Negative regulation of Wnt signalling by HMG2L1, a novel NLK-binding protein


  • Misato Yamada,

    1. Department of Molecular Cell Biology, Medical Research Institute, Tokyo Medical and Dental University, Kanda-Surugadai, Chiyoda-ku, Tokyo 101-0062, Japan
    Search for more papers by this author
  • Bisei Ohkawara,

    1. Department of Molecular Cell Biology, Medical Research Institute, Tokyo Medical and Dental University, Kanda-Surugadai, Chiyoda-ku, Tokyo 101-0062, Japan
    Search for more papers by this author
  • Naoya Ichimura,

    1. Department of Molecular Cell Biology, Medical Research Institute, Tokyo Medical and Dental University, Kanda-Surugadai, Chiyoda-ku, Tokyo 101-0062, Japan
    Search for more papers by this author
  • Junko Hyodo-Miura,

    1. Division of Morphogenesis, Department of Developmental Biology, National institute for Basic Biology, Okazaki 444-8585, Japan
    Search for more papers by this author
  • Seiichi Urushiyama,

    Corresponding author
    1. Department of Molecular Cell Biology, Medical Research Institute, Tokyo Medical and Dental University, Kanda-Surugadai, Chiyoda-ku, Tokyo 101-0062, Japan
    Search for more papers by this author
  • Kyoko Shirakabe,

    1. Department of Molecular Cell Biology, Medical Research Institute, Tokyo Medical and Dental University, Kanda-Surugadai, Chiyoda-ku, Tokyo 101-0062, Japan
    Search for more papers by this author
  • Hiroshi Shibuya

    1. Department of Molecular Cell Biology, Medical Research Institute, Tokyo Medical and Dental University, Kanda-Surugadai, Chiyoda-ku, Tokyo 101-0062, Japan
    Search for more papers by this author

  • Communicated by: Eisuke Nishida

*Correspondence: E-mail:


Background:  Wnt signalling plays a critical role in many developmental processes and tumorigenesis. Wnt/β-catenin signalling induces the stabilization of cytosolic β-catenin, which interacts with TCF/LEF-1 transcription factors, thereby inducing expression of Wnt-target genes. Recent evidence suggests that a specific MAP kinase pathway involving the MAP kinase kinase kinase TAK1 and the MAP kinase NLK counteract Wnt signalling.

Results:  To identify NLK-interacting proteins, we performed yeast two-hybrid screening. We isolated the gene HMG2L1 and showed that injection of Xenopus HMG2L1 (xHMG2L1) mRNA into Xenopus embryos inhibited Wnt/β-catenin-induced axis duplication and expression of Wnt/β-catenin target genes. Moreover, xHMG2L1 inhibited β-catenin-stimulated transcriptional activity in mammalian cells.

Conclusions:  Our findings indicate that xHMG2L1 may negatively regulate Wnt/β-catenin signalling, and that xHMG2L1 may play a role in early Xenopus development together with NLK.


The canonical Wnt/β-catenin pathway is highly conserved among different species and plays crucial roles in a variety of developmental processes, including the specification of cell fate and polarity, body axis formation, and neural development of metazoan embryogenesis (Cadigan & Nusse 1997; Moon et al. 1997; Wodarz & Nusse 1998; Sokol 1999; Polakis 2000). This pathway is now relatively well understood, as a result of extensive genetic and biochemical experiments in Caenorhabditis elegans, Drosophila, Xenopus and mammalian cells. Wnt ligand stimulation via the Frizzled receptors has been shown to promote inactivation of GSK3β (glycogen synthase kinase 3β), a kinase that phosphorylates cytoplasmic β-catenin and tags it for degradation. This leads to the stabilization and accumulation of β-catenin, which subsequently interacts with the TCF (T cell factor)/LEF (lymphoid enhancer factor) family of transcription factors to activate various Wnt/β-catenin target genes (Cadigan & Nusse 1997; Wodarz & Nusse 1998).

