TCF/LEF (HMG box)
TCF/LEF proteins are the DNA-binding end of the canonical Wnt pathway. Humans have four: TCF7 (TCF-1), TCF7L1 (TCF-3), TCF7L2 (TCF-4) and LEF1. Each binds DNA through a single SRY-type HMG box plus an adjacent basic region, reading the minor groove of the Wnt response element CTTTGWW and bending the helix sharply. None activates transcription on its own: an N-terminal domain recruits beta-catenin, which supplies the coactivators, and without nuclear beta-catenin the same proteins repress through TLE/Groucho. Wnt response elements control intestinal crypt renewal, hair follicle and limb development, and T cell differentiation; they are constitutively occupied in APC-mutant colorectal cancer, and an intronic variant at TCF7L2 is the largest-effect common risk allele for type 2 diabetes.
The factors
Four human genes encode TCF/LEF proteins: TCF7 (originally TCF-1), TCF7L1 (TCF-3), TCF7L2 (TCF-4) and LEF1. Each carries one HMG box of the same SRY-related, sequence-specific type used by SOX proteins, together with an adjacent basic region. The HMG box binds entirely in the minor groove and bends DNA sharply; the LEF-1 solution structure showed the box in a widened minor groove of a markedly distorted helix, with the basic region lying across the narrowed major groove and adding to recognition 7651541. Because of that geometry these proteins were first described as architectural factors that bring proteins at flanking sites into contact, the original example being Ets-1, RUNX and ATF/CREB at the TCR-alpha enhancer 7651541. Transcriptional output comes from partners rather than from the HMG box: an N-terminal domain binds the armadillo repeats of beta-catenin, whose groove accommodates the extended TCF peptide 11136974, and without nuclear beta- catenin the same proteins repress through TLE/Groucho and histone deacetylases. Alternative splicing produces "E-tail" isoforms of TCF7 and TCF7L2 carrying a 30-residue cysteine clamp (C-clamp), a second and independent DNA-binding module 17893322. TCF7 and LEF1 are highest in T lymphocytes and progenitors, TCF7L1 is broad and prominent in embryonic stem cells, and TCF7L2 is broad with high levels in intestine, pancreatic islets and brain.
The motif
The Wnt response element, also called the TCF binding element, is written CTTTGWW (W = A/T), most often CTTTGAT or CTTTGAA, and AWWCAAAG on the opposite strand 1639073. The medoid of the 18 models collected here is AGATCAAAG, the reverse complement of CTTTGATCT; the databases differ in which strand they display, not in what they measured. Cyclic selection with full-length LEF-1 and TCF-1E gave the extended element TYYCTTTGATSTT 17893322. The invariant part is the CTTTG pentamer, where the HMG box inserts into the minor groove 1639073; the two positions after it are read with help from the basic region and from residues C-terminal to the box, and site selection on Tcf1 showed the functional element to be about 12 base pairs, not the 6-8 usually assumed for an HMG box 10867006. E-tail isoforms additionally bind a GC-rich helper element, RCCG, near the core 17893322. The C-clamp does not change which WRE is recognized, but it restricts occupancy to sites carrying the extra block 17893322, and that is visible here: the two longest models (HOCOMOCO TCF7L1, CAGCGCCTCGATGT, 18.0 bits; Vierstra AC0308 LEF/TCF, CAGCGCCCCGATG, 17.4 bits) pair a GC-rich block with the core instead of extending the AT-rich flanks. All four proteins give the same motif - cohesion 0.84, the tightest family in the HMG class - so sequence cannot separate TCF7, TCF7L1, TCF7L2 and LEF1, and the archetypes are named "LEF/TCF" and "TCF7L/LEF". The boundary that does matter is with SOX: the reverse complement of CTTTGWW is WWCAAAG, and the SRY/SOX consensus is WWCAAWG, identical except that SOX tolerates T where TCF/LEF requires A 1639073,7774012.
