Regulation of TCF3 by Wnt-dependent phosphorylation during vertebrate axis specification

Abstract

A commonly accepted model of Wnt/β-catenin signaling involves target gene activation by a complex of β-catenin with a T-cell factor (TCF) family member. TCF3 is a transcriptional repressor that has been implicated in Wnt signaling and plays key roles in embryonic axis specification and stem cell differentiation. Here we demonstrate that Wnt proteins stimulate TCF3 phosphorylation in gastrulating Xenopus embryos and mammalian cells. This phosphorylation event involves β-catenin-mediated recruitment of homeodomain-interacting protein kinase 2 (HIPK2) to TCF3 and culminates in the dissociation of TCF3 from a target gene promoter. Mutated TCF3 proteins resistant to Wnt-dependent phosphorylation function as constitutive inhibitors of Wnt-mediated activation of Vent2 and Cdx4 during anteroposterior axis specification. These findings reveal an alternative in vivo mechanism of Wnt signaling that involves TCF3 phosphorylation and subsequent derepression of target genes and link this molecular event to a specific developmental process.

Figure 1. TCF3 is phosphorylated by HIPK2 in response to Wnt signaling

(A) Wnt8 stimulates TCF3 phosphorylation in animal pole explants. Xenopus embryos were injected at 4 cell stage with Wnt8 DNA (100 pg) or extracellular domain of Frizzled 8 RNA (ECD8, 500 pg). Cell lysates were prepared at indicated gastrula stages for western analysis using antibodies against Xenopus TCF3, unphosphorylated β-catenin (ABC), and β-tubulin (control for loading). Ratios of TCF3 and β-tubulin levels are shown at the bottom. An apparent decrease in TCF3 protein is likely due to reduced TCF3 transcription (see below, Figure S2A and Figure 4B). (B) Wnt8 DNA (100 pg) causes TCF3 to migrate slower in animal pole explants (st. 12.5), after immunoprecipitation. This effect is sensitive to alkaline phosphatase (AP) treatment. (C) TCF3 is phosphorylated in ventral (VMZ), as compared to dorsal (DMZ) marginal zone explants, dissected at stage 10 and harvested at stage 13. RNAs encoding Wnt antagonists DKK1 (300 pg), ECD8 (500 pg) and dominant negative Wnt8 (dnWnt8, 1 ng) inhibit TCF3 phosphorylation, reduce β-catenin levels, but do not significantly affect phospho-Smad1. (D) Wnt8 MO (15 ng) blocks TCF3 phosphorylation in the ventral marginal zone at the late gastrula stage (st. 12.5). (E) TCF3 phosphorylation in ectoderm cell lysates from embryos injected with HIPK2 RNA (80 pg) (compare to B). (F and G) HIPK2 MO (HK2MO, 20 ng) and a dominant negative HIPK2 (HK2KD, 400 pg) block TCF3 phosphorylation in ectoderm explants (st. 12) expressing Wnt8myc DNA (100 pg).

Figure 2. TCF3-dependent repression and Wnt8/HIPK2-dependent activation of Vent2 require the unique TCF-binding site in the Vent2 promoter

(A) Diagram indicates relative positions of SMAD (S)-, OAZ (O)- and TCF (T)-binding sites and the corresponding point-mutations (*) in Vent2-Luc constructs. Point-mutations were introduced as in (Karaulanov et al., 2004). Two-cell embryos were injected animally with MOs or DNA constructs and harvested at the late gastrula stages (st.12–13) for luciferase activity determination. (B) Reporter with mutated TCF-binding site (TCFm-Luc) has increased activity. (C) Wnt8 DNA activates Vent2-Luc but not TCFm-Luc. (D) Endogenous TCF3 is depleted in the embryos injected with TCF3MO (30 ng). (E) TCF3MO enhances BREm-Luc, but not BREm/TCFm-Luc activity. (F) TCF3-VP16 DNA, a constitutive transcriptional activator, stimulates BREm-Luc through the TCF-binding site.

