The Tbx-files: The truth is out there



T-box factors are critical regulators of embryonic development and have been implicated in several human diseases. This primer describes the basics of how T-box factors work and features a discussion of the state of T-box gene research with three experts in the field. Developmental Dynamics 231:232–236, 2004. © 2004 Wiley-Liss, Inc.

What are T-box genes?

They are a family of genes that encode transcription factors with a conserved DNA binding domain, the T-domain. These factors bind to specific sequences on DNA, named T-sites.

How did the T-box gene family get its name?

The namesake of this gene family, T (tailless) or Brachyury, is one of 18 family members in mammals. T heterozygous mice are viable but have short tails, whereas homozygotes are embryonic lethal due to a lack of mesodermal structures (Herrmann et al., 1990).

Is the family evolutionarily conserved?

There are T-box genes in metazoans, including primitive chordates (ascidian, amphioxus), nonchordate deuterostomes (sea urchin, acorn worm), protostomes (Caenorhabditis elegans, Drosophila), and diploblasts (hydra).

How do they regulate gene expression?

T-box proteins can function as transcriptional activators (Kispert et al., 1995; Zhang and King, 1996; Plageman and Yutzey, 2004) or repressors (Carreira et al., 1998; He et al., 1999; Chen et al., 2004; Plageman and Yutzey, 2004). They bind either as a monomer (Casey et al., 1998) or as homodimers (Muller and Herrmann, 1997; Papapetrou et al., 1997) to the T-site. The consensus site for Brachyury is T(C/G)ACACCTAGGTGTGAAATT, where the underlined sequence is a palindrome of the T half-site (Kispert and Herrmann, 1993). Brachyury orthologs (Casey et al., 1998; Takahashi et al., 1999) and verified T-sites for other T-box factors bind close sequence variations of the T half-site (Carreira et al., 1998; Papapetrou et al., 1999; Hsueh et al., 2000; Goering et al., 2003). The binding sequence can be configured as a single half-site, or as two half-sites in palindrome: head-to-tail, or tail-to-tail. Specific sequences and T-site configurations are preferentially bound by different T-box factors in vitro (Conlon et al., 2001; Goering et al., 2003).

Where are they expressed?

T-box factors often display dynamic expression patterns in multiple cell and tissue types throughout development. One example is Tbx22, which is expressed in developing muscles of the head, limbs, and body wall, and also in mandible and maxilla craniofacial mesenchyme (Fig. 1). Expression patterns of different T-box genes often overlap. Both Tbx2 and 3 are expressed in the anterior and posterior margins of hindlimb and forelimb buds, overlapping with Tbx4 in the hindlimb and Tbx5 in the forelimb (Gibson-Brown et al., 1996). Coexpressed T-box factors can modify one another's function in a combinatorial, additive, or antagonistic manner (Amacher et al., 2002; Goering et al., 2003). Therefore, domains of overlapping expression may reveal regions where a given T-box protein functions differently than when it is expressed alone.

Figure 1.

Whole-mount in situ hybridization analysis of Tbx22 expression during chick development. A: Lateral view, anterior is up, Hamburger and Hamilton (HH) stage 18. Expression is seen in paraxial mesoderm, myotome, tongue muscle precursors (arrowhead), frontonasal mesenchyme (arrow), and in the mandibular and hyoid arches. B: Head region, dorsal is up, HH stage 26. Expression is seen in ring muscles around the eye, palatal maxilla (arrows), and mandibular processes (arrowheads). Tbx22 expression is conserved in chick (Haenig et al., 2002), mouse, and human (Laugier-Anfossi and Villard, 2000; Braybrook et al., 2001). Expression encompasses craniofacial mesenchyme, regions important for development of the disease phenotype observed in X-linked cleft palate and ankyloglossia. e, eye; fl, forelimb, h, heart; hy, hyoid arch; mb, first (mandibular) arch; my, myotome; pm, paraxial mesoderm; spc, spinal cord; tel, telencephalic vesicle. Reproduced with permission from the authors and Elsevier (Haenig et al., 2002).

What do they do?

