Wnt11/β-catenin signaling in both oocytes and early embryos acts through LRP6-mediated regulation of axin

The single-pass transmembrane protein, LDL receptor-related protein 6(LRP6) was first identified as an essential co-receptor for mouse Wnt signal transduction (Pinson et al.,2000). LRP6 is required in the mouse for midbrain, axis and limb patterning (Kelly et al.,2004), and a mis-sense mutation in LRP6 causes folatesensitive neural tube defects in the Crooked tail mutant(Carter et al., 2005). LRP6 is upregulated in several human tumor cell lines(Li et al., 2004; Logan and Nusse, 2004).

Despite its importance in development and disease, the mechanism of LRP6 function is incompletely understood. LRP6 activity in Wnt signaling requires binding of the intracellular protein, axin, to its cytoplasmic domain through a phosphorylationdependent process involving GSK-3β and CK1γ(Davidson et al., 2005; Tamai et al., 2000; Tamai et al., 2004; Wehrli et al., 2000; Zeng et al., 2005). Axin is an essential component of the canonical Wnt signaling pathway, and modulates the level of cytoplasmic β-catenin by targeting it for phosphorylation and degradation. When axin function is blocked by Wnt-pathway activation or by antisense depletion of axin mRNA, β-catenin is stable and enters nuclei to initiate the transcription of target genes(Kofron et al., 2001; Nusse, 2005; Willert et al., 1999a). However, the regulation of axin function by LRP proteins is not clear. LRP may bind to axin and sequester it from the β-catenin-destruction complex(Nusse, 2005), or may promote its degradation, thus inactivating the β-catenin-degradation complex(Mao et al., 2001; Tolwinski et al., 2003; Willert et al., 1999b). Alternatively, in the absence of Wnt signaling, axin may constitutively shuttle β-catenin out of the nucleus; Wnt signaling blocks shuttling by binding LRP to axin, allowing nuclear β-catenin to rise(Cong and Varmus, 2004; Wiechens et al., 2004). To determine which mechanism works in axis formation requires an analysis of endogenous axin in wild-type and LRP6-depleted embryos.

Here, we analyze the function of LRP6 in the dorsal axisspecifying pathway in early Xenopus embryos. In Xenopus, the most recent model of the mechanism of axis formation involves Wnt signaling during the cleavage stage (Tao et al., 2005). This model suggests that the axis-forming pathway is activated by Wnt11 because the protein is necessary and sufficient for dorsal-axis formation, is stored as a maternal transcript in the vegetal cortex of the oocyte, and is enriched in dorsal cells of the embryo at the 32-cell stage(Schroeder et al., 1999; Tao et al., 2005). However,the timing and mechanism of signal transduction is not known. The frizzled family member Xfz7, the EGF-CFC protein FRL1 and heparan sulphate proteoglycan activity are required downstream of Wnt11(Sumanas et al., 2000; Tao et al., 2005). LRP6 has previously been implicated in this pathway because it is maternally expressed(Houston and Wylie, 2002) and because over-expression of LRP6 on the ventral side of the 4-cell-stage embryo causes the formation of a partial second axis(Tamai et al., 2000). However,when injected into early embryos, a dominant-negative form of LRP6, lacking the cytoplasmic domain, does not interfere with endogenous axis formation,suggesting either that this Wnt signaling pathway does not require LRP6, or that signaling occurs earlier in development and cannot be blocked by the expression of the dominant-negative construct(Tamai et al., 2000).

Axin is an essential negative regulator of canonical Wnt signaling pathways, acting by causing the degradation of β-catenin(Hamada et al., 1999; Lee et al., 2003; Salic et al., 2000; Willert et al., 1999a). In Xenopus, depletion of axin mRNA causes an increased nuclear localization of β-catenin, increased expression of Wnt target genes,including siamois and Xnr3, and radially dorsalized phenotypes (Kofron et al.,2001). In Xenopus egg extracts, axin is estimated to be present at low levels (picomolar quantities) compared with β-catenin,GSK3β and disheveled (nanomolar quantities), and β-catenin degradation is extremely sensitive to small changes in axin concentration(Lee et al., 2003; Salic et al., 2000). Injection of only 6 pg of axin mRNA is sufficient to rescue the dorsalized phenotype of axin-depleted Xenopus embryos(Kofron et al., 2001).

