We ascertained three different families affected with oto-dental syndrome, a rare but severe autosomal-dominant craniofacial anomaly. All affected patients had the unique phenotype of grossly enlarged molar teeth (globodontia) segregating with a high-frequency sensorineural hearing loss. In addition, ocular coloboma segregated with disease in one family (oculo-oto-dental syndrome). A genome-wide scan was performed using the Affymetrix GeneChip10K 2.0 Array. Parametric linkage analysis gave a single LOD score peak of 3.9 identifying linkage to chromosome 11q13. Haplotype analysis revealed three obligatory recombination events defining a 4.8 Mb linked interval between D11S1889 and SNP rs2077955. Higher resolution mapping and Southern blot analysis in each family identified overlapping hemizygous microdeletions. SNP expression analysis and real-time quantitative RT–PCR in patient lymphoblast cell lines excluded a positional effect on the flanking genes ORAOV1, PPFIA1 and CTTN. The smallest 43 kb deletion resulted in the loss of only one gene, FGF3, which was also deleted in all other otodental families. These data suggest that FGF3 haploinsufficiency is likely to be the cause of otodental syndrome. In addition, the Fas-associated death domain (FADD) gene was also deleted in the one family segregating ocular coloboma. Spatiotemporal in situ hybridization in zebrafish embryos established for the first time that fadd is expressed during eye development. We therefore propose that FADD haploinsufficiency is likely to be responsible for ocular coloboma in this family. This study therefore implicates FGF3 and FADD in human craniofacial disease.
Otodental syndrome (MIM 166750) is an autosomal-dominant condition characterized by grossly enlarged canine and molar teeth (globodontia), associated with sensorineural hearing deficit (1,2). Ocular coloboma segregating with otodental syndrome has also been reported in one family (3). The phenotype has been described in simplex cases (4,5) and in families of European (1,3,6,7), Chinese (8) and Brazilian descent (9). The underlying aetiology is of ectodermal origin (4,10), but the molecular defects associated with the autosomal-dominant inheritance pattern have not been characterized.
Developmental defects affecting teeth, the ear and the eye either individually or together are very common. Tooth agenesis (missing teeth) is a frequent developmental anomaly of human dentition, occurring in 25% of the population (11); however, otodental syndrome is the only genetic condition causing tooth enlargement. Approximately, one in 1000 children is affected by severe or profound hearing loss at birth or during early childhood (12). The syndromic forms of hearing loss account for 30% of pre-lingual genetic deafness and include several hundred deafness syndromes, with the underlying genetic defect known in only 30 of them (12,13). The incidence of ocular coloboma ranges from 0.5–7 per 10 000 births and has been reported in up to 11.2% of blind children worldwide (14). In some populations, >20% of the cases of coloboma are inherited (15).
The striking dental phenotype of globodontia in otodental syndrome affects the primary (deciduous) and secondary (permanent) dentition and is pathognomonic for the condition (2). The sensorineural high-frequency hearing deficit is bilateral and stable from early childhood. Some patients also exhibit bilateral iris and retinal ocular colobomata. Tooth, ear and eye development are all processes under strict genetic control as revealed by gene mutations that are associated with arrested tooth development (16), deafness (17) and coloboma (14).
In this paper, we localize otodental syndrome to chromosome 11q13 and identify hemizygous deletions as the underlying molecular defect. Using genomic mapping, Southern blot analysis, quantitative expression data and histological analysis, we provide evidence that haploinsufficiency of FGF3 and FADD (Fas-associated death domain) is the likely cause of the otodental and ocular coloboma phenotypes, respectively. These findings provide insight into the molecular aetiology underlying otodental disease.
Some clinical features have already been described for the Brazilian (OD1) (9), British (OD2) (3,18) and Belgian (OD3) (7) families. In addition, here we have compared data with significant intra- and inter-familial differences. Complete penetrance was seen in all affected patients for the dental phenotype which included missing premolars, delayed tooth eruption, abnormal tooth shape, globodontia, multilobular pulp chambers, pulp stones and enamel hypoplasia. Contrary to previous assessments, pathological examination of extracted affected teeth in pedigree OD1 suggested fusion of a number of original teeth rather than true globodontia (Fig. 1A). Other differences between families included marked palate abnormalities, causing narrowing and crowding of the incisor region (Fig. 1B) and micrognathia (Fig. 1C) in many affected individuals from pedigree OD1. Despite detailed ophthalmological assessments, ocular coloboma was seen only in pedigree OD2 in all affected family members (Fig. 1D and 1E). In all pedigrees, the affected patients described partial hearing loss from their earliest memories, and audiometry confirmed high-frequency loss from an early age (Fig. 1F).