T-box factors have been shown to regulate patterning and cell fate (Gibson-Brown et al., 1996; Zhang and King, 1996; Rodriguez-Esteban et al., 1999; Takeuchi et al., 1999; Good et al., 2004) or cell movement (Kimmel et al., 1989; Ahn et al., 2002). One idea is that, generally, T-box factors specify regional “tissue identity” by means of regulation of cell fate and behavior. For example, Tbx4 and Tbx5 may specify regional limb identity, because they are required for bud development and outgrowth in the hindlimbs and forelimbs, respectively (Ahn et al., 2002; Naiche and Papaioannou, 2003; Rallis et al., 2003). Similarly, zebrafish spadetail mutants are deficient in trunk mesoderm, as a result of aberrant migration of trunk paraxial mesoderm progenitors (Kimmel et al., 1989), whereas other mesoderm develops normally. T-box factors tend to directly regulate genes that control patterning and differentiation of cell types in which they are expressed. Xenopus VegT activates derriere and Xnr1, both TGFβ family members that control mesoderm patterning and formation (Hyde and Old, 2000; White et al., 2002). Mouse Tbx2 represses melanocyte-specific trp1 (tyrosinase-related protein), and Ciona Ci-Bra activates the notochord-specific Ci-trop (tropomyosin-like; Carreira et al., 1998; Di Gregorio and Levine, 1999). Based on these findings, T-box factors may specify regional tissue identity by coordinating expression of patterning and cell type-specific genes.

What is the relevance to humans?

Several genetic human diseases are associated with mutations in T-box genes. Haploinsufficiency of TBX3 leads to ulnar–mammary syndrome (UMS, OMIM 181450), affecting limb, apocrine gland, tooth, and genital development (Bamshad et al., 1997). Holt–Oram syndrome (HOS, OMIM 142900), a developmental disease characterized by cardiac and limb abnormalities, is caused by mutations in TBX5 (Basson et al., 1997; Li et al., 1997). Patients homozygous for TBX19 mutations have adrenal insufficiency (OMIM 201400; Lamolet et al., 2001), and X-linked cleft palate with ankyloglossia (forked tongue; CPX, OMIM 303400) is caused by mutations in TBX22 (Braybrook et al., 2001). DiGeorge syndrome (DGS, OMIM 188400) is often associated with deletions in 22q11.2, a region that encompasses TBX1. Patients display variable phenotypes, including heart defects and facial dysmorphism. Whereas TBX1 most likely contributes to DGS (Jerome and Papaioannou, 2001; Lindsay et al., 2001; Merscher et al., 2001; Piotrowski et al., 2003), penetrance of the phenotype may also depend on additional components such as environmental factors or genetic background (Gong et al., 2001).

Do we understand the etiology of T-box factor associated disorders?

In all the syndromes mentioned above, the causative gene is normally expressed in tissues that are affected in diseased individuals (Fig. 1). Therefore, disease phenotypes are most likely caused by misregulation of direct transcriptional targets of T-box factors. This has been shown to be the case in two disorders. In tbx19 mutants, a direct target, pituitary pro-opiomelanocortin, is not expressed, thereby resulting in adrenocorticotrophin and adrenal deficiency (Lamolet et al., 2001; Liu et al., 2001). In addition, missense mutations found in patients with HOS yield a truncated TBX5 that cannot interact with its cofactor, NKX2.5. As a result, atrial natriuretic factor and connexin 40, both direct targets of TBX5 and NKX2.5, are not transcribed, thus contributing toward aberrant cardiomyocyte differentiation (Bruneau et al., 2001; Hiroi et al., 2001). These examples illustrate how basic and patient-oriented research can be combined to understand how T-box factors work.

What is next?

Developmental Dynamics posed this question to three scientists whose main focus of research is examining aspects of T-box genes in development: Jim Smith, Professor of Developmental Biology and Chairman of the Wellcome Trust/Cancer Research UK Gurdon Institute, University of Cambridge, Jeremy Gibson-Brown, Assistant Professor, Department of Biology, Washington University, and David Grunwald, Associate Professor, Department of Human Genetics, University of Utah. The scientists discuss the state of research in the field, and emerging themes that deserve further study. An extended version of this discussion can be viewed at the Supplementary Material page available at

Developmental Dynamics:

Given that the sequences of T-domains are highly conserved and that some T-box genes have overlapping expression patterns, how do you think functional specificity is achieved?

Illustration 1.

Jim Smith:

Interacting proteins are really important in defining specificity, I bet. For example, Nkx2.5 and Tbx5. I think we should all be looking for differences in what interacts with Tbx4 and Tbx5 in regulating limb development.

I also think we should be concentrating more on the structures of T-box proteins, including regions outside of the T-domain.

Jeremy Gibson-Brown:

I think functional specificity is most likely conferred by trans interactions with cofactor proteins. After all, it is the non–T-box domains that have evolved most divergently.

