ABSTRACT
Cortical rotation and concomitant dorsal translocation of cytoplasmic determinants are the earliest events known to be necessary for dorsoventral patterning in Xenopus embryos. The earliest known molecular target is β-catenin, which is essential for dorsal development and becomes dorsally enriched shortly after cortical rotation. In mammalian cells cytoplasmic accumulation of β-catenin follows reduction of the specific activity of glycogen synthase kinase 3-beta (GSK3β). In Xenopus embryos, exogenous GSK3β suppresses dorsal development as predicted and GSK3β dominant negative (kinase dead) mutants cause ectopic axis formation. However, endogenous GSK3β regulation is poorly characterized. Here we demonstrate two modes of GSK3β regulation in Xenopus. Endogenous mechanisms cause depletion of GSK3β protein on the dorsal side of the embryo. The timing, location and magnitude of the depletion correspond to those of endogenous β-catenin accumulation. UV and D2O treatments that abolish and enhance dorsal character of the embryo, respectively, correspondingly abolish and enhance GSK3β depletion? A candidate regulator of GSK3β, GSK3-binding protein (GBP), known to be essential for axis formation, also induces depletion of GSK3β. Depletion of GSK3β is a previously undescribed mode of regulation of this signal transducer. The other mode of regulation is observed in response to Wnt and dishevelled expression. Neither Wnt nor dishevelled causes depletion but instead they reduce GSK3β-specific activity? Thus, Wnt/Dsh and GBP appear to effect two biochemically distinct modes of GSK3β regulation.
INTRODUCTION
Glycogen synthase kinase 3-beta (GSK3β) is an intracellular serine-threonine kinase implicated in vertebrate axis formation. Overexpressing GSK3β on the dorsal side of Xenopus embryos causes suppression of axis formation while overexpressing a dominant negative mutant GSK3β is sufficient to cause formation of a secondary, ectopic axis (Dominguez et al., 1995; He et al., 1995; Pierce and Kimelman, 1995). A causal role for GSK3β activity modulation in Xenopus axis formation is also implied by its ability to regulate the stability of β-catenin, an essential component of axis formation. GSK3β phosphorylates β-catenin protein and thereby triggers β-catenin degradation (Aberle et al., 1997; Pai et al., 1997; Yost et al., 1996). Evidence for a role for endogenous β-catenin in axis formation is very good. Overexpression of β-catenin induces axis formation (Guger and Gumbiner, 1995), while depletion of maternal mRNA encoding β-catenin results in embryos that lack a dorsal axis (Heasman et al., 1994). Furthermore, dorsal cortical enrichment of β-catenin protein from the 2-cell stage has been demonstrated biochemically and by immunofluorescent staining (Larabell et al., 1997). Thus β-catenin has been shown to be sufficient, necessary and present in the right place and time to initiate the dorsal axis. β-catenin can enter the nucleus (Funayama et al., 1995) and activate or de-repress transcription of a number of target genes, including known dorsal-specific transcription factors (McKendry et al., 1997; Brannon et al., 1997). Wild-type GSK3β and dominant negative mutant GSK3β also have the predicted effects on β-catenin levels, namely suppression and enrichment, respectively (Larabell et al., 1997).
GSK3β and β-catenin are both parts of the canonical Wnt signalling pathway. Wnt proteins are a family of intercellular signalling molecules known to be involved in many processes in animal development and in human cancers (Nusse and Varmus, 1992). The Wnt pathway minimally includes Wnt itself, frizzled (a membrane-spanning Wnt receptor), dishevelled (a transducer of unknown activity), GSK3β, β-catenin and transcriptional co-factors of the TCF/LEF family. Proteins of the axin family and APC are also involved. Axin and its close relatives are sometimes referred to as scaffold proteins, most likely involved in establishing or maintaining the other components in transducing complexes (see Peifer, 1999 for review). The Adenomatous Polyposis Coli (APC) protein is also involved, though its exact function in the normal pathway is unclear (Polakis, 1997). The Wnt pathway has long been thought to be involved in Xenopus dorsal axis formation because several Wnt proteins are potent axis inducers (McMahon and Moon, 1989). Overexpression of dishevelled also induces axis formation (Sokol et al., 1995). Dishevelled protein has recently been shown to be moved from a vegetal position to the dorsal side of the embryo during cortical rotation (Miller et al., 1999), the event known to establish dorsoventral asymmetry. This is consistent with dishevelled being the dorsal determination molecule of Xenopus embryos. However, as the authors of that study point out, in the absence of a functional test, one must be cautious about the role of the dishevelled protein. Studies have been carried out to test whether dishevelled or Wnt are actually necessary endogenously, as well as sufficient experimentally, for axis generation. Unexpectedly, expression in embryos of dominant negative (interfering) dishevelled and Wnt mutants do not supress axis formation (Sokol, 1996; Hoppler et al., 1996). This suggests that, although Wnt and Dsh are experimentally sufficient for axis induction, they may not be endogenously necessary. However, such mutant experiments rely for their interpretation on a negative result, namely failure to ventralize the embryo. Thus, while they are suggestive, their interpretation is always vulnerable to relatively trivial complications such as inadequate expression, inappropriate mutant isoform, etc. (Miller et al., 1999).
