Secretory and inductive properties of Drosophila wingless protein in Xenopus oocytes and embryos

One of the earliest developmental decisions during amphibian embryogenesis is the specification of mesoderm in the marginal zone of the blastula. This process, known as mesoderm induction (Nieuwkoop, 1969), is believed to be triggered by diffusible factors produced by the underlying vegetal pole cells (reviewed by Smith, 1989 and Slack, 1991). The induced mesoderm subsequently differentiates into a variety of different tissues, which occupy distinct positions along the dorsoventral and anteroposterior axes of the embryo. The initial subdivision of the prospective mesoder-mal cell fates within the marginal zone results from the production by dorsal vegetal cells of a mesoderm-inducing factoids) which is qualitatively or quantitatively different from that produced by ventral and lateral vegetal cells (Gimlich and Gerhart, 1984, Dale et al., 1985; Dale and Slack, 1987b). Subsequently the induced dorsal marginal zone (DMZ) “organizes” the adjacent ventral marginal zone (VMZ) by dorsalizing it in a graded fashion according to the relative positions of DMZ and VMZ cells (Spemann and Mangold, 1924; Smith and Slack, 1983, Dale and Slack, 1987b; Stewart and Gerhart, 1990). It is this dorsalization that lays the foundation of axis determination in the developing embryo.

Understandably there has been much interest in identifying the endogenous mesoderm inducers, and a great deal of progress has been achieved recently by analysing the response of isolated animal cap cells to exogenous polypeptide growth factors. In this assay, animal cap cells excised from Xenopus blastulae can be respecified from an ectodermal fate (epidermis) to a range of mesodermal fates (e.g. notochord, muscle, kidney etc) by incubation with purified growth factors of the transforming growth factor-beta (TGF-β) or fibroblast growth factor (FGF) superfamilies (Slack et al., 1987; Kimelman and Kirschner, 1987; Smith, 1987; Rosa et al., 1988; Patemo et al., 1989, Asashima et al., 1990; Smith et al., 1990; Sokol et al., 1990). Apart from mesoderminducing activity, an absolute requirement for an endogenous mesoderm-inducing factor is that it be present within the embryo at the right time and place. At present, the main candidates that have been suggested and that meet at least some of these requirements are basic FGF (Slack and Isaacs, 1989; Kimelman et al, 1989), activin (Thomsen et al., 1990), BMP4 (Koster et al., 1991; Dale et al., 1992) and Vg1 (Weeks and Melton, 1987; Dale et al, 1989, Tannahill and Melton, 1989). Recently, using a different approach, evidence has been obtained for the participation in mesoderm induction of another class of growth factor, that encoded by the wnt gene superfamily (McMahon and Moon, 1989; Christian et al., 1991b; Smith and Harland, 1991; Sokel et al., 1991).

The murine proto-oncogene wnt-1 encodes a 370 amino acid, cysteine-rich glycoprotein (van Ooyen and Nusse, 1984), which is secreted very poorly from cultured cells into the surrounding media (Papkoff et al., 1987, Papkoff, 1989) Recent evidence suggests that secreted wnt-1 is associated with the extracellular matrix, an interaction that can be prevented by the addition of negatively charged polyanions such as heparin or suramin (Bradley and Brown, 1990; Papkoff and Schryver, 1990). In Drosophila there is compelling genetic and biochemical evidence that the wnt-1 homologue, wingless (wg, Rijsewijk et al., 1987), is secreted and acts on neighbouring cells (Morata and Lawrence, 1977; van den Heuvel et al., 1989; Gonzalez et al., 1991). A large family of wnt-related genes have now been identified in both mouse and frog (Gavin et al., 1990, Christian et al., 1991a), although their roles remain to be defined. In mice, wnt-1 appears to be involved in the development of the central nervous system (McMahon and Bradley, 1990; Thomas and Capecchi, 1990), whilst in Drosophila it is implicated in segment polarity and imaginai disc development (Morata and Lawrence, 1977; Nusslem-Volhard and Weischaus, 1980; Baker, 1988). Expression of murine wnt-1 (McMahon and Moon, 1989) and overexpression of Xenopus wnt-8 (Christian et al., 1981b) in early Xenopus embryos results in a duplicated dorsal axis. These latter results suggest that wnt proteins may act to establish a secondary organizer within the VMZ of early Xenopus embryos, although splitting of the primary organizer cannot be ruled out.

