TCF/LEF proteins are the major mediators of Wnt-responsive gene transcription in the nucleus. In the absence of Wnt ligands, β-catenin is phosphorylated by the destruction complex containing Glycogen Synthase Kinase 3β (GSK3β), Casein Kinase 1 (CK1), Adenomatous Polyposis Coli (APC) tumour suppressor protein, and Axin. Phosphorylated β-catenin is targeted towards ubiquitination and further proteasome-mediated degradation. In the nucleus, inhibitory TCF/LEF members are bound on Wnt Cis-Regulary-Motifs (W-CRM), interact with co-repressors such as Gro/TLE proteins, and act as transcriptional repressors. Conversely, in Wnt-stimulated cells, the destruction complex is inhibited leading to the cytoplasmic accumulation of β-catenin and further nuclear translocation. Increase in the nuclear β-catenin levels transiently converts TCF/LEF into transcriptional activators (Figures 1A,B) (reviewed in MacDonald et al., 2009).

Observations from mammalian cells (Schuijers et al., 2014), flies (Franz et al., 2017), and amphibian (Nakamura et al., 2016) among others, reported the requirement of TCF/LEF for the transcriptional regulation of most β-catenin target genes, supporting the classical model of Wnt transcriptional regulation. Mammalian cells lacking all four genes encoding TCF/LEF proteins display perturbations in the association of β-catenin with DNA. In such cells, β-catenin was found to regulate different transcriptional targets (Doumpas et al., 2019), revealing that only when TCF/LEF is absent, β-catenin autonomously regulates a subgroup of genes whose transcription does not initially require TCF/LEF. Genome-wide analysis methods identified Wnt/TCF target genes that are available at http://www.stanford.edu/group/nusselab/cgi-bin/wnt/.

TCF/LEF sequence alignment and phylogenetic trees in species such as the hemichordate Saccoglossus kowalevskii (S. kowalevskii), Caenorhabditis elegans (C. elegans), Drosophila melanogaster, Hydra magnipapillata, and Ciona intestinalis reveal the presence, in the TCF/LEF structure, of the four major binding domains found in vertebrate (Atcha et al., 2007; Žídek et al., 2018), indicating that the TCF/LEF in invertebrate is probably the ancestral precursor of that described in vertebrate. Further complexity has been added through evolution following the emergence of the different TCF/LEF isoforms in mammals which are generated through alternative transcription, translation start sites, and alternative splicing (reviewed in Hoppler and Waterman, 2014).

Structural and functional analysis of TCF/LEF provided important cues on the domains mediating their transcriptional activities (Figure 2A). On their N-terminal region, all TCF/LEF isoforms have a β-catenin-binding domain (BCBD), which contains 50 amino acids (aa). Three sets of aa are involved in the TCF/LEF-β-catenin interactions: residues 2–15 (known as the β-hairpin module) fit into the groove of the central Armadillo (Arm) repeat domain (the homologue of the vertebrate β-catenin and signal transducer of wingless (Wg) signaling in flies). Residues 16–29 form an extended strand, and residues 40–51 form an α-helix (Graham et al., 2000). TCF/LEF interaction with β-catenin is necessary for their activity (Kratochwil, 2002). Isoforms of TCF7 and LEF1 expressed from alternative promoters, and encoding proteins lacking BCBD, behave as dominant-negative (van de Wetering et al., 1997; Hovanes et al., 2001). Recognition of the specific DNA sequence motif (CCTTTGAT(G/C)) by TCF/LEF is mediated by a highly conserved High Mobility Group (HMG)-box, whose DNA-binding domain structure and general mechanisms of DNA binding and bending, have been extensively studied (van Beest et al., 2000; reviewed in; Malarkey and Churchill, 2012). This HMG-box is followed by a Nuclear Localization Signal (NLS). The BCBD and the DNA-binding domain are separated by a less conserved context-dependent regulatory domain (CRD), partly encoded by an exon (exon VI), which can be alternatively spliced in all TCF/LEF except TCF7L1. Two conserved aa motifs, LVPQ and SxxSS, flank exon VI in Xenopus Tcf7l1 and Tcf7l2, but not in Tcf7 or Lef1, and can be alternatively spliced in Tcf7l2. Mutations in these two motifs validate their strict requirement for Tcf7l1 repressive activity. Furthermore, their insertion into the lef1 sequence abolishes Lef1 activator capacity, as detected through its inability to induce an ectopic secondary axis when injected ventrally in Xenopus embryos (Pukrop et al., 2001; Gradl et al., 2002; Liu et al., 2005). Other studies suggest that the Gro/TLE-binding domain encompasses the entire CRD and part of exon VII. Indeed, alternative splicing within the CRD (exon V to exon VII) modifies the interactions of TCF/LEF with Gro/TLE (Young et al., 2019) (reviewed in Hoppler and Waterman, 2014). The C-terminal tail is the most variable region among the TCF/LEF, where much of the aa sequence exhibit a low level of conservation. The C-terminal tail exists either as a long C-terminal extension, referred to as E tail, that contains additional domains, or as a short C-terminal extension, referred to as B tail, lacking the additional transcriptional regulators’ binding domains. Whereas TCF7L1 only carries an E tail, the LEF1 gene lacks the E-tail-encoding exon (Atcha et al., 2003). The E-tails encode two copies of a specific short motif (PLDLS) that binds the evolutionarily conserved co-repressor phosphoprotein C-terminal-Binding Protein (CtBP). Indeed, CtBP binds to both Tcf7l1, and the Tcf7l2 isoforms carrying an E tail (Brannon et al., 1999; Valenta et al., 2003; Fang et al., 2006). An additional small, highly conserved 30 aa motif (CRARF) is present in invertebrate TCF/LEF and in vertebrate splice variants Tcf7-E and Tcf7l2-E, but not in Tcf7l1. CRARF is required for the β-catenin-mediated transcriptional activation of the lef1 promoter, and forms a C-Clamp (Cysteine-rich domain) that allows TCF/LEF to bind an additional DNA motif known as the Helper site (5′-RCCGCCR-3′) (Atcha et al., 2007; Ravindranath and Cadigan, 2014).

TCF/LEF are metazoan inventions (Adamska et al., 2010). In choanoflagellates, which are unicellular eukaryotes considered the closest known relatives to metazoans, there is no evidence supporting the existence of any TCF/LEF protein and the only found component of the Wnt pathway is GSK3 (King et al., 2008). In invertebrate genome, only one Tcf/Lef gene is detected (reviewed in Hoppler and Waterman, 2014). One exception is found in the phylum of Platyhelminthes, in which five Tcf/Lef have been found in the genome of the flatworm Schmidtea mediterranea. Only two of these Tcf/Lef have a putative BCBD, which suggests a function in mediating Wnt transcription (Brown et al., 2018).

Most of our knowledge about TCF/LEF activity in invertebrate derives from studies performed in Drosophila and C. elegans. As in vertebrate, their Tcf/Lef is converted from a transcriptional repressor to activator by increasing nuclear levels of β-catenin. Drosophila Arm/β-catenin promotes transcriptional activation by binding Pangolin (Pan), the Tcf/Lef in fly (referred to as Tcf) (van de Wetering et al., 1997). Consequently, co-repressors such as Gro are displaced, allowing Arm binding to transcriptional co-activators such as Pygopus (Pygo) (Parker et al., 2002). As in vertebrate, in the absence of Arm, Tcf acts as a transcriptional repressor (Cavallo et al., 1998). Transcriptional repression appears to be directly mediated by the Tcf/Arm interactions with a specific DNA sequence motif (AGAWAW). The exchange of the AGAWAW motif into a standard Tcf-binding site (CCTTTGAT(G/C)) reversed the mode of regulation, resulting in Wnt-mediated activation instead of repression. Whereas both transcriptional activation and repression require binding of Arm to the N-terminal part of Tcf, allosteric regulation has been proposed to explain differences in Tcf/Lef transcriptional capacity. Indeed, Tcf binding to different DNA motifs may allow its interaction with distinct co-regulators, which subsequently controls its transcriptional activity (Blauwkamp et al., 2008).

