see Supplementary Material online).

RESULTS

Clinical Studies

Figure 2 presents the multi-generational pedigree consisting of 24 members of the family who underwent detailed physical examination by the senior author.

The evaluations of family members, performed over several months, confirmed significant variability in the phenotype. For example, some individuals only had hypertelorism (Fig. 2: II-2, III-8), in contrast to some of their children (Fig. 2: III-1, III-5, IV-10, and IV-15) who had more severe manifestations. Preauricular sinuses were seen in six individuals, including the propositus and his father. In all except one individual (Fig. 2: III-1), the preauricular sinuses were bilateral. Among the six individuals with preauricular sinus, three also had deafness. Deafness also was seen in one additional individual. As the medical records were not available for the family members, we were unable to determine if the deafness was conductive, sensorineural, or mixed. However, all individuals with deafness wore hearing aids. Variation of the palmar flexion creases and the presence of a vertical crease in the interdigital area were seen in many members of the family. Interdigital vertical creases have been observed as a minor manifestation of cutaneous syndactyly (Y.L. unpublished work). Lacrimal duct obstruction and punctal pits were seen only in the propositus and his affected father. The major phenotypic features and additional minor anomalies are presented in Table I. Based on the information provided by the mother of the propositus, there may be other affected individuals who were inaccessible for this study. For example, II-5, the sibling of II-2 (paternal grandmother of the propositus), is the common ancestor of two generations of individuals with hypertelorism as the sole manifestation; however, we were not able to examine her offspring and their children. The pedigree suggests an autosomal dominant pattern of inheritance with variable phenotypic expression.

Table I. Summary of the Major Clinical Features
ID/sexAgeHead CirIPICPreauricular sinusDeafPunctal pitsPalmar creaseOthers
cmCentilemmCentilemmCentileLtRt+/−LtRt
  1. IP, interpupillary distance; IC, inner canthal distance; Lt, left; Rt, right; −, absent; +, present; >, greater than; >>>, much greater than; <, less than; td, tendency to; I2, interdigital area 2; I4, interdigital area 4; t′′, axial triradius in distal position; sc, simian crease; R2F, right 2nd finger; R5F, right 5th finger; L5F, left 5th finger; y, years; m, months; numbers in bold indicate hypertelorism.

II-2/F59 y555063>973475–97
II-5/F58 y555062>973150–75
III-1/M40 y56<5065>9737>97+?Telecanthus, facial palsy (Rt), fusiform fingers, lupus
IV-1/F13 y 3 m532–505875–973475–97Café-au-lait spots, hypo-hyperpigmented hemiabdomen
IV-2/F10 y 11 m525058753475–97Disruption of transverse proximalCafé-au-lait spots
IV-3/F8 y 10 m53>5061>973597++Td to special form (Rt), single palmar flexion crease, vertical crease in I4Café-au-lait spots, curly eyelashes
IV-4/F6 y 3 m48<255>9733>97Vertical crease in I4Almond shaped, asymmetric upslanting palpebral fissures, curly eyelashes, and sparse distal eyebrows
III-5/M43 y 8 m565079>>>9746>>>97+++++Vertical crease in I4 (bilateral), t′′Telecanthus
IV-5/M17 y565069>>>9741>>>97Retinal degeneration (mother has cataracts)
IV-6/M12 y54506075–9732>50Cubitus valgus
IV-7/F10 y5450–9864>>>973475–97+++Flat nasal bridge, epicanthal fold, cubitus valgus
IV-8/M1 m36<5047>972575–97+++++Vertical crease in I4, td special form, bridge between the vertical and transverse proximalBregma 5 × 5 >2SD, penoscrotal inversion
III-8/M30 y57>5063>9733>75High palate
IV-9/M10 y522–505775–9732<75Epicanthus
IV-10/F10 y512–505575–9728>25+Extra crease in R2F, td vertical crease in I4Epicanthus, bluish sclera, trait for sc anemia, minor cubitus valgus, neuroblastoma
IV-11/F9 y51<5062>>>973275–97Short 4th metacarpal
IV-12/F7 y 2 m502–505575–972975–97Vertical crease in I2 (Lt and Rt)Bluish hue in sclera
IV-13/M1 y 7 m45<254>>>9731>97Disruption of transverse distal, td I2 (Rt), mesobrachydactyly R5FMicrocephaly, bluish sclera
IV-14/F4 m415057>>>9730>97?Telecanthus
IV-15/F4 m415049>9729>97++?Clinodactyly L5FFlat nasal bridge (coumarin-like phenotype)
III-2/F 56>5066>9736>97
III-6/F32 y55506075–973475–97
III-7/F 5650–9865<973475–97
III-9/F 5650–9868>>>9739>>>97

Molecular Studies

Evaluation of BOR/BO syndrome candidate genes

Phenotypic features seen in this family such as preauricular sinus, deafness, and lacrimal duct obstruction suggest a possible diagnosis of BOR/BO syndrome. Thus, EYA1, SIX1, and SIX5 were evaluated as possible gene candidates by PCR-based amplification and sequencing of their exons. No pathogenic mutation was found in any of the exons of the three candidate genes. Additionally, parametric linkage analysis of intragenic SNPs gave negative lod scores at all three loci (data not shown), thus ruling out linkage to these regions, and suggesting that a mutation at a novel locus underlies the phenotype in this family.