Figure 3. Opposite roles of TCF3 and HIPK2 in dorsoanterior development

(A) CoMO (60 ng), HK2MO (60 ng), or TCF3MO (40 ng) were injected four times into the marginal zone of four-cell embryos as indicated. Anterior views are shown, dorsal is up. HK2MO enhances, whereas TCF3MO inhibits the dorsoanterior fate, assessed by head and cement gland development at stage 28 (top). Effects of HIPK2 and TCF3 MOs on Vent2 (middle) and Otx2 (bottom) were analyzed by whole mount in situ hybridization at stage 14 and 17, respectively. CoMO does not affect Vent2 and Otx2 gene expression (95%, n=56, and 96%, n=43, respectively). HK2MO suppresses (68%, n=56), whereas TCF3MO expands (56%, n=45) Vent2 expression in the anterior and ventral region. HK2MO expands (56%, n=50), while TCF3MO suppresses (72%, n=50) the Otx2 expression domain. Arrowheads demarcate cement gland (CG) boundaries (top), and the boundaries of the Vent2 (middle), and the Otx2 (bottom) expression domains. (B) Inhibition of anterior development by TCF3MO is rescued by TCF3 RNA. Embryos shown represent phenotype classification based on CG size at stage 28. Class I, CG is between 1/3 to 1/2 of normal size; Class II, CG is less than 1/3 of normal size. (C) HK2MO enhances anterior development, this defect is rescued by RNA encoding HIPK1, a related protein kinase, and partially reversed by TCF3MO. Phenotype classes are based on cement gland size at stage 26. Class I, CG size is increased less than twice; Class II, CG size is increased more than twice. RNAs or MOs were injected into the animal pole (B) or the marginal zone (C) of four-cell embryos at indicated dose. (D) HK2MO inhibits Vent2-Luc reporter activation via the TCF-binding site. MOs and reporter DNA were injected six times into the animal pole and the marginal zone for luciferase activity determination analysis at late gastrula stages.

Figure 4. Identification of Wnt8-dependent TCF3 phosphorylation sites

(A) Xenopus TCF3 contains 16 serine/threonines that are adjacent to proline and conserved among vertebrate TCF3 proteins (highlighted in red in the aligned sequence). These residues were substituted for alanine in eight TCF3 constructs: P1, S120; P2 (S147, S149); P3 (T170, S181, S184); P4, S190; P5 (S206, S209, S215); P6, S290; P7, S434; P8 (S445, T448, S450, S455). βcat, β-catenin binding domain; HMG, high mobility group domain; NLS, nuclear localization signal. (B) Ectoderm lysates at stage 13, co-expressing TCF3 RNAs (4 pg) with Wnt8 (W) or dnWnt8 (D) DNAs (100 pg), were precipitated with anti-flag beads and probed with anti-TCF3 antibody to assess TCF3 protein mobility. (C) HIPK2 phosphorylates wild-type (WT), but not P2/3/4 mutant TCF3 at the P4 site. TCF3 proteins were precipitated from 293T cells by anti-flag beads and detected by anti-flag or anti-TCF3P4 antibodies. Doses of transfected DNAs: flagTCF3 constructs, 8 μg; mycHK2KD, 8 μg; mycHK2, 6 μg. (D) In vitro kinase assay was performed after immunoprecipitation with anti-flag beads. The products were analyzed with anti-flag or anti-TCF3P4 antibody. (E) Wnt3a triggers P4 site phosphorylation in HEK293T cells. Cells were transfected with flagTCF3 constructs (0.3 ug), treated with Wnt3a-containing medium for 10 hrs followed by anti-flag precipitation and analysis with anti-TCF3P4 antibody.