In C. elegans, loss-of-function phenotypes indicate a dual regulatory mode for the Tcf/Lef termed POsterior Pharynx defect (POP1). An interesting mechanism has been reported for mesoderm and endoderm fate specification during embryogenesis (Rocheleau et al., 1997; Thorpe et al., 1997). At the four-cells stage, two sister cells, the anterior (MS) and the posterior (E) are fated to respectively generate the mesoderm and the endoderm. Higher levels of pop1 are detected in the MS blastomere (Lin et al., 1995), where it represses the transcription of Wnt-responsive endoderm-determining gene end1 through the recruitment of the histone deacetylase HDA-1 and UNCoordinated (UNC)-37 (the homologue of the Gro/TLE) (Calvo et al., 2001). In a POP1 mutant, both blastomeres adopt an endoderm-like fate. However, in the E blastomere receiving Wnt signals, WRM1/β-catenin binds to the N-terminal domain of POP1 protein and decreases its nuclear levels, alleviating POP1 repressive activity, which will then activate the expression of end1 and induce the specification of the endodermal fate (Rocheleau et al., 1997; Shetty et al., 2005). Another model proposes that the switch of POP1 from a transcriptional repressor to an activator depends on its DNA-binding site. The C-terminal tail of POP1 contains a C-clamp, which enables POP1 to recognize another DNA motif (the Helper site). When Wnt signaling is activated, β-catenin stabilizes the interaction between the C-Clamp of POP1 and the Helper sites found in the end1 sequence, which enables end1 transcription (Bhambhani et al., 2014).

In vertebrate, the founder members of the TCF/LEF family are TCF7 (van de Wetering et al., 1991) and LEF1 (Travis et al., 1991), initially identified as lymphocyte-regulators in mice. The two other members, TCF7l1 and TCF7l2 have been characterized few years later (Castrop et al., 1992).

TCF/LEF are largely expressed during vertebrate embryogenesis in some overlapping but also distinct regions including the central nervous system, suggesting a functional redundancy of the TCF/LEF members. For instance, in mice, Tcf7l2 and Lef1 transcripts are detected in the mesencephalon and the diencephalon (Korinek et al., 1998). In zebrafish, tcf1 and lef1 expression overlaps in the tail bud, fin buds and paraxial mesoderm (Veien et al., 2005). Observations made in lung epithelial progenitors also supports redundant and additive functions between the different TCF/LEF members (Gerner-Mauro et al., 2020). However, genetic mutants lacking a single Tcf/Lef gene, as well as double knockout (KO) mutants, exhibit severe developmental alterations (van Genderen et al., 1994; Galceran et al., 2000), indicating expanded and diversified roles for each TCF/LEF. Based on these findings among others, a specific activity as Wnt transcriptional activator and/or repressor has been attributed to each TCF/LEF.

Lef1 and Tcf7l2 KO mice show reduced Wnt transcriptional activity and are considered to mostly act as activators of the pathway (Korinek et al., 1998; Kratochwil, 2002). Similarly, analysis in zebrafish reveals activating functions for Tcf7, Lef1, and Tcf7l2. Loss of Lef1, expressed in several embryonic tissues, specifically the neural crest, decreases β-catenin activity (Dorsky et al., 1999, 2003). Additional observations from Tcf7l2 mutants show that it maintains proliferation of the intestinal epithelium through activating Wnt target genes’ transcription (Muncan et al., 2007).

In contrast, numerous studies strongly argue that Tcf7l1 mediates Wnt repressive activity. Mice depleted of Tcf7l1 gene phenocopy those with ectopic activation of Wnt signaling, suffering severe forebrain abnormalities in addition to perturbations in the midbrain and hindbrain (Merrill et al., 2004). Similarly, the zebrafish genome contains two tcf7l1 genes, headless hdl/tcf7l1a (Kim et al., 2000) and tcf7l1b (Dorsky et al., 2003), giving a total of five tcf/lef genes. The two Tcf7l1 appear to normally act as transcriptional repressors. hdl/tcf7l1a mutants exhibit truncated Tcf7l1 protein, which cannot undergo nuclear translocation. Such mutants show severe head defects including a lack of eyes, forebrain, and a part of the midbrain, a hallmark of Wnt overactivation. This phenotype could be rescued by overexpressing tcf7l1b, which in this context also act as a negative regulator of the Wnt pathway (Dorsky et al., 2003). Compared to zebrafish, the medaka genome contains a single tcf7l1 gene. Medaka lacking tcf7l1 have the same phenotype as the double-mutant zebrafish hdl/tcf7l1b (Doenz et al., 2018).

Some of the most informative studies regarding transcriptional activities of the four TCF/LEF members came from investigating the development of Spemann organizer (SO) in the amphibian Xenopus (also see next section). The early Xenopus embryo expresses three maternally inherited tcf/lef mRNAs: tcf7, tcf7l1 and tcf7l2 (Molenaar et al., 1998; Houston et al., 2002; Roël et al., 2002). Tcf/Lef activities are not redundant during mesoderm induction in amphibian. At late blastula/early gastrula stages, maternally encoded tcf7l1 represses the dorsal organizer genes’ expression (Houston et al., 2002), whereas both tcf7 and tcf7l2 act as transcriptional activators of SO genes (Standley et al., 2006). In this developmental context, whereas an activating form of tcf7l1 can rescue the Tcf7-morphant phenotype, only a constitutive repressor form of tcf7l1 rescues the Tcf7l1-morphant phenotype (Liu et al., 2005). Taken together, these observations indicate that during early Xenopus mesoderm induction, Tcf7l1 is mostly required for transcriptional repression, whereas Tcf7 and Tcf7l2 mostly mediate transcriptional activation. Interestingly, lef1 transcripts are first detected after the mid-blastula transition (MBT) (Molenaar et al., 1998), and during mesoderm induction, Lef1 transcriptional activity appears to be redundant with that of Tcf7 (Liu et al., 2005).

All the TCF/LEF members need to bind with nuclear co-factors to regulate target genes’ transcription. A key insight into the mechanism of Wnt transcriptional inhibition mediated by the TCF was the finding that they can directly bind to members of the Gro/TLE family of transcriptional co-repressors.

Gro/TLE are evolutionary conserved nuclear proteins. The invertebrate genome encodes a single member: Gro, initially identified in Drosophila, and UNC-37 in C. elegans, both of which antagonize signaling by Wnt (Cavallo et al., 1998; Calvo et al., 2001). Four members have been identified in human, known as TLE1-4, and in mice, named the Groucho-Related Genes (GRG1-4) (reviewed in Jennings and Ish-Horowicz, 2008; Turki-Judeh and Courey, 2012). In mice, a fifth family member (GRG-5) has also been identified as a gene encoding a shorter variant. GRG-5 is thought to act as a naturally occurring dominant negative (Table 1) (Brantjes, 2001; Wang et al., 2004).

A conserved structural organization comprising five domains characterizes Gro/TLE proteins (Figure 2B). Lacking a DNA-binding domain, Gro/TLE rely on their interaction with transcription factors for their specific recognition of promoter and/or enhancer DNA sequences. The highly conserved N-terminal glutamine-rich (Q) domain contains two motifs termed the amphipathic α-helices (AH1 and AH2), which mediate both Gro/TLE homo-oligomerization and their interactions with various transcription factors, including TCF/LEF (reviewed in Jennings and Ish-Horowicz, 2008). The central portion of Gro/TLE contains three less well-conserved domains. Gro/TLE was found to bind to the Drosophila Hdac known as Rpd3 (Chen et al., 1999), and with the mammalian HDAC1 an interaction mediated by the glycine (G) and proline (P)-rich domain (GP) (Chen et al., 1999; Arce et al., 2009). Second, the central (CcN) domain which includes a NLS, and third, a Serine (S) Proline (P)-rich domain (SP) generally involved in repression. The CcN and SP domains contain phosphorylation sites, which can modulate Gro/TLE-mediated repression (reviewed in Jennings and Ish-Horowicz, 2008). Of note, GRG5 contains the TCF/LEF binding domain and a GP domain that carries mutations impeding its ability to interact with HDAC (Brantjes, 2001). At their C-terminal end, Gro/TLE have a four tryptophan-aspartic acid repeat domain (WD), which is highly conserved across evolution. The WD motif is involved in nucleosome binding and condensation (Sekiya and Zaret, 2007), and mediates Gro/TLE interactions with repressor proteins. The WD motif of Gro/TLE interacts with two distinct peptidic motifs, the Engrailed Homology-1 (EH1) motif, and the WRPW (Trp-Arg-Pro-Trp) motif.

