Sequence comparisons and 3D structures of metazoan β-catenins
In analyzing early evolution of the β-catenin complex, we first examined conservation of β-catenin sequences of early-branching nonbilaterians (Fig. 1). This is important for understanding post-translational modifications and direct protein interaction sites. As a reference structure, we included ARM6, a β-catenin-like protein from the choanoflagellate, Salpingoeca rosetta. Choanoflagellates are the sister group of Metazoa. Choanoflagellate ARM6 is ancestral to metazoan β-catenin and Adenomatous Polyposis Coli (APC) groups and has fewer armadillo repeats (Fig. 1A, B) [27]. Overall, cnidarian and poriferan β-catenins share many functional motifs with bilaterian β-catenin, but the proportion decreases in ctenophores, and except for a few phosphorylation sites, most functional motifs are not conserved in ARM6 of Choanoflagellate.
In our comparison, GSK3β and CK1α phosphorylation sites critical for β-catenin degradation are apparently conserved in all metazoans, except for Hydra, which lacks the site corresponding to S37 (Fig. 1B, S1, S2). Interestingly, in unicellular S. rosetta, two residues corresponding to S33 & T41 were also conserved. The central domain is the leading interaction site for several proteins, including TCF and Cadherin1(CDH1) [28, 29]. A high degree of conservation was observed in Bilateria, Cnidaria, and Porifera in the central region (Fig. 1B). Cnidarian and poriferan β-catenins displayed full conservation of the three critical lysine residues (K312, K335, and K435 in mouse β-catenin) and others (Y331, D390, and R582) that are required for E-cadherin binding [28, 29]. Among these sites, only two lysines, K312 and K435, were conserved in Ctenophora (Fig. 1B, S1).
Three arginine residues of mouse β-catenin, R474, R582, and R612, are important in binding to TCF [28, 30]. These amino acids are conserved in Cnidaria and Porifera, but in Ctenophora, the residue corresponding to R474 is replaced with a lysine residue, and R582 & R612 are not conserved. To gain further insight into the influence of these differences on binding of TCF, we investigated other amino acids, K312, N426, K435, R469, H470, K508, H578 and Y654, that are thought to be involved in β-catenin-TCF interactions in mice [28, 30, 31] (Fig. 1B, S1). Like the three important arginine residues (R474/R582/R612) mentioned above, these amino acids are also common among Bilateria, Cnidaria, and Porifera, and all except H578 and Y654 are conserved in Ctenophora. Interestingly, most of these amino acids that function in binding of β-catenin to TCF are not conserved in S. rosetta. However, S. rosetta ARM6 has asparagine corresponding to N426 and arginine at a position corresponding to lysine (K312), which has chemical properties similar to those of lysine. The C-terminus of β-catenin, which has a transactivation domain, is important for regulation of gene expression by the β-catenin/TCF complex [32, 33]. Its signaling ability is probably due to two motifs, A and B [33–36]. Alignment combined with Multiple Expectation maximization for Motif Elicitation (MEME) analysis confirmed that motif A is present only in Bilateria and Cnidaria, whereas motif B is also present in Porifera (Fig. 1B, S1, S3). On the other hand, neither motif A or B exists in ctenophore β-catenin and choanoflagellate ARM6.
Finally, we compared amino acids required for binding to α-catenin. The α-catenin binding site in β-catenin was previously narrowed down to amino acids 118–146 (in mice) [37].
Metazoan-wide conservation across this region is not very high, except for a few residues. In mouse β-catenin, Y142 is vital for α-catenin binding, since mutation to alanine eliminates β-catenin-α-catenin interaction [37]. The site corresponding to Y142 is conserved in all metazoans (Fig. 1B, S1, S4). Two acidic residues, D144 and E147, also affect interaction with α-catenin [38]. D144 is not conserved among poriferans. S. rosetta has a large insertion sequence in this region, which precluded accurate verification.
