Register for Mechanobiology 2016!

Wnt signalling: variety at the core
Stefan Hoppler, Claire Louise Kavanagh


The Wnt/β-catenin pathway is a conserved cell-cell signalling mechanism in animals that regulates gene expression via TCF/LEF DNA-binding factors to coordinate many cellular processes. Vertebrates normally have four Tcf/Lef genes, which, through alternative splicing and alternative promoter use give rise to a variety of TCF/LEF isoforms. Recent evidence from several experimental systems suggests that this diversity of TCF/LEF factors is functionally important in vertebrates for mediating tissue- and stage-specific Wnt regulation in embryonic development, stem cell differentiation and associated diseases, such as cancer.


Wnt signalling is a conserved molecular mechanism in metazoan animals. This pathway enables cells in these multicellular organisms to converse with each other to coordinate a remarkable variety of cellular processes, such as cell fate, cell proliferation vs differentiation, cell survival vs apoptosis, cell behaviour and migration during morphogenesis (reviewed by Logan and Nusse, 2004). The extracellular Wnt signal stimulates numerous intracellular signal transduction cascades, including the essentially linear canonical pathway, which regulates gene expression in the nucleus (reviewed by Cadigan and Liu, 2006), and what seems to be a network of non-canonical pathways, which regulate many other aspects of cell biology (reviewed by Kohn and Moon, 2005). A hallmark of the canonical Wnt pathway is the stabilisation and nuclear localisation of β-catenin (e.g. Schohl and Fagotto, 2002). Without Wnt signalling, β-catenin is targeted for degradation by the so-called β-catenin destruction complex. Active Wnt signalling disrupts this, which results in β-catenin stabilisation and nuclear localisation (reviewed by Willert and Jones, 2006).

How does nuclear β-catenin regulate gene expression? It is now about a decade since T-cell factor/lymphoid enhancer factor (TCF/LEF) proteins were identified as nuclear-binding partners of β-catenin that mediate the transcriptional regulation of Wnt/β-catenin target genes (Behrens et al., 1996; Molenaar et al., 1996; van de Wetering et al., 1997). This constituted a major breakthrough in our understanding of Wnt signalling. TCF/LEFs form a subfamily of the high-mobility group (HMG)-box-containing superfamily of transcription factors. They were originally identified in the mammalian immune system (see below). However, the discovery of the interaction between TCF/LEFs and β-catenin revealed a much wider role in Wnt signalling in many different tissues.

Strictly speaking, TCF/LEFs are not transcription factors by themselves. Although they bind to a conserved DNA sequence – the Wnt-response element (WRE: C/T-C-T-T-T-G-A/T-A/T) (e.g. van Beest et al., 2000) – via their HMG domain, they must bind to other cofactors to influence transcription. Binding of β-catenin to an N-terminal domain of TCF/LEFs (Behrens et al., 1996; Molenaar et al., 1996; van de Wetering et al., 1997) facilitates assembly of multimeric complexes containing transcriptional co-activators, such as CBP/p300 (Bienz and Clevers, 2003) and BCL9/LGS and Pygo (Stadeli and Basler, 2005), which can activate transcription of target genes.

TCF/LEFs are not only transcriptional activators; without β-catenin they assemble alternative complexes with transcriptional co-repressors, which act as multimeric transcriptional repressors – CtBP (see Brannon et al., 1999), groucho/TLE (reviewed by Daniels and Weis, 2005) and others (reviewed by Willert and Jones, 2006). There are of course exceptions (e.g. Jamora et al., 2003), but as a general rule nuclear β-catenin, and therefore ultimately the canonical Wnt signalling pathway, regulates whether TCF/LEFs function in β-catenin-free complexes as transcriptional repressors (no Wnt signalling) or in β-catenin-containing complexes as transcriptional activators (active Wnt signalling).

Canonical Wnt signalling thus hinges on four protein complexes (Fig. 1): (1) the Wnt receptor complex in the membrane, (2) the β-catenin destruction complex, (3) the nuclear TCF/LEF/β-catenin transcriptional activator complex and (4) the nuclear TCF/LEF transcriptional repressor complex. In uninduced cells, β-catenin destruction complexes assemble and, as a consequence, TCF/LEFs assemble repressor complexes at target genes. When Wnt ligands bind and assemble receptor complexes, β-catenin destruction complexes are disrupted and instead TCF/LEFs form complexes with nuclear β-catenin that generally activate transcription (but see Jamora et al., 2003). This is of course a simplified view of a highly dynamic and complex process (for a review, see Willert and Jones, 2006), but it illustrates how TCF/LEFs are key to regulation of gene expression by Wnt signalling. Note also that β-catenin has nuclear activities independent of TCF/LEFs (e.g. Zorn et al., 1999), and TCF/LEF have functions independent of Wnt/β-catenin signalling (e.g. Carlsson et al., 1993).