The EH1 motif is a Phenylalanine/Isoleucine/Leucine (FIL)-rich domain (FxIxxIL), required for transcriptional repression in vitro and in vivo (Smith and Jaynes, 1996; Muhr et al., 2001; Jennings et al., 2006). The EH1 motif is found in a large number of HD-containing TFs involved in neuronal specification such as Gastrulation Brain homeobox 2 (GBX2), Orthodenticle homeobox 2 (OTX2) (Heimbucher et al., 2007), Forkhead box (FOX) family of TFs (Yaklichkin et al., 2007), Engrailed (EN) (Jimenez et al., 1997) and BARHL that are notably the only Gro/TLE partners containing two EH1 domains (Offner et al., 2005). The TF Dorsal is involved in DV axis patterning in Drosophila. Dorsal was found to physically interact with Gro. Interestingly, in embryos lacking Gro, Dorsal functions as a transcriptional activator rather than as a repressor (Dubnicoff et al., 1997). It has been demonstrated that Gro interacts with a motif with partial homology to the EH1, located in the C-terminal part of Dorsal (Flores-Saaib et al., 2001). This interaction is weak and is stabilized by the presence of additional Gro-binding repressors (Valentine et al., 1998).

The second Gro/TLE-interacting motif is the WRPW present in basic-helix-loop-helix (bHLH) proteins including the Hairy/Enhancer of Split (E(spl))/HES proteins, transcriptional repressors that function as downstream targets of activated Notch receptors (Grbavec et al., 1998) (reviewed in Cinnamon and Paroush, 2008; Turki-Judeh and Courey, 2012). In the absence of Notch signaling, Gro/TLE is recruited via Hairless to a complex containing Suppressor of Hairless (Su(H)) and CtBP, which represses Notch target genes, including E(spl). Upon activation of Notch signaling, the Notch intracellular domain (NICD) enters the nucleus, displaces the Gro-containing complex, recruits Mastermind (Mam) on Su(H) an interaction which further results in the transcriptional activation of E(spl). E(spl) encoded factors interact with Gro/TLE to repress proneural genes (reviewed in Cinnamon and Paroush, 2008; Turki-Judeh and Courey, 2012). In Drosophila, and mammals association of Gro/TLE to bHLH proteins is required in cell fate decisions during tissue development including neurogenesis, segmentation, sex determination and myogenesis (Paroush, 1994; Jimenez et al., 1997). The WRPW motif has been demonstrated to be a functional transcriptional repression domain. It is sufficient to confer active repression to Hairy-related proteins or a heterologous DNA-binding protein through its ability to recruit Gro/TLE to target gene promoters (Fisher et al., 1996). Similar to Dorsal, in Drosophila, the Runx family member Lozenge that contains a WRPW motif exhibit low affinity for Gro/TLE and requires the Cut HD protein to form a stable repressive complex (Canon, 2003).

The Gro/TLE-binding site in the central portion of TCF/LEF extends and overlaps the β-catenin binding site (Daniels and Weis, 2005). Therefore, association of Gro/TLE with TCF/LEF counteracts the TCF/β-catenin transactivation activity (Cavallo et al., 1998; Roose et al., 1998; Brantjes, 2001). Together with other observations, these data lead to the generally accepted model where β-catenin activates Wnt-responsive genes by simply displacing Gro/TLE. Whereas recent studies provide arguments for a more complex regulation of Wnt-driven transcriptional switch (reviewed in Ramakrishnan et al., 2018), a large spectrum of genes are regulated by both β-catenin and Gro/TLE through their respective interactions with TCF/LEF. Chromatin immunoprecipitation sequencing (ChIP-seq) data from Xenopus embryos provide over 80% correlation between β-catenin and Gro/TLE-binding sites (Nakamura et al., 2016). In mouse hair follicle stem cells, more than half the genes occupied by TCF/LEF are also occupied by Gro/TLE (Lien et al., 2014).

The way Gro/TLE mediate transcriptional repression is still a matter of debate. Recent observations indicate that Gro/TLE could act either short distance via modulating RNA-polymerase II (RNA-Pol II) activity, and/or long distance via chromatin remodeling. ChIP-seq analysis combined to RNA-seq data performed in Drosophila identified the Gro/TLE direct targets. Such analysis suggested that Gro/TLE doesn’t affect the recruitment of RNA-Pol II to the transcription start sites but further increases RNA-Pol II pausing time (Kaul et al., 2014). Other studies indicate that in some context, Tcf/Gro complex promotes compaction of the chromatin when the canonical Wnt pathway is switched off. As previously mentioned, Gro/TLE interact with Hdac. In the presence of an Hdac-inhibitor, Wnt target genes are de-repressed (Billin et al., 2000). It is therefore possible that Gro/TLE interaction with Hdac drives long distance, transmittable changes in the chromatin state. Other studies argue that Hdac recruitment does not account for full co-repressor activity, suggesting that another Gro/TLE-dependent silencing could occur via tetramerization of Gro on a Tcf7l1/Gro complex, thereby promoting structural transitions of chromatin leading to transcriptional repression (Sekiya and Zaret, 2007; Chodaparambil et al., 2014).

In Xenopus embryos, injection of gro represses transcription of Wnt target genes (Roose et al., 1998), and mutations in Gro/TLE-binding sites of tcf7l1 reduces Tcf7l1 repressive activity (Liu et al., 2005; Tsuji and Hashimoto, 2005). Analysis performed on the Xenopus siamois (sia) promoter demonstrated that Tcf/Lef-binding sites mediate both basal repression and β-catenin-dependent activation at the W-CRM (Brannon et al., 1997; Fan et al., 1998). More recently, large-scale analysis demonstrates that in the dorsal blastomeres, Gro/TLE binds to the same W-CRM as β-catenin (Yasuoka et al., 2014; Nakamura and Hoppler, 2017; Afouda et al., 2020). In this context a few lines of evidence indicate that β-catenin activates Wnt-responsive genes by displacing the whole Tcf7l1/Gro repressor complex and replacing it with an activator complex, containing β-catenin in association with Tcf7 (Chambers et al., 2017) (reviewed in Cinnamon et al., 2008; Sokol, 2011; Ramakrishnan et al., 2018)

In conclusion, the mechanisms by which Gro/TLE mediate transcriptional repression in the presence and/or absence of TCF/LEF are still not fully understood. To add complexity, both TCF/LEF and Gro/TLE proteins are targets for developmental signals, which influence the affinity of Gro/TLE to TCF/LEF and/or W-CRM. Thereby, the developmental and cellular contexts in which Gro/TLE repression causes epigenetic regulations via the binding of HDAC by Gro/TLE as well as the exact role(s) of such transcriptomic regulations during development are still poorly understood.

One of the earliest, well-documented, and evolutionarily conserved functions of Wnt/β-catenin signaling is the induction of the blastopore lip organizer. The discovery made by Spemann and Mangold in 1924 has revolutionized our understanding of embryonic axis formation. In their classic transplantation experiment in newt, the authors showed that a mesodermal region - the dorsal lip of the blastopore - of a gastrula embryo induces a secondary axis including a complete nervous system when grafted ventrally (reviewed in De Robertis et al., 2000). This primary embryonic organizing center known as SO determines the dorso-ventral (DV) body axis. Requirement of canonical Wnt signaling for axis formation has been demonstrated following overexpression of Wnt signaling components. For instance, in Xenopus, wnt1 (McMahon and Moon, 1989), wnt8 (Sokol et al., 1991), and β-catenin (McCrea et al., 1993) can induce a complete dorsal axis when overexpressed ventrally. A similar phenotype has been observed when two Wnt inhibitors are depleted: Tcf7l1 (Merrill et al., 2004) and Axin2 (Zeng et al., 1997). More recently, it was shown in non-bilaterian metazoan species that the same molecular mechanism was used for inducing secondary axes as in chordates: the Wnt/β-catenin signaling, indeed demonstrating that the emergence of the Wnt/β-catenin driven blastopore-associated axial organizer predates the cnidarian-bilaterian split, which occurred over 600 million years ago (Kraus et al., 2016).