Here, we test the hypothesis that LRP6 is the necessary receptor for Wnt11 signal transduction and that it acts by degrading axin. We show first that LRP6 is essential for Wnt11-activated dorsal-axis formation in Xenopus. Second, we show that axin protein levels are increased when LRP6 is depleted, indicating that LRP6 regulates the degradation of axin. Third, we identify a specific time in early development, the 8-cell stage,when axin levels are reduced on the dorsal side of the wild-type embryo,indicating the time of transduction of the Wnt signal. Finally, we present evidence that Wnt11 and LRP6 also act in oocytes to regulate axin, maintainingβ-catenin in a low, steady-state level. This suggests that canonical Wnt signaling pathways may have three states (off, steady-state and transcriptional activation) rather than the generally accepted two (off and transcriptional activation).

Full-grown oocytes obtained from adult-frog ovary were manually defolliculated and cultured in oocyte culture medium (OCM). Oocytes were injected with antisense oligo or mRNA in the doses described in the text and cultured for 72 hours at 18°C. Oocytes were matured by the addition of 2μM progesterone in OCM, cultured for 12 hours after the addition of progesterone, labeled with vital dyes and fertilized using the host-transfer technique (Zuck et al.,1998).

Oocytes were washed in Ca²⁺-free 1×MMR (100 mM NaCl, 2 mM KCl, 15 mM HEPES) and nutated in 0.02 mg/ml collagenase (Sigma Type 1) in Ca²⁺-free MMR +1.6 mM CaCl₂. Denuded oocytes were washed in 1×MMR and then in OCM.

Oocytes were treated with MG132 (Sigma) as described by Dupont et al.(Dupont et al., 2005). A concentration of 1:100 MG132 was made in sterile water from a stock of 10 mM MG132 in DMSO and 10 nl was injected into oocytes. Injected oocytes were cultured in OCM containing 10 μM MG132 for 2-3 days. Oocytes were harvested for western blots in lysis buffer containing 10 μM MG132.

Antisense oligo for LRP65′-T^*C^*G^*AGGCTGATCCAG^*C^*T^*C-3′;for Wnt115′-G^*T^*C^*GGAGCCATTGGT^*A^*C^*T-3′;and for Axin5′-T^*T^*C^*CTCGCCAGGAAC^*T^*G^*G-3′,where phosphorothioatemodified residues are indicated by an asterisk(^*).

Oligos were resuspended in sterile distilled water and injected in doses of 4-5 ng per oocyte. Mouse LRP6 was subcloned into pRN3, linearized with SfiI and transcribed with T3 message machine (Ambion). Xenopus pCS2+ Myc-β-catenin plasmid was linearized with NotI, and pSP64T-Wnt11 with XbaI, and transcribed with SP6 message machine kit (Ambion). Analysis of gene expression using real-time RT-PCR was as described by Tao et al. (Tao et al., 2005). The β-catenin constructs used for Fig. 7F,G,H were either wild-type or stabilized β-catenin in pCS2-MT(Yost et al.1996). The stabilized β-catenin lacks the four N-terminal Gsk3β phosphorylation sites because of their mutation from serine to alanine.

In total, five embryos or oocytes were homogenized in 200 μl of ice-cold PBS plus protease inhibitors (PIC P8340 Sigma 1:100, PMSF 10 μg/ml) and spun for 5 minutes at 700 g, 4°C. Supernatants were precipitated with acetone for 20 minutes on ice. Samples were spun for 5 minutes at 3000 g, 4°C. Pellets were resuspended in 50μl 2×SDS sample buffer and boiled for 5 minutes. A total of 18 μl was loaded onto 7.5% Tris-glycine ready gel (Bio-Rad) and electrophoresed for 2.5 hours at 75 V. Protein was transferred to nitrocellulose membranes and blocked overnight in PBS-Tween 0.1%, 5% dry milk at 4°C. Antibody conditions were anti-axin 1067 AP 1:400 (gift from Peter Klein, University of Pennsylvania, Philadelphia, PA), anti-β-catenin: Sigma C2206 1:2000,Roche anti-c-myc clone 9E10 1:2000. Membranes incubated with HRP-conjugated antibodies were incubated with ECL developing solution (Amersham) and exposed to X-ray film at variable times to avoid the saturation of bands (Hyperfilm,Amersham). Quantification was performed using IPlab densitometry software,normalized to α-tubulin levels and values expressed as the percentage change relative to controls. Blots were re-probed with a FITC-conjugated anti-α-tubulin antibody (Sigma F2168) at 1:4000 for loading control, and imaged on a Typhoon variable mode imaging system.