The mechanisms outlined above are essentially conserved in all animals. The ancestral TCF that gave rise to invertebrate and vertebrate TCF/LEFs was already able to mediate both transcriptional activation and repression. However, whereas invertebrates tend to have one Tcf gene and rarely display alternative transcripts, vertebrates normally encode four Tcf genes (Tcf-1, Lef-1, Tcf-3 and Tcf-4), and each gives rise to a variety of specialised isoforms. Wnt/β-catenin signalling conveys remarkably specific instructions to different tissues at different stages in embryonic development and in adult stem cell differentiation. Here we examine the evidence that the diversity of vertebrate TCF/LEF isoforms plays an important role in mediating the diverse functions of Wnt/β-catenin signalling in vertebrates.

Fig. 1.

Wnt/β-catenin signalling as a tale of four multiprotein complexes. Two snapshots of the dynamic processes in the Wnt/β-catenin signal transduction cascade that involve multiprotein complexes. (A) In the presence of extracellular Wnt signalling, Wnt receptor complexes assemble in the cell membrane (Complex 1); as a consequence, β-catenin is able to form transcription activation complexes with TCF/LEFs in the nucleus (Complex 3). (B) In the absence of extracellular Wnt signalling, β-catenin destruction complexes assemble (Complex 2), leaving TCF/LEF to form transcription-repressing complexes with transcriptional co-repressors (Complex 4). A, Axin protein; β, β-catenin protein; TCF, TCF or LEF protein; W, extracellular Wnt signal. For further details, see main text, recent reviews (e.g. Cadigan and Liu, 2006; Willert and Jones, 2006) and the Wnt homepage (

The divergent structure of vertebrate TCF/LEFs

An invertebrate TCF consists of four domains (Fig. 2): (1) an N-terminal β-catenin-binding domain; (2) a central domain; (3) a well-conserved HMG DNA-binding domain, including a nuclear localisation signal (NLS); and (4) a long C-terminal tail. This generalised picture of TCF structure is conserved in vertebrate TCF/LEFs. Vertebrate TCF-1E isoforms are remarkably similar in overall domain structure to invertebrate TCFs; other vertebrate TCF isoforms have lost parts of these domains and/or include novel peptide motifs.

Alternative promoters in mammalian Tcf-1 and Lef-1 genes result in the use of alternative downstream translation start sites to produce short TCF/LEF isoforms (ΔNTCF and ΔNLEF) that lack the β-catenin-interaction domain (e.g. van de Wetering et al., 1996; Hovanes et al., 2001). These isoforms are therefore refractory to Wnt/β-catenin regulation and constitutively assemble repressor complexes (e.g. Molenaar et al., 1996; Roose et al., 1999; Hamilton et al., 2001). Indirect evidence indicates that other vertebrate Tcf genes might encode similar inhibitory isoforms (Duval et al., 2000; Shulewitz et al., 2006).

Alternative splicing in the central domain creates further diversity of all vertebrate TCF isoforms apart from TCF-3. In Tcf-1 and Lef-1, this produces transcripts with or without a central alternative exon (van de Wetering et al., 1996; Roel et al., 2003). The equivalent exon is always included in invertebrate TCFs and in vertebrate TCF-3 and TCF-4. However, this exon is differently spliced to the preceding exon and to the subsequent exon in TCF-4 to give rise to isoforms with or without LVPQ and SxxSS motifs (Duval et al., 2000; Pukrop et al., 2001; Young et al., 2002). These motifs appear always to be present in vertebrate TCF-3 (e.g. Molenaar et al., 1996) but do not seem to be present in invertebrate TCFs or vertebrate TCF-1 or LEF-1.