In this paper, we extend the experiments of McMahon and Moon (1989), using the Drosophila wg gene which differs from other known wnt-1 genes in encoding a protein containing an extra hydrophilic domain (Rijsewijk et al., 1987) We show that the wg gene is capable of generating a complete duplicate dorsal axis following injection of its RNA into ventral blastomeres of the Xenopus embryo. Furthermore, it is able to completely rescue UV-irradiated embryos, which would otherwise fail to develop dorsal mesoderm. These effects can be executed by wg without the protein being expressed in the responding cells, suggesting that wg acts by inducing non-expressing cells into alternative pathways. We also show that, although wg is able to induce ventral mesoderm in isolated animal caps, the presence of FGF is required for it to induce dorsal mesoderm. These inductive effects of wg are discussed in relation to the tendency reported by others (Bradley and Brown, 1990; Papkoff and Schryver, 1990) and observed here in oocytes and embryos, for wnt proteins to remain within, or closely associated with, the cells in which they are synthesized.

Oocytes were obtained from females of the frog Xenopus laevis, maintained and where necessary microinjected as described by Colman (1984). Following injection, oocytes were incubated in modified Barths’ saline (MBS) supplemented with 1% fetal calf serum and an antibiotic/antimycotic solution (GIBCO/BRL). They were labelled by addition of 1 mCi/ml of [³⁵S]methionine (1415 C1/mmol, Amersham), incubated for 18 hours at 20°C, and then chased in medium supplemented with 2 mM cold methionine. When required, suramin (Mobay Chemical Company) at 5 mM or heparin (Sigma) at 50 μg/ml were added to the media For analysis, media was only taken from incubations containing healthy oocytes. The preparation of oocyte homogenates, wg and chick oviduct protein immune precipitation (1 μl antisera per 40 μl homogenate or medium) and electrophoresis on reducing, 12 5% SDS-polyacrylamide gels have been described previously (Dale et al, 1989) The rabbit anti-wingless antisera was a kind gift of Drs M. van den Heuvel and R Nusse, Stanford, USA Anti-chick egg white antiserum is described in Cenotti and Colman (1988)

Eggs were obtained from females injected with 500-750 i u of human chorionic gonadotrophin (‘Chorulon’ Intervet) the previous evening and fertilized artificially using a macerated testis. Embryos were dejelled in 2% cysteine hydrochloride (pH 7 8-8 1), thoroughly washed and transferred into 10% MBS All embryos were staged according to Nieuwkoop and Faber (1967) Where necessary, dorsoventral polarity was ascertained at the 4-cell stage when dorsal blastomeres are often less pigmented than ventral blas-tomeres The dorsal marginal zone was then stained with a crystal of Nile Red (Kirschner and Hara, 1980) At the 1-, 4- or 32-cell-stage, single blastomeres were injected with 5-15 nl of synthetic RNA of the required concentration in water, in some cases RNA was injected as a 50:50 mixture with rhodamine-dextran-amine (RDA, Gimhch and Braun, 1985) at a final concentration of RDA of 50 mg/ml During and after injection, embryos were kept in 10% MBS containing 3% Ficol (Type 400. Sigma) to prevent leakage of cytoplasm from injected embryos In some cases, embryos were placed on a quartz slide and irradiated at the vegetal pole, 35-40 minutes after fertilisation (1st cleavage approx 90 minutes) using a Miner-alite UV source (approx 260 nm wavelength), the length of exposure was determined empirically to couple good survival with a high proportion of severly defective embryos.

UV-irradiated embryos were scored using the dorsoantenor index (DAI) of Kao and Ehnson (1988) Briefly, DAI=0 indicates no dorsoanterior structures and radially symmetric embryos composed entirely of ventoposterior structures; DAI=l-4 indicates an increasing degree of dorsoantenor structures and a concommitant reduction of ventroposterior structures; DAI=5 indicates normal morphology, DAI=6-9 indicates an increasing degree of dorsoantenor enhancement, coupled with increasing seventy of ventroposterior trunk reduction, DAI=10 indicates a radially symmetnc embryo composed almost exclusively of dorsoanterior structures, particularly in neural and notochordal tissues

Animal caps were isolated from embryos at stage 8 in full-strength MBS using ground forceps and electrolytically sharpened tungsten needles They were incubated individually in 96-well plates lined with agar and containing 100 μl of MBS plus 1 mg/ml BSA and either 0, 5 or 50 ng/ml of Xenopus basic Fibroblast Growth Factor (FGF; Kimelman and Kirschner, 1987). When control embryos had reached stage 40, animal caps were fixed (see below)

Embryos were prepared for confocal microscopy as described by Dent and Klymkowsky (1989) with the following modification: 0.1% Tween 20 was added to the Tris-buffered saline, pH 7 5 (TTBS). The first antibody, rabbit anti-wg, was used at 1/100 dilution, the second antibody used was a goat affinity-purified, biotinylated anti-rabbit IgG (Vector Laboratories) and was used at 1/200 dilution Finally the embryos were stained with fluorescein Avidin D (Vector Laboratories at 1/1000 dilution. In some experiments, this procedure was then repeated using a cocktail of eight anti-haemagglutinin mouse hybridoma culture media (each at 1/125 dilution), followed by rhodamine (TRITC)-labelled goat anti-mouse antisera (Sigma) at 1/100 dilution. The hybridoma supernatants were a kind gift of DrJ Skehel, NIMR, London, UK Whole mounts were examined using a Biorad MRC 600 confocal microscope