Whole-genome genotyping and linkage analysis

Following SNP quality control, 5,673 genotypes were analyzed by parametric linkage analysis. For the linkage analysis, we excluded II-5, as none of her progeny underwent phenotypic evaluation. In addition, individuals with two or more phenotypic features were designated as affected (Fig. 2: III-1, III-5, IV-3, IV-4, IV-7, IV-8, IV-10, and IV-15) and individuals with one phenotypic feature were designated as unknown (those in generation IV were not included in the analysis), while IV-1, IV-6, and IV-9 and all married-ins were identified as unaffected. Parametric multipoint linkage analysis yielded three loci with lod scores >1.0 (Fig. 3). The largest (3.14, P = 0.004), was found on 14q31.1–q31.3; the two other regions were 11q25 (1.87, P = 0.03) and 8p23 (1.18, P = 0.004).

Figure 3.

Genome-wide parametric multipoint linkage analysis. Multipoint lod scores (y-axis) are shown for markers on each of the 22 autosomes (x-axis). The size of each chromosome in cM is shown at the top. The vertical lines in the chromosome panel delineate the series of SNPs analyzed in sets of 100 each. The solid horizontal line in the y-axis shows a lod score of 1.0. Markers were also analyzed in sets of 50 and 200 markers with similar results.

Subsequent linkage analyses were performed using only SNPs that spanned the candidate intervals at 14q31, 11q25, and 8p23 (33, 8, and 40 SNPs, respectively). For this analysis, individuals with at least one phenotypic feature were considered affected (Fig. 2: II-2, III-1, III-5, III-8, IV-3, IV-4, IV-5, IV-7, IV-8, IV-10, IV-11, IV-12, and IV-15). Individuals IV-1, IV-6, IV-9, and all married-ins were identified as unaffected. Because of GENEHUNTER restrictions on the number of individuals that can be included in an analysis, three individuals in generation IV with only one phenotypic feature (IV-2, IV-13, and IV-14) were excluded from this analysis. At 14q31 the lod score increased slightly to 3.3 (P = 0.0005); at the 11q25 locus the lod score increased to 2.8 (P = 0.01) while linkage at 8p23 was excluded (results not shown).

Positive linkage results at 14q31 and 11q25 loci were followed up by genotyping microsatellite markers. Linkage analysis was repeated with microsatellite genotype data using the broad definition of the phenotype. The lod score at the 11q31 locus reduced from 2.8 to 0.95 (P = 0.05), whereas at 14q31 the lod score of 3.3 (P = 0.0003) was similar to that obtained with the SNP markers. Additionally, on chromosome 11 no genes were found distal to the proximal recombination event (153 cM), suggesting the exclusion of this locus as a potential candidate. Haplotype analysis of the 14q31 locus identified proximal (80.3 cM) and distal (86.54 cM) recombinant events showing a critical interval of 8.3 Mb (Figs. 2 and 4A). The 8.3 Mb region was relatively gene poor with only seven known genes (NRXN3, DIO2, TSHR, GTF2A1, STON2, SEL1L, and FLRT2) and one gene of unknown function (C14orf145) (Fig. 4B).

Figure 4.

A: Multipoint linkage analysis using microsatellite markers at Chr14q locus. The suggestive recombination events are at markers D14S1008 and D14S1033, giving a critical interval of 8.3 Mb. The microsatellite markers on the x-axis are arranged relative to each other (left to right; proximal to distal) and their corresponding genetic positions are tabulated on the right. The relationship between the genotyped SNPs and the microsatellite markers is also shown. Note that the lod scores at the proximal and distal recombination events did not reach −∞ even though the markers were fully informative because incomplete penetrance (90%) was assumed. The physical map of the critical interval (hg18; chr14:78,971,665–87,400,000) between markers D14S1008 and D14S1033 is shown below (B) along with the genotyped markers, the SEL1L −1249 variant, and genes.

Candidate gene evaluation

All exons of seven known RefSeq genes and one unknown protein coding gene within the linkage interval were sequenced prioritized in the following sequence: FLRT2 (OMIM 604807), GTF2A1 (OMIM 600520), SEL1L (OMIM 602329), C14orf145, TSHR (OMIM 603372), NRXN3 (OMIM 600567), STON2 (OMIM 608467), and DIO2 (OMIM 601413). We examined for the presence of missense, nonsense, splice-site mutations and indels causing a frame-shift leading to loss of protein function. No pathogenic mutation in the exons or splice junctions was found in any of these genes.

