LRP6ICD Promotes β-Catenin Stabilization and Axin Degradation in Xenopus Egg Extract.

To establish a cell-free system that would facilitate biochemical analysis of LRP6 signaling, we tested whether recombinant LRP6ICD, which activates Wnt signaling in vivo, prevents degradation of β-catenin in Xenopus egg extract. We find that LRP6ICD protein prevents degradation of radiolabeled, in vitro-translated (IVT) β-catenin in Xenopus egg extract (Fig. 1E). Consistent with a proposed mechanism for LRP6 signaling, we demonstrate that LRP6ICD also stimulates degradation of IVT Axin and Axin2 (Fig. 1E). We also tested whether LRP6ICD induces phosphorylation of Axin. We find that λ phosphatase reverses the LRP6ICD-mediated upward mobility shift of the Axin protein detected by SDS/PAGE, suggesting that LRP6ICD promotes Axin phosphorylation [supporting information (SI) Fig. S1]. However, in the presence of LRP6ICD, the total Axin signal is decreased even after λ phosphatase treatment, consistent with LRP6ICD mediating Axin degradation.

The ability of LRP6 to stabilize β-catenin depends on GSK3's phosphorylation of the serine residue on at least one of five Pro-Pro-Pro-Ser-Pro (PPPSP) motifs on LRP6 (21, 22). If LRP6ICD accurately reconstitutes endogenous LRP6 signaling in extract, LRP6ICD's activity should depend on intact PPPSP motifs. An LRP6 construct in which all five PPPSP motifs have been mutated to PPPAP (PPPAPX5) does not bind Axin or stabilize β-catenin in cultured cells (21). This construct also fails to activate Wnt target genes in Xenopus ectodermal explants (21). To test whether LRP6ICD signaling in egg extract requires intact PPPSP motifs, we expressed and purified LRP6ICD(PPPAPX5) protein from bacteria. In contrast to LRP6ICD, LRP6ICD(PPPAPX5) does not inhibit β-catenin degradation or stimulate Axin degradation in egg extract (Fig. 1F), even when added at a concentration 2-fold higher than that used for LRP6ICD (data not shown). We also find that LRP6ICD, but not LRP6ICD(PPPAPX5), is phosphorylated at PPPSP Ser-1490 in egg extract (21, 22) (Fig. 3C). Requirement of these PPPSP motifs suggests LRP6ICD in extract functions in a manner that is similar to that of LRP6 in cultured cells and Xenopus embryos.

Open in a separate window

Fig. 3.

LRP6ICD's inhibition of GSK3-mediated β-catenin phosphorylation is sufficient to promote β-catenin stabilization in the absence of Axin degradation. (A) LRP6ICD inhibits β-catenin degradation in extract where endogenous Axin is replaced by nondegradable Axin1-713. Addition of IVT Axin or Axin1-713 restores the ability of Axin-depleted extract to degrade radiolabeled β-catenin. LRP6ICD inhibits both IVT Axin and Axin1-713-induced β-catenin-degradation. (B) LRP6ICD does not affect the ability of Axin to bind GSK3 or β-catenin in egg extract. Endogenous Axin was immunoprecipitated from extract and immunoblotted for GSK3, β-catenin, and Axin. (C) Incubation of LiCl (50 mM) or LRP6ICD [but not LRP6ICD(PPPAPX5)] in egg extract (30 min) inhibits phosphorylation of endogenous β-catenin at GSK3 target sites P33/37/41. Immunoblot of LRP6ICD from the same gel reveals that LRP6ICD, but not LRP6ICD(PPPAPX5), is phosphorylated at the PPPSP Ser-1490. All samples were blotted from a single gel. (D) LRP6ICD inhibits GSK3-mediated β-catenin phosphorylation in extract in which endogenous Axin is replaced by nondegradable Axin1-713. Axin depletion did not affect total β-catenin levels as assayed by immunoblot (Fig. S7). Depletion of endogenous Axin prevents β-catenin P33/37/41 phosphorylation. Addition of IVT Axin1-713 restores β-catenin phosphorylation in Axin-depleted extract. LRP6ICD inhibits IVT Axin1-713-induced β-catenin phosphorylation. Extracts were analyzed after 2-h incubation. All samples were blotted from a single gel; intervening lanes were removed for clarity.