Recent genetic and biochemical studies indicate that a mitogen-activated protein kinase (MAPK)-related pathway composed of TAK1 (TGF-β activated kinase 1) MAPK kinase kinase (MAPKKK) and NLK (Nemo like kinase) MAPK regulates the canonical Wnt/β-catenin pathway (Arias et al. 1999; Behrens 2000). In C. elegans, MOM-4 (more of MS-4) and LIT-1 (loss of intestine-1) were identified as proteins related to the vertebrate TAK1 and NLK, respectively. They cooperate with the canonical Wnt/β-catenin pathway to down-regulate POP-1 (posterior pharynx abnormal-1), a protein related to vertebrate TCF/LEF-1, to establish anterior-posterior polarity during early embryogenesis (Meneghini et al. 1999; Rocheleau et al. 1999; Shin et al. 1999). On the other hand, in mammalian cells, TAK1 activates NLK, which in turn phosphorylates and down-regulates TCF/LEF-1 transcriptional activity (Ishitani et al. 1999, 2003). Moreover, in Xenopus embryos, injection of mouse NLK (mNLK) or Xenopus NLK (xNLK) suppresses the axis duplication induced by ectopic Wnt/β-catenin signalling (Ishitani et al. 1999; Hyodo-Miura et al. 2002). Thus, the TAK1/MOM-4-NLK/LIT-1 MAPK-related pathways appear to regulate Wnt/β-catenin signalling pathway by same system between these species.

The high-mobility-group-box (HMGB) proteins are one of three HMG chromosomal protein superfamilies. HMGB proteins are further classified into two major subgroups. The first subgroup consists of proteins having more than one HMGB domain and a long acidic C-terminal tail. HMGB proteins of this subgroup are usually expressed abundantly, and bind to bent or distorted DNA without particular sequence specificity. The second subgroup consists of HMGB proteins that contain only a single HMGB domain and do not contain an acidic C-terminal tail. HMGB proteins of the second group exhibit a high degree of DNA binding sequence specificity, and act as transcription factors (Laudet et al. 1993; Bustin 1999; Thomas & Travers 2001). TCF/LEF-1 proteins belong to this second subgroup of sequence-specific, single HMGB-domain proteins (van de Wetering et al. 1991), and function as transcription factors to regulate expression of Wnt/β-catenin signalling target genes via interaction with additional coactivators or corepressors, such as β-catenin (van de Wetering et al. 1997) or Groucho/TLE1 (transducin-like enhancer of split 1) (Cavallo et al. 1998; Roose et al. 1998; Levanon et al. 1998), respectively. Recent evidence suggests that other HMGB proteins also modulate Wnt/β-catenin signalling. The Sox (Sry box protein) proteins containing a single HMGB domain seem like to bind specific DNA sequences, and function alone as classical transcription factors or together with other transcriptional regulators to regulate developmental processes (Pevny & Lovell-Badge 1997; Wegner 1999). Experiments in Xenopus embryos and mammalian cells revealed that two members of Sox proteins, xSox17α/β, which were first identified as mediators for endoderm differentiation in Xenopus (Hudson et al. 1997), and another factor xSox3 (Koyano et al. 1997; Penzel et al. 1997) can repress Wnt/β-catenin signalling via their physical interactions with β-catenin (Zorn et al. 1999). SOX7 also blocks β-catenin-stimulated transcriptional activity (Takash et al. 2001). Moreover, SON-1 (sheath-to-neurone transformation-1), an abundant HMG1/2-like nonspecific DNA-binding protein in C. elegans, has been genetically shown to act in opposition to POP-1 in regulating the Wnt/β-catenin pathway (Jiang & Sternberg 1999).

In the present study, we isolated xHMG2L1 as an xNLK-interacting protein. xHMG2L1 is a Xenopus homolog of human HMG2L1, whose function is unknown. We demonstrate here that xHMG2L1 suppresses axis duplication induced by β-catenin over-expression, and inhibits the transcription of Wnt/β-catenin signalling target genes during early Xenopus development. Moreover, xHMG2L1 was found to inhibit β-catenin stimulated-transcriptional activities in mammalian cells. Our results indicate that xHMG2L1 may negatively regulate Wnt/β-catenin signalling.