Investigations from the past decades lead to the current model of SO development. Before initiation of zygotic transcription, Tcf7l1 represses gene transcription throughout the embryo (Figure 3) (Molenaar et al., 1996; Houston et al., 2002). Accumulation and stabilization of β-catenin by maternal determinants in the nucleus of the dorsal cells, inhibit Tcf7l1 repressors’ activity (Schneider et al., 1996; Larabell et al., 1997), and activate the transcription of sia and siamois homologue 2 (twin) (Lemaire et al., 1995; Carnac et al., 1996), which in turn activate the transcription of 123 genes, including goosecoid (gsc) and chordin (chd), leading to the formation of the SO territory. All genes de-repressed by β-catenin in this region have been identified (Ding et al., 2017b). sia and twin are directly regulated by binding Tcf/Lef to their promoters, and poised for transcriptional activation by β-catenin before the Mid blastula transition (MBT) (Brannon et al., 1997; Laurent et al., 1997; Fan et al., 1998; Blythe et al., 2010). In the absence of β-catenin, Tcf7l1, together with Gro/Tle, inhibit sia and twin transcription (Roose et al., 1998). More recently, a thorough transcriptomic analysis, combined with genome-wide β-catenin association using ChIP-seq, identified stage-specific direct Wnt target genes. The direct comparison of genome-wide occupancy of β-catenin with a stage-matched Wnt-regulated transcriptome reveals that only a subset of β-catenin-bound genomic loci are transcriptionally regulated by Wnt signaling. The differences in classes of direct Wnt target genes appear to be context specific, and dependent on the presence of co-factors such as FoxH1, Nodal/TGFβ signaling (Afouda et al., 2020), Bone Morphogenetic Protein (BMP), and FGF signaling (Nakamura et al., 2016). These studies reveal that the cellular transcriptional responses to Wnt signal are highly dependent on the context, and thereby on the tissue, the developmental steps, the presence of co-factors and/or activation of co-signaling pathways (reviewed in Nakamura and Hoppler, 2017).

The Bar-class HD, BarH1 and BarH2, are HD-containing transcription factors initially discovered in Drosophila (Kojima et al., 1991; Higashijima et al., 1992). Barhl genes have subsequently been identified in fish (zebrafish, medaka), amphibian (Xenopus), birds (chicken), mammals (mouse, human), nematode (C. elegans) and S. kowalevskii among others (Lowe et al., 2003; Pani et al., 2012; Yao et al., 2016). Phylogenetic analysis shows that BARHL1 and BARHL2 proteins are extremely well conserved in the chordate phylum and are predominantly expressed in the central nervous system (CNS), where their expression patterns are distinct but partially overlapping (Figure 4) (Bulfone, 2000; Patterson et al., 2000; Offner et al., 2005; Colombo et al., 2006) (reviewed in Schuhmacher et al., 2011). BARHL1 and BARHL2 are involved in diverse processes such as the acquisition of a neural identity in the retina, specification of commissural neurons in the spinal cord and cell migration in the cerebellum and the hindbrain (Chellappa et al., 2008; Ding et al., 2009; Jusuf et al., 2012) (reviewed in Reig et al., 2007).

BARHL proteins are characterized by a conserved HD sequence of about 60 amino acids, which forms a three-dimensional helix-loop-helix structure required for their fixation to DNA (Figures 4A,B) (Gehring et al., 1994). Unlike other homeoproteins, BARHLs contain a tyrosine (Y) at site 49 as opposed to phenylalanine (F) at this site of the HD. Whilst the biological significance behind this substitution is unknown; it is thought that there could be a difference in the specificity of the DNA recognition motif. BARHL sequence also contains an NLS, and at their amino-terminal region, two EH1 domains. Sequence comparison reveals a conserved domain with an unknown function at the C-terminal part of BARHL proteins.

Biochemical experiments performed in both mammalian cells and Xenopus embryo validate the physical interaction between BARHL2 and Gro/TLE. Surprisingly, BARHL2 was found to interact with TCF/LEF, more specifically TCF7l1, and dramatically enhance the ability of TCF7l1 to co-immunoprecipitate Gro4/TLE4, at least in mammalian cells. This interaction is independent of TCF7L1 binding to Gro/TLE. Functional observations confirm that Barhl2 enhances the capacity of to repress transcription, and abolishes the β-catenin-driven activation of TCF/LEF target genes (Figure 1C) (Sena et al., 2019).

In Xenopus, barhl2 is not expressed maternally. Whereas W-CRM have been identified in the barhl2 loci (Nakamura et al., 2016), barhl2 is neither part of the early dorsal β-catenin signature, nor induced by overexpression of RNA coding for wnt8b. It is however expressed following the initiation of early β-catenin induction, and its expression increases following sia1 mRNA overexpression (Owens et al., 2016; Session et al., 2016; Ding et al., 2017b; Sena et al., 2019), suggesting that at these developmental stages, barhl2 transcription is under the control of both sia1 and β-catenin (Figure 3).

In Xenopus, overexpression of barhl2 generates massive developmental defects including loss of the SO territory and all anterior structures, including the cement gland and the head. In contrast, Barhl2 depletion expands both the organizer territory and its signaling activity, as detected through a massive increase in neuroepithelium size, and patterning alterations (Offner et al., 2005; Sena et al., 2019). Experimental evidence demonstrates that these developmental defects are direct consequences of Barhl2 normally enhancing Tcf7l1-mediated transcriptional repression. These observations lead to a model in which stabilization of β-catenin first de-represses Tcf7l1, and then initiates the dorsal developmental program through activating Tcf7 and/or Tcf7l1. The presence of Barhl2 locks Tcf7l1 and/or Tcf7 in an inhibitory state, and consequently limits induction of the dorsal development program. In this way Barhl2 participates in progression of the blastula development, and normally limits SO formation in time and/or in space.

Analysis of Barhl2 proteins that are mutated either in their ability to interact with DNA, or to bind Gro/TLE, indicate that its normal role requires both. As previously stated, Gro/TLE can silence target genes by tetramerizing on a Tcf7l1-Gro complex (Chen et al., 1999; Chodaparambil et al., 2014). It is therefore possible that Barhl2 enhances the binding of the complex to histones, associated with the long-term silencing of Tcf/Lef target genes through increasing Gro/TLE stoichiometry in a protein complex containing Tcf7l1. Moreover, the presence of Hdac1 is detected in a protein complex containing Barhl2, Tcf7l1 and Gro4. Hdac1 depletion promotes SO development. In parallel, Barhl2 depletion promotes key organizer genes’ acetylation. Thereby, Hdac1 activity could contribute to the Barhl2-mediated repression of Wnt target genes. ChIP-qPCR observations on the promoter of gsc indicate that both Barhl2 and Tcf7l1 can interact with the same Tcf-W-CRM in the absence of an adjacent Barhl2-W-CRM.

Overall, these observations are consistent with Barhl2 acting over long distance via its specific binding on DNA, perhaps on super-enhancers as previously suggested (Lin et al., 2016), and inducing long-term silencing of SO target genes maybe via Hdac1 activity and/or direct interaction with chromatin. In this way Barhl2 irreversibly locks cells in a SO identity.

The forebrain (telencephalon and diencephalon) is derived from the most anterior part of the neuroepithelium: the prosencephalic neural plate. Fate mapping analysis revealed that the telencephalon emerges from the most anterior part of the neural plate, whereas the diencephalon is formed within the caudal forebrain.Whereas inhibition of Wnt pathway is strictly necessary for telencephalic development (Glinka et al., 1998) (reviewed in Wilson and Houart, 2004), growth and patterning of the diencephalic territories (thalamus and epithalamus) require high levels of Wnt. While the Wnt1 or Wnt3A KO mice lose both the midbrain and hippocampal areas, double Wnt3A/Wnt1 mutant embryos exhibit an additional reduction in the diencephalon, caudal hindbrain, and rostral spinal cord (Thomas and Capecchi, 1990; Lee et al., 2000). Conversely, ectopic expression of Wnt1 or Wnt3A induces the enlargement of the neural tube along the DV axis, without altering the cellular identities of diencephalic neurons (Megason and McMahon, 2002; Panhuysen et al., 2004). Zebrafish masterblind (mbl)-mutant embryos carrying a mutation in the GSK3-binding domain of Axin1, which constitutively activates Wnt signaling, show a net reduction in the telencephalic and retinal territories in favor of the diencephalic territory (Heisenberg, 2001). Indeed, the diencephalic primordium, more specifically the diencephalic alar and roof plates, express Wnt ligands such as Wnt3, Wnt3A, Wnt8B, Wnt4 and Wnt2B (Colombo et al., 2006; Juraver-Geslin et al., 2011, 2014; Schuhmacher et al., 2011). Wnt target genes’ expression as well as the Wnt signaling machinery are enriched in the thalamus of all vertebrate analyzed so far (Jones and Rubenstein, 2004; Shimogori et al., 2004; Quinlan et al., 2009; Mattes et al., 2012).

