Statistics from Altmetric.com
Treacher Collins syndrome (TCS, MIM *154500) is an autosomal dominant craniofacial disorder characterised by malar hypoplasia, micrognathia, downward slanting palpebral fissures, lower eyelid coloboma, malformed auricles, conductive deafness, and cleft palate. The estimated incidence is 1/50 000, with 60% of the cases resulting from sporadic mutations.1 There is marked phenotypic variability among patients, ranging from perinatal death because of a compromised airway to those that go undetected by medical examination. The gene underlying this condition, TCOF1, was mapped in 19962 and since then mutation detection studies have concluded that: (1) the majority of pathogenic mutations are small deletions and insertions causing frameshifts that are predicted to result in a truncated protein; (2) mutations (both polymorphic and pathogenic) can be found throughout the 25 coding exons of the gene; (3) most mutations are family specific with the exception of a commonly occurring 5 bp deletion in exon 24 (found in approximately 16% of families); and (4) there is no correlation between type and/or localisation of the mutation and phenotypic expression.3–6 Furthermore, the observation that more than 50% of all described pathogenic mutations known to date are clustered in five exons (10, 15, 16, 23, and 24) has led to the hypothesis that these five exons are mutational hot spots, suggesting that any effort to identify mutations in TCOF1 would benefit from testing these five exons before extending the analysis to the rest of the gene.6
The protein encoded by the TCOF1 gene, treacle, has repetitive motifs encoded by exons 7-16, which are subject to phosphorylation by casein kinase II7 and a nucleolar localisation signal on its C-terminus.8 It bears weak similarity to a family of nucleolar phosphoproteins and its precise function remains unknown.9,10 In the present work, screening for TCOF1 mutations in a large cohort of TCS patients from different populations confirms that exons 10, 15, and 16 of the repeat region and exons 23 and 24 of the nucleolar localisation signal domains are mutational hot spots, responsible for more than half of the pathogenic mutations in TCS. The identification of a pathogenic missense change in exon 2 suggests that a functionally important domain of treacle exists near the N-terminus.
MATERIALS AND METHODS
Twelve patients were referred directly to the Centro de Estudos do Genoma Humano in São Paulo, Brazil, and samples from 38 patients evaluated by a clinical geneticist or photos were received at the Johns Hopkins University School of Medicine. Of these patients, six were familial cases that showed linkage consistent with the disease locus at the chromosome 5q3 region11 and 29 had previously been tested for mutations in exons 4-16 through heteroduplex analysis.4
Screening of mutations by SSCP of PCR amplified products was performed as described elsewhere.6 Products displaying a mobility shift were sequenced in an ABIPrism Model 377 (Applied Biosystems) according to the manufacturer's instructions. Paternity was confirmed for sporadic cases where parental DNA was available by means of five highly informative tetranucleotide markers (D3S1754, D5S820, D6S477, D7S821, D12S391) following standard procedures. Segregation analysis with markers flanking the TCOF1 gene at 5q31-34 (D5S434, D5S413, IL9, D5S820, D5S816, D5S209, D5S210, D5S211) was conducted for one family displaying a seemingly autosomal recessive mode of inheritance.
Of the 12 patients referred directly to CEGH, we detected alterations in SSCP pattern in 10. All alterations were confirmed to be pathogenic after sequencing, and one patient was found to have two pathogenic mutations in exon 24. The two remaining patients were subjected to direct sequencing of the 25 coding exons of the gene. One alteration was thus identified, a C→T transition at the splice acceptor region in intron 22 (c.3370-3 C→T). This alteration abolishes a restriction site for MboII, and a 100 chromosome control sample was tested for the presence of this alteration, which was found in 12% of chromosomes. Both patients without an identifiable pathogenic mutation were familial cases, with the first (TCS31) comprising an affected father and son, and the second (TCS37) two affected brothers born to normal parents. In this last case, haplotype analysis showed that both brothers shared the same markers surrounding the TCOF1 locus on both chromosomes.
Patients from JHUSM were sorted into two groups: those who had not been previously screened for mutations (group I) and those who had had exons 4-16 previously analysed by heteroduplex analysis (group II).4 Of the 19 patients in group I, 13 shifts were detected through SSCP. Through sequencing of abnormally migrating fragments, we determined that 11 patients had mutations causing frameshifts and two patients had missense changes. Patient G1568 had a c.149A→G transition in exon 2 (Y50C) and patient G1428 a c.3389 T→A transversion in exon 22 (M1130K). Neither alteration was detected in 100 control chromosomes in the Brazilian population. Patient G1568 is a sporadic case and analysis of parental DNA indicated that the mutation arose de novo, thus suggesting its pathogenic nature. Parental samples were not available for patient G1428. Of the 29 patients in group II, pathogenic mutations could be identified in 11; two of these mutations were located in exons that had been previously screened by heteroduplex (one in exon 4 and another in exon 12). All mutations detected are detailed in table 1.