Isolation of xHMG2L1 as an xNLK-interacting protein

To identify NLK-interacting proteins, we performed yeast two-hybrid screening using the C-terminal half of xNLK (see Experimental procedures) as bait. Thirteen positive clones were sequenced and one was found to encode a partial sequence containing a novel HMGB domain. We isolated the full-length cDNA from a Xenopus oocyte cDNA library, and found a single open reading frame of 554 amino acids that has remarkable homology (65% identity) with the human HMG2L1 (HMG protein 2-like 1 or HMGBCG) gene product (Fig. 1). HMG2L1 was originally identified from the human genome sequence (Seroussi et al. 1999; Dunham et al. 1999), but its function has not been examined. xHMG2L1 contains a single HMGB domain near the C-terminal region, and lacks an acidic C-terminal tail, similar to human HMG2L1. This suggests that xHMG2L1 is one of the sequence-specific HMGB DNA binding proteins, such as the TCF/LEF-1 and Sox proteins (Laudet et al. 1993; Thomas & Travers 2001). Database searches identified several closely related cDNA sequences in mouse and rat, but not in Drosophila and C. elegans (data not shown).

Figure 1.

Primary structure and conserved regions of xHMG2L1. Alignment of the deduced amino acid sequences of xHMG2L1 and human HMG2L1. Identical residues are shaded. Dashes denote gaps. The High Mobility Group (HMG)-box is boxed. Underline indicates the region corresponding to the fragment which was obtained by the yeast two-hybrid screening. The GenBank accession number for xHMG2L1 is AY281348.

xHMG2L1 associates with xNLK

We next sought to confirm that the interaction between xNLK and xHMG2L1 could also be observed in mammalian cells. HEK293 cells were transiently transfected with different combinations of T7-tagged-xHMG2L1 and Flag-tagged-xNLK expression vectors. We found that Flag-xNLK could be co-immunoprecipitated by anti-T7 antibody in lysates from cells co-transfected with T7-xHMG2L1 (Fig. 2, lane 3). In the reciprocal experiment, T7-xHMG2L1 could also be co-immunoprecipitated by anti-Flag antibody from lysates of cells co-transfected with Flag-xNLK (Fig. 2, lane 6). These results show that xHMG2L1 specifically interacts with xNLK in mammalian cells, consistent with the results of the yeast two-hybrid system.

Figure 2.

Association of xHMG2L1 with xNLK.HEK293 cells were transiently transfected with the indicated expression vectors. Cell lysates were immunoprecipitated with anti-Flag antibody or anti-T7 antibody, and then immunoprecipitates (IP) and cell lysates were immunoblotted (IB) with the indicated antibody.

xHMG2L1 was expressed maternally and throughout early embryogenesis

Given the interaction between xNLK and xHMG2L1 observed in both yeast and mammalian cells, we next tested whether the expression of xHMG2L1 might overlap with those of xNLK in early Xenopus embryogenesis. RT-PCR analysis of xHMG2L1 expression in early Xenopus embryos revealed that xHMG2L1 mRNA was expressed maternally and throughout early development (Fig. 3). We next examined localization of xHMG2L1 mRNA expression in embryos at the gastrula stage by RT-PCR, using total RNAs obtained from four microsections, dorsal mesoderm, ventral mesoderm, endoderm and ectoderm at stage 11. xHMG2L1 mRNA was found to be widely expressed at the early gastrula stage (data not shown). Previously we have shown that xNLK is expressed maternally and throughout development, and is expressed widely in the ectoderm during the early gastrula stages (Hyodo-Miura et al. 2002). Taken together, these results suggest that xHMG2L1 might act in concert with xNLK in Xenopus early embryogenesis.

Figure 3.

Developmental expression of Xenopus HMG2L1 by RT-PCR analysis. Each lane number indicates developmental stage: 0 unfertilized egg; 7, 8 blastula; 10.5, 11 gastrula; 15 neurula; 24 tailbud tadpole; 40 swimming tadpole. Histone was used as an internal control.