Besides its role in fate determination, Wnt promotes cell-cycle progression, and cell growth. Its ability to modulate the activity of GSK3β promotes a general increase in protein stability, specifically that of β-catenin (Taelman et al., 2010), and through activation of Target of Rapamycin (TOR) pathway, it stimulates growth and protein synthesis. β-catenin nuclear accumulation induces TCF/LEF-mediated expression of the proto-oncogene c-Myc (He et al., 1998), which encodes a bHLH leucine zipper (bHLHZip) TF that has two distinct roles in the G1 progression. On one hand, it increases the expression of CyclinD1 and CyclinD2 that promotes progression from the G1 to the S phase; on the other, it represses the cell cycle inhibitors p27Kip1 and p21Cip1, thereby promoting cell cycle progression, and enhancing cell proliferation (reviewed in Juraver-Geslin and Durand, 2015).

The Sonic hedgehog (Shh)-secreting Mid-Diencephalic Organizer (MDO), also known as the Zona Limitans Intrathalamica (zli), develops within the diencephalic primordium (Larsen et al., 2001). Within the thalamic complex, Shh secreted by zli cells participates in the survival, growth, and patterning of neuronal progenitor subpopulations (Hashimoto-Torii et al., 2003; Scholpp et al., 2006, 2007; Vieira and Martinez, 2006). Mice lacking Shh show severe defects in most of the diencephalic territory (Chiang et al., 1996; Ishibashi and McMahon, 2002). Investigation of the chick neural tube growth revealed an epistatic relationship between Shh and Wnt in progression of the G1 cell cycle phase: Shh permits transcriptional activation of Tcf7l1 and Tcf7l2, which then induces β-catenin dependent expression of Cyclin-D1 (Alvarez-Medina et al., 2008). Phenotypic observations of Shh mutated mice suggest a conservation of these interactions in the diencephalon. Such mice develop a reduced diencephalon with decreased Tcf7l2 expression (Ishibashi and McMahon, 2002).

Shh and Wnt synergistically promote proliferation in the alar diencephalon, whereas cell-cycle analysis in chicken and mice reported slow proliferation kinetics in the diencephalon compared to its neighboring territories (reviewed in Martínez and Puelles, 2000). Moreover, diencephalic changing patterns observed upon manipulation of Wnt activity appear to be primarily due to altered fate specification rather than changes in proliferation (reviewed in Wilson and Houart, 2004). barhl2 transcripts are detected in the diencephalic histogenic field at late gastrula/early neurula stages in Xenopus (Offner et al., 2005; Juraver-Geslin et al., 2011), zebrafish (Staudt and Houart, 2007), and mice (Mo et al., 2004). In the diencephalic anlage, Barhl2 acts upstream of Shh in establishment of the zli and its absence generates massive defects specifically in the patterning of the alar diencephalon (Juraver-Geslin et al., 2014; Yao et al., 2016; Ding et al., 2017a) (reviewed in Sena et al., 2016). Besides its role in zli formation, Barhl2 normally limits diencephalic progenitors’ proliferation: Barhl2-depleted Xenopus embryos exhibit both a dramatic hyperplasia, and a neuroepithelial architectural disorganization in the caudal forebrain (Juraver-Geslin et al., 2011, 2014). In depth analysis of Barhl2-depleted embryos revealed an excessive Wnt transcriptional activation that stimulates neuroepithelial cell proliferation and induces defects in β-catenin intracellular localization together with an upregulation of axin2 and cyclinD1. Measurement of the relative velocity of the cell cycle in Barhl2-depleted embryos reveals a shortening of the cell cycle length (6 versus 8 h). As the length of the S-phase in these cells remains unchanged (1.5 h), and CyclinD1 is part of the G1-S cell cycle checkpoint, Barhl2 probably acts on the length of G1 phase (Juraver-Geslin et al., 2011).

Interestingly, in the developing diencephalon, a non-apoptotic function of the effector caspase, Caspase-3, limits neuroepithelial cell proliferation by inhibiting the activation of Tcf/Lef by the β-catenin (Juraver-Geslin et al., 2011). In this context, Caspase-3 acts either in parallel, or downstream of Barhl2, and its activity does not depend on its apoptosis-effector function. In addition, in the neuroepithelium, Caspase-7 acts as the executioner Caspase leading to cell death (Sena et al., 2020). Indeed, how Barhl2 regulates Caspase-3 non-apoptotic activity in Xenopus and limits β-catenin levels and stability in the developing diencephalon is unknown.

In conclusion, in the caudal forebrain, Barhl2 acts as a brake on Wnt transcriptional activation, probably through the stabilization of the inhibitory Tcf/Gro complex. Barhl2 could increase the length of diencephalic progenitors’ G1 phase, thereby modulating neuronal progenitors’ response to extracellular signals, including those of Wnt and Shh.

What are the extracellular signals influencing Barhl2 expression and activity in the caudal forebrain? Wnt3a is expressed in E9.5 mice (Louvi et al., 2007) at the onset of Barhl2 expression in the same territories. Pioneer studies performed in Drosophila presented the wg pathway as a positive regulator of barhl2 expression in the notum. barhl2 expression was lost in clones mutated for Arm (reviewed in Reig et al., 2007). Conversely, the expression of a constitutively active form of arm induces an ectopic expression of barhl2 in the pre-scrutum, associated with a decrease of wg (Sato et al., 1999). In Xenopus, RNA-sequencing analysis revealed that both morpholino-mediated depletion of Tcf7l1, and pharmacological activation of Wnt canonical signaling, induce an increase in barhl2 transcripts (Wills and Baker, 2015; Stevens et al., 2017).

Taken together, these observations suggest a model where Barhl2 could be a direct, or an indirect, target of the canonical Wnt signaling pathway. In return, Barhl2 would establish a negative feedback loop that limits Wnt’ activity.

Beside Barhl2, other co-repressors influence Tcf7l1 inhibitory activity. Indeed, cDNA expression screens performed in mammalian cells, combined with functional analysis in Xenopus, identified 45 inducers and 96 inhibitors of Tcf/Lef activity (Freeman et al., 2015). Co-repressors’ modes of action are diverse, sometimes divergent between vertebrate and invertebrate, and involve protein–protein interactions, changes in Tcf7l1 affinity for Wnt-target gene promoters, recruitment of co-repressors or co-activators, modulation of protein stability, and nuclear translocation.