Copy-number analysis

Deletions or duplications in the intergenic or intragenic regions could explain the absence of any mutation in the coding sequences of genes within the critical interval. Thus, a copy-number analysis was performed using custom oligonucleotide aCGH analysis spanning the 14q31 critical interval. Four individuals in the family (Fig. 2: III-5, III-6, IV-7, and IV-8) were evaluated for the presence of heritable copy-number changes. We did not find any segregating copy-number change in the four individuals that were analyzed.

Promoter analysis

Mutations in the promoter region can affect gene expression levels, and thus may explain the pathogenesis of the phenotype seen in this family. We sequenced regions of more than 3 kb upstream from the start codon for all genes within the critical interval, leading to the identification of a novel polymorphism (NM_005065.4:c.−1249A>C) 1,249 bp from the SEL1L start codon (hg18, chr14:81,071,090). In addition, two previously described SNPs (rs74064461 and rs61980901) were segregating in this family. The SEL1L −1,249 bp C allele (ss184956559) segregated with the disease haplotype. Subsequent haplotype analysis incorporating the rs61980901 genotypes narrowed the candidate interval, moving the proximal recombination boundary distal to the GTF2A1 gene (Fig. 4B).

In order to determine the frequency of the SEL1L −1,249 bp C allele in the control population, we sequenced 66 African-American and 96 Caucasian samples. One African-American and three Caucasian samples were heterozygous and none were homozygous, giving C allele frequencies of 0.008 in African-Americans and 0.016 in Caucasians. We tested the novel SNP for regulatory potential by cloning 1,631 bp of the SEL1L 5′-flanking region (hg18, chr14:81,069,833–81,071,472) upstream of a firefly luciferase reporter construct. We generated two constructs (pGL3-BASIC-A and pGL3-BASIC-C) for transient transfection experiments. The C allele resulted in an approximately 20% reduction in transcriptional activity in both HeLa and HEK293 cells (see Supplementary Fig. 1 online).

DISCUSSION

The family reported here manifests a distinctive set of phenotypic features with significant variability among different members that has not been previously documented. The prominent features are hypertelorism, preauricular sinus, and hearing loss. Thirteen individuals had hypertelorism, not counting the two individuals (Fig. 2: III-2 and III-9) who married into the family. Among these 13 individuals, 7 had hypertelorism as the only manifestation, 2 of whom (Fig. 2: II-2 and III-8) had children with additional manifestations. This suggests that hypertelorism may be the mildest expression of the causative mutation.

The presence of preauricular sinus and hearing defects and variable expressivity within the family is suggestive of a differential diagnosis of BOR/BO syndrome, although the presence of hypertelorism is new. So far, mutations in three genes, EYA1, SIX1, and SIX5 have been found in patients with BOR/BO syndrome, though this only accounts for 60% of the BOR/BO patients. Direct sequencing of the coding exons and SNP-based linkage analysis eliminated the involvement of all three candidate genes. This suggests that a new candidate gene/locus underlies this constellation of manifestations. Given the overlap in the phenotype seen between our family and BOR/BO syndrome, it is possible that mutations within the EYA-SIX gene network [Rebay et al., 2005] could underlie the manifestations seen in our pedigree. In addition, paralogs of EYA and SIX are ideal candidates for mutation screening as genes with overlapping functions and expression patterns or gene products that interact with each other could lead to the development of similar disease phenotypes [Hoskins et al., 2007]. Conversely, these candidate gene approaches to mutation screening would fail to identify any new genes that have not been previously characterized. Therefore, we employed an unbiased approach to gene mapping by combining positional information from whole-genome linkage analysis with functional candidates.

Using SNP-based genome-wide linkage analysis we mapped the disease gene to a novel locus at 14q31. Microsatellite marker genotyping confirmed this result and narrowed the critical interval to an 8.3 Mb region. All but one of the individuals with only one phenotypic feature, such as hypertelorism or palmar variation, carried the disease haplotype (Fig. 2); the one exception is individual IV-13, who did not inherit the disease haplotype from her father (III-8), but may have inherited her hypertelorism from her hyperteloric mother (III-9). The 8.3 Mb interval is a gene desert containing only seven annotated genes and one unknown protein coding gene. None of the genes within the critical interval implied any functional overlap with EYA or SIX proteins or served as biological candidates based on paralog or protein–protein interaction mapping. Neither sequencing of the exons and the overlapping splice-donor and splice-acceptor sites of all genes within the critical interval nor a copy-number analysis of the 20 Mb region surrounding the critical interval showed any mutations, deletions, or duplications. However, a novel variant (NM_005065.4:c.−1249A>C) was identified within the SEL1L promoter sequence. The SEL1L promoter C allele segregated with the disease haplotype and all affected individuals were heterozygous for the polymorphism. Unaffected individuals within the family were homozygous for the A allele.

