Results

As shown in table 1, we identified eight common polymorphisms and 10 rare variants. Only one of the common polymorphisms results in an amino acid change (1189C→A, His397Asn). In the mouse, this amino acid residue is a histidine. Of the remaining seven polymorphisms, one is a C→T transversion in the 5′UTR, the second is a base pair substitution in intron 7, and the last five polymorphisms are all single base pair changes in the coding region that do not alter the amino acid sequence. These common SNPs may be useful for association studies. Among the rare variants, we identified three deletions, one insertion, one duplication, and five single nucleotide substitutions (table 1, fig1B). To determine if these variants were low frequency polymorphisms, we screened over 100 normal controls (>200 chromosomes). Only one of the infrequent changes, the C→A transversion at nucleotide 886, was identified in one of the control samples, giving a frequency of ∼0.5%. Interestingly, two control samples had deletions similar to the del(9 bp) at nucleotides 1404-1412. All three 9 bp deletions occurred in a tract of 15 alanines. However, when the sequences were aligned, it was possible to show that each 9 bp deletion occurred in a slightly different region of the alanine repeats. Thus, although the nucleotide sequence in these variants is not identical, the resulting change in the protein sequence is unlikely to be associated with alteration in the protein function. None of the remaining eight rare variants were found in the control samples.

For the rare variants, only two of the five single base pair substitutions are predicted to cause a change in the encoded amino acid. The 1049 G→A changes the Gly at 350 to Asp (G350D) in a patient with AAA. However, the sequence of the mouse at this position is not conserved (the mouse has a single Gly whereas the human has two). The second missense mutation is a C→T transition at nucleotide 1187, resulting in Pro→Leu at amino acid residue 396 (P396L) in a patient with AAA. A proline is found at this position in the mouse. Unfortunately, DNA from the parents of these two patients is not available to clarify whether the changes are de novo. Of the three remaining single base pair substitutions, one is the C→A transversion at position 886 discussed above, one is a silent change (576C→T, R192R), and the last is a C→T transition 39 bp 5′ of the presumed start codon. The patient with the –39C→T (p23 in table 1) also has a 15 bp deletion (1334-1348del) that results in an in frame deletion of five amino acids (GYHPH) starting at codon 445. The five amino acids are highly conserved with 100% identity to mouse and 40% toDrosophila TBX1sequence (CAA76529). Analysis of samples from this patient's parents and his unaffected brother showed that he had inherited the 15 bp deletion from his father. The −39C→T change was found in his unaffected mother. Sequence analysis shows that his normal brother has the same two changes, the −39C→T and the del(15 bp). Patient 23 had clinical findings that are consistent with DGS, including absent thymus with partial T cell insufficiency, low parathyroid levels, minor ear anomalies and complex conotruncal heart defect (double outlet right ventricle with pulmonary atresia, VSD, and ASD).

One of the remaining three variants was the result of a small insertion within exon 9C in a patient (p22 in table 1) with a phenotype suggestive of VCFS (cleft palate, dysmorphic facial features, autism with moderate-severe mental retardation, and developmental delay). The patient has a 3 bp insertion (1371-1372insCAC), resulting in an additional histidine within a run of five. The mouse has five histidines in this region but the sequence immediately flanking the histidines is somewhat variable. Family analysis indicated that the 1371-1372insCAC was transmitted to the patient from her unaffected mother. This variant was not found in 139 control samples. In a second patient (p46), we identified a 30 bp in frame duplication (1399-1427dup30), resulting in a tract of 25 alanines instead of the 15 normally seen in this position. The comparative region of the mouse sequence has 10 alanines. DNA from the patient's family is not available and so we are not able to determine whether this duplication is de novo. Analysis of 139 controls did not detect this change in any samples. Finally, we found an 8 bp deletion in a patient with the diagnosis of DGS. The deletion occurs between nucleotides 1274-1281 and is predicted to cause a frameshift at amino acid 425, resulting in the replacement of the terminal 71 amino acid residues from TBX1 with 187 amino acids of novel sequence. This patient has the hallmark features of DGS, including facial dysmorphism, IAA with a VSD, hypocalcaemia as a result of hypoparathyroidism, absent thymus with virtually no T cells, and developmental delay. The same 8 bp deletion was also observed in the patient's unaffected mother but not in 139 normal controls.