Given that the roles of GSK3β and β-catenin in axis formation are well established but those of Wnt and dishevelled are in some doubt, it remains an open question how GSK3β is regulated in vivo. There has not hitherto been a direct biochemical study of endogenous GSK3β. In this paper, we carry out such a study. We demonstrate that, while exogenous Wnt and dishevelled can reduce GSK3β-specific activity as expected, there is no detectable reduction of GSK3β-specific activity on the dorsal side of the embryo. Instead, endogenous GSK3β is downregulated by a depletion of the protein. Furthermore, the time, place and magnitude of the dorsal depletion of GSK3β correspond precisely with the accumulation of endogenous β-catenin. UV and deuterium oxide treatments that are known to prevent or enhance the distribution of dorsal determinants from the vegetal pole of the embryo (Scharf and Gerhart, 1983; Scharf et al., 1989; Holowacz and Elinson, 1993; Rowning et al., 1997) also prevent and enhance GSK3β protein depletion. We go on to investigate the role of GBP, a GSK3β-binding protein that has recently been shown to be sufficient and necessary for dorsal axis formation in Xenopus (Yost et al., 1998). We find that ectopic GBP can mimic the endogenous depletion of GSK3β. This effect of GBP identifies it as part of a new, apparently Wnt-independent signalling pathway for GSK3β regulation.
MATERIALS AND METHODS
Embryo manipulation
In vitro fertilization and embryo culture were carried out in 0.1′ MMR as described (Peng and Kay, 1991). Staging was according to Nieuwkoop and Faber (1967). For RNA microinjections, embryos were injected at stage 2 or 3 in 2% Ficoll (Pharmacia), 0.5′ MMR and injected with in vitro-transcribed XWnt-8 (2 pg), XDsh (0.7 ng) and GBP (0.5 ng) (Christian et al., 1991; Sokol et al., 1995; Yost et al., 1998). Embryos were bisected with an eyelash knife at various stages. UV-ventralization and D2O dorsalization were carried out as described (Scharf and Gerhart, 1983; Scharf et al., 1989) with an egg batch-by-batch titration for UV or D2O efficacy.
Embryonic extractions and immunoprecipitations
For activity measurements, embryo halves were homogenized in lysis buffer (1% Triton X-100, 25 mM Tris-HCl (pH 7.5), 50 mM NaCl, 2 mM EDTA, 1 mM DTT, 10 mM NaF, 1 mM vanadate 25 mM β-glycerophosphate, 10 μg/ml of leupeptin, aprotinin and pepstatin, 1 mM PMSF). Lysates were cleared by centrifugation at 14,000 g for
20 minutes. Cleared lysates were immunoprecipitated with anti-GSK3β antibody (Transduction Laboratories, no. G22320) for 2 hours at 4°C, then incubated for an additional 30 minutes with protein A-agarose. Beads were washed three times with lysis buffer and twice with kinase buffer (30 mM TrisHCl (pH 7.5), 5 mM MgCl2, 1 mM DTT). For protein abundance measurements, cleared lysates or embryo halves homogenized in SDS-PAGE sample buffer without fractionation were loaded directly on gels.