Animal cells of embryos were injected with 500 pg Wnt11-HA mRNA and 1.5 ng LRP6-N-myc RNA (lacking the carboxyl intracellular domain(Tamai et al., 2000) at the 4-to 8-cell stage. Adjacent blastomeres were injected such that only secreted protein could interact in order to avoid non-specific interactions between the two proteins in the ER/golgi complex. Embryos were harvested at stage 10.5 and centrifuged to remove yolk protein. Co-immunoprecipitation was carried out as described previously (Tao et al., 2005). The anti-HA high-affinity rat antibody (3F10, Roche) was used to pull-down Wnt11-HA protein. Lrp6-N-myc protein was detected using the anti-c-myc antibody (9e10, Roche). Detection of proteins in western blots used HRP-conjugated anti-rat or -mouse antibodies and ECL plus (GE healthcare, RPN2132). Blots were visualized with ECL fluorescence on a Molecular Dynamics typhoon phosphor imager.

TOPflash DNA (60 pg) containing three copies of the TCF-binding site upstream of a minimal TK promoter and luciferase open reading frame, together with 20 pg pRLTK DNA (Renilla reniformis luciferase used as an internal control), was injected into two dorsal cells at the 4-cell stage of control and LRP6-depleted embryos. Three replicate samples, each of four embryos, were frozen at the late blastula stage and luciferase assays were carried out using the Promega luciferase assay system.

To study the role of maternal LRP6 in Wnt signaling in the early Xenopus embryo, we first determined whether LRP6 mRNA was localized. Real-time RT-PCR of oocytes cut into animal and vegetal halves, or into animal, equatorial and vegetal thirds, showed that, compared with vegetally localized Wnt11 mRNA, LRP6 mRNA was distributed throughout the oocyte (Fig. 1A). Next, we showed that an antisense oligonucleotide caused a dose-dependent depletion of LRP6 mRNA from full-grown oocytes,without significantly affecting the levels of expression of the related family member LRP5 (Fig. 1B).

When LRP6-depleted oocytes were fertilized, they developed normally until gastrulation, but then phenocopied maternal β-catenin- or Wnt11-depleted embryos, displaying delayed gastrulation (data not shown) and the loss of axial structures at the tailbud stage - the so called `ventralized' phenotype(Fig. 1C,D; 121/132 cases [92%]had this appearance in ten experiments). Over-expression of mouse LRP6 mRNA in the oocyte had the opposite effect, causing dorsalization (Fig. 1C). The effect of antisense depletion was specific, because ventralization was rescued by the reintroduction of LRP6 mRNA into LRP6-depleted oocytes(Fig. 1D; 80/83 embryos [96%]rescued in this way in five experiments). The expression of the early zygotic Wnt11 target genes siamois, Xnr3, chordin and goosecoid was severely reduced in LRP6-depleted embryos at the early gastrula stage(Fig. 1E), and remained so throughout gastrulation (Fig. 1F, Fig. 2). Wnt target gene expression was rescued by the injection of LRP6 mRNA before fertilization, showing that the effect was specific(Fig. 1E). The expression of the endoderm marker XSox17 and the ventral marker Xwnt8 was delayed, but reached wild-type levels (Fig. 1F). Expression of the mesodermal marker Xbra was also reduced in LRP6-depleted embryos (Fig. 1E). Although embryos and molecular markers were not completely rescued to the wild-type state, Wnt target gene expression and dorsal-axis formation were consistently recovered by the injection of mouse LRP6mRNA (Fig. 1G), showing the requirement for LRP6 in this process.