Genome-wide linkage search
In order to locate the defective gene, a genome scan was conducted in the OD1 family, using an Affymetrix platform. The average genotype call rate obtained was 91% (range 84–98%) providing approximately 9285 genotypes per individual. This was slightly lower than the expected call rate (>95%); however, the observed concordance rate between the SNP genotypes obtained from repeated samples (II-1 and II-8) was 99.6 and 99.3%, respectively. Parametric MERLIN calculations revealed a single LOD score peak of 3.9 on chromosome 11q13 (Fig. 2). Haplotype sharing of SNPs between family members combined with microsatellite marker analysis (Fig. 3) identified a 4.8 Mb region defined by two recombination events with D11S1889 (unaffected individuals III-1 and III-10) and one recombination with SNP rs2077955 (affected individual II-1). In the OD2 family, we genotyped nine microsatellite markers which identified recombination events with D11S2006 (individual III-4) and D11S2002 (individual IV-2) defining a 20 Mb linked region (Fig. 4). In the OD3 family, we genotyped the same markers and found a segregating haplotype (Fig. 4, inset).
Fine mapping and loss of heterozygosity
In all three families, marker D11S4136 gave genotype data of affected individuals consistent with the appearance of non-inheritance from the affected parent, i.e. they had inherited an allele from the affected parent that was deleted for D11S4136. All other microsatellite markers in the region showed perfect segregation within the family.
The size of the deletion in each family was estimated by loss-of-heterozygosity, using 10 SNPs in the region (Fig. 5A). All three families were heterozygous for SNP rs9666584 in the 5′-untranslated region (5′-UTR) of the FGF4 gene, indicating that both alleles of FGF4 were present, thereby identifying the centromeric boundary of the deletions (position 69299140: March 2006 freeze, http://genome.ucsc.edu/). All families showed non-inheritance of the affected parental D11S4136 allele, implying that this marker was deleted and that the patients were hemizygous at this position. Affected individuals in OD3 were heterozygous for a novel 5′-UTR SNP of FGF3 (rs41408348), delineating the telomeric boundary of the deletion (genome position 69342896), indicating that the maximum size of the deletion was ∼43 kb. In OD1, two SNPs in FGF3 (rs41408348 and rs41538178) and rs1940238 in TMEM16A were hemizygous, but SNP rs3740720 in the 5′-UTR of the FADD gene was heterozygous, indicating the telomeric extent of the deletion (genome position 69731195), inferring that the deletion was a maximum of 432 kb. OD2 had the largest deletion of up to 490 kb encompassing FGF3, C11orf78, TMEM16A and FADD genes flanked by the heterozygous SNP rs11235675 (genome position 69789255) in affected family members.
To confirm the deletions in each family, we carried out Southern blotting (Fig. 6). The FGF3 hybridization signal, detecting a single 9.4 kb EcoR1 fragment encompassing the whole gene, was reduced in affected patients in all three families compared with unaffected family members, indicating hemizygous deletion of the FGF3 gene. When the blots were stripped and re-probed with the FGF4 probe, the hybridization signal detecting a 2.4 kb EcoR1 fragment was similar in all family members, indicating that FGF4 was not deleted, supporting the presence of heterozygous SNP rs9666584 in FGF4. A 7.5 kb Kpn1 FADD signal was decreased only in affected OD2 patients, confirming hemizygosity for FADD in this family.
Position effect and genomic analysis
Evidence in the literature from other disease-associated deletions suggests that the otodental phenotype could be due to a position effect of the deletions (down-regulation or silencing of flanking genes through loss of essential cis elements). To investigate this, we identified SNPs in flanking genes that were heterozygous in genomic DNA and then sequenced them in RT–PCR products from lymphoblast cells lines of affected OD2 patients (i.e. if the SNPs were heterozygous at the transcriptional level, this would imply that the deletion did not have a position effect).