Stained, whole-mount embryos were taken out of Murray’s medium into methanol and then embedded in Paramat paraffin wax (Gurr). 10 μm sections were dewaxed with Histoclear (Cell path plc), rehydrated, and counterstained with Hoechst 33258 (1 μg/ml, Calbiochem) for 1 minute. They were viewed in a Leitz Diaplan microscope equipped with epifluorescent optics, and photographed with Kodak T MAX 400 film

Embryos were fixed at 4°C with 4% paraformaldehyde in phosphate-buffered saline, pH 7 5, whilst animal caps were fixed in Susa’s fixative Samples were dehydrated and embedded in Paramat. 10 μm sections containing RDA lineage label were mounted on gelatin-subbed slides, stained with Hoechst 33258, and observed and photographed as above All other sections were stained with periodic Schiff’s, then 0 5% naphthalene black and finally 1% methyl green

A full-length cDNA encoding wg, obtained from A. Martinez Arias (Cambridge University, UK) as a 3kb BamHI fragment, was inserted into the BglII site of the transcnption vector pSP64T (Krieg and Melton, 1989). pSP64T clones containing influenza haemag-glutinin [pSP64HA], and bovine prolactin (BP4) have been described previously (Ceriotti and Colman, 1989). All DNAs were transcribed with SP6 polymerase (Pharmacia) using standard protocols (Krieg and Melton, 1989). After transcription, the reaction mix was digested with 20 μg/ml DNAase 1 (Calbiochem) before phenol:chloroform extraction and ethanol precipitation Poly(A)⁺ RNA was prepared from chick oviducts as described by Ceriotti and Colman (1988) For injection the RNA was resuspended in stenle, deionized water at the required concentrations

Previous reports (McMahon and Moon, 1989; Christian et al., 1991b) have shown that injections of Xenopus embryos with murine wnt-1 or Xenopus wnt-8 (Xwnt-8) mRNAs result in the duplication of the dorsal axis. Our preliminary experiments were designed to test whether a similar phenotype resulted from Drosophila wg mRNA injections. In these experiments we injected wg mRNA into either dorsal or ventral blastomeres of the 4-cell Xenopus embryo, and the results confirm that this molecule, like its murine and Xeno-pus homologues, is able to induce a secondary dorsal axis (Fig. 1; Table 1). Co-injection of wg mRNA and the lineage label rhodamine-dextran-amine (RDA) shows that progeny of the injected blastomeres make a substantial contribution to this new axis (Fig. 2A,B) Embryos were injected with either 125 pg or 5 ng of wg mRNA and although the results are broadly similar, the highest concentration of RNA tended to give the most extreme phenotype. However, in other experiments as little as 5 pg of wg mRNA proved effective in producing multiple dorsal axes, many with completely formed heads. As controls, embryos were injected with synthetic mRNA encoding bovine prolactin (BP4), a secretory protein, and duplications of the dorsal axis were never observed (Fig. 1 A; Table 1).

In the experiments described above, ventral injections proved to be the most effective at inducing secondary dorsal axes (Table 1) and the earliest sign of this induction is the appearance of a second dorsal lip in the ventral quadrant of the early gastrula (Fig. 3). In many cases and at both RNA concentrations, bifurcation occurred in the hind trunk with the embryos developing an almost complete secondary dorsal axis (Fig. 1C). This is particularly the case following injection of 5 ng of RNA. Occasionally, triplications were also observed (Fig. ID). Histological examination of sectioned material has shown that these secondary axes usually possessed the correct spatial array of mesodermal, as well as neural and endodermal tissues (Fig. 2A). However, embryos were occasionally observed in which the secondary axis possessed two parallel notochords (see Fig. 2B,C), indeed similar duplications of the notochord were observed in embryos that were apparently quite normal externally. In contrast, following the injection of 125 pg of wg mRNA into dorsal blastomeres, duplicated axes were rarely obtained, and all of these either bifurcated in the head or possessed enlarged or duplicated cement glands. The frequency and extent of duplicated axes were much greater following injection of 5 ng of wg mRNA into dorsal blastomeres (see Table 1), although once again in the majority of cases (70%) the duplications bifurcated in the head (Fig. IB). These results suggest that injection of wg mRNA does indeed induce a secondary organizer in recipient embryos, and that the site at which the primary and secondary axes fuse is determined by the angle separating the two organizers. They also clearly demonstrate that as a result of wg RNA injection and presumably its translation, ventral blastomeres have changed their fate to that expected of théir dorsal counterparts, and suggests the induction of an ectopic dorsal organizers) in these blastomeres.