LRP6ICD Signals Independently of Disheveled in Xenopus Egg Extract and Embryos.

Disheveled (Dsh) is a cytoplasmic protein required for signaling downstream of Fz and upstream of the β-catenin destruction complex (1). In cultured mammalian cells, overexpression of LRP6 that lacks its extracellular domain promotes Wnt signaling despite down-regulation of Dsh by RNAi or overexpression of a dominant-negative form of Dsh (23), suggesting that the intracellular domain of LRP6 can signal independently of Dsh. More recently, it was shown that Dsh is required for LRP6 oligomerization and phosphorylation (24), which are necessary for LRP6-mediated activation of Wnt/β-catenin signaling. Interestingly, LRP6 expressed without its extracellular domain bypasses this requirement for Dsh and is constitutively oligomerized and phosphorylated (24). These data suggest that LRP6ICD may mimic Dsh-activated LRP6 and circumvent the requirement for Dsh in Wnt/β-catenin signaling.

To test whether LRP6ICD signaling in Xenopus egg extract bypasses its requirement for Dsh, we immunodepleted endogenous Dsh from egg extract (16). Depletion of Dsh (Fig. S2A) did not affect the ability of LRP6ICD to stabilize β-catenin or promote Axin degradation (Fig. S2B). To determine whether Dsh is required for LRP6ICD signaling in vivo, we tested whether Xdd1 (a dominant negative form of Dsh) (25) prevents LRP6's activation of the Wnt/β-catenin pathway in Xenopus embryos. In mRNA coinjection experiments, Xdd1 inhibits Wnt8-induced secondary axis formation but has no effect on the ability of LRP6ICD to induce secondary axes (Fig. S2C). Thus, our data in Xenopus egg extract and embryos demonstrate that LRP6ICD signals independently of Dsh and are consistent with a model in which LRP6ICD mimics Dsh-activated LRP6 in Wnt/β-catenin signaling (24).

Axin-bound GSK3 has been suggested to play a role in phosphorylation and activation of LRP6 (26). Because phosphorylation of LRP6 is a prerequisite for its binding to Axin (21), however, the initial phosphorylation of LRP6 may occur by a pool of GSK3 that is not bound to Axin. In egg extract where Axin has been immunodepleted, we find that LRP6ICD still becomes phosphorylated at PPPSP Ser 1490 as assayed by immunoblot (data not shown), suggesting that initial LRP6 phosphorylation may occur independently of Axin.

LRP6ICD-Mediated Axin Degradation Occurs via the Ubiquitin/Proteasome Pathway and Is Distinct from GSK3-Regulated Axin Degradation.

To identify the mechanism by which LRP6 promotes Axin degradation, we tested whether LRP6ICD induces Axin degradation via a ubiquitin-mediated, proteasome-dependent process. We find that LRP6ICD promotes Axin ubiquitination in Xenopus egg extract (Fig. 2A). Furthermore, we show that inhibition of the proteasome with MG132 prevents Axin degradation, leading to accumulation of a more slowly migrating form of Axin (Fig. 2B). Thus, our data indicate that, consistent with results from intact Xenopus oocytes (8), Axin degradation is proteasome dependent in egg extract.

Open in a separate window

Fig. 2.

LRP6ICD mediates Axin degradation independently of GSK3 inhibition. (A) LRP6ICD stimulates addition of GST-ubiquitin to radiolabeled, IVT Axin in egg extract. GST-ubiquitin conjugates were pulled down with glutathione beads at indicated times and analyzed by using SDS/PAGE and autoradiography. Asterisk indicates full-length Axin. (B) Addition of the proteasome inhibitor MG132 (1 mM) to egg extract inhibits LRP6ICD-mediated Axin degradation. (C) Degradation of Axin mutants in egg extract in the presence of LRP6ICD. The indicated LRP5/6-binding region on Axin is based on previous Axin-LRP5 and Axin-Arr yeast two-hybrid studies (6, 7); dotted lines represent large deletions of Axin that were not further mapped, and the borders of Axin-LRP5/6 interaction likely reside within the dotted lines. RGS, RGS domain; GBS, GSK3 binding site; βBS, β-catenin binding site; DIX, DIX domain. (D) LRP6ICD promotes degradation of Axin and Axin375–869, whereas the GSK3 inhibitor BIO (50 μg/ml, Calbiochem) promotes degradation of Axin but not Axin375–869. (E) Inhibition of GSK3-mediated Axin phosphorylation [by LiCl (50 mM) or mutagenesis (AxinSA)], but not incubation with LRP6ICD, alters the trypsin proteolysis pattern of IVT Axin after incubation in egg extract for 30 min (note bands at level of asterisk). All experiments used equal concentrations of IVT Axin. An SDS/PAGE autoradiograph of IVT Axin and AxinSA before trypsin treatment is shown at right.