Discussion

Mutation analysis of our cohort of non-deleted patients did not detect any cases with unambiguous loss of TBX1 function. Considering the strong experimental evidence for a major role of TBX1 in the 22q11 deletion syndrome from the deletion studies in the mouse,14-16 our findings are somewhat surprising. In contrast to our results, the majority of mutations inTBX5 in association with Holt-Oram syndrome are predicted to produce haploinsufficiency by nonsense or frameshift mutations within the T box region.12 13 23 Similarly, five of the 10 published mutations in TBX3, found in patients with ulnar-mammary syndrome, are predicted to disrupt the DNA binding domain.24 It is likely that the aetiology of non-deleted DGS/VCFS patients is heterogeneous. Malformations similar to those seen in the 22q11 deletion syndrome have been associated with prenatal exposure to retinoic acids and ethanol, maternal diabetes, and deletions of the short arm of chromosome 10. Therefore, it is possible that mutations showing clear loss of function were not found in TBX1 because they are very rare. Similarly, if loss of function of more than one gene within the 22q11 deletion region were necessary to mimic the 22q11 phenotype, then the chance of such an infrequent event occurring would be quite small.

Alternatively, it is possible that the rare variants that are characterised in this report are responsible for the patients' phenotypes. In particular, the 8 bp deletion that results in a frameshift in the terminal exon, the del(GYHPH), and the 30 bp duplication that causes the expansion of the alanine tract, are good candidates for altering TBX1 function. The unaffected parents of four of the patients were available for analysis and in all cases the variant was present in one of the parents, indicating that the changes are not de novo (table 1). These results would appear to rule against a role for the variant in the patient's phenotype. However, studies in mice show that haploinsufficiency forTbx1 does not result in cardiovascular defects in all animals. Rather there is reduced penetrance, ranging from 19 to 50%.7 14 16 Further, the percentage of mice with cardiovascular anomalies appears to vary depending on the genetic background.8 14 Thus, it is possible that the mutations in human TBX1 might also show variable penetrance.

Seven of the 10 rare variants occur in the last exon ofTBX1, outside the DNA binding domain (fig1B). Based on results of T protein, TBX2 and TBX3, the three proteins that have been examined, the C-terminal domains appear to be necessary for repression and/or transactivation of transcription.25-27 Mutations in the C-terminus distal to the DNA binding domain of TBX3, TBX5, andTPIT have been associated with ulnar-mammary syndrome, Holt-Oram syndrome, and isolated ACTH deficiency, respectively. In these instances, the terminal mutations resulted in phenotypes similar to patients with mutations within the DNA binding domain.13 24 28 The expansion of alanine tracts in nuclear proteins, similar to that seen in patient 46, have been associated with four human genetic disorders: synpolydactyly/HOXD13,29 30cleidocranial dysplasia/CBFA1,31oculopharyngeal muscular dystrophy/PABP2,32 and holoprosencephaly/ZIC2.33 The parents of patient 46 were not available for DNA testing and thus we were not able to determine if this change is de novo. Lastly, as has been seen for other genetic disorders,34 it is possible that the single base change variants affect the correct processing of the TBX1 transcript. SinceTBX1 has restricted expression and is not detectable by RT-PCR in lymphoblastoid cell lines, it will be necessary to develop in vitro assays to determine the effect the rare variants have on the RNA processing and protein function.

In this study, only the exons and splice junctions ofTBX1 were screened. Thus, mutations in the regulatory regions would be missed. In addition to the promoter, there are at least six blocks of sequence greater than 130 bp in the intronic regions of TBX1 that share 75-93% of similarity to the corresponding mouse sequence (fig 1C). These highly conserved regions may harbour regulatory elements. Mutation screening of these control regions will provide more complete analysis. Finally, although TBX1 may be a critical gene for the aetiology of DGS/VCFS, the possibility that haploinsufficiency of more than one gene in 22q11.2 is necessary for the expression of the DGS/VCFS phenotype cannot be excluded at present.