In a previous study using the same methodology as the one used here, we were able to identify pathogenic mutations in 26/28 (93%) patients through SSCP.6 These findings indicate that the SSCP conditions we are currently using are a good method for mutational screening of TCOF1. Indeed, we detected in the group II sample from JHUSM two pathogenic changes that were not detected by heteroduplex analysis in a previous work.4
In the present study, we were able to identify a pathogenic mutation in 33 out of 60 patients, 10/12 (83%) among CEGH patients and 23/48 (48%) among JHUSM patients. The overall mutation detection rate was 33/60 (55%). Among patients from JHUSM, a much lower mutation detection rate was obtained when compared to that achieved in the group from CEGH. This could be because of less stringent diagnosis among the patients, as there is wide phenotypic variability observed in TCS. Also, since there are some unmistakable cases of TCS which do not display a mutation in the coding region of TCOF1,6 it is possible that mutations in the promoter region or 3` UTR, gross rearrangements, or other pathogenic mechanisms could have different frequencies in these two populations.
In this study, we confirmed our previous observations that five exons accounted for over 50% of pathogenic mutations, with exons 23 and 24 being the main hot spots. As in our previous publication,6 the most frequent kind of alterations were deletions (21), followed by insertions (7), splice mutations (3), nonsense mutations (2), and one missense change. Of all 31 different pathogenic mutations found in 33 patients, 27 are novel and four have already been described. Two of the mutations already described were observed in more than one patient independently: the frequent 5 bp deletion in exon 24 (c.4135-4139delGAAAA) was observed in three patients, and the insertion of a guanine residue in exon 23 (c.3987-3988insG), already described in two unrelated cases,6 was observed in two index patients. One patient (TCS39) had two different pathogenic changes in exon 24 (c.4135-4136insG and c.4156-4157delAA), being the second reported case with two mutations in the same exon. The only other TCS patient with two mutations in the same exon, in this case both in exon 16 (c.2552delA and c.2561delA), was described by Wise et al.4
One of the Brazilian families that did not have an identifiable pathogenic mutation after direct sequencing of the whole coding region of TCOF1 displays a seemingly autosomal recessive mode of inheritance. This is the second instance of a seemingly autosomal recessive form of TCS where the affected sibs share the same haplotypes on chromosome 5,6 even though there was not a haplotype in common shared by the two families. Therefore, TCOF1 could not be exempted from having a causative role in either an autosomal dominant or autosomal recessive mode of inheritance.
One of the novel missense mutations (M1130K) lies in exon 22, in a region of the protein that bears 68.8% identity with the mouse and canine homologues, and causes the substitution of a hydrophilic amino acid (lysine) for a hydrophobic one (methionine in humans and mice, valine in dogs) (fig 1, above). Even though this mutation was not observed in 100 control chromosomes in the Brazilian population, we cannot definitely classify this change as either pathogenic or polymorphic since the control population tested may not reflect the variation found in the American population. The M1130K mutation may be a rare, non-pathogenic change, like others that have already been described.6
The missense mutation in exon 2 deserves special attention. This is only the second pathogenic missense change described associated with TCS among the 75 known to date. The first description of a pathogenic missense change was made by Edwards et al,5 also in exon 2 (W53R). This missense change was regarded as pathogenic because it was not detected among 200 control chromosomes and the proband's parents were not carriers. The authors considered the possibility that, because this change falls close to the end of exon 2, it could interfere with normal splicing.
We describe 27 novel pathogenic mutations in TCOF1 and confirm that exons 10, 16, 15, 23, and 24 are hot spots, accounting for over 50% of all known pathogenic mutations.
The identification of a second missense pathogenic mutation in TCOF1, also located in exon 2, suggests that the N-terminus of the protein is functionally important and less tolerant of changes.
As previously noted,6 the treacle protein tolerates a fair amount of amino acid substitutions in its coding region, especially in the repetitive domains. However, the 55 amino acid sequence encoded by human exons 1 and 2 has 94.5% and 90.9% identity with mouse and canine sequences respectively, being the most conserved region between species9,10,12 (fig 1, below). The two amino acids encoded by exon 2 involved in human pathogenic mutations are conserved in both mice and dogs; furthermore, both pathogenic changes result in the substitution of an amino acid with a short side chain for another with an aromatic side chain. These observations support an important functional domain present at the N-terminus of the protein, and future work attempting to unravel treacle's function or determine interacting molecules should especially focus on this region.
The authors would like to thank the TCS families for their collaboration in this study; Drs Martin Hergersberg, Jan Miertus, Elias Olveira da Silva, Têmis Maria Félix, and Nivaldo Alonso; and Constância Gotto Urbani and Elisângela Quedas for secretarial and technical help. The financial support of FAPESP, CNPq, and Pronex is also gratefully acknowledged. MRPB is supported in part by an International Research Scholars grant from the Howard Hughes Medical Institute. This work was supported in part by NIH P60 DE13078, NIH HD24061, and NIH RR00052 (EWJ).
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.