At the centromeric end of the deletion, ORAOV1 (rs1789166) was heterozygous in genomic DNA and in the cell lines, confirming that the deletion in OD2 did not have a position effect on this gene (Fig. 7A). For the other surroundings genes, they were either not expressed in the cell lines (FGF19, FGF4, SHANK2) or no SNPs were present in patients for the analysis (PPFIA1, CTTN). However, since CTTN and PPFIA1 were expressed in the cell lines, we used real-time quantitative RT–PCR (qRT–PCR) to detect transcriptional expression levels of these genes compared with normal cell lines. We found that expression levels of both genes were comparable with the normal cell line (Fig. 7B), hence no position effect by the OD2 deletion on the transcriptional expression of CTTN or PPFIA1 was apparent. The real-time qRT–PCR analysis also revealed that the expression levels of TMEM16A and FADD were reduced, confirming the SNP mapping and Southern blot analyses.
To investigate whether homologous recombination between repetitive sequences could be a cause of these deletions, we analysed the sequence surrounding the deleted regions. We found one low copy repetitive element (REP2) repeat (387 bp in length) between FGF3 and FGF4 and two REP2 repeats between exon 2 and 3 of FGF3 (Fig. 5B). We found multiple copies of REP1 repeats (270 bp in length) between FGF3 and FGF4. REP1 hits with up to 85% homology were also found in the SHANK2 and TMEM16A genomic regions (Fig. 5A). Genome analysis for the presence of Alu repeat elements also identified 16 Alu repeats clustered between FGF4 and FGF3, 12 repeats between TMEM16A and FADD (breakpoint region in OD1) and 66 repeats between FADD and PPFIA1 (breakpoint region in OD2).
From the genomic analysis, we also searched for regulatory elements in the minimal hemizygous region of OD3 that could be involved in cis regulation of FGF3. Four conserved genomic elements (A–D, Fig. 5B) were identified in human, mouse, rat and dog in the region spanning the 5′-UTR region of FGF4 to exon 3 of FGF3. Element A is a 551 bp sequence with 62% identity between species containing 10 evolutionary conserved binding sites: AHR, Hes1, LRF, two NF1 sites, MEF2 and NFMUE1 and three YY1 sites. Element B is a 57 bp sequence with 63% identity and element C is a 99 bp sequence with 61% identity between species, the latter containing conserved E12, E47 and TAL-binding sites. Element D is a 133 bp element with 51% identity between species.
Candidacy of FADD in eye development
Comparing the genes in the deletion regions in OD1 (otodental) and OD2 (otodental + coloboma), we found that the only difference was the additional hemizygous deletion of FADD in OD2. To provide evidence that haploinsufficiency of FADD could be a cause of the coloboma phenotype, spatiotemporal expression patterns of fadd in zebrafish embryos were investigated by in situ hybridization. In zebrafish wholemounts, we observed that fadd expression at the two-somite stage was throughout the embryo (Fig. 8A and B); however, by 24 h post-fertilization (hpf), spatial expression was restricted to the developing brain, the eyes and otic vesicles (Fig. 8C). High levels of fadd expression were seen in the developing eye at the optic fissure (Fig. 8D). By 36 hpf, there was high-level expression of fadd in the midbrain, hindbrain and optic stalks (Fig. 8E). At 48 hpf, the optic fissure was closed but there was still strong expression in the mandibular mesenchyme and the otic vesicles (Fig. 8F). In comparison, a riboprobe to fgf3 showed clear expression at the tailbud stage in the forebrain and rhombomere 4, from where fgf3 is known to induce formation of the otic placode during the earliest stages of ear development (Fig. 8G and H). At 36 hpf, fgf3 expression is seen in the pharyngeal endoderm, otic vesicle, the midbrain–hindbrain boundary and in the retina (Fig. 8I).