β-catenin sequences from Xenopus laevis (Vertebrata), Nematostella vectensis (Cnidaria), Ephydatia fluviatilis (Porifera), Bolinopsis mikado (Ctenophora), and ARM6 of S. rosetta were then used in homology modeling with the crystal structure of mouse β-catenin (PDB ID: 4ev8) to investigate structural evolution. Phi & Psi distributions of Ramachandran plots generated with PROCHECK indicated that over 90% of amino acids in modeled structures were in favored positions (Fig. S5). Despite amino acid variations of β-catenin observed among metazoans (Fig. S6), the structure of the predicted armadillo repeat region of β-catenin appears to be stable since emergence of the Metazoa. Interestingly, between residues 452–878, the S. rosetta ARM6 protein possesses a structure that resembles metazoan β-catenins (Fig. S7). However, S. rosetta ARM6 has a short loop, whereas metazoan β-catenins have a long loop between armadillo repeats 10 and 11, indicating that this is probably a metazoan invention (Fig. 1A). The armadillo repeat region contains multiple amino acids that are or may be involved in TCF binding regions (Fig. 1B). These amino acids are conserved in β-catenins from nonbilaterians. We also demonstrated steric conservation of these amino acids. Thus, in addition to sequence similarities, orientations of TCF-binding amino acids are highly conserved in β-catenin from Bilateria, Cnidaria, and Porifera (Fig. 1C). In Ctenophora, homology was observed only in a few amino acids.
Given their structural similarities, the next question is, “To what extent are functions of β-catenins conserved?” Functional analysis of nonbilaterian β-catenins, using mainly Nematostella, shows that β-catenin is important in the blastoporal organizer and during gastrulation and subsequent endoderm fate determination. On the other hand, in poriferans and ctenophores, knowledge of β-catenin function is much more limited [12, 15, 16, 39]. During developmental of Mnemiopsis leidyi (Ctenophora), perturbation of classical β-catenin signaling had limited effect [12]. Classical β-catenin complexes have also been reported in adult Ephydatia, but their functions remain unknown [16].
Comparison of axis-inducing activity of metazoan β-catenins in Xenopus embryos
To investigate conservation of β-catenin activity, we performed a secondary axis induction assay using an ectopic expression system with Xenopus embryos. In vivo embryonic systems, which require a variety of cellular developmental events such as epithelial formation, active cell division, and formation of signal centers necessary for body axis patterning, are ideal for comprehensive functional surveys of the β-catenin complex.
Differences in codons in mRNAs of different organisms can affect rates of translation when expressed in other organisms. Therefore, we began by optimizing the amount of mRNA injected to achieve expression of relatively similar levels of FLAG-tagged proteins (Fig. S8A). For all metazoan β-catenins, FLAG-tagged β-catenin protein was observed at the expected band sizes. However, expression of S. rosetta ARM6 in Xenopus could not be confirmed even when the amount of injected mRNA was increased (Fig. S8B). Therefore, the following analysis was performed using only metazoan β-catenins.
To test whether basal metazoan β-catenins are functional, they were injected into the ventral equatorial region of one blastomere of Xenopus 4-cell embryos. Figure 2B shows that injection of 100 pg mRNA resulted in expression of FLAG-tagged β-catenin proteins. These results showed that N. vectensis and E. fluviatilis β-catenin induced a secondary body axis similar to that induced by X. laevis β-catenin (Fig. 2A-C). In contrast, B. mikado β-catenin did not induce a secondary axis. Even when B. mikado β-catenin mRNA was injected at high concentrations (> 500 pg), only protrusion-like structures without head characteristics appeared in some embryos (Fig. S9). This suggests that the lack of secondary axis-inducing ability of B. mikado β-catenin is due to differences in signals it can activate, rather than to weakness of activation.
To further confirm whether basal metazoan β-catenins can drive β-catenin/TCF signaling, a TOPflash luciferase assay using Xenopus embryos was performed [40, 41]. Consistent with secondary axis induction experiments, there was a significant increase in luciferase activity following expression of X. laevis, N. vectensis, and E. fluviatilis β-catenin (Fig. 2D), confirming that signaling activity of β-catenin is conserved in Cnidaria and Porifera. On the other hand, expression of B. mikado β-catenin did not increase luciferase activity. This reflects reduced conservation of amino acids necessary for TCF binding in ctenophore β-catenins (Fig. 1B). Additionally, together with the finding that β-catenin inhibition in Mnemiopsis does not clearly affect conserved functions, such as body axis formation [12], these data suggest that ctenophore β-catenin does not contribute significantly to canonical Wnt signaling (Wnt/β-catenin/TCF signaling).