SELIL is a human ortholog of C. elegans sel-1 that was first identified through a genetic screen for mutations that suppress lin-12/Notch family of proteins [Grant and Greenwald, 1997]. SEL1L, cloned as a pancreatic specific gene [Biunno et al., 1997] is highly conserved across several vertebrate species [Biunno et al., 2002] and encodes a 794 amino acid protein with multiple domains. Combined information from in situ hybridization and immunohistochemical experiments in humans suggests that in addition to its high expression levels in the pancreas, SEL1L is also expressed at varying levels in many human fetal tissues at different gestational stages, including the neuroepithelium, neural-tube, branchial epithelium, gastrointestinal tract, urogenital tract, cartilage, skeletal and smooth muscles [Biunno et al., 2006]. SEL1L expression in many embryonic tissues suggests its involvement in growth and differentiation of several tissues. In fact, homozygous deletion of Sel1l in mice leads to impaired growth and differentiation of pancreatic epithelium possibly mediated through its negative regulation of Notch signaling [Li et al., 2010]. Mounting evidence also points to the role of SEL1L in endoplasmic reticulum (ER) homeostasis by facilitating the removal of misfolded or unfolded protein from the ER lumen into the cytosol for proteosomal degradation [Mueller et al., 2006; Cattaneo et al., 2008]. Despite SEL1L expression in the embryonic branchial epithelium, its function in the growth and development of branchial structures has not been studied.

The SEL1L promoter region was shown to be highly specific to pancreas with modest activity in other cell types [Cattaneo et al., 2001a]. Using luciferase reporter constructs, sequence variants within the SEL1L proximal promoter region (−366 and −354 bp relative to start codon) have been shown to influence transcript levels. The presence of the C allele at −366 and −354 base positions resulted in a 40% increase in the expression of luciferase [Cattaneo et al., 2001b]. In our luciferase reporter constructs, the presence of the C allele at the −1,249 base position resulted in a ∼20% reduction in expression level relative to the A allele. Given that all individuals in this pedigree were homozygous for the T allele at −366 and −354 base positions, our reporter constructs also carried the T allele at these two positions. Such a small effect on expression level could be due to the HeLa and HEK293 cell lines used in our study. Alternatively, different cell types could be sensitive to minor changes in the SEL1L dosage. The −1,249 bp A nucleotide is situated in a stretch of 11 homopolymer A nucleotides, and the A to C transversion creates an additional copy of a hepta-nucleotide (CAAAAAA) simple tandem repeat polymorphism (see Supplementary Fig. 2 online). The −1,249 bp A to C transversion did not alter or create any cis-acting elements based on searches in TRANSFAC databases. Given that the C allele was also found at low frequencies in control populations, and has a relatively minor effect on expression levels, it is probably not the causative mutation, but may be in linkage disequilibrium with it. If this is the case, extensive sequencing around this promoter variant is likely to identify the mutation. Potential targets for sequencing would be the SEL1L intronic sequences and sequences 5′ and 3′ to the gene. Recently, massively parallel next-generation sequencing has been used to identify rare variants by partitioning the genome using hybridization-based capture of all human exons [Ng et al., 2010] or by the targeted capture of large genomic regions identified by linkage [Rehman et al., 2010; Volpi et al., 2010]. The fact that we did not find a mutation in any of the transcribed sequences within the critical interval suggests that a next-generation sequencing study would most likely find additional variants in the noncoding intronic or intergenic regions, which could be regulatory, but demonstrating biological causation is likely to be complicated. Alternatively, the role of SEL1L in the development of branchial structures and in the pathogenesis of the phenotype seen in this family can be studied in mouse models by analyzing its expression in branchial epithelial cells and in tissues of clinical relevance.

The constellation of phenotypic findings in this family, including hypertelorism, preauricular sinus, deafness, and punctal pits has not been previously reported and we believe represents a new autosomal dominant syndrome. This is particularly likely because the disease locus is linked to a region of the genome that has not been implicated in syndromes with phenotypic similarities, although a causative mutation in this region could not be identified. We hope that other clinicians will report similar patients, thus confirming this new clinical entity.

Acknowledgements

We thank the members of the family for their participation and collaboration with this study; Dr. Floyd Buras for referral; Dr. Regina Zambrano and Sammeta V. Raju for their help with collecting blood samples; Dr. San San Ng (LSUHSC genomics core facility) for microsatellite markers genotyping, Dr. Diptasri Mandal for assistance with the linkage programs, and Dr. Wanguo Liu for support with promoter sequencing and luciferase assays. We appreciate Kelly Allerton's assistance with editing.

Ancillary