Kinase assays
Immune complex kinase assays were carried out in the presence of 25 μM [32P]ATP, (4 Bq/pmol) and 10 μM peptide substrate (‘2B-(S)’of Welsh et al. (1997)), RRAAEELDSRAGS*PQL (where S is the target serine and S* is pre-phosphorylated). Control peptide (2B-(A)), RRAAEELDSRAGAPQL was also used in some experiments. One fifth of each reaction mixture was subjected to SDS/15% PAGE followed by autoradiography. Peptide bands from the gel were excised and counted in an scintillation counter. Immunoprecipitates were also subject to 8% SDS-PAGE followed by western blotting and densitometry to determine the amounts of GSK3β immunoprecipitated in each sample.
Quantitative western assays
For western analysis, one quarter embryo equivalent was loaded per lane and electrophoresed on an SDS/8% polyacrylamide gel and electroblotted onto PVDF membranes (Millipore). The membrane was blocked with 10% goat serum, and then incubated with anti-GSK3β antibody (either Transduction Laboratories, no. G22320 as above or Santa Cruz Biotechnology no. 0011-A: sc-7291), anti-spectrin (Sigma) or anti-tubulin (Sigma) followed by horseradish peroxidase-conjugated secondary IgG antibodies (Jackson ImmunoResearch). Note that the anti-GSK3β antibodies recognize different epitopes since the Transduction Laboratories antibody does not recognize mammalian GSK3-alpha, while the Santa Cruz antibody does (manufacturers’ data). Quantitation in Xenopus was based exclusively on the GSK3β band. (In mammalian cells and Xenopus, GSK3-alpha runs at a higher apparent molecular weight (I. D., unpublished data). Incubations were 1 hour each, separated by three 15 minute washes with phosphate-buffered saline (PBS), 0.05% Tween 20. Peroxidase activity was visualized with an ECL system (Amersham) and preflashed film (Kodak). Exposed films were scanned with a flatbed scanner and quantitated using NIH image. For each band, integrated pixel density minus local background density was determined. Incubation times, substrate and enzyme concentrations, gel loading and densitometry were optimised to be within the linear range of assay (see Fig. 1).
Immunostaining
Embryos were fixed exactly according to the protocol of Larabell et al. (1997) at the 4-to 8-cell stage and left intact or bisected with a razor blade in the horizontal plane prior to antibody incubations. Fixation was carried out overnight at room temperature with the embryos animal pole uppermost and without agitation. The same anti-GSK3β mouse monoclonal antibody as used for western blots (Transduction Laboratories) was used at a dilution of 1:25 or 1:50 with the same result. No additional non-GSK3β bands are seen at this concentration compared to lower antibody concentrations in western blots (data not shown). Anti-β-catenin rabbit polyclonal antibody kindly provided by Dr B. Gumbiner (McCrea et al., 1993) was used at a dilution of 1:250. Secondary antibodies were diluted to 1:500. Three 1 hour washes at room temperature and one overnight wash at 4°C overnight with nutation (rocking) after antibody incubations were carried out in Superblock (Pierce). The secondary antibodies used were Cy-3 and Cy-5 conjugates (Jackson ImmunoResearch), which fluoresce in the rhodamine and far-red channels, respectively. Stained specimens were cleared and mounted between coverslips. Optical sections were taken with a Zeiss inverted laser scanning confocal microscope. Channels were analyzed in series (i.e. not simultaneously) to avoid any ‘bleed through’ between them. Autofluorescence of yolky cytoplasm in the fluorescein channel was used to reveal the extent of the embryo.
RESULTS
Endogenous GSK3β abundance but not specific activity is lower dorsally
We set out to determine whether GSK3β is downregulated endogenously on the dorsal side of the Xenopus embryo (and if so, when and how) by directly measuring GSK3β activity. Embryos were cut into dorsal and ventral halves (as evaluated by sperm entry point and pigmentation differences), homogenized and subjected to immunoprecipitation with a specific anti-GSK3β antibody (see Methods). This antibody is specific in Xenopus to GSK3β, marking a single band both in western blot and immunoprecipitation (Fig. 1A,B). Western blots show that GSK3β protein is present maternally and throughout all the stages analyzed in this study (data not shown). Precipitated GSK3β was assayed for kinase activity by incubation with radiolabelled ATP and a synthetic peptide known to be a sensitive, specific substrate (Welsh et al., 1997). A control peptide, substituted at a critical prephosphorylation site, was not phosphorylated in our assays (Fig. 1B; see Methods for details of peptides), ruling out contaminating non-GSK3β kinase activity. After incubation, enzyme and phosphorylated substrate were separated by gel electrophoresis. Phosphorylated peptide was excised and quantitated in a scintillation counter. GSK3β was detected by western blotting and quantitated densitometrically. Rigorous quantitative controls established the linearity of the measurements (Fig. 1C,D).