Furthermore, to determine whether intragenic mutations in FADD could be a cause of isolated coloboma in other families, we screened a defined cohort of 49 DNA samples (19) from patients with coloboma. We found sequence variants in two unrelated patients. However, these were single-nucleotide substitutions, known to be polymorphic in the population (rs10898853, rs1131677), hence we considered these as not disease causing.
The underlying genetic defect causing otodental syndrome in all families in this study was hemizygous microdeletion at chromosome 11q13.3, a region of the human genome that is particularly prone to recombination (20). There is compelling evidence that homologous recombination between repetitive elements underlies hemizygous deletions in a number of autosomal-dominant disorders (21,22). In silico genomic analysis revealed that this region of the genome has a major repetitive sequence (23) that might explain the deletions. However, until the breakpoints are cloned in each case, the mechanism surrounding the cause of the deletions remains to be determined.
The fortuitous arrangement of overlapping hemizygous microdeletions in the three families gives an insight into the underlying molecular mechanisms of the disease. The smallest deletion of ∼43 kb in the OD3 family only deleted the FGF3 gene (also deleted in the other two larger microdeletions), implying that haploinsufficiency of FGF3 could be the cause of the dental and hearing defects in all three families. Functional studies have shown that FGF signalling is crucial to proper otic placode development, including six ligands (Fgf-3, -4, -8, -10, -16, -19) and three receptors (Fgfr1, -2B, -3) (24–27). Fgf3 expression is required for inductive signals leading to otic placode formation (24,28). Similarly, studies in dental development have implicated at least five ligands (Fgf-3, -4, -8, -9, -10) and three receptors (Fgfr-1, -2, -3) (29,30). Fgf3 is expressed in the dental mesenchyme and in the mouse enamel knot, a localized region of dental epithelium that is thought to control cusp morphogenesis (30). Interestingly, Fgf3 expression is absent in the dental mesenchyme in Runx2 null mutant mice, suggesting that Fgf3 is a direct target gene of Runx2 (31). In humans, RUNX2-dominant mutations cause cleidocranial dysplasia (32), which includes multiple supernumerary teeth, delayed tooth eruption, enamel hypoplasia and enamel pearls, characteristics common with the otodental syndrome phenotype.
Very recently, recessive mutations in FGF3 gene have been reported (Fig. 5C) in a new form of severe syndromic deafness, microtia and small teeth, without eye abnormalities (33). It was proposed that the mutations were loss-of-function alleles and that heterozygous carriers of the FGF3 mutations did not show any obvious clinical phenotype. This report is however still consistent with our findings of otodental disease associated with FGF3 hemizygosity for several reasons. First, since no functional data was reported, it cannot be assumed that the FGF3 mutations are loss-of-function alleles, as they may retain some aberrant functional activity leading to the different phenotype. Although conventionally assumed to be loss-of-function mutations, gain-of-function recessive alleles have been reported. For example, missense and frameshift mutations in the CNGB3 gene result in an enhanced channel function in recessive achromatopsia (34). Similarly, Dejerine–Sottas disease is caused by a PMP22 recessive gain-of-function allele (35). Secondly, it is unclear how comprehensive the clinical examination was in the FGF3 heterozygote carriers. In particular, we note that the phenotype in the otodental families included a milder deafness, in comparison with the profound deafness described for the recessive FGF3 mutations (33). Thirdly, it is possible that the microdeletions we identified may have unmasked a recessive allele of FGF3 on the normal chromosome in the otodental families, reminiscent of hemizygous deletion of the PMP22 gene unmasking a recessive PMP22 mutation causing Charcot–Marie–Tooth disease (36). We have screened the FGF3 coding sequence, splice sites and UTRs in all our families and found no sequence changes or variants that segregated with disease. However, there may still be an abnormality in the regulatory regions of FGF3 on the normal chromosome.