Alternative splicing at the C-terminus in Tcf-1 and Tcf-4 gives rise to isoforms that have a variety of C-terminal tails with or without three conserved regions: the CRARF motif; the RKKKCIRY motif; and proposed binding sites for the CtBP cofactor (e.g. van de Wetering et al., 1996; Duval et al., 2000; Hovanes et al., 2001; Young et al., 2002; Roel et al., 2003). The long TCF-4 isoforms contain all three regions; TCF-1E has only the CRARF and RKKKCIRY motifs; LEF-1 has none of them; and TCF-3 has only the proposed CtBP-binding sites. Invertebrate TCFs contain CRARF- and RKKKCIRY-like motifs, but do not appear to have identifiable CtBP-binding sites; yet, CtBP is able to regulate Wnt signalling in TCF-independent ways in vertebrates and invertebrates (Hamada and Bienz, 2004; Fang et al., 2006). The involvement of CtBP in Wnt/β-catenin signalling might therefore predate the evolution of CtBP-binding sites in vertebrate TCFs.

This molecular variety of TCF/LEF isoforms in vertebrates suggests that they should be capable of mediating diverse functions. We examine below the evidence that this is the case in normal development, stem cell maintenance and immune function.

Establishment of the dorsal development in the early embryo

The clearest evidence for diversity of TCF/LEF protein activity is arguably from analysis of Xenopus development. Wnt signalling normally functions on the prospective dorsal side of the early embryo, where it is required for induction of dorsal development (reviewed by Croce and McClay, 2006). Overexpression of Wnt signalling components (such as certain Wnt ligands or β-catenin) on the opposite ventral side results in the induction of ectopic dorsal development.

TCF/LEFs mediate both endogenous and artificial induction of dorsal development. TCF-3 represses this (Fig. 3): endogenous TCF-3 is required to repress dorsal development and gene expression where Wnt signalling is not activated – for instance on the prospective ventral side (Houston et al., 2002) (see also Fig. 3) – and artificially high levels of TCF-3 expression are able to repress dorsal development and gene expression even on the dorsal side (Brannon et al., 1999). Endogenous Tcf genes, such as Tcf-1, are also required for normal activation of dorsal gene expression and development in the embryo (Standley et al., 2006) (see also Fig. 3). Furthermore, artificial LEF-1 expression on the opposite ventral side is able to induce ectopic dorsal development, which suggests that LEF-1 can act as a transcriptional activator (e.g. Behrens et al., 1996).

Fig. 2.

The structural variety of vertebrate TCF/LEFs. Protein-coding exon structure is shown for a generic invertebrate Tcf gene and generic vertebrate Tcf-1, Lef-1, Tcf-3 and Tcf-4 genes or in HUGO nomenclature, TCF7, LEF1, TCF7L1 and TCF7L2, respectively (from top to bottom, with N-terminal exons on the left and C-terminal exons on the right). Note the four broad domains of TCF proteins: N-terminal domain, central domain, DNA-binding domain and C-terminal tail. Described protein-protein interaction domains and polypeptide motifs that are conserved between invertebrate and vertebrate TCF proteins are indicated between the invertebrate and the vertebrate TCFs. The N-terminal β-catenin-binding domain (BCBD) is marked in green, the interaction domain with the groucho/TLE transcriptional co-repressors in red, the HMG DNA-binding domain in dark blue, the nuclear localisation signal (NLS) in turquoise, the CRARF motif in mid blue and the RKKKCIRY motive in pale blue. Note the central exon coloured yellow, which is an obligatory exon in invertebrate TCFs and in vertebrate TCF-3 and TCF-4, but an alternative exon in vertebrate TCF-1 and LEF-1. Described protein-protein interaction domains and polypeptide motives in certain Tcf genes that are conserved among vertebrates but apparently not with invertebrate Tcf genes are indicated below the vertebrate TCFs. They are the LVPQ and SxxSS motives in the central domain and the C-terminal domain described as a CtBP-binding domain (purple) present in alternatively spliced vertebrate TCF-4 isoforms, but always present in vertebrate TCF-3. The identified SUMOylation sites in LEF-1 (K25) and in TCF-4 (K297) are indicated in grey. Alternative translation initiation sites for TCF-1 and LEF-1 are indicated, which produce ΔN-TCF isoforms lacking the β-catenin-binding domain (BCBD). For further detail, see main text and recent molecular review (Arce et al., 2006).

Closer inspection of the dorsal-inducing activity of LEF-1 reveals its isoform specificity (Gradl et al., 2002). Efficient induction of dorsal development and activation of dorsal gene expression is only achieved by LEF-1 isoforms that contain the central alternative exon mentioned above (Fig. 2). Alternative splicing in the central domain of Tcf-4 is also decisive (Pukrop et al., 2001): only isoforms containing the LVPQ motif (see above and Fig. 2) repress a Wnt target gene.