We measured GSK3β-specific activity during the 2-cell to 32-cell (stage 6) period. This period was chosen because by stage 2 (2-cell stage) dorsal enrichment of β-catenin has been observed by immunofluorescence (Larabell et al., 1997) while, by stage 6 (32-cell stage), a dorsal enrichment of β-catenin about 30% has been measured biochemically (Larabell et al., 1997). Surprisingly, we found no significant differences in the in vivo specific activity of GSK3β between dorsal and ventral halves at any of the stages analyzed (Fig. 2A,B), even though a specific activity difference sufficient to generate an axis should have been detectable in our assay (see below). These data suggest that GSK3β-specific activity does not change dramatically during the establishment of the dorsoventral axis. However, in the course of the immunoprecipitation experiments, we noticed that dorsal yields of GSK3β consistently appeared lower than ventral yields (Fig. 2A). To control for the possibility that this was due to differential immunoprecipitation, we loaded cleared embryonic lysates equalized for total protein content onto gels and analyzed by western blotting and probing for GSK3β. Relative to total amount of protein, dorsal halves contained approximately 20% less GSK3β protein than ventral halves (data not shown, but see below). This difference was observed in all the stages studied and occurs at least as early as the earliest detectable dorsoventral differences in β-catenin abundance (Larabell et al., 1997). This dorsal deficit in GSK3β abundance is thus consistent, both in timing and magnitude, with it being responsible for the dorsoventral difference in β-catenin levels (Larabell et al., 1997) in the absence of a dorsoventral-specific activity difference.
One explanation for differential yields of GSK3β in embryonic extracts might be that dorsal GSK3β is somehow less extractable than ventral GSK3β. To bypass potential effects of differential extractability, embryo halves were lysed in sample loading buffer and the total unfractionated lysates loaded directly onto gels for western analysis. Abundance of GSK3β was normalized to that of the ubiquitous protein spectrin (Larabell et al., 1997). A 20-30% reduction of GSK3β on the dorsal side of the embryo was observed (Fig. 2C,D), consistent with the results above. Normalizing by a specific protein also rules out the possibility that the measured difference is merely a consequence of asymmetric distribution of bulk protein (mostly yolk) which is known to move with cortical rotation (Denegre and Danilchik, 1993). Similar results were obtained when we normalized to a second protein, tubulin or used a second anti-GSK3β antibody that recognizes a different epitope (data not shown; see Methods). To control for variability in identification of the dorsoventral axis (which is well but not completely correlated with the position of sperm entry), we also did measurements on embryos that had been held at an angle and stained at the uppermost point with Nile blue vital dye. This method forces the formation of the dorsal axis at the uppermost, marked meridian and makes dorsoventral dissection close to 100% reliable (Peng and Kay, 1991). An identical result was obtained (data included in Fig. 2D). Overall, the dorsoventral abundance difference was found to be statistically highly significant (P<0.0001 or 99.9999% confidence using either Wilcoxon signed rank test or Student’s t-test). Thus, the lower abundance of GSK3β on the dorsal versus the ventral side of the embryo, whilst small, is consistent. This observed dorsoventral difference in abundance and hence total activity quantitatively matches the concurrent accumulation of cytoplasmic β-catenin and consequent axis formation.