From the experiments described above, we have concluded that wg expression changes the fate of injected blastomeres, such that a second dorsal organizer is formed in the lateral or ventral marginal zone of the gastrula. We wished to confirm this conclusion by rescuing embryos which as a result of expenmental intervention, lack an organizer. This is achieved by irradiating embryos at the vegetal pole with UV light (260 nm) early in the first cell cycle. These embryos subsequently develop with reduced dorsoanterior structures, many developing as radially symmetric “ventral belly pieces” (Scharf and Gerhart, 1980). If wg protein were inducing a secondary organizer, we would expect that injection of its mRNA would rescue these UV-irradiated embryos. To test this, embryos were irradiated at the vegetal pole soon after fertilization and then injected with either wg, influenza haemagglutinin (HA), or BP4 mRNA prior to the first cleavage division The results were scored using the dorsoanterior index (DAI) of Kao and Elinson (1988; see materials and methods) Following injection of either HA or BP4 mRNAs into UV-irradiated embryos, no increase in dorsoanterior development was seen over uninjected embryos (DAI < 2 in both cases, Table 2), many remained completely ventralized (Fig. 4B). In contrast, injection of wg mRNA greatly increases the extent of dorsoanterior development (DAI > 5, Table 2 and Fig. 4C) with some of the resulting tadpoles near normal in appearance (Fig. 4C), in others dorsoantenor tissues were over-represen ted in the final body pattern (Fig. 4D). The range of body plans observed following wg mRNA injection into these embryos is similar to those previously described for rescue of UV-irradiated embryos following treatment with lithium (Kao et al., 1986). Coinjection of RDA with wg mRNA has confirmed that the injected blastomere makes an extensive contnbution to the developing dorsal axis in UV-irradiated embryos (data not shown). These results confirm our conclusion that wg protein is capable of inducing an organizer in competent marginal zone cells.

From the results described above, it is not clear if wg needs to be expressed in the marginal zone cells that will ultimately form the organizer, or whether expression in neighbouring animal or vegetal cells is sufficient. To test this we have injected its mRNA into progressively smaller portions of the embryo, localizing it to either inducing or responding cells. In the first of these experiments wg mRNA was injected into ventral blastomeres at the 8-cell stage, when the 3rd cleavage plane has divided the animal and vegetal hemispheres, and the results presented in Table 3 show that injections into vegetal blastomeres were much more effective at inducing duplicated dorsal axes than similar injections into animal blas-tomeres. Following animal injections duplicated axes were rarely obtained, and anterior structures, such as the head, were never observed. Coinjection of lineage label, demonstrated that wg expression had no effect on the fate of injected animal blastomeres (data not shown). In contrast, following vegetal injections, well developed secondary axes were usually obtained, and in most cases these included complete, or nearly complete, secondary heads.

To further clarify the most effective site of wg expression for axial duplication we have injected single blastomeres of the 32-cell embryo with wg mRNA. At this stage blastomeres are frequently arranged as four tiers of eight cells, tiers A-D running from the animal to the vegetal pole (Nakamura and Kishiyama, 1971), and we have concentrated on injections into ventrovegetal blastomeres of tiers C and D. Fate maps have shown that, whereas C-tier blastomeres contribute to both mesoderm and endoderm, D-tier blastomeres contribute most of their volume to endoderm, making little contribution to mesoderm (Dale and Slack, 1987a; Moody, 1987). Following injection of wg mRNA into either tier, duplicated axes were observed with equal frequency (Table 3), and in both cases complete secondary heads were obtained. By coinjecting lineage label, followed by histological examination of the resulting embryos, it was clear that wg mRNA-injected C-tier blastomeres made a substantial contribution to mesoderm of the second dorsal axis, most of the progeny populating the notochord and somite of this axis (Fig. 5B). In contrast, wg mRNA-injected D-tier blastomeres contributed most of their volume to endoderm of the second axis, and it was clear that the fate of this endoderm had been respecified to a more dorsoanterior form (Fig. 5D). Our failure to observe a consistent contribution of injected D-tier blas-tomeres to the mesoderm, suggests that expression of wg within this germ layer is not necessary for the observed duplications. We conclude that injection of wg RNA changes the inductive properties of ventral D-tier blastomeres; normally these blastomeres would only induce ventral mesoderm in the adjacent marginal zone (Gimlich and Gerhart, 1984; Dale et al., 1985, Dale and Slack, 1987b), but following wg mRNA injection they are able to induce dorsal ‘organizer’ mesoderm.