As further confirmation that LRP6 was required to activate the canonical Wnt signaling pathway, we used the TOPflash reporter to examine the effects of the loss of LRP6 activity on Tcf3- and/or LEF-mediated transcription. Fig. 1H shows that LRP6 depletion, at doses that cause ventralization, abolished the activity of this reporter in embryos at the late blastula stage.

β-catenin in the axis-forming pathway LRP6-depleted embryos resemble Wnt11-depleted embryos (Fig. 1) (Tao et al.,2005). To analyze this further, we compared the expression of Wnt target genes in sibling Wnt11- and LRP6-depleted embryos derived from the same batch of oocytes. Fig. 2 shows that, in both depletions, the expression of siamois and Xnr3was severely reduced. Because ventralized phenotypes occur because of either reduced dorsal gene expression, or the upregulation of ventral gene expression, we examined the expression of the BMP target genes Xvent1and Xvent2, as well as of BMP4. These markers were not upregulated at stage 9.5-10.25, confirming that LRP6 depletion causes ventralization by preventing dorsal gene activation.

To confirm that LRP6 acts in the maternal Wnt11 signaling pathway, we tested whether LRP6 mRNA could rescue Wnt11-depleted embryos. Oocytes were depleted of Wnt11 using an antisense oligo(Tao et al., 2005), cultured for 2 days and injected with mouse LRP6 mRNA before fertilization. Mouse LRP6 mRNA (75 pg) rescued Wnt11-depleted embryos in producing neural fold formation, although the size of the neural folds and head formation were abnormal, whereas higher doses caused embryos to develop with a dorsalized phenotype (Fig. 3A). This abnormal morphology may be explained by the fact that Wnt11 depletion causes a loss of a localized asymmetric signal, which is not replaced by LRP6,so that the rescued axis is not normal. LRP6 mRNA also rescued the expression of siamois and Xnr3 in Wnt11-depleted embryos at the gastrula stage (Fig. 3B).

By contrast, the effects of maternal LRP6 depletion were not rescued by the injection of 100-300 pg Wnt11 mRNA(Fig. 3C,D), although these doses of Wnt11 RNA caused the dorsalization of wild-type siblings(data not shown). Thus, LRP6 lies downstream of Wnt11 in the signaling pathway.

To place β-catenin relative to LRP6 in the pathway, we asked whether LRP6-depleted embryos could be rescued by the injection ofβ -catenin mRNA. β-catenin mRNA (50 pg) was injected into one dorsal cell in LRP6-depleted embryos at the 4-cell stage,and rescued the formation of axial structures as well as siamois and Xnr3 expression (Fig. 3E,F). Thus, LRP6 is upstream of β-catenin in the axis-forming pathway.

To determine whether Wnt11 physically interacts with LRP6, HA-tagged Wnt11(500 pg) and LRP6 lacking the C-terminus(Tamai et al., 2000)(LRP6-N-myc 1.5 ng) were injected into adjacent cells at the 4- to 8-cell stage, and were frozen at the early gastrula stage for co-immunoprecipitation analysis. Fig. 3G shows that LRP6-N-myc interacts with Wnt11-HA.

The above data raise the question of the mechanism of LRP6 function. Current models suggest that LRP6 either sequesters or degrades axin protein,or shuttles it continuously out of the nucleus. To determine how LRP6 acts in the early Xenopus embryo, we compared the amounts of endogenous axin in western blots of wild-type and LRP6-depleted embryos using an affinity purified anti-Xenopus axin antibody in embryos at the 8-cell and early blastula stages. Remaining sibling embryos were cultured to the tailbud stage to confirm that they were ventralized. Fig. 4A shows that LRP6 depletion resulted in increased levels of endogenous axin in whole-embryo lysates. In three experiments, increases ranged from 25-47% greater than control levels. Although these changes were not extreme, axin is known to be extremely active and present at limiting amounts in wild-type embryos(Lee et al., 2003; Salic et al., 2000), such that small changes can dramatically alter Wnt signal transduction. This finding supports the model that LRP6 transduces Wnt signals by causing axin degradation.