Mesoderm induction and midbrain patterning

Wnt/β-catenin signalling also functions in a gene regulatory network that is responsible for mesoderm induction (Galceran et al., 1999; Vonica and Gumbiner, 2002; Schohl and Fagotto, 2003). Both TCF-1 and TCF-3 regulate mesoderm induction in Xenopus embryos (Liu et al., 2005). However, whereas TCF-1 is required for mesoderm induction, this is only partially defective in embryos in which TCF-3 expression is suppressed. Rescue experiments show that TCF-1 and TCF-3 have distinct functions in mesoderm induction: TCF-1 is needed for assembly of transcription-activating complexes, and TCF-3 mediates transcriptional repression (Fig. 3). Their different C-terminal domains are not responsible for this functional difference (see discussion of skin differentiation below). Instead, the characteristics responsible turn out to be the short LVPQ and SxxSS motifs, which TCF-1 lacks (see Fig. 2).

Fig. 3.

Diverse functions for Xenopus TCF/LEFs. Required roles for TCF/LEFs in early Xenopus development from fertilisation (left in figure) to gastrulation stages of development (right in figure). Maternal expression of Tcf genes is required for regulating dorsal axis induction, with TCF-3 expression being required for repressing transcription and TCF-1 expression being required for activating transcription. However, TCF-1 is also required in a repressing function and TCF-3 and TCF-4 function partially redundantly with TCF-1 in activating transcription of target genes of dorsal axis development (see Standley et al., 2006). Zygotic expression of Tcf/Lef genes is required for regulating mesoderm induction and subsequent mesoderm patterning, with TCF-3 required as a transcriptional repressor and TCF-1 required as a transcriptional activator for mesoderm induction and TCF-1 and LEF-1 required as transcriptional activators for mesoderm patterning (see Liu et al., 2005).

Fig. 4.

TCF/LEF function in intestine and colorectal cancer. The stem cell population at the base of the crypt in the intestine produces a proliferating progenitor cell population that migrates up the side of the crypt. Upon reaching the crypt-villus junction, cells begin to differentiate and continue moving up the villus until mature cells are shed into the lumen at the tip of the villus. The levels of nuclear β-catenin, and therefore active TCF/LEF, are highest at the base of the crypt in the stem cell and proliferating progenitor population. In this population it is proposed that full-length TCF-4 (activating, green) and ΔNTCF-1 (repressing, red) are the main active TCF/LEFs. However, when mutated Wnt signalling leads to colorectal cancer, ectopic expression of full-length LEF-1 is readily detected. As a result the predominant TCF/LEF isoforms present in colorectal cancer are activating (green) as opposed to repressive (red) TCF/LEFs. The levels of nuclear β-catenin in colorectal cancer are very high because Wnt signalling pathway mutations mean that β-catenin cannot be degraded. Figure modified from Radtke and Clevers (Radtke and Clevers, 2005).

In Xenopus neural development, endogenous TCF-4 is required for normal midbrain and hindbrain development (Kunz et al., 2004). Rescue experiments reveal functional diversity of TCF-4 isoforms: TCF-4 proteins containing LVPQ and SxxSS motifs rescue dorsal midbrain development; isoforms without these have a different activity and restore normal formation of the midbrain-hindbrain boundary (Kunz et al., 2004).

Stem cells in the intestine and colorectal cancer

TCF/LEF-mediated Wnt/β-catenin signalling also plays an essential role maintaining stem cells in the crypts of the intestine, the skin (see below) and haematopoietic stem cells (reviewed by Reya and Clevers, 2005). This is often taken advantage of by cancers, which can manipulate the pathway to maintain undifferentiated cancer cells; a well-studied example of this is colorectal cancer.

TCF/LEF-mediated Wnt/β-catenin signalling is active in the crypts of both the small and large intestine, where it maintains the stem cell population and the proliferating progenitors (Fig. 4). The importance of TCF/LEF transcription factors is confirmed by the fact that loss of Tcf-4 in mice proves lethal and results in loss of the intestinal proliferative compartment (Korinek et al., 1998). Mutations in components of the Wnt signalling pathway are often found in colorectal cancer (reviewed by Schneikert and Behrens, 2007). These mutations lead to the stabilisation of β-catenin, which becomes Wnt independent and can accumulate in the cytoplasm and enter the nucleus to activate target genes. However, the role that different TCF/LEF transcription factors play in colorectal cancer is only just beginning to be elucidated.