All our results so far suggest that wg protein acts to induce dorsal mesoderm in the ventral and/or lateral marginal zone of injected embryos. One way to test this point further is to apply wg protein to animal caps isolated from blastulae, an assay that has identified molecules such as basic FGF and activin as mesoderm-inducing factors (Slack et al., 1987; Kimelman and Kirschner, 1987; Asashima et al, 1990; Smith et al., 1990; Sokol et al., 1990). Unfortunately, purified, active wg protein is not available, so we have had to adopt the alternative strategy of isolating animal caps from embryos previously injected in the animal hemisphere with wg RNA. Embryos were injected with 3 ng of either BP4 or wg RNA before the first cleavage division and cultured until stage 8 (mid blastulae), at which point animal caps were isolated and incubated in a simple salt solution until control embryos had reached stage 40, before being examined histologically. Whereas control, BP4 mRNA-injected animal caps differentiated as atypical epidermis (Fig. 6A), wg mRNA-injected animal caps differentiated as fluid-filled epidermal vesicles containing a few mesenchyme-like cells and occasionally mesothelium (Fig. 6B). Such cases are usually regarded as weak ventral inductions, suggesting that wg protein can be classified as a ventral mesoderm-inducing factor alongside molecules such as FGF (Slack et al., 1987; Kimel-man and Kirschner, 1987) and BMP4 (Koster et al, 1991; Dale et al, 1992), as opposed to a dorsal mesoderm-inducing factor such as activin (Asashima et al., 1990; Smith et al., 1990, Sokol et al., 1990).

Since no dorsal mesoderm, i.e. notochord and muscle, was found in the experiments described above, the inductive potency of wg protein alone cannot account for duplications of the dorsal axis that we have described above; in these cases, notochord as well as muscle was clearly differentiated in the secondary dorsal axis (see Figs 2A, 5). This suggests that other factors, perhaps FGF or members of the TGF-β family, are required to act in concert with wg protein to induce dorsal mesoderm. To test this point, we have isolated animal caps from embryos previously injected with 1-5 ng of either wg or BP4 mRNA and incubated them in media containing either 0, 5 or 50 ng/ml of Xenopus bFGF. Animal caps were incubated until control stage 40 before being analysed histologically and, in three separate experiments, wg mRNA-injected animal caps differentiated dorsal mesoderm in the presence of bFGF (Fig. 6D). Although in most cases only muscle was differentiated, in one experiment, util-ising the highest concentration of both RNA and bFGF, a few caps (2/11) also differentiated a small amount of notochord (Fig. 6F). In these same experiments, animal caps injected with BP4 mRNA only differentiated small amounts of muscle in the presence of 50 ng/ml bFGF (Fig. 6E). The difference between these animal caps was evident as early as control stage 10.5-11. In the presence of bFGF, wg mRNA-injected animal caps underwent a greater degree of convergent extension than BP4 mRNA-injected animal caps. These results show, that whereas wg protein alone cannot act as a dorsal mesoderm-inducing factor, it can act synergistically with a ventral mesoderm-inducing factor such as bFGF to induce dorsal mesoderm. However, the near absence of notochord differentiation in these experiments might suggest a requirement for yet another factors) to explain complete axial duplication in wg mRNA-injected embryos. This factoids) would be present in either the marginal zone or vegetal pole, but not in the animal cap.

Results described above suggest that wg protein can act to alter the fate of blastomeres not previously injected with its mRNA. This is not surprising since it is known to be secreted by Drosophila embryonic cells and is frequently found 2-4 cell diameters from its site of synthesis (Gonzalez et al, 1991). In previous studies (McMahon and Moon, 1989, Christian et al, 1991b), the spatial distribution of the mammalian or amphibian Wnt proteins in Xenopus embryos was not reported, probably because of the unavailability of the appropriate antisera. To analyse the distribution of wg protein in Xenopus embryos, we have injected 5 ng of its RNA into single blastomeres at the 8-cell stage and examined the embryos by immunomicroscopy at the early gastrula stage after staining with anti-wg antibodies. Initially, embryos were examined using whole-mount confocal microscopy and Fig. 7B shows a typical optical section seen in such an embryo, whilst Fig. 7A shows a superimposition of all the optical sections that contained any wg protein signal. The fluorescing region shown represents approximately the volume occupied by the blastomere injected initially. It is clear that wg protein remains localised to the region of injection and some is found close to the plasma membrane of individual blastomeres.