To uncover structural elements of Axin required for its LRP6-mediated degradation, we analyzed a panel of truncated Axin polypeptides. We identified a minimal Axin fragment (Axin375–809) that degrades in response to LRP6ICD (Fig. 2C and Fig. S3). This minimal fragment includes the GSK3, β-catenin, and PP2A binding sites on Axin (27). However, deletion of the GSK3 or β-catenin binding domain from full-length Axin does not prevent its LRP6ICD-mediated turnover (Fig. 2C); thus, binding of Axin to GSK3 or β-catenin may not be required for LRP6-mediated degradation of Axin. Interestingly, amino acids 375–427 appear to be required in the large N-terminal truncation mutants (compare Axin375–809 and Axin427–809) but not in the internally truncated AxinΔGBS; we believe this may result from abnormal folding of certain truncation mutants, redundancy within Axin regarding sequences required for LRP6ICD-mediated Axin degradation, and/or dimerization of certain Axin mutants with endogenous Axin (28). Notably, we find that the region of Axin identified as binding LRP5/6 by yeast two-hybrid assays (6, 7) also appears to be required for its LRP6-mediated degradation (Fig. 2C). These data are consistent with a model in which LRP6/Axin binding is required for LRP6-mediated Axin degradation.

Several models for Wnt pathway activation involve inhibition of GSK3, positing global inhibition of GSK3 within the cell or specific inhibition of GSK3 within the β-catenin destruction complex. Either mechanism would allow β-catenin levels to rise because its phosphorylation, which is necessary for its degradation, is blocked. Experiments suggest an inherent feed-forward mechanism whereby GSK3 inhibition also stimulates Axin degradation by preventing phosphorylation of Axin, which is normally required for its stability (29). Thus, we tested whether LRP6ICD promotes Axin degradation by inhibiting GSK3-mediated phosphorylation of Axin.

If LRP6ICD promotes turnover of Axin by inhibiting its GSK3-mediated phosphorylation, Axin mutants that degrade in response to LRP6ICD should also be able to degrade in response to GSK3 inhibition. Alternatively, if LRP6ICD-mediated Axin turnover does not occur via GSK3 inhibition, certain Axin mutants may degrade in response to LRP6ICD but not in response to GSK3 inhibition. We find evidence in support of the latter model. Both LRP6ICD and the GSK3 inhibitor BIO (Fig. 2D) promote turnover of full-length Axin; in contrast, Axin375–869 degrades in response to LRP6ICD but not the GSK3 inhibitor BIO (Fig. 2D). Another GSK3 inhibitor, lithium (50 mM), also promotes turnover of full-length Axin but not Axin375–869 (data not shown). Furthermore, Axin mutants lacking previously identified GSK3 phosphorylation and binding sites as well as an Axin mutant (AxinSA) in which predicted GSK3 phosphorylated serines are mutated to alanines (29) degrade in response to LRP6ICD (Fig. 2C and Fig. S3). Together, these data indicate that LRP6 is unlikely to promote Axin degradation via a mechanism that inhibits GSK3-mediated stabilization of Axin.