David Grunwald:

Jim already pointed out that one level of specificity is due to synergistic interactions with other factors—I think he showed this first. One reason that so many T-box mutants are haploinsufficient may be due to synergistic interactions with cofactors. Our work is developing the idea that one common partner may be other members of the T-box family.

We have started to look widely at endogenous targets—by looking for large numbers of genes that are down-regulated in T-box zebrafish mutants and then checking if they are direct targets by ChIP (chromatin immunoprecipitation) analysis. We are finding that there is variation among the T-binding sites; that different endogenous T-sites interact with subsets of T-box factors. Previously, Jim and others have found that different arrangements of T-sites affect binding—so cis elements with differing affinities are likely to be one level of distinguishing T-specificity.

I suspect that there is also a quantitative control element—this may be demonstrated by phenotype haploinsufficiency, and we see this in the additive interactions among T-box genes. I think one particularly fun thing to come out of finding the endogenous targets will be to see how levels of expression are regulated.


I think David is absolutely right about this. It also emphasizes the need to understand how T-box gene expression is regulated. A change in level of a factor of two is clearly enough to change completely the way a cell behaves.

Dev Dyn:

In your opinion, what are some exciting ideas that are emerging in the field?


We are interested in the recurrent finding that multiple members of the T-box family are found coexpressed in many developmental fields—they seem to make quilt-like overlapping patterns often—this is true in Drosophila too—so one of our interests is, Why? Are they interacting when coexpressed? I think it is interesting as to whether they are often subdividing fields in a way that is independent of tissue type—patterning hindlimb or forelimb; or in dorsal vs. ventral.


I agree that this is an important issue. Why are there so many of these genes? C. elegans has 20, I think, which is lots. It would be interesting to knock these out one by one.

Dev Dyn:

Do you agree, Jeremy? For example, do you see overlapping expression of T-box genes in the basal chordate amphioxus?


Ahh! This overlapping expression business is undoubtedly due to the conservation of certain cis-regulatory modules between sister genes. Remember, there are two or three vertebrate genes for most amphioxus genes, so these duplicates will initially have had completely overlapping expression patterns and redundant functions. Divergence of these cis modules, and the evolution of novel modules, will be the basis of divergent, nonredundant expression.

Amphioxus gene expression domains tend not to overlap nearly as much as those of vertebrate genes. This finding is not surprising, because the gene duplications that gave rise to the amphioxus paralogs occurred much earlier, before the divergence of humans and Drosophila, and in some cases before the divergence of diploblasts and triploblasts.


I also think that it is dangerous to concentrate too much on one gene family! I draw diagrams of how T-box genes interact, and put them at the middle of all my hierarchies, but there are many other genes out there, and we must see the forest as well as the trees.


I think Jim and Jeremy have pushed the view that T-box factors are gene expression controllers, like any other transcription factor—I am wondering whether they in fact often have a role in regionalization—differentiation of anterior from posterior limb, different heart chambers, dorsal vs. ventral skin, regions of the mesoderm (not tissue types). This may be an overstatement, but I am throwing it out as an idea.


I certainly do not regard T-box genes only as regulators of tissue types, nor even as solely transcription factors. I agree that they must be involved in regionalization (Xbra is a good example), although regionalization may involve the activation of genes, of course.


I see regionalization as merely the process by which different downstream batteries of target genes are activated/repressed in different places (such as the anterior or posterior part of a limb bud). Until functions other than transcription factor functions have been found, I DO regard T-box genes as just transcription factors. There is no intrinsic “leggy” property about Tbx4, for example. That does not in any way reduce their importance or interest.

Dev Dyn:

In your opinion, what are some important questions that remain to be answered?


I think that any biological system involving the regulation of gene expression could have T-box genes involved, and I think one of the exciting emerging areas in this field is the role of T-box genes, and/or their mutations, in the onset and progression of various types of cancer.


Tbx2 is implicated in breast cancer, and of course Colin Goding has found that Tbx2 protein appears to have the potential to bind mitotic chromatin.


There is another area we have not begun to address at all, and it is an important open question for the future. What physiological processes are T-box genes involved in in regulating adult tissues? Because we are all developmental biologists, we are not so interested in this, but we know that most of the genes (except T and maybe TbxT) are expressed in adult tissues. There could be roles in regulating hormone production, germ cell proliferation/differentiation, neurotransmitter synthesis, etc.


Jeremy is right on with the comment on adult functions.


That is a good point, and worth thinking about.