An alternative explanation for the difference in phenotype and inheritance pattern could be that the deletions are having a position effect, i.e. interruption in the control of expression of other nearby genes is causing the phenotype. We excluded a positional effect on several flanking genes (ORAOV1, CTTN, PPFIA1), but not for FGF4 or FGF19, as these are only expressed in early development. FGF4 is known to have a proliferative role in the dental epithelium (29) and FGF19 is expressed in the developing ear (37), and so is it possible that there could be a contribution to the otodental phenotype by a position effect on these genes. To date, no mutations have been identified in either FGF4 or FGF19 for phenotype comparison. Targeted deletion of Fgf4 in mice is embryonic lethal at E4.5, and heterozygote carriers of the targeted allele are phenotypically normal, implying that one copy of Fgf4 is sufficient (38). Furthermore, targeted deletion of murine Fgf15 (homologous to human FGF19) does not result in any otic vesicle abnormalities (39). The conclusion drawn from both of these functional studies is that other factors act in a redundant fashion to compensate, which is documented in FGF signalling pathways (26,27,30). Thus, a position effect by the deletions we identified on the expression of either FGF19 or FGF4 is debatable, at least from the mouse data.
The deletions we identified may also have an effect on the cis elements regulating FGF3 itself. It is interesting to note that an insertional mutation between Fgf3 and Fgf4 loci in Bey mice causes bulging eyes due to lens overgrowth. Fgf3 gene expression in this mutant was upregulated, suggesting that there may be important regulatory elements in the intragenic region that influence Fgf3 activity (40), which could be affected by the deletions in our families. Our analysis has shown that one of the four evolutionary conserved elements we identified (element D) was deleted in all three families, which may have an effect on the cis regulatory control of FGF3 or FGF4. Recently, a new autosomal recessive, non-syndromic deafness locus (DFNB63) without eye or tooth defects (41) was identified which overlaps with the whole 11q13.3 genomic region shown in Figure 5A. It will be important to determine whether mutations in FGF3 or FGF19 account for this phenotype, as it may help delineate the genotype–phenotype association with this genetic region.
A key feature of the OD2 family phenotype is ocular coloboma. The deletion in this family included the novel TMEM16A and C11orf78 genes of unknown function and the FADD gene, which encodes an adaptor molecule that interacts with death receptors upon stimulation by apoptosis. TMEM16A or C11orf78 deletion would seem unlikely candidates for the coloboma defect, because they are also in the deletion region of the OD1 family, with no eye phenotype. FADD is an attractive candidate since its physiological role has been determined by gene targeting in mice (42). Deletion of murine Fadd leads to embryonic lethality at E11.5 and failure to develop eyes, confirming the importance of FADD in eye development. Furthermore, over expression of Xenopus fadd transgene in tadpoles results in either pin-head or small eyes (43). These data suggest that precise control of FADD expression is critical during eye development, supporting the FADD deletion data we observed in the OD2 family and our expression analysis in zebrafish eye development.
Our results clearly demonstrate hemizygous microdeletions as the underlying genetic defect causing oto-dental syndrome, and this suggests that FGF3 haploinsufficiency could be responsible for the dental and hearing defects. Additionally, FADD hemizygosity appears to affect eye development, in particular, its role in coloboma formation. We are currently investigating the specific role of FADD in optic fissure morphogenesis.
MATERIALS AND METHODS
We recruited three families to this study of Brazilian, British and Belgian descent designated OD1, OD2 and OD3, respectively. Informed consent was obtained from all participants in accordance with the Local Research Ethics Committee. The diagnosis of otodental syndrome was established on the basis of clinical, histopathological and audiometric criteria. Affected patients underwent dental examination including radiography. The oral–dental findings of the British and Belgian families were documented using the Diagnosing Dental Defects Database record form (http://www.phenodent.org). All subjects had a full ophthalmic examination of anterior segments and retina. Deafness was reported by all affected individuals, and selected affected individuals had this confirmed by formal audiometry.
Genomic DNA from peripheral blood samples or buccal swabs was isolated by standard methods using a Nucleon kit (Scotlab). The DNA from buccal swabs tended to be of low yield or poor quality DNA as assessed by NanoDrop® spectrophotometry. Therefore, these DNA samples were linearly amplified by multiple displacement amplification (MDA) (44). MDA was conducted using the GenomiPhi DNA amplification kit according to the manufacturer’s instructions (Amersham Biosciences).