Where it acts
Wnt response elements are predominantly distal. Genome-wide mapping of TCF7L2 in colorectal cancer cells placed most occupied sites well away from transcription start sites, often in clusters of several elements per responsive gene, and recovered the CTTTG core as the enriched motif 18268006. Canonical targets include AXIN2, LGR5, ASCL2, SP5, NKD1, CDX1, MYC and LEF1 itself 18268006,16413481. In intestinal epithelium TCF7L2 sites co-occur with CDX2, HNF4A, GATA6 and SOX9 sites; in T cells TCF7 sites co-occur with RUNX and GATA3. With two degenerate positions, CTTTGWW is expected on the order of 1.5 million times in a 3.1 Gb double-stranded genome, so the element is not what limits output - nuclear beta-catenin level, chromatin state and the presence of a C-clamp helper site are.
Biology and disease
TCF/LEF proteins convert Wnt signaling into transcription. Beta-catenin was shown to bind LEF-1 and convert it into an activator 8757136, and the Drosophila experiment with Armadillo and dTCF established that this is the pathway's output step 9118222. In colorectal cancer with APC loss, beta- catenin accumulates and forms a constitutive complex with TCF7L2 that holds Wnt targets on 9065401; activating CTNNB1 mutations produce the same result. LEF1 is required for hair follicle, tooth and limb development; TCF7 for thymocyte development, and TCF1 marks the stem-like CD8 T cell population that responds to checkpoint blockade; TCF7L1 restrains differentiation in embryonic stem cells. The intronic TCF7L2 variant rs7903146 was identified as a common risk allele for type 2 diabetes 16415884 and remains the largest-effect common variant for that disease, acting through islet function.
Reading a hit
A CTTTGWW match identifies a Wnt response element, not which TCF/LEF protein uses it: all four are identical on DNA. Choosing among them needs expression data; deciding whether the site is active needs evidence about beta-catenin. The bigger practical hazard is the SOX family: CTTTGTT-type sequences satisfy both consensuses, several HOCOMOCO SOX models (SOX4, SOX11, SOX13, one SOX2 model) read CCTTTGTT, and there is no sequence-only way to resolve them - an AT- rich flank leans SOX, a GC-rich RCCG block leans an E-tail TCF. The element is common enough to occur by chance more than a million times, so isolated matches are weak evidence. Three of the seven HOCOMOCO models here are ChIP-only; prefer the HT-SELEX and methyl-HT-SELEX models when the question is whether a protein binds a sequence. Finally, do not read the GC-rich block of the long TCF7L1 models as a WRE - it is the helper site, and alone it is not a TCF/LEF motif.
How this was worked out
- 1992mediumMethylation and carbethoxylation interference footprinting, base-analogue substitution1639073
The HMG boxes of TCF-1 and SRY are shown to recognize the heptamer (A/T)(A/T)CAAAG through contacts in the minor groove, with no major-groove guanine contacts: the first definition of the element now called the Wnt response element.
Still open then: Nothing yet connected the site to a signaling pathway; TCF-1 was a T cell factor of unknown input.
The solution structure of the LEF-1 HMG box with its adjacent basic region shows the box in a widened minor groove of a sharply bent helix and the basic region across the narrowed major groove, supporting an architectural role in assembling the TCR-alpha enhancer complex.
- 1997highYeast two-hybrid, reporter assays, Drosophila genetics, cell-line reconstitution8757136 9118222 9065401
TCF/LEF proteins are established as the output step of Wnt signaling: beta-catenin binds LEF-1 and converts it to an activator, Armadillo does the same for dTCF in Drosophila, and APC-mutant colon carcinoma cells carry a constitutive beta-catenin-TCF4 complex.
Two complementary results: site selection shows Tcf1 reads a 12-base-pair element rather than the 6-8 bases assumed for an HMG box, and the beta-catenin/Tcf3 crystal structure shows the TCF peptide extended along the armadillo-repeat groove.
An intronic microsatellite at TCF7L2, tagging rs7903146, is identified as a common type 2 diabetes risk allele, replicated in three cohorts, which makes this the only TCF/LEF gene with a large common-variant disease signal.