Sections from injected embryos were also examined by conventional epifluorescence microscopy (Fig 7C). Analysis of this section indicates that most of the wg protein expressed by injected blastomeres is found intracellularly although some is at the cell surface (also see Fig. 7 D). The perinuclear localisation and punctate appearance of the signal suggests that intracellular wg is localised to the endoplasmic reticulum and other components of the secretory pathway. In Drosophila, wg protein is found associated with cells in which it is not synthesised (Gonzalez et al., 1991 ). In order to establish whether this occurs in the frog embryos we have co-injected wg mRNA along with haemagglutinin mRNA, as an integral membrane protein, haemagglutinin will act as a cell autonomous marker. We find (Fig. 7 D,E) that wg protein is only seen in cells expressing haemagglu-tinin and that it has a predominantly intracellular distribution which is in contrast to the surface distribution of haemagglutinin. We conclude that all detectable wg protein remains associated with the cells in which it is synthesised.

The results above indicate that some, albeit a very small proportion, of wg protein can get to the cell surface of blas-tomeres, behaviour consistent with its putative role in mesoderm induction. Unfortunately more direct answers on wg protein secretion in embryos are difficult to obtain due to the fragility of the early blastomeres. Instead, we have examined wg protein secretion in oocytes since this cell type has proved a useful system for testing the intrinsic properties of a variety of non-frog secretory and membrane proteins (Colman and Morser, 1979; Colman et al., 1981a; Cemoti and Colman, 1988). In addition, the behavior of the wg protein in oocytes might provide a useful guide to its behavior in the cleaving embryo.

Initially, oocytes were coinjected with mRNAs encoding wg protein and several chick oviduct secretory proteins, including ovalbumin and lysozyme Oocytes were then cultured with radioactive methionine for various times before immune precipitating oocyte homogenates and culture media with the relevant antisera. It is clear from Fig. 8 (A and B) that whereas the oviduct proteins are secreted into the media, wg protein is not. Thus wg protein, like its murine counterpart wnt-1, is poorly secreted by expressing cells. Of course these results do not tell us where the wg protein isolated from homogenates is localized m the oocyte; it may be intracellular as suggested for the embryo above, or it may be associated with components of the extracellular matrix as suggested for murine wnt-1 (Bradley and Brown, 1990; Papkoff and Schryver, 1990).

To analyse this point further, we have incubated oocytes injected with wg mRNA in media containing either 5 mM suramin or 50 μg/ml heparin. It has previously been shown that some murine wnt-1 is secreted if transformed cells are incubated in media containing these reagents (Bradley and Brown, 1990; Papkoff and Schryver, 1990). When we included these reagents in oocyte incubation media, we found that suramin, but not heparin treatment, led to the appearance of wg in the medium (Fig. 8C), and this occurred whether suramin was present during or after synthesis of radioactive wg protein. In fact, more protein was found in the media than in the Triton extract of the oocyte prepared at the end of the incubation. Since the amount of extracted material was the same, regardless of treatment, we conclude that suramin treatment allows the escape from the cell of protein which would otherwise be degraded The possibility that a significant proportion of wg protein is lost in the Triton X-100-insoluble oocyte debris was excluded by experiments where the Triton-insoluble material was extracted with 1 % SDS at 80°C prior to immune precipitation, conditions observed to solubilise matrix-associated murine wnt-1 (Bradley and Brown, 1990); less than 10% of the radiolabelled wg protein was recovered by these procedures (data not shown). The reasons for the failure of heparin to release wg protein into the media are unclear; unlike murine wnt-1 it is not known if wg protein binds to heparin, in which case its release would not be expected. Alternatively heparin and suramin may release murine wnt-1 from different compartments of the cell, suggesting the absence of wg protein from the heparin release compartment in oocytes.

In mammalian cells, it is not clear how suramin acts to release proteins from cells into the media; whereas it may displace or interfere with the binding of proteins to the extra-cellular matrix, as suggested for murine wnt-1 (Papkoff and Schryver, 1991), it has been claimed that suramin can enter cells by endocytosis promoting the secretion of intracellular material (Huang and Huang, 1988). We have attempted to resolve the situation in oocytes in two ways: First oocytes expressing wg in the absence of suramin, were exposed to trypsin at 4°C; we have previously shown that the conditions used resulted in the proteolysis of surface-associated, but not intracellular, oocyte proteins (Ceriotti and Colman, 1989). As shown in Fig. 8A, trypsin treatment had no effect on oocyte-associated wg protein, although after disruption of the oocytes by homogenisation, all the wg protein was sensitive to trypsin (data not shown) Second, oocytes expressing wg were incubated with suramin at 4°C, conditions known to severely inhibit both endocytosis and secretion. Fig. 8D shows that at 4°C no wg appears in the medium and also that oocyte-associated wg is stabilised; in controls incubated at 20°C, wg protein is efficiently secreted into the media in the presence of suramin. If suramin can displace extracellular wg from the surface, we would have expected to have seen this occur at 4°C. These results suggest that the majority of wg protein is located intracellularly and its release is triggered by the presence of externally added suramin, although we cannot rule out the possibility that, in the absence of suramin, wg protein is constitutively secreted but immediately degraded extracellularly.