CtBP, first described in Xenopus and later in rodents and human, binds to the C-terminal part of Tcf7l1-E and Tcf7l2-E isoforms. In fly, CtBP appears to be required for both activation of some Wnt targets and the repression of others, in parallel to, and independently of Tcf/Lef (Fang et al., 2006). However, the vertebrate CtBP acts as a co-factor for Tcf7l1, enhancing its repressor activity (Brannon et al., 1999; Xia et al., 2011). Lack of both Gro/TLE-binding domain and of the C-terminal region of Tcf7l1 leads to target genes’ transcriptional activation (Gradl et al., 2002). Notably, during Xenopus SO formation, the C-terminal part of Tcf7l1, which recruits the CtBP, is not required (Liu et al., 2005). In colorectal cancer cells, TCF7l1 recruits both CtBP and HDAC1 to repress expression of the Wnt antagonist DICKKOPF4 (DKK4) (Valenta et al., 2003; Eshelman et al., 2017). Besides CtBP, Tcf7l1 directly interacts with the methyl-CpG-dependent transcriptional repressor Kaiso in Xenopus. This interaction results in their mutual disengagement from the respective DNA-binding sites in such a way that Tcf7l1 can be inhibited following Kaiso overexpression both in cell lines, and Xenopus embryos (Daniel and Reynolds, 1999). Kaiso cooperates with Tcf7l1 to repress β-catenin target genes such as sia, through epigenetic regulation (Park et al., 2005). The interaction of Kaiso with Tcf7l1 depends on Kaiso zinc-finger domains, and on the HMG-box DNA-binding domain of Tcf/Lef factors (Ruzov et al., 2009). The LIM (Lin-11, Islet-1, and Mec-3; the three original members of the family) protein HIC-5 [Hydrogen Peroxide-Induced Clone 5, also termed ARA-55 (Androgen Receptor Activator of 55 kDa)] has been also identified as a binding partner to Tcf7l1 and Tcf7l2. Overexpression of HIC-5 acts as a negative regulator of a subset of Tcf/Lef family members, and can suppress secondary axis formation in Xenopus (Ghogomu et al., 2006). Important modulators of TCF/LEF activity are also found in the family of SOX (SRY-related HMG box) factors containing over 20 members (reviewed in Kormish et al., 2009; Bernard and Harley, 2010). In both mammalian cells and Xenopus, SOX17 and SOX4 directly bind to the HMG-box of TCF7l1, TCF7l2 and LEF1, an interaction that modulates the stability of the TCF/β-catenin complex (Sinner et al., 2007). More recently, SOX17 was shown to functionally cooperate with Wnt/β-catenin to specify an endodermal fate while repressing the meso-ectodermal fate. In this context, SOX17 and β-catenin co-occupy hundreds of key enhancers. In some cases, SOX17 and β-catenin synergistically activate transcription, apparently independently of TCF/LEF, whereas on other enhancers, SOX17 represses β-catenin/TCF-mediated transcription to spatially restrict gene expression domains. In this context, SOX17 acts as a tissue-specific modifier of the TCF/LEF responses (Mukherjee et al., 2021). Another modulator of the canonical Wnt signaling is SOX9, which was found to associate with β-catenin and further inhibit its activity (Topol et al., 2009). Further observations show that SOX9 proteins, together with Krüppel-like factor 4 (KLF4), suppress the Wnt-induced transcription through competing with TCF/LEF for the same β-catenin promoter sites, inhibiting the β-catenin-TCF/LEF (more specifically TCF7l2) binding and transcriptional activity (Sellak et al., 2012).

Besides the spatial and temporal distribution of repressor partners, PTM, including ubiquitination and/or phosphorylation of TCF/LEF and Gro/TLE, influence positively or negatively, Gro/TLE-TCF/LEF interactions (reviewed in Cinnamon et al., 2008; Turki-Judeh and Courey, 2012; Ramakrishnan et al., 2018). In gastrulating Xenopus embryos and in mammalian cells, phosphorylation of TCF7l1 by the Homeodomain Interacting Protein Kinase 2 (HIPK2) inhibits its capacity to bind its target genes (Figure 5A). β-catenin was found to serve as a scaffold that promotes HIPK2 interaction with TCF7l1 and the subsequent dissociation of TCF7l1 from the target gene promoter, thereby opening the way for β-catenin interaction with the non-phosphorylated TCF7, and activation of Wnt target genes’ transcription (Figure 5Ab). Mutated TCF7l1 proteins resistant to Wnt-dependent phosphorylation function as constitutive inhibitors. HIPK2-dependent phosphorylation also causes the dissociation of LEF1 and TCF7l2 from their targets’ promoter (Figure 5Ac) and its effect is thereby highly context specific: HIPK2 up-regulates transcription by phosphorylating TCF7l1, a transcriptional repressor, but inhibits transcription by phosphorylating LEF1, a transcriptional activator (Hikasa et al., 2010; Hikasa and Sokol, 2011). Alternatively, in mouse embryonic stem cells (mESCs), β-catenin inactivates TCF7l1 by removing it from DNA, which is followed by TCF7l1 protein degradation. Interestingly, in this context, genetic cues indicate that TCF7l1 inactivation appears to be the only required effect of the TCF7l1/β-catenin interaction (Shy et al., 2013).

Gro/TLE co-repressors are targets of PTM, which modulate their affinity not only for Wnt effectors, but also Notch, and Epidermal Growth Factor Receptor (EGFR) signaling cascades (Figure 5B). One example comes from studies in Drosophila demonstrating that EGFR signaling, mediated via the Mitogen-Activated Protein Kinase (MAPK), phosphorylates Gro/TLE, and leads to the weakening of its repressor function, and attenuation of Gro/TLE-dependent transcriptional silencing by the E(spl) proteins, which are the effectors of the Notch cascade (Figure 5Bb). Reversibly, when RAS/MAPK signaling is impeded, Gro/TLE-mediated repression is enhanced both in vitro and in vivo. Thus, downregulation of Gro/TLE-dependent repression by MAPK modulates the transcriptional output of the Notch pathway, and possibly of other pathways (reviewed in Cinnamon et al., 2008). In both Drosophila and human cell lines, the E3 ubiquitin ligase UBR5 is required for Wnt cellular response. In this context, Wnt signaling induces the ubiquitination of Gro/TLE by UBR5, which happens downstream of β-catenin stabilization (Figure 5Bc). In vivo observations argue that ubiquitination inactivates Gro/TLE, thereby enabling Arm/β-catenin to activate transcription (Flack et al., 2017). Interestingly, inactivation of Gro3/TLE3 occurs via the activity of AAA ATPase Valosin-containing protein (VCP, also known as p97). VCP unfolds ubiquitinated proteins via its ATPase activity and disrupts ubiquitinated Gro3/TLE3 tetramerization, a process required for Gro/TLE to repress Wnt targets (Chodaparambil et al., 2014). Moreover mono-ubiquitination of Gro3/TLE3 by the E3 ubiquitin ligase XIAP (X-linked Inhibitor of Apoptosis) at its N-terminal Q-rich domain disrupts the ability of Gro3/TLE3 to bind TCF7l1, and consequently inhibits TCF7l1 repressor activity. XIAP is recruited to the Gro/TCF complex upon Wnt pathway activation, which enhances β-catenin/TCF complex assembly and the initiation of a Wnt-specific transcriptional activation program (Hanson et al., 2012). Because UBR5 and XIAP ubiquitinate Gro3/TLE3 in distinct ways (poly vs. mono) and at different locations on the Gro3/TLE3 protein, it is possible that the two E3 ligases modulate the Wnt transcriptional switch either in parallel, or simultaneously, depending on the cellular context. In addition to its ubiquitination activity, XIAP has been shown to play a role in inhibiting Caspases. In vertebrate, XIAP directly binds to and functionally blocks Caspase-3, Caspase-7 and Caspase-9 proteolytic activity (reviewed in Liston et al., 2003). However, there is no evidence for the XIAP-mediated degradation of vertebrate Caspases in vivo, which appears to depend on the type of ubiquitination and on the cell type.

As previously described, BARHL1 and BARHL2 proteins have a strong degree of homology between one another. BARHL2 is highly conserved amongst distant species in the evolutionary scale, as observed through the high aa sequence conservation throughout its entire sequence (Figure 4). Besides, they are amongst the TFs essential for patterning the body axis of the developing embryo that are conserved in simpler organisms beyond the phylum of chordates. For example, genetic programs ancestral to the ones required for vertebrate development were found conserved in hemichordates. In S. kowalevskii, which is thought to be the closest species to the common ancestor at the base of the phylogenetic tree of chordates, Barhl2 ortholog gene shares close similarities in its distribution and expression patterns compared to chordates (reviewed in Röttinger and Lowe, 2012; Sena et al., 2016). A conserved Shh Brain Enhancer (SBE1) has been discovered in mice with an equivalent function to that described in the S. kowalevskii. SBE1 directly regulates Shh expression in the zli through binding the second intron of the Shh gene. Diverse transcription factors, including Otx2 and Barhl2, directly regulate SBE1 within the zli. Functional analysis in both species demonstrated sufficient conservation between Barhl2 and one of the S. kowalevskii barH HD for both binding, and activating CRM, thereby controlling Shh expression (Yao et al., 2016) (reviewed in Sena et al., 2016).