Co-localization of β-catenin enrichment with GSK3β depletion at the cortex
We reasoned that the 20-30% gross dorsoventral difference in GSK3β might disguise a much greater effective difference in a cortical domain. Early β-catenin enrichment is restricted to the most cortical dorsal region of the embryo (Larabell et al., 1997), which is known to be the location of dorsal determinants following cortical rotation in the first cell cycle (Fujisue et al., 1993; Kikkawa et al., 1996). We wished to know whether this spatial restriction also applied to GSK3β depletion. We performed in situ immunostaining according to the protocol of Larabell et al. (1997). Fig. 3 shows the results of staining from two different egg batches. Similar results were obtained from a total of sixteen embryos from five independent egg batches. We found that GSK3β is depleted in a narrow cortical fringe corresponding to the region of β-catenin enrichment. The region of GSK3β depletion usually appears to be more cortically restricted than that of β-catenin enrichment, although this difference seems to vary between batches. However, it is consistent with diffusion of hypo-phosphorylated β-catenin away from the region of GSK3β depletion. The dorsal restriction (as marked by β-catenin) confirms the biochemical data. The β-catenin also provides a useful technical control, since it shows that the cortical region is stainable and thus lack of GSK3β staining is not due to non-specific blocking of stainability or edge effects due to physical bisection or optical sectioning. The narrowness of this region explains the relative smallness of both β-catenin enrichment and the GSK3β depletion measured in homogenates of embryo halves ((Larabell et al., 1997) and Fig. 2D above). It suggests that locally the depletion of GSK3β may be close to 100%.
UV ventralization abolishes and D2O dorsalization enhances GSK3β depletion
Vegetal cortical cytoplasm has been shown to contain the endogenous determinants whose dorsal displacement during cortical rotation is both sufficient and necessary for normal axis formation (Yuge et al., 1990; Fujisue et al., 1993; Holowacz and Elinson, 1993; Kikkawa et al., 1996; Sakai, 1996). If dorsal GSK3β depletion is functionally significant, preventing dorsal axis initiation should prevent the dorsoventral difference from arising. To test this prediction, we analyzed GSK3β abundance in embryos ventralized by UV irradiation of the vegetal pole in the first cell cycle (Scharf and Gerhart, 1983). This treatment abolishes the cortical rotation that brings dorsal determinants from the vegetal pole to the dorsal side of the embryo opposite the sperm entry point (Kikkawa et al., 1996; Fujisue et al., 1993). As predicted, UV-ventralization results in equalization of GSK3β levels on ‘ventral’ (sperm entry point) and ‘dorsal’ (opposite) sides (Fig. 4A).
Another prediction is that dorsalization of embryos by deuterium (D2O) treatment should cause an increased depletion of GSK3β. D2O treatment is unlike lithium-dependent dorsalization in that it acts during the first cell cycle and usually causes axis duplication rather than radial dorsalization (Scharf et al., 1989). D2O causes chaotic polymerization of cortical microtubules, which disperses the spread of vegetal cortical materials (Scharf et al., 1989; Rowning et al., 1997). Fig. 4B shows that as predicted, D2O-induced double-axis embryos (DAI = 7.0, see Scharf et al., 1989) show roughly double the net depletion seen in single-axis controls. This result in whole embryos argues against translocation as a possible mechanism for dorsal GSK3β depletion since translocation would not result in a net disappearance of GSK3β.
Wnt and Dsh mimic GSK3β depletion quantitatively by specific activity reduction
Reduction in GSK3β activity is known to cause β-catenin enrichment and β-catenin enrichment is known to cause axis formation. This chain of causality is confirmed by the axis-forming effects of expression exogenous dominant negative mutant GSK3β. It might be argued, however, that the reduction in endogenous GSK3β that we observed is too small to account for endogenous axis formation. To address this question, we used a biochemically well-characterized method for reducing GSK3β activity, namely Wnt8 signalling, to test how much of a reduction would be sufficient to generate an axis. Wnt8 and dishevelled regulate GSK3β by reduction of specific activity (Cook et al., 1996; Fisher et al., 1999). We therefore measured GSK3β abundance and specific activity in embryos injected ventrally with mRNA encoding XWnt8 or Xenopus dishevelled (XDsh), a transducer of Wnt signalling. We used the minimum amounts of RNA sufficient to induce complete ectopic (secondary) axes in 80% of recipient embryos. As shown in Fig. 5, the effect both of XWnt8 and of XDsh was a 25-40% reduction in ventral GSK3β-specific activity, with no observable change in abundance. This result shows that specific activity differences sufficient to generate an axis are measurable in this assay, an important positive control for the measurements showing no dorsoventral GSK3β-specific activity differences in normal embryos (Fig. 2A,C above). It also shows that protein depletion is not merely an effect (direct or indirect) of specific activity reduction by Wnt or dishevelled.