Distinct mechanisms underlying LRP6-mediated and GSK3 inhibition-mediated Axin degradation may induce different Axin conformations. Because changes in a protein's conformation may expose or conceal certain tryptic cleavage sites, a protein's tryptic proteolysis pattern is traditionally used to detect conformational changes (12, 30, 31). Incubation of radiolabeled, IVT Axin in egg extract followed by partial trypsin proteolysis results in a characteristic Axin digestion pattern upon analysis by SDS/PAGE and autoradiography (Fig. 2E). Trypsin digestion of Axin lacking GSK3 phosphorylation [either via mutation (AxinSA) or incubation with a GSK3 inhibitor (LiCl)] results in a proteolysis pattern distinct from that of wild-type Axin. Incubation of Axin with LRP6ICD, however, yields a digestion pattern that is indistinguishable from that of Axin alone. Because addition of LRP6ICD and inhibition of GSK3 phosphorylation have distinct effects on Axin conformation as assayed by trypsin digest, we propose that LRP6 signaling and GSK3 inhibition affect Axin through different mechanisms. These data are consistent with evidence in Drosophila embryos that Arrow, the Drosophila LRP5/6 ortholog, can promote Axin degradation in the absence of GSK3 activity (7).

LRP6ICD-Mediated β-Catenin Stabilization Does Not Require Axin Degradation.

Although we hypothesize that LRP6-mediated degradation of Axin, a required component of the β-catenin destruction complex, leads to β-catenin stabilization, we wanted to determine whether this is the only mechanism by which LRP6ICD stabilizes β-catenin. To test this model, we assessed whether LRP6ICD can stabilize β-catenin in egg extract in which endogenous Axin is replaced by a nondegradable Axin mutant, Axin1-713 (Fig. 3A). Axin1-713, like full-length Axin, ventralizes Xenopus embryos (indicative of inhibition of Wnt/β-catenin signaling) (data not shown), stimulates β-catenin degradation in egg extract (Fig. 3A), and binds LRP6ICD in egg extract (Fig. S4). Thus, Axin1-713 retains all measurable activities of full-length Axin except that it is not degraded in response to LRP6ICD (Fig. 2C). Consistent with the requirement of Axin for destruction complex formation, immunodepletion of endogenous Axin from extract (Fig. S5A) prevented β-catenin degradation (16) (Fig. 3A). Addition of IVT Axin1-713 to Axin-depleted extract restored β-catenin degradation to an extent similar to that of addition of full-length Axin. We then tested whether LRP6ICD inhibits β-catenin degradation in Axin1-713-rescued extract. As shown in Fig. 3A, LRP6ICD inhibits β-catenin degradation in extract where endogenous Axin is replaced by either full-length Axin or non-degradable Axin1-713. Thus, LRP6ICD can inhibit β-catenin degradation independently of Axin degradation in Xenopus egg extract.

LRP6ICD Prevents GSK3-Mediated Phosphorylation of β-Catenin.

We next sought to identify the mechanism by which LRP6 stabilizes β-catenin independently of Axin degradation. It has been proposed that LRP6 might inhibit β-catenin degradation by promoting dissociation of the β-catenin destruction complex (32). To test this model, we immunoprecipitated Axin from egg extract incubated in the presence or absence of LRP6ICD and immunoblotted for GSK3 or β-catenin. As shown in Fig. 3B and Fig. S6, LRP6ICD (at a concentration that inhibits β-catenin degradation in Xenopus egg extract) does not affect Axin's ability to bind GSK3 or β-catenin. Thus, our data suggest that LRP6 does not sequester Axin from GSK3 or β-catenin.

Alternatively, LRP6 could stabilize β-catenin by directly preventing its phosphorylation within the destruction complex. CK1α phosphorylates β-catenin at Ser-45 (P45) to prime it for GSK3's phosphorylation at Ser-33/Ser-37/Thr-41 (P33/37/41), which is required for β-catenin polyubiquitination and degradation (33). Previous studies showed that Wnt signaling inhibits GSK3-mediated β-catenin phosphorylation but does not inhibit CK1α-mediated β-catenin phosphorylation (33). We therefore tested whether LRP6ICD inhibits the appearance of GSK3-phosphorylated β-catenin in egg extract. Significantly, LRP6ICD, like the GSK3 inhibitor lithium, inhibits GSK3-mediated phosphorylation of β-catenin (Fig. 3C). If LRP6 stabilizes β-catenin through inhibition of β-catenin phosphorylation, LRP6ICD's requirement for intact PPPSP motifs to stabilize β-catenin should extend to LRP6ICD's inhibition of β-catenin phosphorylation. Indeed, LRP6ICD(PPPAPX5), which does not inhibit degradation of β-catenin (Fig. 1F), does not inhibit GSK3's phosphorylation of β-catenin (Fig. 3C). Notably, we find that LRP6's PPPSP serine Ser-1490 is phosphorylated in extracts (Fig. 3C). Thus, LRP6ICD inhibits phosphorylation of β-catenin, likely through a mechanism that requires serine phosphorylated PPPSP motifs.