The Drosophila wg gene was first identified as a mutation that transformed the adult wing into a duplication of the notum, a part of the thorax derived from the same imaginai disc (Sharma and Chopra, 1976; Morata and Lawrence, 1977). Subsequently, mutations which inactivate wg function were shown to be embryonic lethals of the segment polarity class (Nusslein-Volhard and Wieschaus, 1980), the posterior part of each segment being replaced by a mirrorimage duplication of the anterior part (Baker, 1988). The gene encodes a protein with homology to the murine protooncogene wnt-1 (Risjewijk et al., 1987), and is expressed ina narrow stripe of cells within each segment (van den Heuvel et al., 1989; Gonzalez et al., 1991). It is then secreted and moves into adjacent cells, in particular it enters a narrow stripe of cells posterior to the wg stripe where it controls expression of the engrailed gene (Gonzalez et al., 1991; Bejsovec and Martinez Arias, 1991,Heemskerketal., 1991). The gene is therefore an important regulator of cell fate during Drosophila embryogenesis.

Our results show that wg, like its vertebrate homologues, murine wnt-1 and Xwnt-8 (McMahon and Moon, 1989; Christian et al., 1991b; Smith and Harland, 1991;Sokol et al., 1991), causes the formation of a second dorsal axis following injection of synthetic RNA into early Xenopus embryos, demonstrating that the ability of wnt proteins to induce organizer activity is conserved across a broad phylogenetic spectrum. Prior to these experiments it was not immediately obvious that this would be the case, wg differs from other wnt proteins in that it possess an additional domain of 85 amino acids which are remarkably hydrophilic (Risjewijk et al., 1987). Furthermore, not all wnt proteins induce secondary axes following injection into Xenopus embryos, it has been reported that Xwnt-5 does not have this property (Christian et al., 1991b).

In agreement with others (Sokol et al., 1991; Smith and Harland, 1991; Christian et al., 1992), our results suggest that wnt proteins achieve this axial duplication by acting as a dorsal mesoderm inducing signal. During normal development this signal is only emitted by dorsovegetal blastomeres of the blastula (Gimlich and Gerhart, 1984; Dale and Slack, 1987b; Kageura, 1990), inducing the overlying marginal zone to form Spemann’s organizer Our evidence is threefold: First, wg can rescue embryos that had previously been irradiated at the vegetal pole with ultraviolet light, a treatment known to inhibit the formation of the dorsovegetal inducing centre and consequently Spemann’s organizer (Gerhart et al., 1989). Identical results have been obtained for both wnt-1 and Xwnt-8 (Sokol et al., 1991 ; Smith and Harland, 1991 ; Christian et al., 1992), and argues against a mechanical explanation in which the normal organizer is split. Second, a secondary dorsal axis is reliably formed only following the injection of wg mRNA into ventrovegetal blastomeres, injections into the animal hemisphere have little or no effect. At the 32-cell stage injections into the D-tier induce a secondary dorsal axis without any substantial contribution to the mesoderm from the injected blastomere, which differentiates as dorsal endoderm This is the normal fate of dorsal D-tier blastomeres (Dale and Slack, 1987a; Moody, 1987) and suggests that wg expressing ventral D-tier blastomeres have changed their fate towards that of their dorsal counterparts, acquiring the ability to induce an organizer in the adjacent C-tier. The same conclusion was reached by Smith and Harland (1991) following similar experiments with Xwnt-8.

Our third line of evidence is that animal caps isolated from wg expressing blastulae differentiate mesoderm, suggesting that wg protein is a mesoderm inducing factor. However, in contrast to the ability of wg protein to induce dorsal mesoderm in the marginal zone of an embryo, it only induces ventral mesoderm in isolated animal caps. The simplest explanation for this is that wg protein normally acts in concert with other factors that are not present in our animal caps. We have shown that this other factor could be FGF, in its presence wg expressing animal caps differentiate dorsal mesoderm. Interestingly, it has recently been shown that whereas FGF is found within the marginal zone of blastulae it is absent from animal caps (Shiurba et al., 1991). Similar conclusions to ours have also been reached by Christian et al. (1992) using Xwnt-8 expressing animal caps. These authors also showed that when larger animal caps were taken, some dorsal mesoderm was found in absence of exogenous factors, a result that might explain the observations of muscle within Xwnt-8 expressing animal caps by Sokol et al. (1991).