GBP causes GSK3β depletion, mimicking the endogenous axis formation mechanism
The above result constitutes biochemical evidence suggesting that Wnts are not the endogenous regulators of GSK3β in the early Xenopus embryo. We therefore decided to test a recently identified GSK3β-binding protein (GBP) as a candidate endogenous regulator. GBP has been shown not only to bind GSK3β but also to be required maternally for dorsal axis formation in Xenopus (Yost et al., 1998). Yost and colleagues have shown that when injected into oocytes, GBP RNA reduces the (total) activity of exogenous tagged GSK3β. The reported measurements did not distinguish between effects on specific activity or abundance and so we tested whether GBP reduces GSK3β-specific activity (as Wnt does) or amount of protein (as endogenous factors do). To demonstrate this effect in a normal physiological context and to correlate it with axis formation, we injected embryos on the ventral side with just sufficient GBP to make a duplicate axis. We then measured GSK3β activity and abundance as before. As a control, we used a GBP mutant that does not bind GSK3β (Yost et al., 1998)? Fig. 6 shows that GBP reduced the ventral abundance of GSK3β to that of the dorsal side but did not reduce the GSK3β-specific activity. Thus, GBP can mimic the endogenous axis-forming process at a biochemical level making it a good candidate for the in vivo regulator of GSK3β and a critical component of the dorsal determinant.
DISCUSSION
The above data show that endogenous GSK3β is downregulated on the prospective dorsal side of embryos as an early consequence of cortical rotation and that the mechanism is protein depletion rather than specific activity reduction. This is the first study to examine in detail the endogenous GSK3β protein rather than rely on overexpression of exogenous protein to elucidate mechanism. The weakness of simple measurement alone, of course, is that observations on unperturbed embryos cannot by themselves demonstrate mechanistic causation. Thus, a priori, the change could merely be a corollary or epiphenomenon of some other mechanism of axis formation. (This would fit with a view that a dorsal reduction of GSK3β would have no significant effect on dorsal β-catenin abundance or that dorsal β-catenin enrichment is not the mechanism of axis formation.) The ideal experiment to show causal significance would be restoration of equal GSK3β activity to the dorsal side with consequent loss of dorsal specification. This experiment is in fact fraught with complications. Since depletion appears to be highly restricted to a small cortical domain, exogenous enzyme would have to be delivered to this domain. This domain cannot be targeted quantitatively with microinjection because microneedles and fluid flow from them cannot be so accurately controlled. The only resort is to inject relatively large amounts of GSK3β mRNA or protein. In this crude form, this experiment has already been done. The result – that dorsal overexpression of GSK3β does indeed suppress not only axis formation, but also early dorsal gene transcription and dorsal β-catenin accumulation – is well documented (Dominguez et al., 1995; He et al., 1995; Pierce and Kimelman, 1995; Larabell et al., 1997). We have obtained similar results by injecting purified recombinant GSK3β protein (I. D., unpublished observations). However, even in this crude form, complete ventralization is rarely achieved because flooding the embryo with GSK3β is toxic to embryos by gastrula stages (our unpublished observations). This is not surprising, since GSK3β is a multifunctional kinase that would be expected to interfere with multiple pathways when overexpressed at high levels. Perhaps the best indication of the specificity of the GSK3β effect is the experiment of Larabell et al. (1997) in which dorsal overexpression of GSK3β suppresses the dorsal accumulation of β-catenin without affecting cell division or morphology. The dramatic effect of axis duplication elicited by ventral expression of a kinase-dead GSK3β (Dominguez et al., 1995; He et al., 1995; Pierce and Kimelman, 1995) is also highly suggestive of a role for the enzyme, although the mechanism of action of such a putative dominant negative mutant may be complex.
It might be argued that the ‘small’ dorsal depletion of GSK3β revealed by the data makes only a small contribution to dorsal character, and that the real mechanism involves some other process. However, the data show that a similarly ‘small’ reduction in GSK3β-specific activity caused by exogenous Wnt or dishevelled expression is quite sufficient to increase β-catenin abundance and generate a complete ectopic axis. (Incidentally, S6 protein kinase p90-rsk also reduces GSK3β-specific activity and increases β-catenin abundance, though it sends the β-catenin to the membrane where it does not act as a dorsalizing signal; Torres et al., 1999.) Comparison of our data with those of Larabell et al. (1997) show that the dorsoventral differences in GSK3β protein abundance are quantitatively similar to ‘small’ differences observed in β-catenin abundance. In other words, if other mechanisms for regulating β-catenin abundance are operating, they are redundant. The ‘smallness’ of the changes is, in any case, an artefact of measuring abundance in a homogenized half embryo. The immunofluorescence data show that the depletion is cortically localized and may be close to 100% in the cortical region.