The C-clamp of E-tail TCF7 and TCF7L2 isoforms is shown to be a second DNA-binding module: cyclic selection gives the same core element, TYYCTTTGATSTT, with or without it, but wild-type sites are additionally enriched for a GC-rich RCCG block.
Still open then: The helper block's own sequence requirements, and whether LEF1 and TCF7L1 compensate for lacking a C-clamp, were not settled.
Genome-wide mapping of TCF7L2 occupancy in colorectal cancer cells finds 6,868 sites, mostly far from transcription start sites and clustered around responsive genes, with ACA/TTCAAAG recovered as the occupied motif.
In liver tumor organoids, beta-catenin occupies largely distinct genomic regions depending on whether it partners LEF1 or TCF7L2, and the TCF7L2-specific regions are marked by a longer GC-rich helper motif, GGRCCRTACCAC, rather than by extra copies of the classical element, which complicates the view that the four paralogues are interchangeable on DNA.
Still open then: Whether TCF7L2 contacts this helper sequence directly through its C-clamp or is brought there by cofactors is untested.
Still unresolved
- Whether the four TCF/LEF proteins are genuinely interchangeable on DNA is now in question: all four give one motif in these databases (cohesion 0.84), but CUT&RUN in liver tumor organoids puts LEF1 and TCF7L2 at largely separate sites distinguished by a GC-rich helper motif, and it is unresolved whether that reflects intrinsic DNA preference or partner-driven recruitment.
- The sequence of the C-clamp helper site is not agreed: selection with TCF-1E gave the short RCCG block, genome-wide analysis in DLD1 colorectal cells gave GCCGCCR, and the liver organoid work gives the longer GGRCCRTACCAC, so a GC-rich block near a WRE cannot yet be scored against one definition.
- CTTTGTT-type matches cannot be assigned between TCF/LEF and SOX: the two consensuses differ at a single position, several HOCOMOCO SOX models read CCTTTGTT, and no sequence feature has been shown to separate a SOX-occupied from a TCF/LEF-occupied site.
Good confidence that a CTTTGWW match is a Wnt response element; none at all about which of TCF7, TCF7L1, TCF7L2 and LEF1 binds it, since all four share one motif (cohesion 0.84) and the archetypes are named "LEF/TCF". The element overlaps the SOX consensus at all but one position, so CTTTGTT-type hits cannot be assigned to TCF/LEF rather than SOX from sequence alone. Roughly 1.5 million matches are expected by chance genome-wide. Occupancy depends on nuclear beta-catenin and, for E-tail isoforms, on a nearby GC-rich helper site; the GC block in the longest TCF7L1 and LEF/TCF models is that helper, not the core element.
References
- van de Wetering M, Clevers H. Sequence-specific interaction of the HMG box proteins TCF-1 and SRY occurs within the minor groove of a Watson-Crick double helix. EMBO J 1992;11:3039-44. PMID 1639073
- Love JJ, Li X, Case DA, et al. Structural basis for DNA bending by the architectural transcription factor LEF-1. Nature 1995;376:791-5. PMID 7651541
- Behrens J, von Kries JP, Kühl M, et al. Functional interaction of beta-catenin with the transcription factor LEF-1. Nature 1996;382:638-42. PMID 8757136
- van de Wetering M, Cavallo R, Dooijes D, et al. Armadillo coactivates transcription driven by the product of the Drosophila segment polarity gene dTCF. Cell 1997;88:789-99. PMID 9118222