In C. elegans, the cephalic chemosensory neurons (CEM) undergo PCD during hermaphrodite embryogenesis but not in males (Sulston et al., 1983), a process relying on CEH30, a Bar-HD transcription factor (Schwartz and Horvitz, 2007). CEH30 protein interacts with UNC-37, which is the C. elegans homologue of Gro/TLE, through its N-terminal EH1 motif. It thereby prevents cell death (Peden et al., 2007) and inhibits transcription of egl-1 gene, which encodes the executioner cell death protein CED-3, one of the major components of the PCD in worm (Nehme et al., 2010). Sequence comparison between human BARHL2 and CEH-30 proteins reveals 64% identical amino acids in the region including the HD and the motif immediately next to the HD on the C-terminal side called the BARC motif (Bar homeodomain C-terminal motif) (Figure 4). Interestingly, murine Barhl1 or Barhl2 genes compensate for the loss of function of CEH-30 in C. elegans (Schwartz and Horvitz, 2007), which is consistent with a conservation of Barhl genes’ function through evolution.

mESC isolated from the inner cell mass of the blastocyst, the pre-implementation stage mammalian embryo, are characterized by their ability to self-renew and to differentiate into all types of somatic cells, a process referred to as pluripotency (reviewed in Chen et al., 2017). A specific core set of transcription factors including OCT4 (Octamer-binding transcription factor 4), NANOG, SOX2 and KLF4 form regulatory circuitry consisting of autoregulatory and feedforward loops thereby supporting pluripotency and self-renewal of these cells (Boyer et al., 2005). Extracellular signaling including LIF/JAK/STAT3 (Leukemia Inhibitory Factor/Janus Kinase/Signal Transducer and Activator of Transcription) (Williams et al., 1988; Ying et al., 2008), Wnt (Hao et al., 2006; ten Berge et al., 2011), BMP (Ying et al., 2003), and the MAPK/ERK (Yang et al., 2012) cascades, influence mESCs fate decision. Indeed, in mESC, Wnt signaling has been demonstrated to have important, somewhat difficult to interpret, activities (reviewed in Niwa, 2011; Merrill, 2012). From all these studies, a consensus emerges that inhibition of TCF7l1-mediated repression is at the core of mESC self-renewal and pluripotency (Figure 6) (Atlasi et al., 2013) (reviewed in Sokol, 2011; Wray and Hartmann, 2012). Reversibly, enhancement of TCF7l1 repressive activity blocks mESC self-renewal, and allows mESCs to differentiate, even in the presence of Wnt signaling (Wray et al., 2011).

In mESCs, TCF7l1 is the most expressed member among the TCF/LEF protein family (Pereira et al., 2006; Salomonis et al., 2010). Whole-genome approaches including RNA-seq and ChIP-seq show that TCF7l1 transcriptionally represses many genes important for maintaining pluripotency, and self-renewal, as well as those involved in lineage commitment, and stem cell differentiation. TCF7l1 associates with the regulatory regions of 1369 genes (Tam et al., 2008). Among those regions, 1173 bind TCF7l1 and OCT4 (Cole et al., 2008) with more than 940 binding TCF7l1, OCT4, SOX2, and NANOG (Boyer et al., 2005; Pereira et al., 2006; Marson et al., 2008). Depletion of TCF7l1 generates mESC’ refractory to differentiation (Cole et al., 2008). Moreover, both the Gro/TLE and CtBP interacting domains of TCF7l1 are required for OCT4 repression (Tam et al., 2008). Finally, KLF4 gene contains conserved TCF/LEF binding sites, and its expression is downregulated by TCF7l1 (Park et al., 2015). Interestingly, two TCF7l1 isoforms have been discovered, and are expressed equally in mESCs, where they regulate both an overlapping, as well as different sets of target genes. Removal of one of both TCF7l1 isoforms was found sufficient to stimulate self-renewal and delay the differentiation through repression of NODAL and KLF4 (Salomonis et al., 2010). Further analysis revealed that binding of β-catenin to both TCF7l1 and TCF7 contributes to the maintenance of self-renewal and gene expression, at least partly through their recruitment to OCT4-binding sites on ESC chromatin (Yi et al., 2011).

The crucial role of TCF7l1 is reinforced by analysis of mESCs lacking all full-length TCF/LEF. In such cells, re-expression of TCF7l1 makes mESCs capable of differentiating into the three lineages, including neuronal cells (Moreira et al., 2017). In this context, TCF7l1 has been shown to directly interact with OCT4, and compete with SOX2 at some SOX-CRM, a process under the influence of MEK/MAPK (Zhang et al., 2013). Indeed, besides limiting TCF7l1-mediated repression of the pluripotency network, inhibition of the MAPK/ERK pathway participates in maintenance of pluripotency and self-renewal (reviewed in de Jaime-Soguero et al., 2018). In mESCs, inhibition of MEK suppresses LEF1 expression, and depletion of LEF1 partially mimics the self-renewal-promoting effect of MEK inhibitors. In the absence of the exogenous factors, cytokines or inhibitors, depletion of both TCF7l1 and LEF1 enables maintenance of undifferentiated mESCs (Ye et al., 2017).

In agreement with all these data, Gro/TLE, more specifically Gro4, is not required for sustaining pluripotency, and suppressing differentiation genes in mESC. Rather, Gro/TLE activity appears necessary for early differentiation where it acts to suppress the pluripotency network, allowing for the initiation of lineage specific gene expression programs. In mESCs, most of the genes occupied by TCF7l1 were found co-occupied by Gro/TLE (Laing et al., 2015). Through interacting with Gro/TLE, TCF7l1 represses NANOG (Pereira et al., 2006), and repression of OCT4 was found to rely on TCF7l1/Gro2 interactions (Tam et al., 2008). Interestingly, the dominant-negative GRG5 is highly expressed in mESC, and its expression drops once mESCs exit the pluripotent state, to increase again during neuroectodermal cell specification. Whereas overexpression of GRG5 promotes self-renewal, its siRNA-mediated KD deregulates the mESC pluripotent state. Transcriptomic analysis reveals that, in this context, GRG5 represses mesendodermal-related genes, and promotes neuronal specification via inhibition of Wnt and BMP signaling. Moreover, GRG5 maintains the self-renewal of NSCs by sustaining the activity of Notch/HES and STAT3 signaling pathways (Chanoumidou et al., 2018).

In contrast to what is reported in mESC, in human ESC (hESCs), Wnt/β-catenin signaling appears to be actively repressed in an OCT4-dependent manner during self-renewal. In these cells, activation of Wnt signaling appears to induce loss of self-renewal, and differentiation into mesodermal lineages (Davidson et al., 2012). Although such discrepancy is a little puzzling, it has been shown that generation of neural lineages from hESCs requires inhibition of Wnt signaling (Tabar and Studer, 2014) and that activation of Wnt signaling in hESCs-derived neural precursor cells promotes transcription of midbrain-like genes through TCF7l2 directly binding the Engrailed-1 (EN1) promoter (Kim et al., 2018).

Besides its role in development of the CNS, Wnt/β-catenin signaling is crucial for NSCs maintenance. NSCs emerge from territories that have kept their neuroepithelium properties and respond to Wnt signals from embryogenesis through adulthood (Selvadurai and Mason, 2011; Garbe and Ring, 2012; Borday et al., 2018). In the subventricular zone of the developing mouse brain, Wnt signaling is a hallmark of self-renewing, specifically of NSCs’ (Kalani et al., 2008). Investigation of the developmental fate of Wnt/β-catenin–responsive cells in embryonic and postnatal mouse brain using a reporter for Axin2, demonstrates the continued importance of persistent Wnt/β-catenin signaling for NSCs and progenitor cells emergence (Bowman et al., 2013). In mouse adult hippocampus, where new neurons are continuously generated from NSCs, expression of the pro-neural TF Neurogenic Differentiation 1 (NEUROD1) is a landmark of cells dropping out of self-renewal and entering neuronal commitment. Overlapping binding sites for the TCF/LEF factors and SOX2, a marker of most uncommitted cells of the CNS, are present in the promoter region of Neurod1. In this context, Wnt signaling together with removal of SOX2 triggers the expression of NEUROD1, demonstrating that the SOX2-TCF/LEF regulatory elements are critical for NEUROD1 expression, and consequently for the switch from the SOX2-mediated repression to the TCF/LEF-mediated activation, towards a neuronal fate ((Kuwabara et al., 2009). In neural precursor cells of the mouse’ neocortex, expression of TCF7l1 was found to repress Wnt activity (Ohtsuka et al., 2011; Kuwahara et al., 2014), and active Wnt signaling in the rodents’ neocortex apical progenitors sustain their fate plasticity (Oberst et al., 2019). In cultured rat adult hippocampal NSCs, fate decision is influenced by the temporal variations of β-catenin. Optogenetic approaches reveal that continuous activation of β-catenin in cultured NSCs specifies neuronal differentiation, whereas short β-catenin signals activate proliferation but remain insufficient to induce neuronal differentiation. Loss of β-catenin signals promotes apoptosis in differentiating cells, which could be due to inappropriate cell-cycle re-entry (Rosenbloom et al., 2020).