What is the likelihood that GSK3β-regulating mechanisms other than the depletion shown in this paper are important for axis formation? The experiment by Yost et al. (1998) in which depletion of maternal GBP leads to complete ventralization of the embryo would seem to show that all the important mechanisms require the action of endogenous GBP. Since exogenous GBP causes GSK3β depletion but apparently not specific activity reduction, it seems likely that depletion represents the predominant endogenous mechanism. It should be noted, however, that while we have shown that GBP causes depletion of GSK3β? we have not ruled out the possibility that this depletion is accompanied by some GBP-dependent change of the specific activity of GSK3β to some physiological substrate, such as an axin-APC-β-catenin complex. However, such a downregulation, tightly coupled to disappearance, would not appear to be necessary and might be hard to demonstrate experimentally.
What we have also shown is that GBP’s action is mechanistically distinct from the actions of Wnt and dishevelled signalling. Specifically, GBP causes GSK3β to disappear while Wnt and dishevelled signalling do not. Conversely, Wnt and Dsh both alter specific activity of GSK3β (this study, Cook et al., 1996; Fisher et al., 1999), though there may be circumstances in which Wnt does not reduce GSK3β kinase activity (Yuan et al., 1999). The lack of effect on GSK3β abundance due to Wnt or Dsh suggests that decreased GSK3β abundance is not a consequence of Wnt or dishevelled signalling, either directly (e.g. protein modifications reduce both specific activity and stability) or via a feedback mechanism (e.g. β-catenin signals to destabilize GSK3β). Furthermore, the sizes of endogenous and experimentally induced decreases in total GSK3β activity necessary to cause axis formation are similar. This implies that the dorsal decrease in endogenous GSK3β protein levels can account for dorsal axis formation without invoking some additional undetected Wnt-or Dishevelled-dependent specific activity change. These results provide direct biochemical evidence that either the Wnt signal transduction pathway is not the endogenous initiator of dorsal axis formation or that endogenous Wnt and dishevelled signalling is very different from signalling by exogenous Wnt or dishevelled. Later roles for Wnt and dishevelled in axis formation are also not ruled out. For example, Wnt or Dsh signalling could come into play after stage 6 (the latest stage analyzed in this study). Experiments with dominant negative mutants of XWnt8 and Xenopus dishevelled suggest a critical role for these molecules in morphogenesis of a fully extended dorsal axis (Hoppler et al., 1996; Sokol, 1996).
There are formally three possible explanations for dorsal depletion of GSK3β: translocation to the ventral side, reduced synthesis and increased degradation. Dorsal depletion by translocation towards the ventral side would appear to be ruled out by the D2O experiment. After D2O dorsalization of the embryo, a net decrease in GSK3β abundance in whole embryos is observed. Since dorsal and ventral halves are measures together in the same homogenate, translocation cannot explain this effect. However, although in the hyperdorsalized D2O embryo net loss of GSK3β occurs, it does not directly exclude translocation as a mechanism in the normal embryo. Reduced synthesis is unlikely because depletion is caused by GBP (GSK-binding protein) which presumably does not bind GSK3β mRNA or the GSK3β gene promoter. However, some inhibition of synthesis by interaction with nascent GSK3β protein at the ribosome is not inconceivable. More indirect signalling from protein to mRNA is also possible. Thus, while the most plausible explanation for dorsal GSK3β depletion is cortically localized protein degradation, multiple mechanisms have not been ruled out.
Acknowledgements
We thank Sergei Sokol for encouragement in the early phases of this project and for XWnt8 and XDsh constructs; David Kimelman and Cynthia Yost for sending GBP clones; Barry Gumbiner for anti-β-catenin antibody; Becky Gelman for statistical advice; Stine Kräft for assistance with the confocal microscope; Amy Kavka, B. Neel, G. Cooper, M. Mercola, T. Roberts, T. Holowacz, R. Shivdasani and M. Rubock for critical reading of the manuscript. I. D. was supported by a fellowship from the Basque Government. J. B. A. G. was supported by the Dana Farber Barr Program.