Because axin is known to cause β-catenin degradation, such increased axin levels would be predicted to result in reduced β-catenin protein in LRP6-depleted embryos. Blots were stripped and re-probed using aβ-catenin-specific antibody, and this prediction was confirmed; the trend for β-catenin levels was opposite to that for axin, particularly at the early blastula stage (Fig. 4A).

Fig. 3A shows that Wnt11 is upstream of LRP6 in the axis-forming pathway. Therefore, we asked whether Wnt11 depletion also affects axin levels in the early embryo. Wnt11 depletion enhances axin levels in the 8-cell-stage embryo(Fig. 4B), with increases ranging from 19-46% in three experiments.

The most recent model of Wnt11 function in the axis-forming pathway proposes that sperm activates the cortical rotation movements of the first cell cycle, which leads to a dorsal asymmetry of Wnt11 mRNA and protein, causing increased Wnt11 signaling dorsally during the cleavage to blastula stages (Heasman, 2006; Larabell et al., 1997; Schneider et al., 1996; Tao et al., 2005). The work here suggests the following model (Fig. 4): by the 8-cell stage, Wnt11 protein is secreted, mostly by dorsal vegetal cells, and binds to LRP6(Fig. 4Cii). Association of Wnt11 with the extracellular domain of LRP6 causes LRP6 phosphorylation(Davidson et al., 2005; Tamai et al., 2000; Tamai et al., 2004; Wehrli et al., 2000; Zeng et al., 2005) and axin degradation. Because less axin is present to degrade β-catenin,β-catenin is stabilized, binds to Xenopus Tcf3 and enters nuclei on the dorsal side. On the ventral side, low levels of Wnt11 signaling occur,maintaining a level of axin that degrades sufficient β-catenin to reduce nuclear accumulation. In LRP6-depleted embryos, Wnt11 molecules are unable to activate the pathway, so axin increases in concentration and β-catenin is degraded on the dorsal as well as the ventral side(Fig. 4Ciii). Next, we tested four predictions of this model, as follows:

To determine whether LRP6 mRNA is asymmetrically localized in embryos, we examined LRP6 mRNA levels in dorsal versus ventral half-embryos at the 32-cell stage (Fig. 5A). Fig. 5A shows that LRP6 mRNA is not enriched dorsally.

We examined total axin protein in western blots of dorsal and ventral wild-type half embryos at the 8-cell stage. Axin levels were reduced dorsally compared with ventral levels in three repeats of the experiment(Fig. 5B; 32% and 59% less than the ventral levels of expression). Blots were stripped and re-probed forβ-catenin levels, and β-catenin expression was found to be increased on the dorsal side compared with ventral levels(Fig. 5B; 14% and 20% more than the ventral side).

To determine whether low levels of axin were maintained dorsally, we examined a series of embryos from the 8- to the 64-cell stage. Only at the 8-cell stage was axin at a low level dorsally compared to ventrally.β-catenin was enriched dorsally compared to ventrally throughout the 16-to 64-cell stages (Fig. 5C). This correlates with the reported time of nuclear localization ofβ-catenin protein in dorsal cells(Larabell et al., 1997).

To test whether the result of LRP6 depletion was a reduction in the dorsal enrichment of β-catenin, we compared the amounts of endogenousβ-catenin in western blots of dorsal and ventral half-embryos bisected at the 32-cell stage. The dorsal enrichment of β-catenin was lost in dorsal compared to ventral halves of LRP6-depleted embryos(Fig. 4D; repeated in three experiments).

The model suggests a causative relationship between LRP6 and its modulation of axin in regulating β-catenin levels, in turn affecting the transcription of Wnt target genes. To explore this further, we tested the effect of depleting both axin and LRP6 mRNA. We have shown previously that maternal axin depletion causes the upregulation of siamois and Xnr3 (Kofron et al., 2001), whereas LRP6 depletion causes their downregulation(Fig. 1E). In embryos depleted of both LRP6 and axin, the loss of LRP6 would be predicted to allow remaining axin to be stabilized, thus attenuating the up-regulation of siamoisand Xnr3. Fig. 5Eshows that this was the case. Depletion of both axin and LRP6 partially rescued the over-expression of siamois and Xnr3 caused by axin depletion. However, sibling LRP6- and axin-depleted embryos did not develop normal axes, most probably due to the lack of asymmetry of dorsal gene expression (data not shown).