The capacity of cnidarian and poriferan β-catenins to induce phenotypes similar to that of Xenopus β-catenin suggests that they all form a common protein complex in Xenopus embryos. Therefore, we next examined the protein complex formed by each β-catenin in developing Xenopus embryos.
Proteomic analysis of β-catenin protein complexes.
To investigate protein complexes made by each basal metazoan β-catenin in developing Xenopus embryos, FLAG-tagged β-catenin was immunoprecipitated from homogenates at gastrula stage 11, and proteomic analysis of resulting protein complexes was performed. Western blotting confirmed that comparable amounts of exogenous FLAG-tagged β-catenin proteins were immunoprecipitated (Fig. S10). Subsequently, interacting proteins were identified using liquid chromatography with tandem mass spectrometry. An IgG control analysis from uninjected embryos was also included. To reduce false positives in identified proteins, only proteins with an abundance ratio greater than or equal to 2 (p-value < 0.05) are generally used as a threshold for “true” interactions. The presence of known bilaterian β-catenin interacting proteins confirmed that identified proteins represented “true” interactions (Fig. 3A, Table S1). Figure 3B shows a schematic of proteins co-immunoprecipitated with each β-catenin species. A number of proteins, including Cadherin (CDH), have been identified as interacting partners with all exogenous metazoan β-catenins. Our BLAST search confirmed that many of them have protein homologs in nonbilaterian metazoans.
Unexpectedly, many metabolism-related proteins, such as DHRS12, PHGDH, SDR39U1, and MOCS3, were commonly detected in bilaterian/nonbilaterian β-catenin complexes (Fig. 4). Although there are no reports that these enzymes interact with β-catenin, and their physiological significance in the β-catenin complex is currently unknown, overexpression of DHRS12 inhibits β-catenin signaling in human cell lines [42]. Another group of highly conserved components of the β-catenin complex includes proteins associated with cell adhesion. Since conservation of CDH1 binding sites on β-catenin is very high among metazoans (Fig. 1B, Fig. S1), it is not surprising that CDH1 was precipitated by all β-catenins.
Interestingly, E. fluviatilis β-catenin was unable to immunoprecipitate X. laevis α-catenin at detectable levels. This was also confirmed in western blotting analysis (Fig. S11). Poriferan β-catenins do not have the conserved amino acids required for binding to α-catenin (Fig. S4). On the other hand, binding of endogenous β-catenin and α-catenin has been observed in the poriferan, Ephydatia muelleri [16]. This suggests that poriferan ancestors evolved a unique interaction of β- and α-catenin.Given that β-catenin proteins in a broad array of nonbilaterians bind α-catenin and phylogenetically distant CDH1 proteins, indicates that β-catenin complex functions involved in the adherens junction were acquired among the earliest metazoans (even if there were unique modifications in the Porifera) and remain highly conserved. Since organizer-inducing activity was observed in bilaterian, cnidarian, and poriferan β-catenins in developing Xenopus embryos, we expected to detect protein complexes specific to these β-catenin immunoprecipitates. However, proteomic analysis detected only a few proteins common to bilaterian/cnidarian/poriferan β-catenins. Smoothelin (SMTN) was identified as a binding protein common to β-catenin of X. laevis, N. vectensis, and E. fluviatilis. In bilaterians, SMTN binds Cortactin (CTTN) and stabilizes the cortical actin meshwork of epithelial cell membrane [43]. Cortactin binds β-catenin, α-catenin, and p120 catenin (catenin delta-1/ CTNND1) and is important for the function of adherens junctions [44]. VTG1, which has an unknown function, was also detected in this group; however, even though gene homologs of SMTN and VTG1 exist in X. laevis and N. vectensis, they are not in the genome of E. fluviatilis (Fig. S12).