We genotyped 11 affected subjects, seven unaffected subjects and two spouses from the Brazilian family using the Affymetrix GeneChip Human Mapping 10K 2.0 Array and Assay kit according to manufacturers’ instructions. Allele-specific hybridization was detected by analysing scanned images of the array using the Affymetrix GeneChip DNA Analysis (GDAS) 2.0 software. Detailed information on specific SNPs was accessible through the linked NetAffx™ Analysis Centre (e.g. physical and genetic maps, associated genes, neighbouring microsatellites, population frequency).
Genotypes were called by GDAS 2.0 software and exported in table form to the ALOHOMORA suite of programs to perform linkage analysis (45). For quality control, we verified sample genders by checking for heterozygous SNPs on the X chromosome. Family relationship errors were detected using the Graphical Relationship Representation program (46). PedCheck was used to detect and remove any Mendelian errors (47). We carried out parametric linkage analysis with MERLIN (48) using a fully penetrant dominant model and with the information of all SNPs on a chromosome simultaneously. Owing to the limitations of MERLIN to large pedigrees, we split the Brazilian family in two appropriate smaller parts.
Confirmation of linkage and haplotype analysis was made by the Marshfield Mammalian Genotyping Service (provided in the public domain by Marshfield Clinic, Marshfield, WI, USA) or in-house using microsatellite markers (49). Following amplification of microsatellite markers, 1 µl of a 1/25 dilution of the PCR product was mixed with 10 µl of formamide/ROX350 and resolved on an ABI 3730xl DNA analyser. The results were analysed using the software GeneMapper v3.5 (Applied Biosystems).
Screening for germline mutations in the positional candidate FGF3, and FADD was undertaken in affected family members. A cohort of 49 DNA samples from patients with coloboma as the main phenotypic feature were specifically screened for mutation in the FADD gene (19). All exons, intron–exon boundaries and UTRs were sequenced. Primers are listed in Table 1. Amplified PCR products were purified by ExoSap-IT methodology (USB Corporation) and then bi-directionally sequenced using BigDye Terminator chemistry on an ABI 3730xl DNA Analyzer. Sequences were aligned and compared with consensus data obtained from the human genome databases (http://genome.ucsc.edu; http://www.ncbi.nlm.nih.gov).
|Exon||Primers||Annealing (°C)||Size (bp)|
|FGF3 gene primers|
|FADD gene primers|
|Exon||Primers||Annealing (°C)||Size (bp)|
|FGF3 gene primers|
|FADD gene primers|
High-molecular-weight DNA (10 µg) was digested with EcoRI or KpnI for 18 h and then fractionated through 0.8% agarose gels. Denaturation and alkali transfer of DNA to Hybond-N+membranes followed standard procedures. Probes for FGF3, FGF4 and FADD were as follows: FGF3, 1090 bp generated from the 3′-UTR-Exon 3; FGF4, 694 bp generated from 5′-UTR-exon 1-IVS1-2; FADD, 1240 bp including exon3-3′-UTR-genomic sequence. Probes were radiolabelled with [α-32P]-dCTP (3000 Ci/mmol) using the Megaprime labelling kit (Amersham Biosciences). Hybridization was performed at 42°C in 6× SSC, 50% formamide, 5× Denhardt’s solution, 1% SDS, 300 µg/ml sonicated salmon sperm DNA overnight. The blots were then washed to 0.2× SSC/0.2% SDS at 65°C, before autoradiography (Amersham Hyperfilm).
In situ hybridization
Wildtype zebrafish embryos were maintained at 28.5°C on a 14 h light/10 h dark cycle, raised in the presence of 200 µm 1-phenyl-2-thiourea (Sigma-Aldrich) to inhibit pigmentation and staged by standard methods (50). A fadd cDNA clone (pGEM®-T Easy) was generated by RT–PCR amplification and cloning of a 449 bp product from zebrafish RNA obtained from whole embryos (forward, 5′-AACTTGAGAAAATCGACACC-3′; reverse, 5′- TCTCTCCACGAGATCAGC-3′) using the Superscript III RT–PCR System (Invitrogen). Digoxigenin-labelled riboprobes to fadd and fgf3 (51) were generated using DIG RNA Labeling Kit (SP6/T7) according to the manufacturer's instructions (Roche).