Injection of xHMG2L1 mRNA blocks Wnt/ β-catenin signalling-induced axis formation in Xenopus embryos

The Wnt/β-catenin signalling pathway is essential for the dorsal axis formation during Xenopus embryogenesis (Moon & Kimelman 1998; Wodarz & Nusse 1998; Sokol 1999). Recent evidence suggests that the TAK1-NLK MAPK-related pathway regulates β-catenin-TCF/LEF-1 function in parallel with the Wnt/β-catenin pathway (Ishitani et al. 1999; Meneghini et al. 1999; Shin et al. 1999). Injection of mNLK or xNLK mRNAs into Xenopus embryos was shown to strongly suppress secondary axis formation induced by ectopic Wnt/β-catenin signalling (Ishitani et al. 1999; Hyodo-Miura et al. 2002). If xHMG2L1 functions in the Wnt/β-catenin signalling, we would expect the injection of xHMG2L1 mRNA might also affect axis formation. It has been shown that injection of β-catenin mRNA into the ventral region of four-cell embryos leads to the formation of an ectopic second axis, including a duplicated head structure (Funayama et al. 1995) (Fig. 4a). We observed that co-injection of xHMG2L1 mRNA significantly inhibited ectopic axis formation by β-catenin mRNA (Fig. 4a,b). Moreover, we tested whether xHMG2L1 could inhibit the expression of endogenous Wnt/β-catenin signalling target genes using an animal cap assay. We found that Xwnt8-induced expression of endogenous Xnr-3 and Siamois, which are direct target genes of Wnt/β-catenin signalling (Carnac et al. 1996; Smith et al. 1995), were inhibited by injection of xHMG2L1 mRNA (Fig. 4c). These results suggest that xHMG2L1 inhibits the Wnt/β-catenin signalling pathway in Xenopus embryos.

Figure 4.

Effects of xHMG2L1 in Wnt signalling in Xenopus embryos. (a) Effects of xHMG2L1 on ectopic axis formation. β-catenin mRNA with or without xHMG2L1 mRNA (1 ng) was injected into two ventral blastomeres at the four-cell stage. Embryos were evaluated until the 35-36 stage and then observed. (b) β-catenin mRNA (30 or 50 pg) was injected into two ventral blastomeres at the four-cell stage, with or without xHMG2L1 mRNA (0.25 ng–1 ng). Secondary axes were scored at the tadpole stages. (c) xHMG2L1 inhibits the expression of Xwnt8-induced target genes. xHMG2L1 mRNA was injected into the animal poles of two blastomeres at two-cell stage, with or without Xwnt8 (10 pg). Total RNAs were prepared and analysed by RT-PCR. The inhibition was enhanced by high dose (1 ng) injection of xHMG2L1 mRNA. Histone was used as an internal control.

xHMG2L1 represses β-catenin-induced transcriptional activation

These experiments in Xenopus embryos suggested that xHMG2L1 inhibits the transcription of Wnt/β-catenin signalling target genes, such as Siamois and Xnr-3. Then, we investigated whether xHMG2L1 expression would affect Wnt/β-catenin-mediated transcriptional activity in mammalian cells, similar to mNLK (Ishitani et al. 1999, 2003). We analysed the effects of xHMG2L1 expression on a Wnt-responsive TCF/LEF-1 reporter construct (3xTCF/Luc) in HEK293 cells. Regulation of β-catenin turnover requires the amino terminal region of the protein (Aberle et al. 1997). The β-cateninΔN is deleted this region. We found that expression of β-cateninΔN caused an approximate 60-fold increase in reporter luciferase activity, relative to cells transfected with the empty vector (Fig. 5). Expression of xHMG2L1 alone did not affect the reporter activity significantly (data not shown). However co-expression of xHMG2L1 with β-catenin reduced luciferase activity relative to β-catenin alone (Fig. 5). This suggests that xHMG2L1 inhibits the β-catenin-induced TCF/LEF-1 transcriptional response, corresponding to the in vivo results obtained in Xenopus embryos. Taken together, these results indicate that xHMG2L1 may negatively regulate the Wnt/β-catenin signalling pathway.

Figure 5.

Effect of xHMG2L1 on β-catenin regulated transcriptional activation. HEK293 cells were transfected with a reporter plasmid (0.05 µg) and expression vectors encoding β-cateninΔN (0.02 µg) and xHMG2L1. Cells were harvested and luciferase activity was measured 24 h after transfection.


In this report, we identified a novel xNLK-interacting HMGB protein, xHMG2L1. Injection of xHMG2L1 mRNA into Xenopus embryos demonstrated that xHMG2L1 can inhibit the Wnt/β-catenin signalling pathway. Moreover, in mammalian cells, xHMG2L1 inhibits the β-catenin-induced transcriptional activity of TCF/LEF-1. These results suggest that xHMG2L1 negatively regulates the Wnt/β-catenin signalling in a fashion similar to xNLK.

