The T-box gene family encodes a large family of transcription factors with more than 20 members identified in humans so far, and homologues in many other organisms. A number of human disorders have been linked to mutations in T-box genes, confirming their medical importance. They include Holt– Oram syndrome/TBX5, Ulnar-Mammary syndrome/TBX3, and more recently DiGeorge syndrome/TBX1, ACTH deficiency/TBX19 and cleft palate with ankyloglossia/TBX22. This review describes the key features of these disorders and the involvement of T-box genes in their phenotype.
The eponymous T mutation, or Brachyury, was originally identified in the 1920s but the gene underlying the mutation was not cloned until about 60 years later (1). The existence of other genes similar to T was recognized when a region of homology was identified between T and a Drosophila gene, omb (2). The homologous region was a DNA binding domain, styled the T-box, which was used to identify a family of T-box genes in the mouse (3). Homologues of existing genes, and new family members, have been identified in most organisms examined so far (4,5). Alignment of T-box sequences shows that the family is an ancient one, and has arisen due to gene duplication and cluster dispersion (3,6,7).
The key role played by members of this gene family during many aspects of development has been demonstrated by the generation of targeted T-box gene deletions in the mouse. mTbx6 is important for the decision of paraxial mesoderm to follow a mesodermal or neuronal pathway (8), and eomesodermin is necessary for proper mesoderm formation and recruitment into the primitive streak (9). Mutations in Brachyury affect the development of posterior mesoderm (1), reviewed in (10), and similar phenotypes occur when T homologues in other species such as zebrafish (11) and Drosophila (12) are mutated.
Recent papers show that haploinsufficiency of mTbx1 can recreate many aspects of the DiGeorge Syndrome (DGS, OMIM 188400) phenotype in mice (13–15). DiGeorge is one of a number of syndromes associated with deletions and translocations of 22q11. They are common in humans, affecting about 1 in 4000 live births, and involve more than 80 different reported birth defects (16,17). The main cause of death is a congenital heart defect affecting the pharyngeal arches. Hypoplastic or aplastic thymus and parathyroids are also found, resulting in defects in cell-mediated immunity, and hypoparathyroidism. There is some facial dysmorphism, learning difficulties and behavioural problems. The phenotype can be highly variable between patients with the same deletion (18).
Many of the 22q11 genes in the region of conserved synteny in mouse chromosome 16 have been investigated as potential DiGeorge candidate genes. Tbx1−/− mice show virtually all testable features of DGS (13), but in heterozygous mice the phenotype is less penetrant. This may indicate a difference in dosage sensitivity, or penetrance, between mice and humans, also other genes may be involved in generating the human phenotype. No mutations have been identified in TBX1 (14,19). Patients with clinical features consistent with DGS but without the 22q11 deletion, and patients with cardiac abnormalities commonly seen in DGS, were screened for TBX1 mutations (20). Eight common polymorphisms, and 10 rare variants were identified. In the majority of the rare variants there was no amino acid change, and most had been transmitted from an unaffected parent. Regulatory and intronic regions of TBX1 have not yet been screened to complete the analysis but at this stage it seems unlikely that isolated mutation of TBX1 causes DiGeorge syndrome.
The phenotype of DiGeorge patients is highly variable although ∼80–90% share a heterozygous 3 Mb deletion and a further 8% have a nested deletion of 1.5 Mb. The deleted regions are flanked by low-copy repeats which explains their genetic homogeneity (18,21,22). The penetrance of the phenotype may depend on non-genetic factors as well as different genetic backgrounds and modifier genes. Reduced penetrance of cardiovascular defects is observed in a mouse model of DGS (23,24). The penetrance of the cardiovascular defects varies on different genetic backgrounds and is probably due to the presence of genetic modifier(s) (25). A possible modifier of Tbx1 is Fgf 8 (26). Tbx1 is expressed in the pharyngeal endoderm, as is Fg f 8, but this Fg f 8 expression is abolished in Tbx1−/− mutants (26,27). Also, an Fgf8 mutant mouse phenocopies DGS. This is the first gene outside of the deletion region to do so (28). When Tbx1+/− mice were bred with Fg f 8+/− mice double heterozygous mutants had a higher penetrance of aortic arch defects than Tbx1+/−;Fg f 8+/+ mutants and also had more severe abnormalities. Tbx1+/+; Fg f 8+/− mice were normal (26).
Many of the structures disrupted in DiGeorge syndrome are derived from the neural crest during development; bones of the skull, face and palate mesenchyme, stroma of thymus and parathyroids, and heart outflow tract (29). Ablation of premigratory crest cells results in a phenocopy of 22q11 deletion syndromes (30). This suggests that DGS may be caused by abnormalities in the migration, survival, proliferation or differentiation of neural crest cells. The migration of neural crest cells does not appear to be affected by deletion of the DGS region in mice, or by a lack of Tbx1 (27,31). Tbx1 is not expressed in neural crest cells in the pharyngeal apparatus but may regulate a signal inducing differentiation of post migratory cells, or giving directional information. Fg f 8 is not expressed in neural crest cells, but neural crest cell death occurs when Fg f 8 expression is lost from developing pharyngeal arches (32). Thus Tbx1 may regulate expression of Fgf 8 to ensure correct differentiation of neural crest in the pharyngeal area, making Tbx1 responsible for a subset of DGS features.