Phosphorylation of β-catenin by GSK3 requires its recruitment into the β-catenin destruction complex, which is mediated in part by Axin. Thus, it was possible that LRP6ICD-induced inhibition of GSK3's phosphorylation of β-catenin was a direct consequence of LRP6-mediated Axin degradation. To address this possibility, we tested whether LRP6ICD inhibits β-catenin P33/37/41 phosphorylation in egg extract in which Axin is replaced by non-degradable Axin1-713 (Fig. 3D and Fig. S7). Axin depletion (Fig. S5B) from extract inhibited GSK3's phosphorylation of β-catenin, consistent with Axin's role as a required scaffold for this phosphorylation event. Addition of non-degradable IVT Axin1-713 to Axin-depleted extract restored β-catenin P33/37/41 phosphorylation. LRP6ICD blocked this Axin1-713-induced β-catenin phosphorylation (Fig. 3D), demonstrating that LRP6ICD can inhibit phosphorylation of β-catenin by GSK3 independently of Axin degradation.

LRP6ICD in egg extract could specifically prevent β-catenin phosphorylation or act as a general GSK3 inhibitor (possibly by GSK3 sequestration) (19). If the former is correct, LRP6 should inhibit β-catenin phosphorylation without affecting phosphorylation of another GSK3 substrate (e.g., Tau) (Fig. 4A). In egg extract supplemented with exogenous GSK3, recombinant Tau is phosphorylated at its well characterized GSK3 target site Ser-396 (P396) (34). In contrast to lithium, which robustly inhibits GSK3's phosphorylation of both β-catenin and Tau, LRP6ICD inhibits phosphorylation of β-catenin but not of Tau. Thus, our data indicate that levels of LRP6ICD that stabilize β-catenin in egg extract inhibit GSK3-mediated β-catenin phosphorylation without affecting global GSK3 activity. Our finding that LRP6ICD does not act by inhibiting total GSK3 activity is also supported by our data demonstrating that LRP6ICD and lithium have distinct effects on Axin's trypsin proteolysis pattern and that LRP6 promotes Axin degradation independently of GSK3 inhibition (Figs. 2C, ​,22D, and 2E). Although previous experiments suggested that LRP6 inhibits global GSK3 activity, the concentration of LRP6 intracellular domain in those experiments was not reported and may have been significantly greater than the concentration of LRP6ICD used in our experiments (19). Indeed, we detected inhibition of GSK3's phosphorylation of both β-catenin and Tau at higher concentrations of LRP6ICD than those required to inhibit β-catenin phosphorylation in our assays (data not shown).

Open in a separate window

Fig. 4.

LRP6ICD directly and specifically inhibits GSK3's phosphorylation of β-catenin. (A) LiCl (50 mM) inhibits phosphorylation of β-catenin P33/37/41 and exogenous Tau P396 by GSK3, whereas LRP6ICD inhibits phosphorylation of β-catenin, but not Tau, by GSK3. Extract was supplemented with GSK3 to enhance detection of phosphorylated Tau as previously described (32). β-catenin and Tau from the same reaction sample were immunoblotted from a single gel; intervening lanes were removed for clarity. (B) In an in vitro kinase assay containing purified, recombinant Axin (0.1 μM), GSK3 (0.79 μM), CK1 (1.37 μM), Tau (0.34 μM), and β-catenin (0.22 μM), LRP6ICD (4.1 μM) inhibits phosphorylation of β-catenin P33/37/41 by GSK3 without inhibiting the phosphorylation of Tau P396 by GSK3. (C) In a kinase reaction in which recombinant Axin is absent, phosphorylation of β-catenin by GSK3 is inhibited by LRP6ICD but not by LRP6ICD(PPPAPX5) (D). For (B), (C), and (D), β-catenin and Tau were incubated in the same reaction and immunoblotted from a single gel.