Together, these results are consistent with the model shown in Fig. 4C, in which LRP6 transduces Wnt11 signals dorsally by causing axin degradation at the 8-cell stage. When this process is blocked by LRP6 depletion, axin levels rise,leading to increased β-catenin degradation and reduced expression of Wnt target genes.

The Wnt pathway is assumed to be inactive in full-grown oocytes, because transcription is repressed, and chromatin is methylated and hypo-acetylated(Bird and Wolffe, 1999; Landsberger and Wolffe, 1995). However, because axin protein levels act as an early indicator of LRP6 function, we examined the effects of LRP6 depletion on endogenous axin protein levels in oocytes. Unexpectedly, we found that endogenous axin protein levels were increased in LRP6-depleted oocytes compared to controls, suggesting that maternal LRP6 has a role in maintaining oocyte axin at a steady-state level(Fig. 6A; repeated in three experiments).

This raises the query as to whether axin levels in the oocyte are controlled by Wnt signaling. Because the manually defolliculated oocytes used in these experiments retain some follicle cells, these cells could secrete a Wnt signal. We removed all remaining follicle cells using collagenase treatment, and compared endogenous axin levels in control and denuded oocytes after culturing for 3 days (Fig. 6B). No differences were seen between collagenase-treated oocytes compared to control oocytes, suggesting that follicle cells do not play a role in regulating oocyte axin levels.

An alternative possibility is that Wnt11 acts in an autocrine fashion. Thus, we compared the effects on axin levels in oocytes depleted of maternal Wnt11 mRNA with that of controls. Fig. 6C shows that Wnt11-depleted oocytes had increased levels of axin compared with controls,indicating that Wnt11 normally stimulates LRP6-mediated degradation of axin in full-grown oocytes.

We then investigated whether this mechanism regulates oocyte β-catenin levels by examining endogenous β-catenin levels in wild-type and LRP6-depleted oocytes. Levels were also compared 8 hours after progesterone-stimulated oocyte maturation. Fig. 6D shows thatβ-catenin levels were low in oocytes compared with matured oocytes, and that new β-catenin synthesis occurred during oocyte maturation.β-catenin protein in matured oocytes decreased as a result of LRP6 depletion (by 23% compared with controls).

Because β-catenin levels in non-matured oocytes are low, we next asked whether maternal axin is required to maintain this state. We depleted axin mRNA using an antisense oligo(Kofron et al., 2001), and cultured the oocytes for 3 days to reduce axin protein levels. Fig. 6E shows that axin-depletion caused an increase in β-catenin levels in non-matured oocytes compared with controls. Axin-depleted oocytes did not transcribe siamois and Xnr3 even with these increased β-catenin levels (data not shown). These results suggest a role for maternal axin in maintaining β-catenin at low levels in the oocyte, and suggest that Wnt11 signals in an autocrine fashion through LRP6 to degrade sufficient axin to maintain a steady-state level (Fig. 6F).

One prediction of the model that maternal axin regulates endogenousβ-catenin in oocytes is that exogenously introduced β-catenin should also be degraded. To test this, we injected myc-tagged β-cateninmRNA into oocytes, which were then cultured in the presence or absence of the proteasomal inhibitor MG132. The amount of myc-tagged β-catenin was enhanced when proteasomal degradation was prevented, suggesting that exogenous protein is degraded (Fig. 7A). To test the role of LRP6, we co-injected LRP6 mRNA together withβ -catenin-myc mRNA and cultured oocytes for 3 days. Fig. 7B shows a substantial increase in the amount of myc-tagged β-catenin protein in oocytes when LRP6 mRNA was co-injected with β-catenin-myc mRNA,showing that LRP6 enhances the stability of the exogenous β-catenin-myc protein.