Ulnar-mammary syndrome (UMS, OMIM 181450) is caused by mutations in TBX3, see Figure 1 (33–35). It affects the ulnar ray of the limb with phenotypes ranging from hypoplasia of the terminal phalanx of the fifth digit, to the complete absence of forearm and hand (36). Patients with UMS also have abnormal development of breasts, teeth and genitalia (37,38). Male patients typically have delayed onset of puberty (39). TBX3 is expressed in a wide range of tissues including those where developmental abnormalities occur in UMS (3,34,40). There does not appear to be a correlation between TBX3 mutations and UMS symptoms, suggesting that haploinsufficiency of TBX3 causes UMS.
TBX3 has a role in specification of posterior limb mesoderm and in setting up the dorso/ventral limb axis. In some cases of UMS posterior structures are lost or duplicated, and in others the ventral surface of posterior digits becomes dorsalized (35). Breast, tooth and genital development all rely on inductive interactions between epithelial tissue and underlying mesenchyme (41–43) and TBX3 may have a common role in these developmental processes.
TBX5/HOLT– ORAM SYNDROME
Holt–Oram syndrome (HOS, OMIM 142900) is an autosomal dominant disorder affecting ∼1 in 100 000 live births. It is completely penetrant with a highly variable expression and causes both cardiac and skeletal congenital abnormalities. The skeletal abnormalities affect the forelimb (Fig. 2), and include clinodactyly, limited supination, sloping shoulders and phocomelia. They affect the radial ray and are bilateral and asymmetrical, affecting the left side more severely than the right. Defects observed in the heart affect the conduction system, atrial and ventricular septation and tetralogy of Fallot. A high proportion of patients have an absence of the pectoralis major and an ocular defect (44).
Mutations in TBX5 have been identified in both familial and sporadic cases of HOS, see Figure 1 (45–50). The majority of mutations in TBX5 are null alleles that cause the HOS phenotype by haploinsufficiency. They cause both severe cardiac and skeletal phenotypes. Phenotypes resulting from missense mutations affect the heart and limbs differently. Missense mutations at the amino terminus of the DNA binding domain cause severe cardiac but milder skeletal abnormalities. In contrast mutations at the C-terminal end of the T-box cause only mild cardiac defects but severe skeletal ones (47). It has been shown that neither G80R nor R237Q, missense mutations affecting the amino and C-terminal ends of the T-box respectively, is able to bind to a TBX5 target site identified by successive rounds of selection from an oligonucleotide pool (51). Thus the mutant proteins may exert their effects via altered interactions with cardiac or limb specific co-factors. In an investigation of the synergistic transcriptional activation of the ANF promoter by Nkx2.5 and TBX5 (52) G80R is unable to activate the promoter either alone or with Nkx2.5, whereas R237Q behaves almost like-wild type TBX5. Thus, the severe cardiac abnormalities in G80R patients may be explained by the lack of interaction with Nkx2.5, via the TBX5 N-terminus, during heart development. It is possible that TBX5 interactions with limb specific proteins are disrupted in R237Q patients.
A mouse model of Holt–Oram syndrome has been made (53). Heterozygous Tbx5del/+ mice show both cardiac and forelimb defects, having enlarged hearts with atrial septal defects and conduction defects as well as hypoplastic falciformis bones in the wrist and elongated phalanges in the first digit of the forelimb. The genetic background of the deletion plays an important role with regard to phenotype suggesting that modifier genes may have a significant role in determination of phenotype.
Studies of gene expression in the Tbx5 heterozygotes confirm earlier work showing that ANF is downstream of mTbx5 (51,52). Another target of Tbx5 is connexin 40, expression of which is dramatically reduced in the Tbx5del/+ heterozygotes. Cx40 is important for the conduction of electrical impulses throughout the heart (54,55) which may account for observed cardiac conduction defects. Other gene transcripts reduced in mTbx5 deletion homozygotes include Mlc2v, Irx4, Hey2, Nkx2.5 and Gata4.
Limb defects in HOS range from subtle digit abnormalities to phocomelia suggesting a role for TBX5 in forelimb growth and patterning. Since patients have malformations of the forelimb rather than transformations into hindlimb this must involve regulation of anterior forelimb structures. Much work has been done in the chick to investigate the role of Tbx5 in specifying the identity of the forelimb (56–59). Misexpression of Tbx5 in the leg results in wing-like skeletal pattern and can also give rise to truncated limb phenotypes similar to those seen in HOS. Recently work in zebrafish has suggested that not only is Tbx5 important for determination of forelimb identity, but is also required to initiate formation of the pectoral limb bud (60). This has also been demonstrated in the mouse (M. Logan, manuscript in preparation). A conditional knockout of Tbx5 in the limb mesenchyme results in stillborn pups with no forelimbs, and lacking shoulder elements (Fig. 3).