LRP6ICD Directly Inhibits GSK3-Mediated β-Catenin Phosphorylation.

The simplest model for LRP6 signaling is that it directly inhibits β-catenin phosphorylation by GSK3. Alternatively, LRP6-mediated inhibition of β-catenin phosphorylation may require additional components. To determine whether LRP6ICD is sufficient to inhibit GSK3-mediated β-catenin phosphorylation, we tested whether we could reconstitute LRP6 signaling with purified components.

In a kinase assay with purified, recombinant proteins, LRP6ICD inhibits GSK3-mediated phosphorylation of β-catenin at P33/37/41 without inhibiting CK1's phosphorylation of β-catenin at P45 (Figs. 4B, ​,44C, and Fig. S8). Importantly, the concentration of LRP6ICD tested does not inhibit GSK3's phosphorylation of Tau in the same reaction, demonstrating that inhibition of β-catenin phosphorylation by LRP6ICD is not a result of general inhibition of GSK3 activity (Figs. 4B and ​and44C). Thus, LRP6ICD preferentially inhibits GSK3's phosphorylation of β-catenin in a kinase assay with purified components.

The ability of GSK3 to phosphorylate β-catenin independently of Axin (albeit inefficiently) in our purified system allowed us to test whether LRP6ICD inhibits GSK3's phosphorylation of β-catenin directly or indirectly (via a conformational change of Axin upon its binding to LRP6ICD). Significantly, we find that Axin is not required for LRP6ICD's inhibition of β-catenin P33/37/41 phosphorylation (Fig. 4C). In addition, CK1γ's phosphorylation plays a role in LRP6 signaling in vivo (35), but CK1 is not required for LRP6ICD activity in our kinase assay (Fig. 4 B and C). These results demonstrate that LRP6ICD can directly inhibit GSK3-mediated phosphorylation of β-catenin and that this inhibition does not require other components.

Next, we tested whether LRP6 requires intact PPPSP motifs to inhibit GSK3's phosphorylation of β-catenin in our purified system. Unlike LRP6ICD, LRP6ICD(PPPAPX5) does not inhibit GSK3's phosphorylation of β-catenin (Fig. 4D), demonstrating that LRP6's PPPSP motifs are required for LRP6 to inhibit β-catenin phosphorylation in vitro. In a kinase assay with recombinant proteins, GSK3 phosphorylates LRP6 in a manner that requires intact PPPSP motifs (data not shown) (19). Thus, we infer that phosphorylation of PPPSP serines by GSK3 is required for LRP6's ability to inhibit β-catenin phosphorylation in our purified, reconstituted system. Our purified system exhibits specific properties that are consistent with in vivo and egg extract data: (i) requirement for PPPSP serines (21, 22), (ii) specificity for inhibition of β-catenin and not Tau phosphorylation (Fig. 4A), and (iii) inhibition of β-catenin phosphorylation by GSK3 but not CK1 (33). Thus, we believe these studies recapitulate distinct properties of LRP6 signaling in vivo.

We thank Frank Costantini, Xi He, Randall Moon, and Sergei Sokol for plasmids and; Barry Gumbiner for β-catenin baculovirus and antibody; Janet Heasman for sharing data before publication; and Daniela Drummond-Barbosa, Susan Wente, and Chris Wright for critically reading this manuscript. E.L. is a recipient of a Pew Scholarship in the Biomedical Sciences. This work was also supported by American Cancer Society Research Scholar Grant RSG-05-126-01 (to E.L.), ACS Institutional Research Grant IRG-58-009-46 (to E.L.), National Cancer Institute Grant GI SPORE P50 CA95103 (to E.L.), National Institutes of Health Grant 1 R01 GM081635-01 (to E.L.), National Institute of General Medical Studies Medical-Scientist Training Grant 5 T32 GM007347 (to C.S.C.), American Heart Association Predoctoral Fellowships 0615162B (to K.K.J.) and 0615279B (to C.A.T.), National Institutes of Health Cancer Biology Training Grant T32 CA09592 (to K.K.J.), and Molecular Endocrinology Training Grant 5 T 32 DK007563 (to C.A.T.).