Does BARHL2 play a part in mESC biology via its ability to enhance TCF7l1 repressive activity? Although such a question has not been directly asked, it is known that BARHL2 is expressed in mESCs during their commitment phase (Lee et al., 2006). Global expression profile analysis of mESC lines in which BARHL2 overexpression was induced in a doxycycline-controllable manner, reveals that BARHL2 induces a significant fold-change in more than three thousand genes with more than two thousand genes being upregulated, and more than one thousand genes downregulated. In this context, BARHL2 was one of the most influential TF analyzed. Two days following BARHL2 induction, mESCs start to express mesodermal lineage markers (Yamamizu et al., 2016). In a study using another BARHL2 overexpression design in mESCs, a significant increase in the population of neural cells was observed 14 days post-induction (Teratani-Ota et al., 2016). In this context, Notch signaling pathway played a significant influence in driving neural differentiation, and the majority of neuron-like cells generated by induction of BARHL2 expressed markers of GABAergic neurons.

Work is still needed to understand the context-specific regulation of TCF/LEF activities in the biology of ESC, specifically in hESCs. However, studies from the last 15 years strongly support inhibition of TCF7l1-repression as the necessary downstream effect of Wnt signaling in the promotion of mESCs’ self-renewal and pluripotency. Reversibly, the formation and stabilization of the TCF7l1/Gro complex, and its inhibitory influence on specific chromatin loci, is one of the crucial switches driving ESCs towards cellular commitment (reviewed in Sokol, 2011). Genetic and functional studies demonstrated that the Gene Regulatory Networks (GRN) underlying acquisition/loss of the pluripotent state are similar in the rodent, fish, and amphibian’s blastomeres, and in mESCs, with slight differences observed in hESCs. In early blastomeres, BARHL2-mediated lock of TCF7l1 in an inhibitory state pushes early SO cells towards irreversible commitment and differentiation, arguing for a similar function in mESCs.

In the past years, an increasing number of studies have demonstrated that mutations, loss, or aberrant regulation of Wnt signaling are at the origin of a wide variety of diseases (reviewed in Noelanders and Vleminckx, 2017; Ng et al., 2019). In one of its severest forms, Wnt constitutive activation is associated with diverse cancer types including melanoma, leukemia, breast cancer, gastro-intestinal cancers, and others (reviewed in Zhan et al., 2017). Cancer Stem Cells (CSCs), also known as Tumor Initiating Cells (TICs), are characterized by their “stemness” characteristics that contribute to tumor progression and drug resistance and play deterministic roles in cancer recurrence. Cancer cells exhibit many of the same features as stem cells including self-renewal and their low level of differentiation. Whereas the exact connection between cancer and stem cells is not completely understood, it is well established that both cells use similar signaling pathway machineries, specifically those of the Wnt/β-catenin, Shh, MAPK/ERK and Notch pathways (reviewed in Friedmann‐Morvinski and Verma, 2014). In this section, we focus on the impact of Wnt signaling on CSCs, and on medulloblastoma (MB), a pediatric tumor of cerebellar origins in which contributions of both TCF/LEF and the BARHLs are relevant.

As observed through activity of a TCF/LEF reporter gene, Wnt/β-catenin signaling is highly active in various types of CSCs including colon, lung, breast, and gastric cells (Vermeulen et al., 2010; Horst et al., 2012). A pharmacological antagonist of β-catenin/TCF7l2 interaction blocks CSCs’ self-renewal and suppresses tumorigenesis. Treatment of human colon cancer cells, and mouse salivary gland cells with such compound did not only reduce the β-catenin/TCF7l2-mediated proliferation rate and self-renewal, but also induced the differentiation of tumor cells (Fang et al., 2016), making it a potential therapeutic target of the β-catenin-TCF/LEF-dependent tumors, among other tested drugs (reviewed in Jung and Park, 2020; Walcher et al., 2020; Zhang and Wang, 2020). Non-coding RNAs have also been identified as modulators of Wnt-TCF/LEF activity in CSCs. For example, miR-142, which is absent in normal mammary cells but highly expressed in mammary CSCs, increases Wnt activity by inducing degradation of APC, a negative regulator of canonical Wnt signaling (Isobe et al., 2014). Additionally, the long non-coding RNA IncTCF7 has been characterized in hepatocellular carcinoma cells. IncTCF7 maintains CSCs’ properties via TCF7-dependent activation of Wnt signaling (Wang et al., 2015). Notably, the ability of both normal and CSCs to maintain long telomeres – an important feature to prevent their cellular aging – appears to be under direct transcriptional control of β-catenin-TCF/LEF (Park et al., 2009; Noureen et al., 2021). Whereas the promoter of the telomerase enzymatic subunit, TERT, neither bind TCF7l2 nor TCF7l1, it is enriched with TCF7-binding sites located close to the transcription start site that binds β-catenin specifically (Hoffmeyer et al., 2012).

Medulloblastoma (MB), the most common childhood malignant brain tumor, emerges from the cerebellum and accounts for 30% of pediatric brain tumors. Integrated genomic studies allowed the identification of four types of human MB, corresponding to specific genetic alterations (reviewed in Hatten and Roussel, 2011; Wang et al., 2018). One group is associated with alterations in the Wnt/β-catenin signaling pathway (15% of the cases) and originates from brain stem cells. A second group (25% of the cases) is characterized by the constitutive activation of the Shh pathway and derives from Granule Neuron (GN) progenitors. Group 3 (30% of the cases) is specifically found in infants and is thought to originate from overexpression of the MYC oncogene in cerebellar NSCs. Whereas group 4 is the most common MB subgroup (30% of the cases), its underlying biology is not well understood.

In the rodent brain, the cerebellar upper Rhombic Lip (uRhL) produces the GNs that constitute the largest neuronal population in the brain. The GN population exhibits a unique developmental trait: committed GN progenitors (GNPs) are characterized by a very long period of “quiescence” occurring before birth, followed by a long proliferative phase – i.e., 2 weeks in mouse, 2 years in human - occurring after birth, before their final differentiation step (reviewed in Leto et al., 2016). Due to this developmental specificity, this cell population is at risk when it comes to the appearance of developmental defects, including oncogenic events (reviewed in Hatten and Roussel, 2011). At birth, the RhL stem/progenitor cells become responsive to secreted SHH that stimulates their proliferation and self-renewal. The uRhL exhibits stem cell niche properties and exhibit positive TCF/LEF transcriptional activity (Selvadurai and Mason, 2011; Bowman et al., 2013; Yeung et al., 2014; Borday et al., 2018). Atonal Homologue 1 (ATOH1/MATH1) is the master gene of GNPs’ development (reviewed in Leto et al., 2016). ATOH1 directly induces the expression of Barhl1 and Barhl2 (Kawauchi and Saito, 2008). A thorough single-cell RNA-seq performed on mouse cerebellar cells reveals that Tcf7l1 is expressed strongly early in the GNPs’ differentiation pathway (Wizeman et al., 2019), and that Barhl2 expression is uniquely associated with early fate commitment in the GNPs (Carter et al., 2018). Taken together, these observations argue that BARHL2 could participate in driving GN stem/progenitor cells towards irreversible commitment.

Rodent cerebellar uRhL cells are known to be at the origin of group 2 MB that are associated with deregulation of the Shh pathway (reviewed in Hovestadt et al., 2020; Garcia-Lopez et al., 2021). Some tumor-propagating cells from this subgroup are not only resistant to Shh inhibitors but are also TCF/LEF-dependent for their self-renewal (Rodriguez-Blanco et al., 2017). in silico analysis associates BARHL2 expression with the emergence of group 2 MB, and BARHL1 expression with emergence of group 3 and group 4 MB (Pöschl et al., 2011; Lin et al., 2016). Taken together, these observations are a good starting point for future research that should evaluate whether BARHLs act as roadblocks for de-differentiation that are corrupted in MB.