We have shown that the fate of cells not expressing wg protein can be influenced by its production elsewhere in the embryo, yet our immune histological (embryos) and biochemical (oocyte) data appear to indicate an inability of this protein to spread beyond the producer cells. Although this behavior is consistent with what has been observed for this family of proteins in mammalian systems (Bradley and Brown, 1990; Papkoff and Schryver, 1990), in Drosophila embryos, wg protein has been found 2-4 cells away from its site of synthesis (Gonzalez et al., 1991). In cultured cell lines, suramin or heparin treatment leads to the secretion/release of the murine wnt-1 protein, although the proportion of protein found in the culture medium is quite variable and can be very low (Bradley and Brown, 1990; Papkoff and Schryver, 1990). In oocytes, suramin, though not heparin, treatment leads to the almost quantitative recovery of oocyte-associated material in the medium. This salvaged protein does not seem to have been associated with the cell surface, as evidenced by its trypsin-insensitivity or its resistance to suramin-induced release at 4°C In this way, it differs from its mammalian counterpart, some of which clearly accumulates in the extracellular matrix. The observation that the intracellular steady-state concentration of wg is unaffected by suramin treatment indicates that secretion or degradation are mutually exclusive choices for this protein, a situation in oocytes that we have described previously for murine immunoglobulins (Colman et al., 1981b). Although suramin probably can interact with various classes of secretory protein, it will be interesting to see whether it affects the formation of the second dorsal lip (see Fig. 3) in wg expressing embryos, since this lip is the first phenotypic sign of wg action and reflects the presence of a new organizer region. Previous use of suramin has been restricted to later stages of gastrulation, where arrest of gastrulation movements follows its injection (Gerhart et al., 1989).

Although the seeming lack of movement of wg in our experiments could be just an issue of sensitivity, a small amount of wg protein diffusing away from its site of secretion, an alternative explanation is that this molecule acts primarily upon adjacent cells. There are precedents for paracrine factors being retained at the surface of the cells that produce them; in the case of platelet-derived growth factor (PDGF) B chain dimer, this retention enhances its cell transforming ability (Beckman et al., 1988) and is mediated by a C-terminal stretch of amino acids with claimed homology to sequences in the C termini of murine and human Wnt 1 homologues (La Rochelle et al., 1991). Whatever the explanation for wg protein immobility, it remains difficult to reconcile completely the molecular properties of wg with our perceived understanding of mesoderm induction, where diffusion of inducing factors throughout the prospective mesoderm is anticipated. Of particular interest are the contrasting effects of expressing wg in the animal hemisphere and the D-tier of the vegetal hemisphere; although the injected blas-tomeres are adjacent to the C-tier in both cases, only D-tier injections reliably change the fate of the neighbouring blas-tomeres towards that of their dorsal counterparts. We might reconcile these results if we assume that wg protein is indeed restricted to the vicinity of its secretion, perhaps by an interaction with extracellular matrix materials, and that only cells close to the prospective mesoderm-endoderm border are capable of responding. This is unlikely to be the result of the restricted location of wnt-1 receptors, since animal cap cells are competent to respond to wg, but is more likely the result of the restricted location of some accessory factor(s) required for wg function (see above).

Finally, the molecule that we have studied in our experiments, wg protein cannot normally be involved in the development or function of the amphibian organizer. Like others (McMahon and Moon, 1989; Christian et al., 1991b), our hypothesis is that this molecule activates the receptor for an endogenous member of the wnt family. The endogenous activator is unlikely to be Xenopus wnt-1 which appears to be expressed only from neurula stages onwards (Nordermeer et al., 1989), after the organizer’s functions are over. A more credible candidate would be Xwnt-8 which Sokol et al. (1991) and Smith and Harland (1991) have shown to have the properties of a dorsal mesoderm inducer, using some of the approaches described in this paper. However, although this molecule is present at the right time in development, it is localised to the non-organizer .ventral and lateral marginal zone (Christian et al., 1991b; Smith and Harland, 1991). Despite the fact that wg cannot be the endogenous inducer, we feel justified in pursuing our studies using it for two reasons: First, the provision of good antisera allow the patterns of wg location and phenotypic effect to be compared, and second, the availability of various wg null mutants, especially a temperature-sensitive one (Gonzalez et al., 1991), may generate further insight into how wnt proteins act during development and over what time period they are effective

We particularly thank A. Martinez Arias for constant encouragement, guidance, and cnticism of the manuscript. We wish to thank the following for provision of reagents: A. Martinez Anas, wg cDNA, J. Slack, bFGF, M. van den Heuvel and R. Nusse, anti-wmg-less antisera; J Skehel for anti-haemagglutinin hybndoma supernatants. We thank S Bhamra for the confocal microscopy he performed at the University of Warwick. A Colman is indebted to the British Medical Research Council and Wellcome Trust for the support of this work