Tcf-4 plays an essential role maintaining proliferating cells in the crypt of the healthy colon (Korinek et al., 1998) and also maintains the undifferentiated phenotype in colon cancer. Microarray studies of colorectal cancer cells transfected with constitutively repressive ΔNTcf4 have shown that some TCF-4 target genes are shared between colorectal cancer cells and the progenitor or stem-cell population (van de Wetering et al., 2002). Tcf-1, by contrast, appears to play a tumour suppressor role in the intestine, because Tcf-1-knockout mice develop intestinal neoplasms that resemble polyps in colorectal cancer (Roose et al., 1999). How can two such similar molecules have opposing functions? The predominant TCF/LEF isoforms present in the crypts of the colon are proposed to be full-length TCF-4 and ΔNTCF-1 (Korinek et al., 1997; Mayer et al., 1997; Korinek et al., 1998; Roose et al., 1999), and ΔNTCF-1 could act as a feedback repressor by repressing transcription of TCF/LEF target genes (Roose et al., 1999). Thus we can imagine a scenario in which full-length TCF-4 maintains cells in a proliferative state in the crypts and ΔNTCF-1 promotes a reduction in proliferation and an increase in differentiation (see Fig. 4).

LEF-1 is not normally expressed in the human colon or in the embryonic mouse intestine. However, in colon cancer tissue and cell lines, full-length LEF-1 isoforms are selectively expressed (Hovanes et al., 2001). This means that there is an abundance of active TCF/LEF isoforms present in the cancer cell and so there will be more Wnt target gene expression.

The regulation of Lef-1 expression turns out to be another aspect of TCF-isoform-specific function. The promoters for full-length LEF-1 and ΔNLEF-1 are both targets of Wnt signalling, and the expression of full-length LEF-1 isoforms is due to the selective repression of the promoter that produces the ΔNLEF-1 isoform (Li et al., 2006). Only E tail isoforms of TCF-1 and TCF-4 (see Fig. 2) are able to activate the full-length Lef-1 promoter, the domains responsible being CRARF and WCXXCRRKKKCIRY (Atcha et al., 2003). These isoforms appear to bind to the same Wnt-response element in the full-length Lef-1 promoter as other isoforms but DNaseI footprint experiments suggest that they adopt a slightly different conformation there (Atcha et al., 2003). The E tail helps stabilise the association with a Wnt-response element if the sequence differs from the consensus TCF/LEF-binding site (reviewed by Arce et al., 2006).

Aberrant expression of TCF/LEFs may thus result in greater numbers of activating TCF/LEF isoforms in the nucleus of cancer cells, leading to expression of a wider range of target genes. The idea that specific TCF/LEF isoforms can activate different genes (e.g. only TCF-4E and TCF-1E target the full-length Lef-1 promoter) adds an additional level of complexity to the problem of trying to determine the role of Wnt signalling in a positive feedback loop that drives colorectal cancer progression.

Fig. 5.

TCF/LEF function in skin and hair. Structure and tissues of the mammalian hair follicle (A) and the differentiation pathways (black arrows) of multipotent stem cells in the skin (B). The stem cell niche (SC) resides in the bulge of the hair follicle and gives rise to the tissues of the epidermis of the skin (E), the sebaceous gland (SG), the outer-root sheath (ORS) the inner-root sheath (ORS) and the hair shaft (H). Larger rings in B represent cells, whereas the smaller rings within them represent their cell nuclei. The expression of TCF-3 and LEF-1 are indicated with the letter T or L, respectively, in the cell nuclei of cells. Note that LEF-1 is expressed in cells of the hair differentiation pathway, TCF-3 in the stem cells and the adjacent ORS, while the tissues of the skin (E and SG) are devoid of any TCF/LEF expression. The molecular activities of TCF-3 and LEF-1 are decisively different during hair and skin differentiation. Overexpression of full-length LEF-1 promotes hair differentiation (illustrated in B with green dotted arrow); whereas overexpression of full-length TCF-3 or constitutively repressing ΔNTCF-3 prevents skin differentiation (E and SG, illustrated with red dotted inhibitory arrow) and promotes stem cell (SC) maintenance and differentiation into the adjacent ORS tissue (indicated with red dotted arrow). Overexpression of ΔNLEF-1 interferes with hair differentiation by promoting trans-differentiation into cells that resemble those of the sebaceous gland (SG, indicated with red dotted arrow). TCF-3 therefore functions primarily as a repressor (indicated in red), while LEF-1 functions as an activator (indicated in green). Figure modified from Merrill et al. (Merrill et al., 2001).