How does xHMG2L1 influence Wnt/β-catenin signalling? Zorn and his colleagues reported that xSox17α/β and xSox3, which are also TCF/LEF-1-type transcription factors belonging to the sequence-specific HMGB protein subgroup, have also been shown to repress Wnt/β-catenin signalling via the binding to the same Arm-repeat domains of β-catenin as TCF/LEF-1 (Zorn et al. 1999). They suggested that the binding of Sox-β-catenin may preclude TCF/LEF-1 from interacting with β-catenin, resulting in the repression of Wnt/β-catenin signalling. Alternatively, Sox proteins may be part of a ternary complex, that, together with β-catenin and TCF/LEF-1, inhibits Wnt/β-catenin signalling. Thus, our results suggest that HMG2L1 represses Wnt/β-catenin signalling by interfering with the interaction between β-catenin-TCF/LEF-1, similar to what is observed with the Sox proteins.

NLK has been shown to interact directly with and to phosphorylate TCF/LEF-1, and to interact indirectly with β-catenin to form a complex in a TCF/LEF-1-dependent manner (Ishitani et al. 2003). Moreover, we have shown that xNLK interacts with xSox11 to play an essential role in neural development in Xenopus embryos (Hyodo-Miura et al. 2002). xSox11 is a transcription factor (Hiraoka et al. 1997) of the TCF/LEF-1 type that belongs to the HMGB subgroup, such as xHMG2L1. We also indicated the possibility that phosphorylation and interaction with xNLK could be important for the transactivation of xSox11, and that it functions to promote anterior neural development (Hyodo-Miura et al. 2002). Thus, some TCF/LEF-1-class HMGB proteins may be able to bind to NLK, and possibly their transcriptional activities are modulated through the phosphorylation by NLK. Our results, showing that xHMG2L1 can interact with xNLK in yeast (data not shown) and mammalian cells (Fig. 2), suggest that xHMG2L1 activity is also regulated via phosphorylation by NLK. These observations lead us to hypothesize that NLK, presumably activated by TAK-1, may interact with and phosphorylate xHMG2L1. Thus, the activated xHMG2L1 may then form a complex with NLK and β-catenin, thereby excluding TCF/LEF-1 and preventing the formation of a β-catenin-TCF/LEF-1 complex that would otherwise repress Wnt/β-catenin signalling. However, it has not been shown directly that xNLK phosphorylates xHMG2L1 protein following stimulation of the Wnt signalling pathway. Additional experiments will be required to test this hypothesis, and further molecular analysis will be needed to assess the role of HMG2L1 in Wnt signalling.

Experimental procedures

Plasmid constructs

The C-terminal region encoding amino acids 202-447 of xNLK was fused to the GAL4 DNA binding domain of pGBDU-C2 (James et al. 1996) to construct pGBDU-xNLK-C. This construct was used as bait in the yeast two-hybrid screening. The full-length xNLK cDNA (Hyodo-Miura et al. 2002) was also cloned into pCS2+ Flag (for the Flag tag at the amino terminus) to construct the mammalian expression plasmids pCS2Flag-xNLK. Full-length xHMG2L1 cDNA (see below) was cloned into pRK5 T7 (providing the T7 tag at the amino terminus) and used to construct the mammalian expression plasmid pRK5 T7-xHMG2L1. pCS2-xHMG2L1, which contains full-length xHMG2L1 inserted into pCS2+ (Rupp et al. 1994), was constructed as a template of the mRNA synthesis for injection into Xenopus embryos.

Yeast two-hybrid screening and cDNA cloning

The yeast two-hybrid screening was performed as described by James et al. (1996). A Xenopus laevis oocyte MATCHMAKER cDNA Library (Clontech) was screened using pGBDU-xNLK-C as bait. 3 × 106 clones from a Xenopus oocyte cDNA library were screened, and 13 positive clones were obtained. One positive clone contained a 2.4 kb cDNA insert, which encodes partial DNA of xHMG2L1. The full-length xHMG2L1 cDNA was isolated from a Xenopus oocyte cDNA library in the Lambda Zap vector using a cDNA fragment obtained from the yeast two-hybrid screening as probe. Positive clones were isolated and subcloned into the pBluescript vector. The longest insert cDNA among the subclones was sequenced on both strands by primer walking, and a 2626-bp cDNA sequence was identified. We concluded that this cDNA encodes the full-length xHMG2L1 cDNA, because of the presence of stop codons located upstream of the putative translation initiation site (data not shown).