- Korinek V, Barker N, Morin PJ, et al. Constitutive transcriptional activation by a beta-catenin-Tcf complex in APC-/- colon carcinoma. Science 1997;275:1784-7. PMID 9065401
- Graham TA, Weaver C, Mao F, et al. Crystal structure of a beta-catenin/Tcf complex. Cell 2000;103:885-96. PMID 11136974
- van Beest M, Dooijes D, van De Wetering M, et al. Sequence-specific high mobility group box factors recognize 10-12-base pair minor groove motifs. J Biol Chem 2000;275:27266-73. PMID 10867006
- Hallikas O, Palin K, Sinjushina N, et al. Genome-wide prediction of mammalian enhancers based on analysis of transcription-factor binding affinity. Cell 2006;124:47-59. PMID 16413481
- Atcha FA, Syed A, Wu B, et al. A unique DNA binding domain converts T-cell factors into strong Wnt effectors. Mol Cell Biol 2007;27:8352-63. PMID 17893322
- Hatzis P, van der Flier LG, van Driel MA, et al. Genome-wide pattern of TCF7L2/TCF4 chromatin occupancy in colorectal cancer cells. Mol Cell Biol 2008;28:2732-44. PMID 18268006
- Grant SF, Thorleifsson G, Reynisdottir I, et al. Variant of transcription factor 7-like 2 (TCF7L2) gene confers risk of type 2 diabetes. Nat Genet 2006;38:320-3. PMID 16415884
- Werner MH, Huth JR, Gronenborn AM, et al. Molecular basis of human 46X,Y sex reversal revealed from the three-dimensional solution structure of the human SRY-DNA complex. Cell 1995;81:705-14. PMID 7774012
- Kluiver TA, Nordin A, Lu Y, et al. Differential Wnt/β-catenin signaling via TCF7L2/LEF1 binding specificity shapes cellular and tumor phenotypes. Proc Natl Acad Sci U S A 2026;123:e2528450123. PMID 42268900
18 motifs in this family — JASPAR 4 · HOCOMOCO 7 · CIS-BP 4 · Vierstra 3
| motif | database | consensus | bits |
|---|---|---|---|
| TF7L1.H14CORE.2.S.CSC | HOCOMOCO | CAGCGCCtCGATGT | 18 |
| AC0308:LEF/TCF:Sox | Vierstra | CAGCGcCcCGATG | 17.4 |
| TCF7L1 (MA1421.1) | JASPAR | AAAGATCAAAGG | 16.8 |
| M03512_2.00 | CIS-BP | AAAGATCAAAGG | 16.8 |
| LEF1.H14CORE.0.PSM.APSMA | HOCOMOCO | CTTTGATcT | 14 |
| M08167_2.00 | CIS-BP | AgATCAAAG | 13.5 |
| TCF7L2 (MA0523.2) | JASPAR | AgATCAAAG | 12.8 |
| LEF1.H14CORE.1.P.BPB | HOCOMOCO | TtCCTTTGAarTgt | 12.1 |
| TF7L2.H14CORE.0.P.BPB | HOCOMOCO | cCTTTGATcT | 11.8 |
| TF7L1.H14CORE.1.P.CPC | HOCOMOCO | CTTTGAtaTGca | 11.6 |
| AC0127:TCF7L/LEF:Sox | Vierstra | AgATCAAAG | 11.4 |
| TCF7.H14CORE.0.PSM.APSMA | HOCOMOCO | ACATCAAAG | 10.9 |
| TF7L1.H14CORE.0.PM.APMA | HOCOMOCO | CTTTGATgT | 10.9 |
| M05771_2.00 | CIS-BP | cATCAAAG | 10.8 |
| TCF7 (MA0769.3) | JASPAR | CTTTGAt | 10.6 |
| Lef1 (MA0768.3) | JASPAR | CCTTTGAT | 10.4 |
| M05755_2.00 | CIS-BP | CATCAAAG | 10.1 |
| AC0128:LEF:Sox | Vierstra | CCTTTgannT | 7.7 |
HOCOMOCO tags read as assay + quality grade: P ChIP-seq, S HT-SELEX, M methyl-HT-SELEX, G GHT-SELEX, I SMiLE-seq, B PBM; grade A means two or more assay types agreed. A P-only matrix is a record of where the protein was crosslinked, which need not be a sequence it binds itself.
The factors, elsewhere
These entries are written from the primary literature and every citation is checked against PubMed, but this one can still be out of date, too confident, or simply wrong — especially where a motif’s published site and its database matrices disagree. Corrections and additions are welcome, and they are the fastest way to make this page better for everyone.