Skin and hair

Skin and hair in mammals are produced in the embryo and maintained in the adult from populations of multipotent stem cells through related mechanisms (Alonso and Fuchs, 2006). These give rise to the epidermis (E), the sebaceous gland (SG), the outer-root sheath (ORS), the inner-root sheath (IRS) and the hair and also maintain the stem cell niche itself (Fig. 5). Wnt signalling has multiple roles in normal skin development, promoting stem cell maintenance, proliferation and differentiation (see Lowry et al., 2005), but also acts in cancer formation (Lo Celso et al., 2004). LEF-1 and TCF-3 are both expressed in the skin; LEF-1 is present in the cell lineage that gives rise to the IRS and the hair itself, and TCF-3 is in the stem cells and the lineage that produces the ORS (DasGupta and Fuchs, 1999).

LEF-1 is largely required for normal hair development because Lef-1-knockout mice have very few hair follicles (van Genderen et al., 1994). The requirement for TCF-3 in skin development is difficult to study because Tcf-3-knockout mice stop developing at an early stage, displaying axial and germ layer defects (Merrill et al., 2004). However, skin-specific overexpression experiments show that TCF-3 and LEF-1 have different effects on stem cell differentiation in the skin. Overexpression of LEF-1 (Zhou et al., 1995; Merrill et al., 2001) promotes hair follicle differentiation (i.e. differentiation of cells that display features of the IRS and the hair). TCF-3 overexpression on the other hand promotes differentiation of cells with features characteristic of the ORS and the stem cell compartment but also interferes with epidermal differentiation (Fig. 5). The difference between TCF-3 and LEF-1 that accounts for their distinct activities in the skin maps to the central domain around the LVPQ and SxxSS peptide motifs, which LEF-1 lacks. This is similar to the situation in Xenopus dorsal development and mesoderm induction (see above) and also applies to TCF-3 function in embryonic stem cells (Pereira et al., 2006).

TCF-3 might tend to mediate repressor activity and LEF-1 might have primarily activator activity; however, this does not appear to account for the observed differences in the skin. Overexpression of constitutively repressive ΔNTCF-3 and ΔNLEF-1 constructs still shows a clear difference in their activities (Fig. 5). ΔNTCF-3, in common with its full-length counterpart, still promotes appearance of cells resembling those of the ORS and the stem cell niche and interferes with epidermal differentiation; ΔNLEF-1 interferes with hair development but not epidermal differentiation, but remarkably, promotes differentiation of cells resembling those of the sebaceous gland (Merrill et al., 2001). Notably, human sebaceous tumours are often caused by a specific mutation in the N-terminal β-catenin-binding domain of LEF-1 (Takeda et al., 2006), which creates a molecule with ΔNLEF-1-like activity and causes tumours that express molecular markers normally associated with the sebaceous gland.

The immune system

The TCF/LEFs were actually first identified in the immune system (van de Wetering et al., 1991; Waterman et al., 1991) (reviewed by Staal and Clevers, 2005). Several TCF-1 isoforms are expressed at various stages of T-cell development in the thymus (van de Wetering et al., 1996); LEF-1 is expressed in developing T cells in the thymus and also in developing B cells in the bone marrow (Travis et al., 1991).

Some of the best genetic evidence for functional diversity of different TCF/LEFs comes from studies of T-cell development in mice with deletions in Tcf-1 and/or Lef-1 (van Genderen et al., 1994; Verbeek et al., 1995; Okamura et al., 1998) (see Table 1 and Fig. 6). The data clearly argue for considerable redundancy between Tcf-1 and Lef-1, because double mutants cause more severe defects in T-cell maturation than do any single mutants. However, mutations in the Tcf-1 gene, which encodes many different TCF-1 isoforms (see Fig. 2), cause a more severe phenotype than does deletion of Lef-1, which encodes few isoforms (see Fig. 2). This suggests that certain TCF-1 isoforms have functions that cannot be provided by LEF-1 proteins expressed in the same tissue, whereas the functions of the LEF-1 proteins apparently are redundant with those of TCF-1 isoforms. Further support for this notion comes from studies of other immune cells, foetal-liver-derived natural killer (NK) cells. Again, there is evidence for much redundancy between Tcf-1 and Lef-1 in NK cell development, but Tcf-1 (specifically one of the unique TCF-1 isoforms) is required for expression of particular NK-cell-specific receptors (Held et al., 2003).