Antibodies, immunoprecipitation and immunoblots

Monoclonal antibodies against Flag and T7 were purchased from Sigma and Novagen, respectively. Rabbit polyclonal antibody against Flag was purchased from Sigma. HEK293 cells were transiently transfected by the calcium phosphate precipitate method with 1 µg of the xHMG2L1 expression plasmid and 2 µg of the xNLK expression plasmid in a 6 cm dish. The total amount of transfected DNA was equalized to 3 µg as necessary using empty vector DNA. For co-immunoprecipitation, cells were harvested at 24 h post-transfection and lysed in lysis buffer (50 mm Tris-HCl pH 7.4, 150 mm NaCl, 5 mm EDTA, 50 mm NaF, 0.5% NP-40, 1 mm Na3VO4, 1 mm PMSF, 1 µg/mL aprotinin, 1 µg/mL pepstatin, 1 µg/mL Leupeptin and 1 mm DTT). Lysates were incubated with 1 µg of monoclonal anti-Flag or anti-T7 antibody coupled to Protein G-Sepharose beads (Amersham) for 15 h. The immunoprecipitates were washed three times with lysis buffer. Immunoprecipitates and total cell lysates were resolved by SDS-PAGE, and transferred to Hybond P membrane (Amersham). The membranes were immunoblotted with the appropriate polyclonal antibodies, and visualized with HRP-conjugated rabbit IgG using Western LightningTM Chemiluminescence Reagent Plus (PerkinElmer).

Transcriptional activity assays

HEK293 cells (2-4 × 105 cells per well) were seeded into six-well plates. Indicated plasmids were transfected by the calcium phosphate precipitate method 24 h after seeding, and cells were harvested and assayed for luciferase activity after 24 h. Total amount of DNA per transfection was equalized using empty vector DNA. Renilla rennifrmis luciferase expression construct under the control of the EF-1α promoter was co-transfected to normalize for transfection efficiencies.

Embryonic manipulation and microinjection of synthetic mRNA

Unfertilized eggs were collected and were fertilized in vitro as previously described (Suzuki et al. 1997). The embryos were subsequently dejellied using 3% cysteine, and were staged according to Nieuwkoop & Faber (1967). Capped mRNAs for each protein to be tested were synthesized from linearized constructs using the mMESSAGE mMACHINE SP6 kit (Ambion). For animal cap (AC) assays, mRNAs were injected into the animal poles of 2-cell-stage embryos. Injected embryos were kept in 3% Ficoll/0.1 × Steinberg's solution as previously described (Yamamoto et al. 2000). AC explants were dissected with hair knives at stage 8–9. For second axis induction, synthetic mRNAs were injected into two ventral blastomeres at the four-cell stage. Embryos were examined for ectopic axis formation at the tadpole stage.

RT-PCR analysis

ACs were collected from sibling embryos at stage 12, and total RNA was extracted using TRIzol (Gibco/BRL) according to the manufacturer's instructions. cDNA was synthesized from extracted RNAs as previously described (Yamamoto et al. 2000) for RT-PCR analysis. PCR was performed with the oligonucleotide primer pairs: Xnr-3, 5′-CCATGTGAGCACCGTTCC-3′ (upstream) and 5′-GAGCAAACTCTTAATGTAG-3′ (downstream); Siamois, 5′-GATAACTGGCATTCCTGAGC-3′ (upstream) and 5′-ACAAGTCAGTGTGGTGATTC-3′ (downstream); xHMG2L1, 5′-AAATTGAGAGCAGGTCCAGCCACAAGAAAC-3′ (upstream) and 5′-GCTTGTGGTTGAAGA GGTATGGATGGTGGT-3′ (downstream). The primer sequence of histone H4, an internal input control, was previously described (Iemura et al. 1998).


We thank Marc Lamphier for critical reading of the manuscript. This work was supported by Grants-in-Aid for scientific research from the Ministry of Education, Culture, Sports, Science and Technology of Japan and Yamanouchi Foundation for Research on Metabolic Disorders.