A second test of whether β-catenin is degraded in oocytes compared to embryos is to compare the ability of β-catenin mRNA to rescue LRP6-depleted oocytes in which the mRNA was injected either before or after fertilization. We used a dose of β-catenin mRNA (50 pg) that caused axis-duplication when injected into the ventral side of wild-type embryos at the 4-cell stage (data not shown). Fig. 3E and Fig. 7C,D show thatβ -catenin mRNA rescued axis formation and dorsal marker expression when injected at the 4-cell stage, but the same dose did not rescue LRP6-depleted embryos when β-catenin mRNA was injected into oocytes before maturation. This experiment was repeated three times with the same result. One trivial explanation of this difference might be thatβ -catenin mRNA was broken down when injected into oocytes but not embryos. We confirmed that β-catenin mRNA was not degraded when injected into oocytes (data not shown). Finally, to confirm that more degradation of β-catenin protein occurs in oocytes than in embryos, we examined the levels of β-catenin in western blots of siblings of the embryos shown in Fig. 7C, at the 64-cell stage (Fig. 7E). Embryos injected as oocytes with β-catenin mRNA had less totalβ-catenin protein at the 64-cell stage than those injected at the 4-cell stage. These findings are consistent with the hypothesis that, in the oocyte,axin degrades β-catenin synthesized from the injected mRNA.

To confirm that the failure of β-catenin to rescue the LRP6-depletion phenotype was due to the degradation of β-catenin in the oocyte through an axin- and/or GSK-dependent mechanism, we repeated the rescue experiment comparing the rescuing ability of equivalent doses of wild-typeβ-catenin-myc mRNA and stabilized β-catenin-myc lacking the 4 N-terminal Gsk3β phosphorylation sites(Yost et al., 1996). Fig. 7G shows that the stabilized form of β-catenin was indeed protected from degradation in LRP6-depleted oocytes, and Fig. 7F,H shows that it rescued the expression of siamois and Xnr3, and caused a dorsalized phenotype. Embryos derived from LRP6-depleted oocytes injected with stabilized β-catenin before fertilization were radially dorsalized, unlike those injected withβ-catenin after fertilization (Fig. 3E). This is consistent with the fact that RNA injected into oocytes is able to diffuse widely, whereas that injected into a single site at the 4-cell stage does not diffuse and allows a dorsal-ventral asymmetry to be established.

Finally, to determine whether, acting in parallel to LRP6, the co-receptor Xfz7 was able to activate the axis-forming pathway, we injected Xfz7mRNA into LRP6-depleted oocytes before fertilization. LRP6-depleted embryos were not rescued by the injection of Xfz7 mRNA either phenotypically(not shown) or by the expression of siamois and Xnr3(Fig. 7H).

Taken together, these data provide insight into a novel steady-state role for Wnt11 and LRP6 in regulating axin and β-catenin in the oocyte.

We found, by loss- and gain-of-function analysis, that LRP6 is essential in a canonical Wnt signaling pathway, lying upstream of β-catenin and downstream of maternal Wnt11. Unlike LRP6 function in the mouse(Kelly et al., 2004), we found no evidence of redundancy with LRP5, which continued to be expressed in LRP6-depleted oocytes, but did not rescue the effect of LRP6 depletion.

Both the sequestration (Nusse,2005) and nuclear shuttling(Cong and Varmus, 2004) models of axin function predict that LRP6 depletion or over-expression in oocytes and embryos would cause endogenous axin protein to change in localization rather than in total amount. We show here that LRP6 loss of function caused increases in axin levels in oocytes and embryos, supporting a role for LRP6 in the degradation of axin. However, we cannot rule out that LRP6 function in early embryos also involves the sequestration of axin.

These experiments show that LRP6 is essential for axis formation. However,because LRP6 is depleted by injecting antisense oligos into oocytes, we cannot distinguish whether LRP6 is required for axis-formation after fertilization,or before. The Wnt signaling event responsible for establishing the dorsal axis is likely to occur after fertilization, because blocking β-catenin synthesis after fertilization using a translation-inhibitory morpholino oligo causes ventralization if the morpholino is injected vegetally at the 2- to 4-cell stage (Heasman et al.,2000; Yang et al.,2002). If axis formation depended on oocyte-synthesizedβ-catenin that moved in vesicles to the dorsal side during the first cell cycle (Rowning et al., 1997),then blocking β-catenin translation at the 2-to 4-cell stage should not affect axis formation. Also, dorsal and/or ventral asymmetries of Wnt11 mRNA and protein are evident after, but not before,fertilization (Tao et al.,2005; Schroeder et al.,1999), and a dorsal reduction of axin occurs specifically at the 8-cell stage (Fig. 5B), whereasβ-catenin becomes dorsally nuclearly localized by the 16-cell stage(Larabell et al., 1997).