View this table:
Table 1.

Mouse gene knockout studies reveal redundant and non-redundant functions for Tcf/Lef genes in vertebrate embryonic development and adult stem cell regulation

Recent studies of T-cell development also show that expression of short ΔNTCF-1 and ΔNLEF-1 isoforms is carefully regulated (Weerkamp et al., 2006; Willinger et al., 2006). As T-cells mature, the ratio of long and short TCF/LEF isoforms expressed changes; for instance, whereas resting CD8 T cells express relatively high levels of the short isoforms, upon T-cell stimulation their expression is specifically downregulated. Given the known effects of Wnt signalling on haematopoiesis in general and T-cell maturation in particular (reviewed by Staal and Clevers, 2005), transcriptional regulation of short versus long TCF/LEF isoform expression is likely to play an important role regulating proliferation versus apoptosis in T-cell development and possibly in other tissues, such as the intestine (see above).

The nuclear cell biology of TCF/LEFs

Growing evidence indicates that the functional diversity of TCF/LEFs may be connected to regulation of their nuclear and subnuclear localisation by SUMOylation (Sachdev et al., 2001; Yamamoto et al., 2003). The small ubiquitin-related modifier (SUMO) protein modification mechanism resembles the better-known ubiquitin conjugation pathway but uses distinct enzymatic machinery (e.g. AXAM) (Kadoya et al., 2002). SUMOylation of targets such as p53, IκBα and Jun is thought to regulate their subcellular localization and molecular activity (reviewed by Muller et al., 2001). SUMOylation similarly regulates LEF-1 (Sachdev et al., 2001) and TCF-4 (Yamamoto et al., 2003).

Fig. 6.

TCF/LEF function in the immune system. Stages of B-cell and T-cell development are illustrated in bone marrow and thymus, respectively. Large rings represent cells and the smaller rings within them represent their nuclei. Wnt signalling is thought to regulate cell proliferation and apoptosis during lymphocyte development. TCF-1 and LEF-1 expression is indicated with the letter T or L, respectively, in the cell nuclei of cells. Note that TCF-1 and LEF-1 are both expressed during T-cell development. Red stars indicate defects in lymphocyte development reported in Lef-1 knockout mice; red lightening indicates defects in lymphocyte development observed in Tcf-1 knockdown mice; and the red stop sign indicates the complete block of T-cell development observed in Tcf1/Lef1 double mutant mice. Note that in T-cell development, TCF-1 functions to some extent redundantly with LEF-1, because the double mutant phenotype is more severe than the single Tcf-1 knockout phenotype, yet also non-redundantly, because the single Tcf-1 knockout phenotype is more severe than the Lef-1 knockout phenotype. Stages of B-cell development: HSC, haematopoietic stem cells; MLP, multi-lineage precursors; CLP, common lymphoid precursors; ePB, early pre-B-cells; pPB, proliferating pro-B cells; lPB, late pre-B-cells; iBC, immature B cells; mBC, mature B cells. Stages of T-cell development: DN1-DN3, double-negative thymocytes stages one to three; ISP, immature single positive thymocytes (or double-negative thymocytes at stage four); DP, double-positive thymocites; SP, single-positive thymocyctes (see Staal and Clevers, 2005). Figure modified from Staal and Clevers (Staal and Clevers, 2005).

SUMOylation of LEF-1 and TCF-4 coincides with their sequestration into subnuclear compartments called promyelocytic leukemia (PML) nuclear bodies. Importantly, however, the effects on their functions are TCF/LEF specific: SUMOylation of LEF-1 inhibits its transcriptional activation function; SUMOylation of TCF-4 enhances its transcriptional activity. The sites of SUMOylation also lie in different parts of the proteins: the LEF-1 SUMOylation site maps to its β-catenin-interaction domain, but the groucho/TLE interaction domain is the target for SUMOylation in TCF-4 (see Fig. 2).