Figure 5. Wnt-dependent TCF3 phosphorylation relieves transcriptional repression

(A, B) Xenopus TCF3 phosphorylation mutants enhance the dorsoanterior fate (A) and inhibit Wnt8-dependent reporter activation of BREm-Luc (B). TCF3 RNAs were injected into four animal pole blastomeres at the 8-cell stage. Representative phenotype classes based on cement gland size at stage 28 are shown. Class I; more than the double size of normal cement gland, Class II; more than the triple size of normal cement gland. (C) Mouse TCF3 P2/3/4 mutant inhibits Wnt-dependent reporter activation in HEK293 cells more efficiently than wild-type TCF3 (p < 0.05 for 1 ng and p < 0.01 for 10 ng respectively).

Figure 6. β-catenin is required but not sufficient for TCF3 phosphorylation

(A) Western analysis of lysates of animal caps expressing Wnt8 (DNA, 100 pg, or RNA, 80 pg) or β-catenin (DNA, 100 pg; RNA, 120 pg). Wnt8, but not β-catenin, stimulates TCF3 phosphorylation. Injected RNA is translated immediately, but DNA is only transcribed after midblastula stages, indicating the time of the effect. (B) HK2MO (40 ng) and β-catenin MO (βMO, 20 ng) inhibit TCF3 phosphorylation in ventral marginal zone explants at stage 13. (C) β-catenin MO blocks TCF3 phosphorylation in dissociated ectoderm cells stimulated with Wnt3a protein (50 ng/ml) for 3 hours. (D, E) β-catenin-binding is required for Wnt8-stimulated TCF3 phosphorylation. (D) ΔβTCF3 lacks the amino-terminal stretch of amino acids (7–30), which does not contain S/T residues. Lysates from twenty ectoderm explants per group from embryos injected four times with flagTCF3 RNAs (1 pg) and Wnt8 DNA (100 pg) were precipitated with anti-flag beads at stage 13. (E) β* TCF3 has D16A and E24A substitutions. RNAs encoding TCF3 constructs (3 pg) were coinjected with Wnt8 DNA (100 pg) into the animal pole of four-cell embryos. Ectoderm explants (approximately 150 per group) were used for TCF3 precipitation with anti-flag beads following by detection with anti-TCF3-P4 antibody. (F) β-catenin promotes the binding and phosphorylation of TCF3 by HIPK2. Cells were transfected with: flagTCF3 constructs, 5 μg; mycHK2KD, 1 μg; mycHK2, 0.5 μg; β-catenin-myc, 5 μg, and cultured for 10 hours.

Figure 7. Wnt-dependent TCF3 phosphorylation leads to the dissociation of TCF3 from the Vent2 promoter

(A) TCF3 P2/3/4 binds to the Vent2 promoter stronger that wild-type TCF3 (WT) (p < 0.05 for WT and P2/3/4 with Vent2-2 primers). Four cell embryos were injected four times with 3 pg of TCF3 RNA into the marginal zone and harvested for cross-linking at stage 12.5. Vent2-2 primers amplify the region of the Vent2 promoter that contains the unique TCF binding site; Vent2-1 primers span the region ~3 kb downstream of the transcription start site. (B) HIPK2 inhibits the binding of TCF3 to the Vent2 and Siamois promoters (p < 0.05 for the comparison of KD and ΔP in WT-injected group for Vent2-2 and Siamois), but does not significantly influence the binding of P2/4 and P2/3/4 proteins. Four cell embryos were injected animally with the indicated RNAs: TCF3 constructs, 30 pg; HK2KD, 400 pg; HK2ΔP, 400 pg. ChIP analysis was carried out with anti-flag antibodies at stage 12. (C) Dissociated animal cap cells (DC, stage 10) were treated with Wnt3a-containing or control medium for 20 min or 120 min. WE; whole embryos at stage 11.5. p < 0.05 for Vent2-2 groups with and without Wnt3a (120 min). ChIP was carried out with anti-flag (A and B), anti-TCF3 (C) antibodies and anti-GST as a negative control (C). Quantitative PCR was performed after ChIP. Each group contained triplicate samples. (D) Model: Wnt8 stimulates HIPK2/β-catenin-dependent phosphorylation of TCF3, leading to its dissociation from the Vent2 promoter, and promoting ventroposterior development. TBS, TCF3-binding site.