We show here that LRP6 mRNA does not rescue LRP6-depleted embryos when injected after fertilization. One possibility is that there is insufficient time for LRP6 protein to be made and inserted onto the membrane in oligomeric complexes for the transduction of the Wnt11 signal. There is only a brief time-window after fertilization when the dorsal pathway can be activated by β-catenin mRNA, because β-cateninmRNA can only rescue β-catenin-depleted embryos when injected on the dorsal side at or before the 4-cell stage(Heasman et al., 2000; Yang et al., 2002). One possibility is that β-catenin is required for chromatin remodeling as a prerequisite for transcriptional activation. If Wnt11-LRP6 fails to stabilize dorsal β-catenin early, then such remodeling may not be successfully initiated. A second possibility is that a Wnt inhibitor, such as dickkopf, may be synthesized after fertilization and may actively block all but the earliest Wnt11-LRP6 interaction. Preliminary data suggest that Dkk mRNA is synthesized during oogenesis and may be a candidate for this role.

The fact that axin levels in wild-type embryos are not reduced dorsally after the 8-cell stage may be explained in any or all of three ways: first,synthesis of axin may outpace its degradation; second, axin may sequester, as well as degrade, β-catenin; and third, Wnt signaling may be downregulated by a Wnt inhibitor. Further experiments are needed to distinguish between these possibilities.

We have previously assumed that Wnt signaling is `activated' on the dorsal side of the embryo during the cleavage stages, leading to increasedβ-catenin levels on the dorsal side. However, this begs the question of what mechanism regulates the `steady state' levels of β-catenin before such an activation? In the current model of canonical Wnt signaling, the assumption has been that expression of Wnt target genes or TCF reporter constructs are indicators of Wnt signaling(Boras-Granic et al., 2006; De Langhe et al., 2005; Hens et al., 2005), suggesting that their lack of expression coincides with Wnt pathway inactivation. An alternative model is that Wnt signaling regulates β-catenin constitutively (Fig. 6F), and that the changes that lead to the nuclear accumulation of β-catenin are due to either increased levels of Wnt signaling(Fig. 4C), or to the involvement of other pathway components, rather than de novo activation. In its simplest form, a constitutive low level of Wnt signaling would lead to low levels of axin degradation by LRP6 (Fig. 6F). This in turn would lead to high levels of axin, and lowerβ-catenin levels. In the embryo, increased Wnt signaling on the dorsal side would lead to increased levels of β-catenin, through the same mechanism. The data presented here provide evidence for the existence of a steady-state role for Wnt signaling in modulating axin levels. Specifically,Wnt11 depletion and LRP6 depletion both cause an increase in axin levels in the full-grown oocyte. Our study has shown, in several ways, that the effects on axin are important for β-catenin regulation in the oocyte. First, the amount of β-catenin increased in oocytes when axin was depleted using an antisense oligo (Fig. 6E). Second, exogenous β-catenin was degraded in oocytes in a proteasome-dependent fashion (Fig. 7A) and stabilized by LRP6-mRNA co-injection(Fig. 7B). And third,β -catenin mRNA rescues LRP6-depleted embryos when injected after fertilization, but not before (Fig. 7F). This work changes our assumption that Wnt signaling is an on or off switch, and raises the query as to whether similar constitutive Wnt regulation of axin occurs elsewhere in embryonic and adult tissues, in places that would not be revealed by studying transcriptional target activation.

Thanks to Dr Peter Klein for generously supplying the axin serum and to Dr Ethan Lee for sharing data prior to publication. Thanks to Drs Aaron Zorn,Frank Hess (Merck) and Doug Melton for plasmids, and to Stephanie Lang and Helbert Puck for technical assistance.