TBX19/ISOLATED ACTH DEFICIENCY
Tbx19, also known as Tpit, was identified as required for expression of pro-opiomelanocortin (POMC) in the corticotroph and melanotroph pituitary lineages (61). Expression is restricted to the ventral aspects of the anterior pituitary and the ventral diencephalon (61,62).
In humans the absence of pituitary POMC leads to a lack of adrenocorticotrophin (ACTH) resulting in adrenal insufficiency (OMIM 201400). TBX19 (63) was screened in patients with this phenotype and seven different mutations were identified (Fig. 1) (61,64). Isolated ACTH deficiency is transmitted recessively; in all cases, except for one compound heterozygote, the patients are homozygous for a TBX19 mutation whilst their unaffected parents are heterozygous carriers (64). The missense mutations identified have severely reduced ability to bind their DNA target and activate transcription whilst the other mutations generate prematurely terminated transcripts. This suggests that the TBX19 mutations described are loss-of-function alleles.
Tbx19 activates the POMC promoter synergistically with the homeodomain protein Pitx1 (61); this activation requires both transcription factors to be bound to the DNA. Ectopic expression of Tbx19 in cells expressing high levels of Pitx1 is sufficient to activate POMC, suggesting that it may act as a signal to initiate POMC cell differentiation (61). Tpit-deficient mice model isolated ACTH deficiency (64) and should facilitate our understanding of differentiation of the pituitary corticotroph lineage.
Cleft palate affects ∼1 in 1500 births and is caused by failure of the paired palatal shelves to make contact and form the midline seam which later disappears, allowing the palate to become confluent. Cleft palate causes problems with feeding, speech, hearing, dental and also psychological development, and requires corrective surgery. It is usually sporadic and involves both genetic and environmental factors (65,66). However, cleft palate with ankyloglossia (CPX, OMIM 303400) is a semi-dominant X-linked condition mapping to Xq21 in which environmental factors play little part (67–70).
TBX22 was identified on Xq21 during the human genome project (71,72). It maps to the CPX critical region and is expressed at the correct time and in the appropriate tissue for palatogenesis (73–75). Screening of individuals with CPX has identified eight point mutations in TBX22 including splice site, missense and nonsense mutations, see Figure 1 (74,75). These segregate with the disorder in their respective families confirming TBX22 as the causative gene for CPX.
A possible target for TBX22 is TGFβ3. TGFβ3 homozygous deletion knockout mice have cleft palate but no other craniofacial abnormalities, suggesting that TGFβ3 directly affects palate closure (76). Also, work in the chick has shown that Tgfβ3 can induce fusion in a palate that is naturally cleft (77). Thus, TBX22 may act in a pathway directing the transformation of medial edge epithelium into mesenchyme during formation of a continuous palate.
The five T-box genes discussed above play significant roles in inherited human disorders. In addition to this, some evidence points to a possible role for TBX2 in cancer. TBX2 maps to 17q23 which is frequently altered in ovarian carcinomas (78), and is amplified in about 20% of breast cancers (79). TBX2 is amplified in some breast cancer cell lines, with about 4.5% of sporadic breast cancers having a dosage increase equivalent to one extra copy of the TBX2 allele. A screen designed to identify immortalizing genes in breast cancer showed that TBX2 affects the p53 tumour suppressor pathway by repressing the cdkn2a (p19ARF) promoter (80) implying a role for this T-box gene in cancer. An independent screen identified TBX3 as able to repress the same p19ARF promoter (81) demonstrating the involvement of T-box genes in cell proliferation.
A wide range of mutations have been identified in T-box genes, some of which generate a null allele resulting in haploinsufficiency. Missense mutations may affect interactions with cofactors; an interaction between TBX5 and Nkx2.5 has been demonstrated (51–53) and TBX19 acts synergistically with Pitx1 to activate the POMC promoter (61). It is reasonable to suppose that other interactions between T-box genes and their cofactors will be identified in the future.
Another potentially disease-causing gene, TBX15, is a candidate for acromegaloid facial appearance (OMIM 102150) (82). It will be interesting to see whether it is the causative gene for this disorder, and whether any other human diseases are found to be caused by T-box gene mutations.
The authors' work is supported by the British Heart Foundation. We would like to thank Ruth Newbury-Ecob for Figure 2, Malcolm Logan for use of Figure 3 and Jacques Drouin for sharing TBX19 mutation data.
To whom correspondence should be addressed. Tel: +44 1158493217; Fax: +44 1159709906; Email: firstname.lastname@example.org