In situ hybridization was carried out as previously described (52). Briefly, embryos were hybridized overnight in hybridization mix containing 50–200 ng of digoxigenin-labelled probe, 50% formamide, 5×SSC, pH 6.0 (using 1 m citric acid), 0.1% Tween-20, 5 mg/ml torula (yeast) RNA and 50 µg/ml heparin at 65°C. After extensive washes (down to 0.2× SSC), embryos were incubated in 2% blocking agent for 2–3 h and then incubated with 1:5000 anti-DIG-AP antibody (Roche) overnight at 4°C. Detection of the alkaline phosphatase reaction was using the NBT/BCIP substrate. Embryos were washed in PBS, re-fixed in 4% PFA/PBS for 20 min and then mounted in 70% glycerol for photography (Nikon SMZ1500 fluorescent dissecting microscope).
Expression of SNPs in patient lymphoblast cell line
EBV-transformed lymphoblast cell lines were generated from OD2 family members. Cells were grown in suspension in RPMI1640 (10 mm HEPES, 10% FCS, 5% CO2). RNA was extracted from cell pellets using TRIzol (Invitrogen) and cDNA generated using the Superscript III RT–PCR System (Invitrogen) according to the manufacturer's instructions. SNPs that were heterozygous in affected patient DNA were tested for expression and heterozygosity in RNA from the affected patient cell lines by direct sequencing of RT–PCR products.
Quantitative real-time RT–PCR
Total RNA was extracted from a fixed number of cells (6 × 106) from actively growing cultures of lymphoblast cell lines from normal and affected patients using the Absolute RNA Miniprep Kit (Stratagene), which includes a DNase I step to remove any contaminating genomic DNA. Quality of the RNA was determined using an Agilent 2100 BioAnalyzer and RNA concentration determined using NanoDrop® spectrophotometry. Reverse transcription of 1 µg RNA into cDNA was performed using the RT2 PCR Array First Strand cDNA Synthesis Kit (Superarray Bioscience). Real-time qRT–PCR was performed using the RT2 Real-Time™ SYBR Green PCR Kit (SuperArray Bioscience) with pre-validated primer sets (CTTN, PPFIA1, FADD, GAPDH, 18S rRNA). Thermocycling parameters on an Applied Biosystems 7500 system were 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. Samples from each cell line were run in triplicate for each gene and repeated in several independent samples of RNA. The fold change between normal and deleted cell lines was determined using the ΔΔCt method (53) which calculates normalized expression changes relative to GAPDH or 18S rRNA.
Genomic sequence analysis
To identify the location of tandem repeats and Alu sequences in the 11q13.3-deleted regions, Mreps (http://bioinfo.lifl.fr/mreps/) and RepeatMasker (http://www.repeatmasker.org/) programs were used, respectively. Identification of evolutionary conserved elements in the deleted region of OD3 was carried out in rVista (http://rvista.dcode.org/), repetitive sequences were removed with RepeatMasker and conservation of transcription factor-binding sites was carried out in MULAN (http://mulan.dcode.org/).
This work was supported by The Birth Defects Foundation UK (Grant no. 03/05), the Hammersmith Hospital Trust Research Committee (Grant no. 40625) and partially by the German Federal Ministry of Science and Education through the National Genome Research Network (01GR0463). The D(4)/phenodent investigations were partially supported by INSERM, Réseaux de Recherche Clinique et Réseau de Recherche en Santé des Populations, 2003 and GIS Institut des maladies rares, Réseau Français de Génétique Dentaire, 2003–2005. Histopathology of extracted teeth was performed by Dr A.W. Barrett, Eastman Oral and Maxillofacial Pathology Unit, Eastman Dental Hospital, London. The fgf3 cDNA was generously provided by Dr Lisa Maves, Fred Hutchinson Cancer Research Center, Seattle. Professor David Fitzpatrick, MRC Human Genetics Unit, Edinburgh, kindly provided access to the coloboma patient cohort.
Conflict of Interest statement: The first author declares that none of the authors has a conflict of interest.
- in situ hybridization
- genetic linkage analysis
- southern blot assay
- cell line
- dental clinics
- inner ear
- single nucleotide polymorphism
- recombination, genetic
- reverse transcriptase polymerase chain reaction
- fadd protein
- fgf3 protein, human
- eye development