SUMOylation, sequestration to specific subnuclear regions and alteration of molecular activity are clearly closely linked; however, the precise causal relationship has yet to be established. SUMOylation may independently cause changes to protein activity and subnuclear sequestration, or sequestration to PML bodies could independently cause changes to protein activity and SUMOylation. The fact that acetylation of the equivalent residue in the β-catenin-interaction domain of the TCF/LEF orthologue in Drosophila (Waltzer and Bienz, 1998) causes an inhibition of transcriptional activation function similar to that in LEF-1 may argue for protein modification rather than sequestration being the immediate cause of the observed changes to protein activity. Although recent evidence suggests that other mechanisms do regulate TCF-4 transcriptional activity through sequestration to HIC 1 (hypermethylated in cancer 1) bodies (Valenta et al., 2006), it is not established whether this mechanism is TCF-4-specific.

Conclusion and perspectives

Vertebrate TCF/LEFs have diversified in form and consequently function. What is more, not only do different vertebrate Tcf/Lef genes have different functions (e.g. Tcf-3 is mainly a repressor whereas Lef-1 is mainly an activator) but – apart from maybe Tcf-3 – different transcripts and different splice forms of the same vertebrate Tcf/Lef gene also encode isoforms that have distinctly different activities.

Are these differences qualitative or only quantitative? Are different vertebrate TCF/LEF isoforms just better or worse at assembling transcriptional activation complexes or transcriptional repression complexes, binding more strongly or more weakly to WREs and nuclear β-catenin; or alternatively, do different TCF/LEF isoforms mediate assembly of functionally different multi-protein complexes, respond to different regulatory mechanisms and interact with different promoter contexts to regulate different downstream target genes? The distinction between qualitative and quantitative differences may be useful but could turn out to be somewhat academic, since what biochemical and in vitro assays measure as quantitative differences might result in distinct effects in vivo.

The challenge for research in this area is to demonstrate different regulation of target genes by different TCF/LEFs in vivo and to uncover the molecular mechanisms by which different TCF/LEFs do this. We also need to determine the gene regulatory networks that bring about distinct cellular outputs from subtle differences in molecular activities. Combinatorial signalling with other pathways might play an important role in this. Promising discoveries have already been made in terms of TCF/LEF-isoform-specific target gene recognition – for instance, in the case of the cdx1 gene (Hecht and Stemmler, 2003) and, as discussed above, the gene encoding LEF-1 itself (see Atcha et al., 2003).

It will be equally important to investigate how expression of the different TCF/LEF isoforms is regulated in vertebrates. Differential transcription of full-length and short ΔNTCF/LEF isoforms is now well documented in the mammalian immune system, the intestine and colorectal cancer tissue. Undoubtedly this represents an important mechanism for regulating Wnt/β-catenin signalling in general and during cell proliferation in particular (Staal and Clevers, 2005; Arce et al., 2006). Differential alternative splicing of Tcf-4 has been documented in Xenopus embryos (Gradl et al., 2002), where it regulates the production of functionally distinct TCF-4 isoforms with and without the LVPQ and SxxSS motifs. Alternative splicing is a fundamental process in eukaryotic gene expression, particularly prevalent in vertebrates (reviewed by Blencowe, 2006). Yet, the mechanisms and the consequences of differential splicing in vertebrate embryonic development, stem cell regulation and cancer are relatively little studied. The alternative splicing of transcripts of Tcf/Lef genes is particularly exciting because the alternatively spliced products have decisively different activities.

Differential promoter use in healthy and diseased tissues has recently been demonstrated for LEF-1 (Li et al., 2006). Future findings about the mechanisms of tissue-specific Wnt signalling involving the rich variety of vertebrate TCF/LEFs will therefore be directly relevant for our understanding and ultimately the treatment of human diseases, because defective Wnt signalling is associated with abnormalities in bone formation, in lung, craniofacial, limb and urogenital development and in stem cell differentiation, which can result in several types of cancer.

Finally, it seems to us that available sequence information for several vertebrate and invertebrate genomes now suggests that the increasingly complex roles of Wnt signalling in vertebrates are not reflected in an equally dramatic increase in molecular complexity among Wnt ligands, their receptors or even Wnt signal transduction components. Instead these complex roles are mediated by remarkable molecular and functional innovation in nuclear mechanisms orchestrated by a rich variety of TCF/LEF proteins.


We thank Marian Waterman, Hans Clevers and Lynne Shanley for discussions and comments on the manuscript. C.L.K.'s research is supported by the AICR. S.H. acknowledges additional research support by The Wellcome Trust and the BBSRC.

  • Accepted November 27, 2006.


View Abstract